United States        Office of Research and  U.S. Department of   EPA/600/R-92/033
 Environmental Protection   Development      Energy         February 1992
 Agency          Washington DC 20460  Washington DC 20585
 Heterogeneous  Wastes:

 Methods  and
United States Environmental Protection Agency
Environmental Systems Monitoring Laboratory
Office of Research and Development
P.O. Box 93478
Las Vegas, Nevada 89193-3478
United States Department of Energy
Office of Technology Development
Washington, DC 20585-0002

                             February 1992
          March 26-28, 1991
           Gretchen L. Rupp
           Roy R. Jones, Sr.
        Kenneth W. Brown, U.S. EPA
        S.P. (John) Mathur, U.S. DOE
   Cooperative Agreement No. CR 814701
   Harry Reid Center for Environmental Studies
      University of Nevada - Las Vegas
           Las Vegas, Nevada

   Environmental Monitoring Systems Laboratory
     Office of Research and Development
     U.S. Environmental Protection Agency
           Las Vegas, Nevada
                               Printed on Recycled Paper


       The information in this document  has been funded in part by the U.S. Environmental
Protection Agency under an assistance agreement with the U.S. Department of Energy. It has been
subject to the Agency's peer and administrative review, and it has been approved for publication
as an EPA document.


       The U.S. Environmental Protection Agency and the U.S. Department of Energy conducted
a  workshop in March  1991  to examine methods for characterizing  heterogeneous  wastes
contaminated with hazardous chemicals and/or radionuclides. Sites where the wastes are of large
size or varied  composition, including landfills and drum dumps, present severe difficulties to
investigators attempting to collect representative samples to facilitate site cleanup decisions. This
document serves as the workshop proceedings. It summarizes the study planning tools,  sampling
design strategies, and field and laboratory methods now  in use, identifying the advantages and
disadvantages of each.  In addition,  areas that would benefit  from methodological research or
development, or the adoption of new approaches, are identified. Pertinent regulatory definitions
are assembled  and augmented with practical working  definitions. The discussion of the study
planning process emphasizes the establishment of clear, reasonable goals  and the active participation
of the decision maker, along with program, field, and laboratory specialists. Project planning for
heterogeneous waste  characterization is an iterative process, with each step building on knowledge
gained in previous steps. There are a large number of statistical models that are potentially very
useful for characterizing these sites, though only a small number have seen wide use.  Standard
environmental  QA/QC methods  can be adapted  in several ways to enhance the quality  of
heterogeneous waste data.  A plethora of field methods is  currently employed. These range from
excavation and hand-sorting of large objects to sophisticated  instrumental methods for remote
characterization or contaminant  screening.   Several promising field technologies are now in
development. These emphasize non-intrusive  characterization,  since consideration for worker health
and safety often dictates minimal contact with the heterogeneous  waste. In the laboratory,  the three
basic strategies for handling heterogeneous samples are to separate  them, homogenize  them,  or
analyze the entire sample. Exhaustive documentation of sample appearance and condition, and the
sample  preparation method,  are  essential.   Laboratory  waste management  and  assurance  of
personnel safety are areas needing special care when heterogeneous wastes are handled.

       For each aspect of heterogeneous waste characterization there are new methods that bear
research or are  already under development. In addition,  there are entirely new  approaches to site
characterization that could substantially accelerate the remediation of sites containing hetero-
geneous waste.


Notice	 .X1
Abstract	 m
List of Figures	 vl
List of Tables	 vl
Acronyms	vn
Acknowledgements	  x

Chapter 1. Introduction - Gretchen Rupp	   1
       Background	   1
       Purpose and Scope of this Document	   2
       Contents	   4
       References	   "
Chapter 2. Definitions - Fred Haeberer and Jeff van Ee	   7
       Introduction	   7
       Definitions	   °
       References	  21
Chapter 3. Planning the Study - Leon Bergman, Charlotte Kimbrough,
Mitzi Miller and Dean Neptune	  22
       Introduction	  22
       Preliminary Planning	  25
       Establishing Data Needs and Data Quality Objectives	  29
       Sampling and Analysis Design	  44
       Finalizing the Project  Plan	  49
       Conclusions	  50
       Recommendations	  50
       References	  52
Chapter 4. QA/QC and Data Quality Assessment -  Jeff van Ee and Roy R. Jones, Sr	  53
       Introduction	  53
       QA/QC in Sampling Heterogeneous  Waste	  55
       Assessment of Bias and Variability	  56
       QA/QC Samples	  58
       How Many Observations or Samples are Needed?	  61
       Research Recommendations	  62
       References	  64
Chapter 5. Sample Acquisition -  Janet Angert, Alan Crockett, and Timothy Lewis	  65
       Introduction	  65
       Characterization of Uncontained Heterogeneous Wastes	  66
       Field Screening Methods	  76
       Characterization of Contained Heterogeneous Wastes	  79
       Treatment After Minimal Evaluation	  85
       Recommendations	  85
       References	  86

                                CONTENTS (Continued)

Chapter 6. Analytical Laboratory Requirements -  Clare Gerlach, Wayne McMahon, and
James  Poppiti	
       Introduction	"0
       Project Planning	  "1
       Sample  Receipt, Handling,  and Preparation	  "^
       Waste Disposal  in the Analytical Laboratory  	105
       Reporting Requirements for Analysis of Heterogeneous Waste	106
       Conclusions  and Recommendations	107
       References	10°
Chapter 7. The Larger  Perspective - Roy R. Jones, Sr  	 HO
       Introduction	HO
       Successful Waste Charactertization	 HI
       Methods Development	 H2
       A Changing  Perspective  	 H3
       Reference	H5

Appendix A - Hypothetical Case History:  Drum Characterization - Tom Starks and
Gretchen  Rupp	H7
       Background	H7
       Initial Project Planning	118
       Study Design	I22
       References	l2^
Appendix B  - A Survey of Available Statistical Techniques - Leon Borgman	 l2^
       Introduction	127
       List of Methods	12§
       References	138

                                  LIST OF FIGURES

1-1     The "worst-case" drum of heterogeneous wastes	5
3-1     Generalized scheme of the study process	23
3-2     Steps in the data quality objectives process	30
3-3     Classes of drummed wastes found at DOE sites	35
3-4     Uncertainty limits for specific activity in drummed wastes	43
4-1     Quality assessment samples	60
6-1     Decision tree for sample preparation and analysis	94
6-2     Samplers that allow the entire sample to be analyzed for volatile constituents	98
                                    LIST OF TABLES
Number                                                                             Page

4-1     Elements of measurement bias in environmental sampling	57
4-2    Elements of variability in environmental data	57
5-1     Concentration statistics of specified contaminants and methane in landfill-gas
       samples	6
6-1     Relationship of sample size to maximum particle size	99
6-2    A comparison of several radiation screening devices	102
6-3     Sample disposal options	106


AA           atomic absorption spectrometer
AEC          Atomic  Energy Commission
ANSI         American National Standards Institute
ARAR        Applicable or Relevant and Appropriate Requirement (under CERCLA)
ASME        American Society of Mechanical Engineers
ASTM        American Society for Testing and Materials
CAA          Clean Air Act
CERCLA     Comprehensive Environmental Response, Compensation and Liability Act
CERCLIS     Comprehensive Environmental Response, Compensation and Liability Information
CFR          Code of Federal Regulations
CLP          Contract Laboratory Program
CWA         Clean Water Act
DDA         Differential  die-away
DOD         U.S. Department of Defense
DOE          U.S. Department of Energy
DOT          U.S. Department of Transportation
DQO         data quality objective
DT           Deuterium-Tritium
ELES         external laboratory evaluation sample
EM           electromagnetic
EMSL-LV     Environmental Monitoring Systems Laboratory-Las Vegas
EPA          U.S. Environmental Protection Agency
FD           field duplicate
FES          field evaluation sample
FIFRA        Federal Insecticide, Fungicide, and Rodenticide Act
FEMP        Fernald Environmental Management Project
FOCS         fiber optic chemical sensors
FRB          field reagent blank
FT-IR        Fourier-transform infrared spectroscopy
FTE          full-time equivalent (person-years)
FWPCA      Federal Water Pollution Control Act
GC           gas chromatography
GPR          ground-penetrating  radar
H&S          health and safety
HLRW        high level (radioactive) waste
HMTA        Hazardous Materials Transportation Act
HRC          Harry Reid  Center for Environmental Studies
ICP           inductively coupled  plasma spectrometry
LA-ICP-AES  laser ablation inductively-coupled plasma atomic emission spectrometry
LEAFS        laser excited atomic fluorescence spectrometry
LIBS          laser induced breakdown spectroscopy
LIF           laser induced fluorescence
LLRW        low level  (radioactive) waste
LSC          liquid scintillation counting

                                ACRONYMS (Continued)

MQO         method quality objective
MS           mass spectrometry
NAA         neutron activation  analysis
NACEPT     National Advisory  Council for Environmental Policy and Technology
NAREL       National Air and Radiation Environmental Laboratory
NEPA        National Environmental Policy Act
NIST         National Institute of Standards and Technology
NORW       naturally-occurring  radioactive waste
NPDES       National Pollutant  Discharge Elimination System
NPL          National Priorities  List
NQA         Nuclear Quality Assurance
NRC         Nuclear Regulatory Commission
OAR         Office of Air and Radiation
ORD         Office of Research and Development
ORME       other regulated materials exempted
ORNL        Oak Ridge National Laboratory
OSHA        Occupational Safety and Health Administration
OSW         Office of Solid Waste
OVA         organic vapor analyzer
PAH         polycyclic aromatic hydrocarbons
PCB          polychlorinated   biphenyl
PEG         polyethylene glycol
PID          photoionization  detector
PRB          preparation  rinsate blank
PRP          Potentially Responsible Party (under CERCLA)
PS            preparation  split subsample
PSR          particle size reduction
PVC          polyvinyl chloride
QA           quality  assurance
QAMS        EPA's Quality Assurance Management Staff
QAPjP        Quality Assurance Project Plan
QC           quality  control
RAS          CLP  Routine Analytical Services
RCRA        Resource Conservation and Recovery Act
ROD         Record of Decision
RQ           Reportable  Quantity
RS            routine sample
SARA        Superfund Amendment and Reauthorization Act
SDWA       Safe Drinking Water Act
SERS         surface enhanced Raman scattering
SOP          standard operating procedure
TCLP         Toxicity Characteristic Leaching Procedure
TRU         transuranic  (nuclear material)
TSCA         Toxic Substances Control Act
USATHAMA U.S. Army Toxic and Hazardous Materials Agency

                                  ACRONYMS (Continuec
VOA          volatile  organic  analysis
VOC          volatile  organic  compound
XRF          x-ray  fluorescence  spectroscopy


       This document is  the  product  of a  workshop  sponsored  by the U.S.  Environmental
Protection Agency and the U.S. Department of Energy in March 1991. Project manager for the
EPA was Ken Brown of the U.S. EPA's  Environmental Monitoring Systems Laboratory-Las Vegas.
Project manager for the U.S. DOE was John Mathur. The Harry Reid Center for Environmental
Studies (HRC;  formerly the Environmental Research Center) of the University  of Nevada-Las
Vegas conducted the project under Cooperative Agreement No. CR 814701. The UNLV project
manager was Gretchen Rupp and the workshop coordinator was Kathleen Lauckner.

       Ten work group leaders guided the workshop deliberations and served as the principal
authors of this document.  These were:

Leon Bergman              University of Wyoming
Alan Crockett              EG & G Idaho
Clare Gerlach              Lockheed Engineering and Sciences Co.
Fred  Haeberer              U.S. EPA Office of Research and Development
Charlotte Kimbrough        Martin Marietta Energy Systems
Tim Lewis                  Lockheed Engineering and Sciences Co.
Wayne  McMahon           Martin Marietta Energy Systems
Mitzi Miller                Automated Compliance Systems Inc.
Dean Neptune              US. EPA Office of Research and Development
Jeff van Ee                 U.S. EPA EMSL-Las Vegas.

       The project was overseen by a technical steering committee whose members were:

                     Gretchen Rupp (Committee Chair)
                     University of Nevada-Las Vegas

John Barich                                    Roy  Jones, Sr.
 U.S. EPA                                       U.S. EPA (Co-chairman)
Delbert S. Barth                                John Koutsandreas
HRC/UNLV                                   Florida State University
Michael Connolly                               John Mathur
EG&G Idaho                                    U.S. DOE
Jessie Donnan                                  Joe Pardue
 Westinghouse                                   Martin Marietta Energy Systems
George Flatman                                Paul  Scott
 U.S. EPA                                       Battelle PNL
Terry Grady                                    Ralph Smiecinski
 U.S. EPA                                       U.S. DOE
John Hall                                      Mark Smith
 U.S. DOE                                       SAIC
Ervin Hindin                                    Dennis  Wynne
 Washington State Univ.                           USATHAMA
Janine Arvizu
EG&G Idaho

       The following environmental scientists participated in the workshop. Those individuals
whose names  are  denoted with an asterisk (*)  also  contributed materials for the report.  We
sincerely hope we have not overlooked any participants or contributors.
Actor, David
  Chemical Waste Management

Anderson, Scott A.
 EG&G Rocky Flats, Inc.

*Angert, Janet
  Westinghouse Environmental Management

Arvizu, Janine
 INEL, EG&G Idaho

Baca, Steven L.
  Woodward-Clyde Fed. Serv.

Earth, Delbert
  UNLV/Harry Reid Center

Bartlett, Eric B.
 Iowa State University

Bass, Dean A.
 Argonne National Laboratory

Beaulieu, Patrick L.
 Amoco Research Center

Benner, Henry
 Lawrence Berkeley Laboratory

Bennett,  Joseph T.
 INEL, EG&G Idaho

Benny, Harold L.
  Westinghouse Hanford

Bentley,  Glenn
 Los Alamos National Laboratory

Berdahl,  Donald R.
  General Electric
Blacker, Stan

Bottrell, Dave
  U.S. Department of Energy

*Brinkman, Dennis W.
 Safety-Kleen Corporation

Bryan, Rex C.
  VIAR Corporation

Buck, John
 Battelle Pacific Northwest Laboratory

Butler, Larry

Cahill, Marty
  Chemical Waste Management

Calkin, April L.
 Shell Development Co.

Chambers, William B.
 Sandia National Laboratory

Clark, Glen
 Reynolds Electrical  & Engineering Co.

Connolly,  Michael
 INEL, EG&G Idaho

Crawford, Richard W.
 Lawrence Livermore National Laboratory

Dahl, Dave
 Dames & Moore

Dale, Larry D.
 Rocky Mountain Arsenal

David, Herbert T.
 Iowa State University

Donnan, Jessie G.
  Westinghouse, Savannah River Co.

*Eccleston, George W.
 Los Alamos National Laboratory

Edelson, Martin
 Iowa State University/Ames Laboratory

Elliott, Martin

Flanagan, James B.
 Research Triangle Institute

Flatman, George T.

Flueck, John
  UNLV/Harry Reid Center

Fort,  Les
  Westinghouse, Hanford Co.

Friedman, David

Gilbert, Richard O.
  Battelle Pacific Northwest Laboratory

Grady, Terence M.
 Graham, Thomas A.

 Grazman, Brent
  Colgate-Palmolive Corporation
*Griest, Wayne H.
 Oak Ridge National Laboratory

Guymon, Ronald H.
 U.S. Department of Energy

Haas, William
 Iowa State  University/Ames Laboratory

Hall, John
 U.S. Department of Energy, NV

Hankins, Jeanne

Harmon, Larry
  U.S. Department of Energy

Harvey, Elizabeth A.
 Chevron Research & Technology Co.

Hassig, Nancy
 Battelle Pacific Northwest Laboratory

Hawthorne, Howard A.
 Reynolds Electric & Engineering Co.

Hindin, Ervin
  Washington State University

Hines, Lance A.
  Woodward-Clyde Consultants

Ilias, Ajmal M.
  USAGE North Pacific Division

 *James, Dennis
  Texas A&M/Center for Chemical
  Characteristics & Analysis

Jamison, Alma
 Green, David W.
  Argonne National Laboratory
 Keith, Larry
  Radian Corporation

Khonsary, Yasmine

Kiefer, James M.
  U.S. Department of Energy, CO

Kingsbury, Garrie
 Research Triangle Institute

Kinser, Steven E.
  U.S. EPA, Region VII-Superfund

Klesta, Gene
  Chemical Waste Management

Koutsandreas, John
 Florida State University

Kreutzfeld, Rich
  Sandia  National Laboratory

Krochta,  William  G.
 PPG Industries

Lacayo,  Herbert
  U.S. EPA, Office of Regulatory Mgmt.

Lang, Kenneth T.

Leasure, Craig
 Los Alamos National Laboratory

Lillian,  Dan
  U.S. Department of Energy

Lloyd, Vicki

Lucke, Richard
 Battelle Pacfic Northwest Laboratory

*Marcinkiewicz, Charles J.
  Sanford Cohen & Associates,  Inc.

Marcus,  Mark
  Chemical Waste Management
Marron, Bruce
 Benchmark Environmental Corporation

Mathur, John
  U.S. Department of Energy

McBride, Alexander

McKee, Terry M.
 BFI Houston Laboratory

McKenzie, Raymond  L.
  U.S. Department of Energy, Idaho

Meyer, T.J.
 INEL, EG&G Idaho

Miller, Forest L.
 Desert Research Institute

Nawar, Madeleine

Newey,  John M.
 Lockheed Engineering & Sciences Co

*Nocerino, John

Oversby, Virginia M.
 Lawrence Livermore National Laboratory

*Pardue, Joe
 Martin Marietta Energy Systems

Peters, Mark A.
 EG&G Rocky Flats Inc.

Petullo,  Colleen

*Poppiti, Jim
  U.S. Department of Energy

*Reid, Leah
  VIAR Corporation

Rodriguez,  Leopoldo L.
 Armstrong Laboratory/OEAT

Sailer, Shelly J.
 INEL, EG&G Idaho

Sekot, Mercy
 INEL, EG&G Idaho

Smiecinski, Ralph F.
  U.S. Department of Energy, NV

Smith, Mark A.
 Science Applications International Corp.

Starks, Thomas
  UNLV/Harry Reid Center

Stephens, Marvin  W.
  Wadsworth/ALERT Laboratory

Stewart, Bill M.
  U.S. Bureau of Mines,
 Spokane Research Center

Street, Leah
 INEL, EG&G Idaho

Streets, W. Elane
 Argonne National Laboratory

*Tait, Reid
 Dow Chemical,  U.S.A
Trible, Thomas C.
 INEL, EG&G Idaho

Varchol, Brinley D.

Victery, Winona
  U.S. EPA, Region IX

Vincent, Harold

Wagner, Sandy
 Los Alamos National Laboratory

Watkms, Cliff
 INEL, EG&G Idaho

*Weeks, Stephan
Iowa State University/Ames Laboratory

Weiss, Richard L.
  Westinghouse, Hanford Co.

Whalen, Cheryl L.
 Lockheed Engineering & Sciences Co.
Whitehead, Robert J.
  Compuchem Laboratories, Inc.

Zeisloft, Jon
 INEL, EG&G Idaho

 Chapter 1


                                      Gretchen Rupp


       "Heterogeneous wastes" include those materials otherwise known as debris, solid waste,
trash,  and rubbish.   Because of their relatively large  particle  size and  varied composition,
heterogeneous wastes are often much more difficult to  characterize than more uniform materials
such as soils and sludges. Many types of heterogeneous  wastes are found on industrial, municipal,
and federal waste sites in the United States. These include municipal trash, demolition debris,
waste construction materials,  containers such as drums, tanks, and paint cans, the  solid wastes from
laboratories and manufacturing processes, and post-consumer wastes such as transformers, battery
casings, and shredded automobiles.   These wastes may be inert materials such  as rock, glass, or
concrete;  labile organic matter such as  wood and food wastes; or miscellaneous items such as
rubber, plastic, or asbestos wastes. Potentially-contaminated structures and utilities pose problems
similar to  those of heterogeneous wastes,  and similar strategies may be employed for characterizing
heterogeneous wastes and contaminated  structures.

       Heterogeneous  wastes  are found on tens of thousands of sites across the U.S. The U.S.
Environmental Protection Agency (EPA) has estimated that approximately  6000 municipal landfills
were in operation in 1986 (1);  many more dumps or  landfills had been filled  and closed by then.
Heterogeneous wastes contaminated with  hazardous chemicals are found on National Priorities List
(NPL) sites, sites regulated under the Resource Conservation and Recovery Act (RCRA), and sites
listed in the Comprehensive Environmental Response,  Compensation, and Liability Information
System (CERCLIS). Information compiled within EPA's ROD database indicates that more than
half of all NPL sites contain such wastes; the proportion is probably similar for RCRA sites and
CERCLIS sites (those uncontrolled hazardous-waste sites that have been inventoried but have not
been placed on  the NPL). The U.S. Department of Energy (DOE), which has undertaken a major
cleanup program for its facilities, has jurisdiction  over  thousands of sites contaminated with
radionuclides, hazardous chemicals,  or both.  Many of the sites include contaminated heterogeneous
wastes.  Over 1.4 million drums containing radioactive or hazardous chemical wastes must be
characterized preparatory to remedial action at the DOE sites (2).

       Characterizing heterogeneous wastes presents a number of special problems. The principal
difficulty arises  in attempting to obtain representative samples of a material composed  of disparate
elements.   Customary sample segregation, compositing,  and homogenization  schemes used to
characterize water, soil, or sludge are often completely inappropriate for these materials.  Waste
particle size frequently poses difficulties.  According to  standard sampling  theory, obtaining a
representative sample  of varied items in  the size range of a centimeter or larger may entail
collecting  tens or hundreds of pounds of material. Large  objects cannot be made to fill standard
sample containers, so that bulky items exhumed from within waste piles and placed in standard

 sample jars quickly lose volatile organic compounds (VOCs) to the headspace. Few analytical
 laboratories have the capability of performing leaching tests on raw samples of heterogeneous
 materials because of the large volumes involved and the difficulty of conserving VOCs. Nor are
 laboratories well  equipped to reduce  samples of large, varied items to the tiny, homogeneous
 aliquots used for analysis.  In cases where sample grinding and homogenization are possible, they
 may be inappropriate. The contamination of heterogeneous waste particles is often superficial, so
 that contaminant concentration data expressed on a mass basis do not properly reflect the human
 health risk posed by the waste.

       Many field methods have been  used to characterize heterogeneous wastes, but the available
 methods  are often  insufficient for the task.  While there  are several remote  methods for
 characterizing containerized wastes, they cannot give a complete picture of the hazardous chemicals
 contained within a drum.  Samplers must often resort to opening containers and hand-segregating
 objects in  the  field.    Worker health and safety considerations often preclude thorough
 characterization activities.  Data quality assessment is hampered by the lack of performance
 evaluation samples for heterogeneous materials.

       When a  decision is to be made concerning the disposition of a hazardous waste site, the
 decision-maker first establishes the required confidence level in the correctness of the decision. The
 data needs, data quality objectives,  and all site characterization planning follow from this. When
 heterogeneous wastes are to be characterized, many of the above-mentioned problems may be
 encountered. Project planners frequently discover that a study sufficient to achieve a high level of
 confidence would incur excessive costs, labor requirements, or risks to field  workers.  Clearly,
 methodological research and development, and possibly entirely new approaches to these sites, are
                           Purpose and Scope of this Document
       In recent years hazardous-waste professionals have accumulated a great deal of experience
with heterogeneous wastes. The purpose of this report is to synthesize what they have learned and
the methodological gaps they have identified.  Specifically, it is a response to the need expressed
by EPA and DOE personnel for an examination of:

•      sampling design strategies and field and laboratory methods currently in use, their range of
       application, and their shortcomings;

•      areas where research,  development,  or new approaches could improve the  selection of
       available tools; and

•      technologies or sampling designs used in other disciplines that might be adapted to  the
       characterization of heterogeneous wastes.

       Care has been taken to craft this report in accordance with current and evolving EPA and
DOE policy, especially that promulgated by EPA's Office of Solid Waste. However, the document
does not represent any form of policy decision on the part of either agency.  As they appear herein,

the words "must," "shall,""should," and "may" should be construed as the opinions of technical
experts, not the dictates of law.

       This document originated in an invited workshop, conducted by EPA and DOE in March
 1991, where participants  deliberated in several topical work groups.  The discussions herein reflect
the experiences and opinions of more than 100 technical professionals who have planned and
conducted heterogeneous waste studies. The field is an evolving one, and the workshop  made it
clear that professional opinions vary on the applicability of different methods. Insofar as possible,
this report attempts to cover the full  range of techniques available and experiences with those
techniques. Consequently the reader will not generally find the document espousing the use of any
one approach  or set of methods, and, indeed, there are differences in approach between chapters.

       This document is principally targeted towards the needs of project managers. It should be
of most use to those responsible for planning and conducting waste characterization studies, as  well
as agency personnel who regulate such studies and contractor personnel responsible for oversight.
Agency  decision-makers should also  find  these  materials helpful as they attempt  to frame
answerable questions concerning the contamination of heterogeneous wastes. Finally, the report's
research and development recommendations should be of interest to those responsible for these
activities within EPA and DOE.

       Official  agency guidance  concerning the general techniques  to be used  in  waste-site
investigations is available elsewhere and is not repeated herein. This  document focuses on methods
specific to heterogeneous materials and, where appropriate, the adaptation of standard methods to
these materials. The techniques that are discussed can be applied to characterizing municipal solid
waste and non-hazardous  industrial  waste.   However, there is a strong  emphasis on wastes
contaminated  with hazardous chemicals and/or  radionuclides because of the  added risks  and
characterization difficulties they pose (learning what solid wastes are present is only a small  step
towards ascertaining the nature and degree of contamination),

        The discussion in this document assumes that initial site surveys have  been carried  out and
a detailed waste characterization study is to be performed. No size  limit has been placed on the
waste materials that are  considered. Contaminant characterization  of structures is not explicitly
dealt with because the unique aspects of building decontamination have  been  discussed  elsewhere.
Nor are soil characterization methods dealt with, although on a certain scale soil is a heterogeneous
medium. A great number of soil characterization methods have been developed in recent years,
but this report is concerned with materials that vary on a larger scale than soil.

       The scope of discussion herein has  been circumscribed by the type  of question a waste
characterization study may be designed to answer.  Among the questions  most commonly  asked are:

*      Is the container of waste (or the waste pile) hazardous as  a unit?

*      Does the container or pile include regions of material that are  hazardous?

•      What is the likely effectiveness of each potential remedial measure?

•      Is the treatment process working properly?

 •      Do residuals from the treatment process meet standards?

        This document focuses on characterization efforts designed to answer the second of these
 questions. Generally stated, the question is: Does the unit of waste contain areas of contamination
 that exceed the specified action level? This may be the most difficult of the five questions, and
 advancing the "state of the science" in answering this question will provide benefits in all aspects of
 heterogeneous waste characterization. The discussions herein have been further targeted towards
 two specific waste site scenarios. The first scenario is uncontainerized heterogeneous waste: waste
 piles, dumps, landfills, or buried deposits.  The second scenario is a "worst case" instance of
 containerized  waste (Figure  1-1).  This is a drum of heterogeneous waste similar  to those
 encountered by the  thousand at DOE facilities.  The drum contains metal, plastic, liquid-filled glass
 containers packed  in paint cans, laboratory tissues, and other solid wastes.  The contaminants
 present within the solid wastes may be radioactive or chemical or both, and their  identity is not
 known a priori.
       The next  chapter presents  a  set  of definitions pertinent  to the  characterization  of
heterogeneous wastes.  In addition to the regulatory definitions, there are colloquial or common-
sense definitions for a number of terms. These cannot supercede the regulatory definitions, but
they provide a functional complement to the often-arcane definitions codified in law. This set of
definitions provides a common basis for all of the chapters that follow.

       Chapter  3  deals  with  project planning as  performed  for  heterogeneous  waste
characterization. Establishing data needs and data quality objectives is the first topic handled, The
discussion steps through the "DQO process" that EPA has developed for planning site investigations,
using drummed  hazardous and  radioactive  wastes as an example.   This is followed by an
examination of the more detailed, practical considerations that enter into the selection of a strategy
for characterizing heterogeneous materials.  The issues of formulating a site model,  selecting an
appropriate statistical sampling design,  and protecting worker health and safety are discussed.

       Chapter 4 covers data quality assessment issues specific to heterogeneous wastes. Methods
for evaluating data  bias and precision  are discussed.   The  question  "How many  samples  are
enough?" is addressed,  Heterogeneous-waste quality  assurance issues  in  need  of research  or
development are presented.

       In Chapter 5, the field activities of the waste characterization study are discussed. Both
containerized and uncontainerized wastes are covered.  The various  non-intrusive methods and
those that involve handling the waste  are described.   The strengths and limitations of current
methods are set forth, and developing technologies that may offer promise are described.

       Chapter 6 deals with the laboratory aspects of the problem.  Because analytical  methods  are
the same for heterogeneous wastes and conventional environmental media, this chapter  concentrates
on the laboratory preparation of heterogeneous waste samples:  sample receipt, subsampling, and
extraction.  Quality control procedures unique to these wastes are evaluated, and the special prob-

                                                             Poly Bag Liner (s)
Inner Bags Tied Shut I a
  Taped Metal
                                                      50 ml Liquid
     (e.g., wipes, tongs,
     packing materials
      cans, bottles...)
Gallon Paint Can
                                                              (Lid on - bottle
                                                             inside containing
                                                             unknown liquid)
                         Additionally may contain:
                  asbestos gloves, respirator cartridges,
                          photographic materials

                                  Total "free"
                            liquid < 1% by volumne
               Figure 1-1. The worst-case drum of heterogeneous wastes.


lems they pose for laboratory waste management are discussed.  Methods for handling materials
with radioactive and VOC contamination are examined.

       Each of the foregoing chapters includes recommendations for research that could benefit
that particular stage of a heterogeneous waste characterization study. Chapter 7 enlarges on these
and considers the characterization of heterogeneous waste in a holistic perspective. Specific ways
to aid characterization studies through technology transfer and enhanced communication are dis-
cussed. Regulatory impediments are identified, and suggestions for surmounting them are given.
New ideas for enhancing the overall characterization and remediation process are set forth.

       Appendix A presents  a hypothetical case history, a site containing thousands of waste drums
that must be characterized.  The discussion steps through the process of surveying the drums,
defining  the decision to  be  made, establishing the  data needs and data  quality objectives, and
designing the study itself. The example is intended to reasonably portray the difficulties inherent
in such studies without invoking overwhelming detail or esoteric mathematics.

       Appendix B is a discussion of statistical techniques that may be useful for characterizing
heterogeneous wastes.  The most promising techniqus are rated with respect to their degree of
development and potential utility. Literature references are given for each method.


1.      U.S. Environmental Protection Agency.  1988. Report to Congress: Solid Waste Disposal
       in the  United States.   Volume II. Office of Solid Waste and  Emergency Response,
       Washington,  DC. EPA/530-SW-88-011B.

2.      Harmon, L.H. 1991. Heterogeneous waste characterization needs and requirements of the
       Department of Energy. Panel presentation to the Heterogeneous Waste Characterization
       Workshop, Las Vegas, March 26,  1991. U.S. DOE Office of Technology Development.

Chapter 2

                               Fred Haeberer and Jeff van Ee

       Characterization of hazardous wastes requires a clear understanding of the overall objectives
behind the data collection effort.  Legislation and regulations often form a basis for those objectives
with the definition of key terms  being an important consideration.

       Definitions  of terms can come from a variety of sources.  Legislation and regulations,
dictionaries, publications for the  layperson and professional  publications,  and professional societies
are some of the more common sources. Occasionally, there may be no consensus on the definition
of a term.   This lack of consensus might  be construed to  mean that there is no definition.
Sometimes a term may be so commonly used that little thought is given to whether a documented,
authoritative definition exists. In meeting the objectives of a data collection effort, it is important
that everyone involved in the various aspects of the effort have a common language and that there
be clear definitions of terms.

       Members of the Definitions work group came from several  different organizations.  A
number of different perspectives and professional experiences were represented within the group.
The group's objective was  to develop a common language  for the Workshop on Characterization
of Heterogeneous Waste to facilitate communication among the individual work groups and allow
workshop  participants to  address  the  difficult  issues  involved  in the characterization  of
heterogeneous waste.  Obviously, one of the first tasks was to define "heterogeneous waste."

       Definitions development by the work group began with the examination of pertinent, existing
definitions  found  in laws, regulations, dictionaries, and professional publications. These definitions
were examined for their appropriateness to the  characterization of heterogeneous waste.  If a
legally-defined term was  judged sufficient by  the  scientists  and engineers involved in the
characterization of heterogeneous waste, the definition was accepted. If a legally-defined term was
found to be  unacceptable  to the Definitions work group  membership, then  either a layman's
definition or one  from a professional source was put forward to accompany the legal definition.
Debate and discussion of these definitions was often lengthy, with most participants realizing that
the achievement of consensus for a single definition of a term was neither practical nor wise.

       The list of definitions provided below identifies the source of each definition. In some cases,
definitions provided in the listed references have  been modified slightly to make them more
readable and achieve consensus.  Users of the list of definitions are advised to consult the original
source if the fine  points of a definition become important.

       While considerable effort was expended in obtaining an accurate, up-to-date, complete list
 of definitions of terms that might commonly be used in the characterization of heterogeneous
 wastes, much more effort needs to be expended.  Several courses of action are recommended.

       Present-day computer hardware and software  allow for lists of definitions to be easily
 assembled, updated, and conveyed to interested persons. "Hypertext" is a relatively new term for
 information presented on a computer screen, where specific items can be marked for further, more
 detailed information.  Hypertext allows for documents to be written at different levels for different
 people, aiding in  the communication of terms that have different levels  of understanding and
 different sources.   Furthermore, a definition often may include terms that require  additional
 definition. Consequently, it is recommended that a hypertext version of the list of definitions be
 developed and periodically updated.

       Many of the  terms used in the characterization of heterogeneous waste require further
 clarification and standardization, especially across  agencies.  Consequently,  it is recommended that
 a  follow-up  definitions workshop  be held  for professionals from a variety of backgrounds to
 collaborate in further defining these terms and those from related areas.

       Finally, the work group  felt  that, once a list of definitions is developed, it will be important
 for that list  to  serve as  a  source of definitions in the development or  revision of laws and

       Whether the following list of definitions meets  the needs of users outside of the workshop
 remains to be seen. Further modification of the list can be expected, with new terms being added
 and existing terms being expanded  and clarified. The weight given to this list will depend on how
 often people refer to  the list and incorporate the definitions  into their daily vocabulary. If all of
 the participants in the workshop, in their respective work groups, adopt the list, then considerable
 progress toward establishing a uniform language will  have been made. At the beginning of the
 workshop, there were far more definitions of terms than at the end. Time will tell whether those
 remaining definitions prove to  be workable and worthy  of incorporation into other lists of
 definitions in other situations.

       Past regulatory development practice has resulted in the redefinition of some standard, non-
technical English words.  Furthermore, different government agencies can define the same non-
technical word to mean different things  (e.g., see Ground water and Groundwater below). This
practice may jeopardize the public's right to know and, more importantly, its right to understand
environmental legislation and must be avoided.  The  involved agencies must be encouraged to
negotiate terminology that is consistent with common usage and mutual understanding.

       As a partial remedy, careful usage of technical jargon is proposed. For example, "hazardous
waste" should retain its natural meaning, which is applicable to all wastes that can be potentially
damaging to human health or the environment.  The specialized EPA term should be referred to
as "Hazardous waste according to 40 CFR xx, Part xx," or "RCRA hazardous waste." The definitions

below that are not preceded by a citation were developed by the work group and are intended to
be operational rather than legal or enforceable.

       The primary source of the terms and definitions presented in this glossary is the Code of
Federal Regulations (CFR). Additional sources  include the Environmental Protection Agency's
"Quality Assurance Glossary  and Acronyms" (EPA QA Glossary, reference 3),  Department of
Energy Orders (e.g., DOE 5820), Webster's 9th Collegiate Dictionary,  "Principals of Environmental
Sampling" by Lawrence H. Keith, as well as DOE's "Defense Waste Primer" and "Radiation Primer."
A document that was not available to the work group at the  time of its meeting is the "Glossary of
Environmental Management Terms" compiled by the Training and Management  Systems  Division
of Oak Ridge Associated Universities for the U.S. Department of Energy. This document is highly
recommended for persons active in environmental analysis  and restoration.
Agricultural solid waste:                   EPA 40 CFR 243.101
The solid waste that is generated by the rearing of animals,  and the producing and harvesting of
crops or trees.

Biological wastes:
Non-pathological wastes, waste medical supplies, and non-contaminated media that could support
pathogen  growth.

Bulky waste:                               EPA 40 CFR 243.101
Large items of solid waste such as household appliances, furniture, large auto parts, trees, branches,
stumps, and other oversize wastes whose large size precludes  or complicates their handling by
normal solid waste collection,  processing, or disposal methods.

Candidate method:                         EPA 40 CFR 53.1
A method of sampling and/or  analyzing environmental matrices for environmental pollutants for
which an application for reference method determination or equivalent method determination is
submitted in accordance with the CFR.

The determination of the physical,  chemical, radiological,  and biological properties of a pure
substance, compound, or mixture to the extent necessary to support informed decision making.

Characterization method:
A protocol for determining  physical, radiological, biological, and  chemical properties of a  material.
Proper usage of this  term  for waste materials  couples the subject material with  an appropriate
governing regulation;  for example:
       "ICP-AES is a useful characterization method for metallic cations at levels specified by the
       Safe Drinking Water Act."

Commercial solid waste:                   EPA 40 CFR 243.101
All types of solid waste generated by stores,  offices, restaurants, warehouses, and other  non-
manufacturing  activities, excluding residential and industrial wastes.

 Composite sample:
 A sample composed of several distinct subsamples.  Composite samples are often prepared to
 obtain a more representative sample of the unit or when it is not economically feasible to analyze
 a large number of individual samples.

 Construction and demolition waste:        EPA 40 CFR 243.101
 The waste building materials, packaging, and rubble  resulting from construction, remodeling, repair,
 and demolition operations on pavements, houses, commercial buildings, and other structures.

 Confidence  coefficient:                     QA Glossary (3)
 The probability statement that accompanies a confidence interval and is equal to unity minus the
 associated type I  error rate (false positive rate).  A confidence coefficient of 0.90 implies that 90%
 of the intervals resulting from repeated sampling of a population will include the unknown (true)
 population parameter.  See  also Confidence interval.

 Confidence interval:                       QA Glossary (3)
 The numerical interval constructed around a point estimate of a population parameter, combined
 with a probability  statement  (the confidence coefficient) linking  the interval to the population's true
 parameter value. If the same confidence interval construction technique and assumptions are used
 to calculate future intervals, they will include the unknown population parameter with the same
 specified probability. See also Confidence coefficient.

 Contact-handled transuranic waste:         DOE 5820.2A
 Packaged transuranic waste  whose external surface dose rate does not exceed 200 mrem per hour.

 Control sample:
 A sample introduced into a sampling and  analytical process to monitor the performance of the

 Curie (Ci):                                EPA 40  CFR 190.02
 That quantity of  radioactive material producing 37 billion nuclear transformations per second
 (equivalent to 37 billion becquerels (Bq)).

 Data performance criteria:
 The qualitative and quantitative constraints on the design for data collection that  result from the
 application of the  DQO process. These constraints, commonly referred to as DQOs, are  used as
 the basis for the statistical survey design of the data collection effort to insure that the right type
 and quality of data are collected.

 Data quality objective (DQO) process:
 The up-front interactive process between the data user (i.e., decision maker) and the data generator
 (i.e., supporting technical team)  which defines the error level (uncertainty) acceptable in the
 decision/application.  See also Data performance criteria.

The solid remains of a broken or destroyed item.

 That individual who will represent the regulatory agency or other responsible organization in
 deciding whether site remediation is needed and selecting the appropriate action(s).

 Designated facility:                        EPA 40 CFR 260.10
 A hazardous waste treatment, storage, or disposal facility that has received an EPA permit (or a
 facility with interim status) in accordance with the requirements of 40 CFR 270 and 124, a permit
 from a state authorized in accordance with 40 CFR 271, or that is regulated under 40 CFR 261-
 6(c)(2) or Subpart F of Part 266, and that has been designated on the  manifest by the generator
 pursuant to Section  262.20.

 Detection limit:
 The lowest concentration  or amount of a  target  analyte that  can be determined by a single
 measurement to be different from zero or background level at a defined level of probability. The
 detection limit is generally recognized to be sample-matrix and measurement-method dependent.
 See also Method detection limit.

 The systematic and orderly placement, storage, distribution, or transformation of wastes.

 Disposal facility:
 A facility at which waste is intentionally placed, stored, distributed, or transformed.

                                          EPA 40 CFR 260.10, 270.2
 A facility at which hazardous waste is intentionally placed and at which the waste will remain after
 its closure.

 Disposal site - radiological:                 EPA 40 CFR 192.01
 The region within the smallest perimeter of residual radioactive material  (excluding  cover material)
 following completion of control activities.

 Disposal site - hazardous, including mixed waste:
 A location where hazardous waste is disposed.

 Dose equivalent:                           EPA 40 CFR 190.02
 The product of absorbed dose and appropriate factors  to account for differences in biological
 effectiveness due to  the quality of radiation and its spatial distribution in the body. Also see DOE
 Order 5400.5, "Radiation Protection of the Public and the Environment," February 8, 1990.

 Equivalent method:                        QA Glossary (3)
Any method of sampling and/or analysis demonstrated to result in data having a  consistent and
 quantitatively known relationship to the results obtained with a reference method under specified
 conditions, and formally recognized by the EPA.

Extremely hazardous substance:            EPA 40 CFR 355.20
A substance listed in Appendices A and B of 40 CFR 355.

                                          DOE EH-231-003/0191 (January 1991)
Certain hazardous  substances that, when released at  levels above their CERCLA Reportable
Quantities, require notification  of local and  state emergency response authorities due to the
potential for serious irreversible health effects.  See also  Reportable quantity.

Facility - RCRA:                           EPA 40 CFR 260.10
All contiguous land, and structures, other appurtenances, and improvements on the land used for
treating, storing, or disposing of hazardous  waste.  A  facility  may consist of several treatment,
storage,  or disposal operational units  (e.g., one or more  landfills,  surface impoundments,  or
combinations of them).

Facility - Waste:
All contiguous land, and structures, other appurtenances, and improvements on the land, used for
treating, storing, or disposing of waste.  A facility may consist of several treatment, storage, or
disposal operational units (e.g., one or more landfills, surface impoundments, or combinations of

Field measurement:
A determination that is made on-site.

Field screening:
Rapid, qualitative, or semi-quantitative on-site measurements.

Final closure:                              EPA 40 CFR 260.10
The closure of all hazardous waste management  units at the facility in accordance with all applicable
closure requirements so that hazardous waste management activities under 40 CFR Parts 264 and
265 are no longer conducted at the facility unless subject to provisions in Section 262.34.

Food waste:                               EPA 40 CFR 243.101
The organic residues generated by the handling, storage, sale, preparation,  cooking, and serving of
foods, commonly called  garbage.

Generator:                                EPA 40 CFR  260.10
Any person, by site, whose act or process produces hazardous waste identified or listed in 40 CFR
Part 261 or whose act first causes a hazardous  waste to  become subject to regulation.

Geologic repository:                       NRC 10 CFR 60.2
A system (site) located in excavated geologic media and  intended for disposal of radioactive waste.

Ground water:                             EPA 40 CFR  260.10
Water below the land surface in a zone of saturation.

Groundwater:                              NRC 10 CFR  60.2
All water which occurs below the land surface.

Hazard ranking system:                     DOE  5480.14
The methodology  used  by EPA to evaluate the relative potential of inactive  hazardous waste
facilities to cause health or safety problems or ecological or environmental damage.


 Hazardous air pollutant:
 A substance anticipated to cause either mortality or serious illess when released to the air. The
 eight hazardous air pollutants are asbestos, benzene, beryllium, coke oven emissions, inorganic
 arsenic, mercury, radionuclides, and vinyl chloride.

 Hazardous chemical, material, or substance:
 Any substance which, within a specific concentration range, poses an unacceptable risk to human
 health or the environment.

 Hazardous chemical:                      EPA 40 CFR 355.20, 370.2
 Any chemical defined as hazardous under section  1910.1200(c) of Title 29 of the CFR with certain

                                         OSHA 29 CFR 1910, Subpart Z
 Any chemical which is a physical hazard or a health hazard. Physical hazards include  combustibles,
 liquids, compressed gases, explosives, flammables, organic peroxides, oxidizers, pyrophorics, and
 reactives.  A health hazard is any chemical for which there is good evidence that acute or chronic
 health  effects  occur in exposed employees.

 Hazardous constituent:                   EPA 40 CFR 268.2
 A constituent listed in Appendix VIII of 40 CFR 261.

 Hazardous material:                      DOT 49 CFR 171.8; KMT A,  Sect.  1802
 A substance or material,  including a hazardous substance, which has been determined by the
 Secretary  of Transportation to be capable of posing an  unreasonable risk to health, safety, and
 property when transported in commerce, and which has been so designated.

 Any of the more than 16,000 materials appearing in the Hazardous Materials Table in 49 CFR

 Hazardous substance:                     DOT 49 CFR 171.8
 A material, including its mixtures and solutions, that:

       (1)     Is listed in Appendix to 49 CFR 172.101;
       (2)     Is in a quantity, in one package, which equals or exceeds the reportable quantity
              (RQ) listed in the Appendix to 49  CFR 172.101; and
       (3)     When a mixture or solution
              (i)     For  radionuclides,  conforms  to  paragraph 6  of the Appendix  to
                     49 CFR 172.101.
              (ii)     For other than radionuclides, is in  a concentration by weight which equals
                     or exceeds the concentration corresponding to the reportable quantity  of the
                     material, as shown in the table found in 49 CFR 171.8,

Hazardous substance:                     EPA 40 CFR 300.5
Any substance designated by sections 307(a) or 31 l(b)  of the CWA, section 102 of CERCLA,
section 3001 of the Solid Waste Disposal Act, Section 112 of CAA, or Section 7 of TSCA.

                                          DOE EH-231-003/0191 (January 1991)
                                          "Hazardous" Terminology
Any substance that, when released to the environment in an uncontrolled or unpermitted fashion,
becomes subject to the reporting and possibly response provisions of the Clean Water Act and

Hazardous waste:                         DOT 49 CFR 171.8
Any material that is subject to the Hazardous Waste Manifest Requirements of the Environmental
Protection Agency specified in 40 CFR Part 262.

                                          EPA 40 CFR 243.101, 260.10, 261.3, 302.3
Any waste (or combination of wastes) which poses a  substantial present or potential hazard to
human health or living organisms due to its lethal, non-degradable or persistent nature or because
it may cause or tend to cause detrimental cumulative effects.

                                          DOE EH-231-003/0191 (January 1991)
A solid  waste that  must be treated, stored, transported, and disposed of in  accordance with
applicable requirements under Subtitle C of RCRA.

Hazardous waste constituent:               EPA 40 CFR 260.10
A constituent of a waste that results in its listing as a hazardous waste as per 40 CFR 261, Subpart
D, or a constituent listed in Table 1  of 40 CFR 261.24.

Hazardous waste management facility:       EPA 40 CFR 270.2
All contiguous land, and structures, other appurtenances, and improvements on the land, used for
treating,  storing,  or disposing of hazardous waste.  A  facility may consist of several treatment,
storage,  or  disposal operational units (e.g., one or more  landfills, surface impoundments, or
combinations of them).

Heterogeneous:                            Webster's  9th Collegiate Dictionary
Consisting of dissimilar or diverse ingredients or constituents.

Heterogeneous  samples:                    Keith (1)
Samples  that are not consistent in composition or phase throughout. Heterogeneous samples will
not provide representative data when aliquots of them are analyzed.

Heterogeneous  waste:
A waste for  which a sample of a size suitable for analysis is  not representative of the property of
concern. Thus a series of samples will have to be analyzed to establish a range of  results acceptable
to the data user.

High-level (radioactive) waste (HLRW):     NRC 10 CFR 60.2
(1) Irradiated reactor fuel, (2) liquid wastes resulting from the operation of the first cycle solvent
extraction system, or equivalent, and the concentrated wastes from subsequent extraction cycles, or
equivalent, in a facility for reprocessing irradiated reactor fuel,  and (3) solids into which such liquid
wastes have  been converted.

Homogeneous:                            Webster's 9th Collegiate Dictionary
Of uniform structure or composition throughout.

Imminently hazardous chemical substance or  mixture:     EPA 40 CFR 61.01
A chemical substance or mixture which presents an imminent and unreasonable risk of serious or
widespread injury to health or the environment.

Industrial solid waste:                     EPA 40 CFR 243.101
The solid waste generated by industrial processes and manufacturing.

Infectious waste:                           EPA 40 CFR 241.101
Laboratory wastes, surgical operating room pathologic specimens, disposable material of a medical
nature from patients suspected to have or diagnosed to have a communicable disease.

Inorganic solid debris:                     EPA 40 CFR 268.2
Non-friable inorganic solids that are incapable of passing through a 9.5 mm standard sieve that
require cutting, or crushing and grinding in mechanical sizing  equipment prior to stabilization,
limited to the following inorganic or metal materials:  metal  slag; classified slag; glass; concrete,
masonry, and refractory bricks; metal cans, containers, drums, or tanks; metal nuts, bolts, pipes,
pumps, valves, appliances, or industrial equipment; and scrap metal.

Institutional solid waste:                   EPA 40 CFR 243.101
The solid wastes generated by educational, health care, correctional, and other institutional facilities.

Landfill - RCRA  (Subtitle C):              EPA 40 CFR 260.10
A disposal facility or part of a facility where hazardous waste is placed in or on land and which is
not a pile, a land  treatment facility, a surface impoundment, an underground injection well, a salt
dome formation,  a salt bed formation, an underground mine,  or a cave.

Land treatment facility - RCRA:            EPA 40 CFR 260.10
A facility or part  of a facility at which hazardous waste is  applied onto or incorporated into the soil
surface; such facilities are disposal facilities if the waste will remain after closure.

Leachate - RCRA:                         EPA 40 CFR 241.101, 260.10
Any liquid that has percolated through or drained from hazardous or solid waste and has extracted
dissolved, or suspended materials from it.

Level of uncertainty:
The probability of a wrong answer.

Low-level (radioactive) waste:               DOE 5820.2A
Waste that contains radioactivity and is not classified as high-level  waste, transuranic waste, or spent
nuclear  fuel or by-product  material as defined by this  Order. Test  specimens of fissionable material
irradiated for research and development only,  and not for the production of power or  plutonium,
may be classified as low-level waste, provided the concentration of transuranic is less than 100

A specific subset of media (e.g., surface water, drinking water, kaolinite) in which the analyte of
interest may be contained.

A quantitative determination of one or more properties.

The solid, liquid, or gas that serves as a carrier of the analytes of interest.

Method detection limit (MDL):             40 CFR 136, Appendix B
The minimum concentration of an analyte that, in a given matrix and with a specific method, has
a 99% probability of being identified, qualitatively or quantitatively measured, and reported to be
greater than zero. [MDLs are method and sample matrix dependent. See Detection Limit.]

Mixed waste:                              Defense Waste Primer
Mixtures  containing radioactive and hazardous constituents.

                                          DOE 5820.2A
A waste that is both radioactive as defined by the Atomic Energy Act and hazardous as defined by
the Resource Conservation and Recovery Act.

                                          Radiation Primer
Waste that  satisfies  the  definition of LLRW in the Low-Level  Radioactive  Waste  Policy
Amendments Act of 1985  and contains hazardous waste that either (1)  is listed as  a hazardous
waste in Subpart D of 40 CFR 261 or (2) causes the LLRW to exhibit any of the hazardous waste
characteristics of Subpart C of 40 CFR 261.

Mixture:                                  EPA 40 CFR 355.20, 372.3
A heterogeneous  association of substances where  the various individual substances retain their
identities and can usually be separated by mechanical  means. Includes solutions or  compounds but
does not include alloys or amalgams.

Non-intrusive  characterization:
A non-destructive determination that causes no significant change in the material being examined
and  does  not involve physical  entry. Used interchangeably with "non-invasive."

Open dump:                              EPA 40 CFR 241.101
A land disposal site at which solid wastes are disposed of in a manner that does not  protect the
environment, is susceptible to open burning, and is exposed to the elements,  vectors, and scavengers.

Operable  unit:                             EPA 40  CFR 300.6
A discrete part of the entire response action that decreases a release, threat of release, or pathway
of exposure.

Pile:                                      EPA 40 CFR 260.10
Any non-containerized accumulation of solid, nonflowing hazardous waste that is used for treatment
or storage.


 Pollutant:                                  EPA 40 CFR 122.2
 Dredged spoil, solid waste, incinerator residue, filter backwash, sewage, garbage, sewage sludge,
 munitions, chemical wastes, biological materials, radioactive materials, heat wrecked or discarded
 equipment, rock, sand, cellar dirt,  and industrial, municipal, and agricultural waste discharged into
 the environment.

 Quality assessment:                        EPA (2)
 The overall system of activities that provides  an objective measure of the quality of data produced.

 Quality assurance:                         EPA (2)
 A system of activities whose purpose is to provide to the producer or user of a product or service
 the assurance that it meets defined standards of quality. It consists of two separate, but related
 activities,  quality control and quality  assessment.

 Quality Control:                           EPA (2)
 The overall system of activities whose purpose is to control the  quality of the measurement data so
 that they meet the needs of the user.

 Quantitation limits
 The maximum or minimum levels (concentrations) or quantities of a target variable (analyte) that
 can  be quantified with the required certainty by a single  application of the (quality-controlled)
 measurement  method.

 Radiation:                                 EPA 40 CFR 190.02
 Alpha, beta, gamma, or X-rays; neutrons; and high-energy electrons, protons,  or other atomic
 particles; but not sound or  radio waves, nor visible, infrared, or ultraviolet light.

 Radioactive material:                       EPA 40 CFR 190.02
 Any material which spontaneously emits radiation.

 Radioactive waste:
 The byproducts  of obsolete or discarded products of nuclear activities that emit radiation.

                                           NRC 10 CFR 60.2
 High level waste and radioactive materials other than high  level waste that are received  for
 emplacement in a geologic repository.

                                           DOE 5400.3, 5820.2A
 Solid, liquid, or gaseous material that contains radionuclides regulated under the Atomic Energy
Act of 1954, as  amended, and of negligible economic value considering costs of recovery.

                                           DOE 5480.2
 Solid or fluid materials of no value  containing radioactivity; discarded items such as clothing,
containers, equipment, rubble, residues, or soils contaminated with radioactivity; or soils, rubble,
equipment, or other items containing  induced radioactivity such that the levels  exceed safe limits
for unconditional release.

Reference method:                         QA Glossary (3)
A sampling and/or measurement method which has been officially specified by an organization as
meeting its data quality requirements.

Reportable quantity:
The maximum amount of material that can be stored, transported, or otherwise handled above
which specific regulatory practices are required.

                                          EPA 40 CFR 302 (CERCLA)
The quantity designated for each of 699 hazardous substances under the provisions of section 102
of CERCLA. These spill quantities are for any 24-hour period and include spills on land and in
the air in addition to spills in the water.

Repository:                                DOE 5820.2A
A facility for the permanent deep geological disposal of high level or transuranic waste.

Representative  sample:                    EPA 40 CFR 260.10
A sample of a universe or whole (e.g., waste pile, lagoon, ground water) which can be expected to
exhibit  the average properties of the universe or whole.

Risk:                                     QA Glossary (3)
The probability or likelihood of an adverse effect.

Rubbish:                                  EPA 40 CFR 243.101
Solid waste, excluding food wastes and ashes, taken from residences, commercial establishments,
and institutions.

Runoff:                                    EPA 40 CFR 241.101, 260.10
The portion of precipitation that drains from an area as surface flow.

Sanitary landfill:                          EPA 40 CFR 241.101
A land  disposal site employing an engineered method  of disposing of solid wastes on land in a
manner  that minimizes  environmental hazards by spreading the  solid wastes to the smallest practical
volume, and applying and compacting cover material at the end of each operating day.

Saturated zone:                            NRC  10 CFR 60.2
That part of the earth's crust beneath the regional water table  in which all voids, large and small,
are ideally filled with water under  pressure greater than atmospheric.

A solid  material deposited by water, wind,  or glaciers.

Site:                                      EPA 40 CFR 270.2
The land or water area where any facility or activity is  [or was] physically located or conducted,
including adjacent land used in connection  with the facility or  activity.

Any mixture of solids, semi-solids,  or dense liquid wastes which settle out of solution.


                                           EPA 40 CFR 122.2, 241.101, 243.101, 260.10
Any solid,  semi-solid, or liquid waste  generated from a municipal, commercial,  or industrial
wastewater treatment plant, water supply treatment plant, or air pollution control facility exclusive
of the treated effluent from a wastewater treatment plant.

Naturally occurring geo-organic materials smaller than 2 mm in size, generally found in the surface
layer of the Earth and supporting plant life.

Solid waste:                                EPA 40 CFR241.101, 243.101, 260.10, 261.2
Garbage, refuse, and other discarded solid materials, including solid waste materials resulting from
industrial, commercial, and agricultural  operations, and from community activities, but does not
include  solid or dissolved materials in domestic sewage  or  other significant  pollutants in water
resources, such as silt, dissolved or suspended solids in industrial wastewater effluent, dissolved
materials in irrigation return flows, or other common water pollutants. It generally  does  not include
mining,  agricultural, and  industrial solid wastes; hazardous wastes;  sludges; construction and
demolition wastes;  and infectious wastes.

Spent nuclear fuel:                         EPA 40 CFR 191.02
Fuel that has been withdrawn  from  a nuclear reactor following irradiation, the constituent elements
of which have not been separated by reprocessing.

Standard conditions:
The defined reference point(s) by which ambient measurements are related.

A large receptacle for holding, transporting, or storing fluids.

                                           EPA 40 CFR 260.10
A stationary device, designed to contain an accumulation of hazardous waste which is constructed
primarily of non-earthen materials  (e.g., wood,  concrete,  steel, plastic) which provide structural

Traceable:                                 EPA 40 CFR 50.1
The comparison and certification of a local standard directly  or by no more than one intermediate
standard, to a primary standard such as a NIST Standard Reference Material or a USEPA/NIST-
approved reference material.

Transuranic radioactive waste:              EPA  40 CFR 191.02
Waste containing more than 100 nanocuries of alpha-emitting transuranic isotopes, with half-lives
greater than 20 years, per gram of waste, with certain exceptions  (e.g., high-level wastes).

Unsaturated zone:                          NRC 10 CFR 602
The zone between the land surface  and the regional water table. Generally, fluid pressure in this
zone is  less than atmospheric pressure, and some of the voids may  contain air or other gases at
atmospheric pressure. Beneath flooded areas  or in perched water bodies the fluid  pressure locally
may be greater than atmospheric.

Vadoze zone:
Of, relating to, or being water or solutions in the Earth's crust above the permanent groundwater

Material regarded as damaged, defective, or superfluous by a segment of society.
       Also see the following alphabetical listings:
       Agricultural solid waste:
       Bulky waste:
       Commercial solid waste:
       Construction/demolition waste:
       Food waste:
       Hazardous waste:

       Infectious waste:
       Industrial solid waste:
       Institutional solid waste:
       Mining waste:
       Mixed waste:
       Municipal solid waste:
       Solid waste:
       Street  wastes:
       Transuranic radioactive waste:
       Waste  waters:
EPA 40 CFR 243.101
EPA 40 CFR 243.101
EPA 40 CFR 243.101
EPA 40 CFR 243.101
EPA 40 CFR 243.101
EPA  40  CFR 243.101, 260.10, 261.3, 302.3
DOT  49 CFR 171.8
EPA 40 CFR 241.101
EPA 40 CFR 243.101
EPA 40 CFR 243.101
EPA 40 CFR 243.101
DOE Order 5820.2A
EPA 40 CFR 241.101
EPA  40  CFR 241.101, 243.101, 260.10, 261.2
EPA 40 CFR 243.101
EPA 40 CFR 191.02
EPA 40 CFR 268.3
Waste form:                              NRC  10 CFR 60.2
Radioactive waste materials and any encapsulating or stabilizing matrix.

Waste package:                           NRC  10 CFR 60.2
The waste form and any containers, shielding, packing, and other absorbent materials immediately
surrounding an individual waste container.

Waste treatment unit:                     EPA 40 CFR 260.10
A device which receives and treats or stores aqueous hazardous waste.

Waste unit:
The smallest waste increment which is characterized.

Waste waters:                             EPA 40 CFR 268.3
Wastes that contain less than  1% by weight total organic carbon (TOC) and less than 1% by weight
total suspended solids (TSS), with the exceptions listed in 40 CFR 268.2.

White goods:
Large household appliances:  refrigerators, freezers, washing machines, etc.


1.      Keith, L.H. 1988. Principles of environmental sampling.  American Chemical Society.
       Washington, D.C. 480 pp.

2.      U.S. Environmental Protection Agency. 1990. A rationale for the assessment of errors in
       the sampling of soils.   J.J.  van Ee, LJ. Blume, and T.H.  Starks.  EPA/600/4-90/013.
       Environmental Monitoring Systems Laboratory, Las Vegas, NV.

3.      U.S. Environmental Protection Agency. 1991. Quality Assurance Glossary and Acronyms.
       Quality Assurance Management Staff, Washington, D.C.

 Chapter 3
                               Planning the  Study

                            Leon Bergman, Charlotte Kimbrough,
                               Mitzi Miller and Dean Neptune

       This chapter provides an overview of the heterogeneous waste characterization study, with
particular emphasis on planning the study. While the general activities are the same as those for
any environmental study, the peculiarities of heterogeneous materials dictate that some planning
activities take on critical importance.  This chapter presents a generic planning procedure for the
characterization of sites having either unconfined or drummed heterogeneous wastes. Because
many  existing disposal  sites  contain waste materials stored in drums, the development of a
consensus as to a useful procedure for these sites is of substantial and immediate importance.
Consequently, most of the examples in this chapter are directed toward defining reasonable and
rational protocols for characterizing  drums containing  waste materials, where there is often
significant uncertainty (or even no knowledge at all) as to the actual contents of the drums.

       Figure 3-1 presents a generalized scheme of the environmental study process. This flowchart
and the text of the chapter represent a concatenation of the work  of two groups within the
Heterogeneous Waste Workshop, one of which dealt with data needs and data quality objectives,
the other of which discussed selecting the study strategy. This figure forms the frame of reference
for the chapter. It recognizes that waste characterization is almost always carried out in conjunction
with characterization of the surrounding environmental media. Any specific study scheme must be
devised in  conformity  with applicable  agency  and  company  guidance and  procedures. The
investigation of an NPL site, for example, would include numerous Feasibility  Study,  Risk
Assessment,  and Community  Involvement elements  not  shown on  Figure  3-1.  For general
information on characterizing  hazardous waste sites, investigators should consult EPA and DOE
guidance documents and directives (1, 2, 3, 4, 5, 6, 7).

       Figure 3-1 depicts a site study as occurring in five phases. There is a preliminary planning
stage,  in which project scoping  occurs  and an initial  conceptual model of the site and the
contamination is formulated. This is followed  by  a sequence of steps called the "DQO  (data quality
objectives) process." The DQO process is  a study planning process under development by the EPA
that formalizes the elements of good experimental design. This process breaks the problem down
into discrete, specific questions to be addressed by the study, so that the results can be interpreted
with minimal ambiguity  and the decision-maker knows the uncertainty associated with the decision.

       The DQO  process is followed by  sampling and analysis  design. This involves the consider-
ation of alternative statistical designs for the study, and selection of an optimum design and field
and laboratory methods.   In reality, study planning almost always requires cycling back and forth
more than once between the latter steps of the DQO process and the methods  selection steps. The


                                  (PROJECT INITIATION)
                                  PRELIMINARY PLANNING
                                                PROJECT SCOPING
                                           REVIEW PROCESS INFORMATION
                                           AND ALL AVAILABLE SITE DATA

                                           IDENTIFY CONTAMINANTS OF
                                           CONCERN, MEDIA TO BE
                                           INVESTIGATED, POTENTIAL
                                           EXPOSURE PATHWAYS

                                           IDENTIFY DECISION MAKERS,
                                           KEY TECHNICAL STAFF

                                           IDENTIFY ARARt (REGULATORY
                                           AND REMEDIATION LIMITS)

                                           SPECIFY PROJECT BUDGET AND
                                           TIME CONSTRAINTS
                                     OQO PROCESS




                             SAMPLING AND ANALYSIS DESIGN

             PREPARE PROJECT
                                                   AT LEAST
                                                  ONE DESIGN
                                               MEET DOO, BUDGET,
                                               HiS AND TIME
                             SAMPLE COLLECTION AND ANALYSIS
                                   DATA ASSESSMENT
                                             TAKE CORRECTIVE ACTION

                                             COLLECT ADDITIONAL DATA

                                        REVISE SI A PLAN, COLLECT MORE DATA

                                         REVIEW/REVISE STATISTICAL DESIGN

                                                  REVIEW DQOs

                                          REVISE SITE CONCEPTUAL MODEL
       Figure  3-1.  Generalized scheme of thte study process.

 selection of a statistical design entails a trade-off between what is best from the point of view of
 statistical theory and what is within the project's cost and risk constraints.  These can be driving
 factors in  characterizing heterogeneous materials.   It appears  that  there are many  statistical
 techniques  that are potentially applicable to sampling heterogeneous wastes, but most of these have
 not been tested in this or similar applications. These methods are listed and discussed in Appendix

       The sampling and sample analysis steps  follow.   Methods for field characterization or
 collection of heterogeneous materials are dealt with in Chapter 5. Laboratory analysis is handled
 in Chapter 6. Most laboratory problems unique to heterogeneous  materials are encountered in the
 sample preparation steps rather than during chemical analysis.

       The final study phase shown in Figure 3-1 is data assessment. Chapter 4 suggests specific
 methods for evaluating accuracy, precision, and completeness of heterogeneous waste data. As in
 any environmental study, the assessment (both the data validation and all other project validation
 activities)  may  lead  the project manager to  conclude that corrective action is needed. With
 heterogeneous materials, both inherent and sampling/analysis variability may be much higher than
 anticipated. Unless this has been discovered and  compensated for during the study, it may result
 in a  failure to achieve the  DQOs. On reaching the data assessment phase,  the project team will
 have to backtrack and re-evaluate the procedures used, the applicability of the statistical design, the
 data quality objectives or even the conceptual waste model on which the entire project was based.

       This chapter emphasizes periodic re-evaluation throughout the study, to allow for mid-study
 changes  and avoid the need to re-design  and re-do the  entire study should the DQOs not  be
 attained at the end. It is critical to define early all the questions that are to be answered and  all
 the associated data and informational  needs as well as applicable regulations.  Furthermore, this
 is (and should  be) an  iterative  process throughout project planning and implementation. As
 additional project components are  completed and new information becomes available, new questions
 may  arise.  If appropriate, these questions should either be incorporated into the plan or become
 additional projects or  project phases. For instance, a drum of liquid solvent waste may be analyzed
 for a single solvent and water. When the percentage of solvent plus water accounts for only 40%
 of the drum volume,  a new question, "What makes Up the remainder of the  waste?" emerges.
Appropriate actions can then be defined to ensure full characterization of the drum.

       At each stage  where the judgment is made that more information is  needed to answer the
 question, the project manager will estimate the funding and other resources needed to collect that
 information. One possible action that is not explicitly shown on the flowchart is always available:
 the decision to terminate the characterization study, designate  all of the waste as exceeding the
 action levels, and treat or dispose of it  accordingly.   This  decision  may be based on cost
 considerations, concern for worker safety,  time constraints, or some combination of these factors.

                                   Preliminary Planning

Initial Activities

        The project begins when someone is required to make a decision concerning the disposition
of a waste deposit.  The decision maker may represent a regulatory agency or the organization
responsible for site management. Typical reasons for initiating the waste characterization include:
determining whether a regulation is met, whether the material can be stored or shipped, or whether
clean-up activities are warranted.  The preliminary planning and establishing data quality objectives
and sampling design are critical steps prior to beginning the data collection process. Without these
steps, considerable effort can be wasted collecting data that do not answer the question or allow a
decision to be made.

        When characterizing heterogeneous wastes, the early project activities are the same as those
for any other site study. The core project team reviews available information about the site and
the waste. The  decision-maker considers whether  an emergency containment/removal action is
warranted. In the case of uncontainerized wastes, the primary concern is likely to be whether
chemical contaminants can be leached from the wastes by rainwater. For drummed  wastes,  obvious
leakage of fluids is a major concern, as are signs of impending container failure that could allow

        An early site visit by  as many members of the project  team as possible is of critical
importance.  Only by seeing the debris pile  or the drum dump  can the investigators begin to
comprehend the heterogeneity that will have to be dealt with in the study. On the  site, the project
team should search  for patterns in the waste placement, begin to  think about how to classify the
waste into types (populations), and  consider what waste handling can and cannot reasonably be
done on site. Health and safety hazards posed by unstable debris  deposits and leaking containers
should  be identified. As much  information as  possible should be gleaned from drum labels.
Project Scoping

       Assembling the full project team is an early scoping activity. The ultimate success of the
characterization program is directly traceable to the effort that has gone into the project plan. The
planning is best accomplished by a team representing several  different, well-chosen disciplines.
While the team makeup may vary from project to project, it is  incumbent on the project manager
to choose individuals whose skills and experience cover all of the technical, health and safety, and
information aspects that will arise throughout the project, from planning through data reduction to

       The basic technical team for  a waste  characterization  study usually includes a chemist,
sampler, statistician, QA/QC specialist, health and safety officer, transportation specialist, and risk
assessment evaluator. Other disciplines that may be required (particularly if environmental media
are also to be characterized) are meteorology, toxicology, ecology, hydrogeology, various engineering
specialties,  and computer and data specialties.  A  civil engineer or construction/demolition specialist
may be needed if large debris, structures, piping, or tanks are to be disturbed or moved during the

       Throughout project planning, it is essential that both management and regulators be included
in the process and that they concur with the questions identified and the plan of action required
to answer those questions.  Management must commit resources, while the regulators will review
and approve management's determinations and plans; they may also be issuing permits based on
the plan.

       The pertinent regulations must be identified early in the planning process. They will drive
the final  cleanup levels, and hence have important ramifications  for data accuracy and precision and
required  analytical detection limits.  In  addition,  they may prescribe the waste sampling and analysis
methods to be used and circumscribe the available remedial methods. In the current regulatory
climate, waste is usually assumed to be RCRA hazardous,  either by being listed or characteristically
hazardous, until it is proven to be otherwise; therefore, RCRA usually applies. The same is true
for radioactivity, if there is any evidence that the waste could have radioactive components by
association from generative processes or if the source of the waste is simply unknown. When the
waste has the potential to be either mixed or radioactive, the number of applicable regulations is
multiplied significantly. Normally, the regulations that  should be considered are the environmental
regulations; however, others may also  apply.   Those applicable  or  relevant and  appropriate
requirements (ARARs) which may apply to the characterization of heterogeneous materials are:

•      Resource Conservation  and  Recovery Act (RCRA), which defines both solid and hazardous
•      Toxic Substances  Control Act (TSCA),  which, for example, regulates a number of hazardous
       chemicals, including polychlorinated biphenyls  (PCBs).
•      Clean Water Act (CWA), which is the source of many water quality limits and contains
       regulations for wastewater treatment and release to receiving bodies of water.
•      Federal Water Pollution Control  Act (FWPCA), which  governs the treatment of industrial
       wastes and sets standards.
•      Safe Drinking Water Act (SDWA), which is the source  of groundwater standards and
       standards for the  protection of drinking water sources.
•      Comprehensive Environmental  Response, Compensation,  and Liability Act (CERCLA), also
       known as "Superfund," which covers emergency responses and the  public right-to-know, as
       well as clean-up at inactive  hazardous waste sites,
•      Department of Transportation (DOT)  regulations that are  concerned with  the  movement
       of samples, sampling materials, and waste via public roads,  railroads, and air spaces.  While
       there are certain exclusions  for routine  laboratory samples in these regulations, radioactivity
       is covered for all materials, and the regulations are very  specific concerning data and
       packaging requirements.
•      Atomic Energy  Act,  interpreted through  Department of Energy  Orders and  Nuclear
       Regulatory Commission requirements.
•      Occupational Safety and Health Act (OSHA) regulations which  must be  met to ensure
       worker safety.

Other regulations which sometimes apply arise from:

•      Federal Insecticide,  Fungicide, and Rodenticide Act (FIFRA).
•      National Environmental Policy Act (NEPA).
•      State Superfund laws.
•      Various Corps of Engineer permits  and requirements.


•     Federal and state facility agreements, memoranda of understanding, complaints, and  orders.

       As a measure  of potential hazard, standards for waste surface contamination  levels (as
measured by wipe tests) would be very useful for debris.  Risk-based action levels of this type have
been developed for individual sites, but there are no national regulations embodying such standards.

       The project budget and a rough  schedule need to be established during project scoping.
Heterogeneous waste characterization generally requires more samples to achieve a given level of
certainty than does characterization of other materials. Furthermore, the effort required for sample
handling and preparation can be considerable, especially if wastes must be hand-sorted in the field.
Thus per-sample costs can be high, and the budget may be a serious constraint on the scope of the
project. The extra time required  to process  heterogeneous materials must also be taken  into

       Depending on  the site, there may be only a few or very many contaminants of concern.
Sites located in industrial districts have often been occupied by multiple businesses producing  a wide
range of wastes.  Privately operated dumps that were permitted to accept only municipal  wastes
often co-disposed chemical wastes from commercial  waste  haulers along with  garbage and
demolition debris. Defining the contaminants of concern at a debris dump entails a combination
of screening-level sampling and investigation of site records.

       Issues of worker health and safety need to be identified during scoping. The public's right
to participate in decisions, the regulator's need for data, and the  worker's (sampler's and  chemist's)
health must always be considered when developing waste characterization strategies. Indeed, health
and  safety,  both radiological and non-radiological, must be a primary concern throughout the
characterization process. While the in situ characterization may be conducted to determine  risk to
the public and the environment, worker exposure during field survey, sampling, analysis,  and waste
disposal usually engenders a larger, though shorter-term, risk.

       Historical and process information is gathered at this  point and used  in developing the
preliminary conceptual model of the waste to be sampled. Under federal regulations, knowledge
may be  substituted for  actual sampling and chemical analysis of the waste for the purposes of waste
characterization.  If a great deal of information is potentially available, it is appropriate for the
project manager to do at least a minimal cost/benefit  analysis  concerning the good to be gained
versus the time and resources required to  assemble  historical information and data instead of
sampling and analyzing the waste.  In  general,  the  more heterogeneous  the waste,  the more
worthwhile it is to seek out historical and process information.  A subjective  measure of certainty
can be attached to these data, and a comparison with DQOs can be carried out to see whether the
initial characterization objectives have been met. However, if a high degree of certainty is required,
at least  several confirmatory samples characterized by actual chemical analyses will usually be
required.   It should  be noted that,  without measurements,  only  process knowledge allows
determination that a  waste is a listed hazardous waste under RCRA.

       Land titles and survey data are basic to establishing the boundaries of the waste deposit.
Sources of land use records, soil types, and aerial photo surveys include:

•      County, state, or municipal tax estate and registry offices
•      State highway departments and the Federal Highway Administration


        County Extension Agents
        U.S. Soil Conservation  Service
        U.S. Geological Survey
        Tennessee Valley Authority
        U.S. Army Corps of Engineers
        U.S. Bureau of Mines

        Aerial photographs are particularly useful when they can be compared over time for land
 disturbances, fencing, and vehicular traffic.  Inferences  may then be drawn concerning disposal
 areas, types of wastes and containers, or sources of waste from truck markings and direction of

        For wastes generated by specific, known facilities, plant  records and interviews are another,
 more specific source of information.  These include:

        purchase records, inventories, and shipping and receiving records
        laboratory analyses and production reports
        industrial hygiene and health physics  files
        interviews  with plant managers, engineers, process operators, truck drivers, and maintenance

        An algorithm can be postulated which states that the percent assurance of historical waste
 data is inversely proportional  to its age.  All of the major  laws governing waste disposal were
 promulgated in or after 1975; hence, information generated after 1980 has a much higher validity
 than earlier information. It was around 1980  before the need  for waste records with any degree of
 detail was recognized.

        Historical  and site data are used to develop a conceptual model of the site and waste (2).
 Critical questions  to be answered in formulating the model are:

        What is the source of the waste?
        How was it emplaced?
        What matrices are affected?
        What is the potential for contamination and with what?
        What was the source and purity of raw materials?
        Who is/was the generator?
        What was the process?
        Why is it a waste?

The model, linking contaminant sources  to receptors via pathways, should be as detailed as current
information justifies.  It should include the contaminated waste  as  well as  the surrounding
environmental media. Insofar as possible it should define the actual flux of contaminants from the
waste: are they actively leaching into the ground, or is  there only  potential  for  leaching?  If
necessary, the model should distinguish between contaminants emanating from  solid wastes and
those originating in adjacent or underlying tanks or pits.  Dump operators frequently disposed  of
liquid wastes by allowing them to seep  out of unlined pits, and later filled the  pits with debris.

There would be little point in devoting the principal study resources to characterizing the debris at
such a site; most effort should be focused on the soil and groundwater.

       As they develop the  site  conceptual model, the study planners need to begin considering
potential waste management strategies.  These will strongly influence the nature and scope of the
characterization study.   For example, an NPL site that is a municipal landfill with co-disposed
hazardous wastes may pose a low-level long-term threat, or waste treatment may be impracticable,
according to the expectations of  the National Contingency Plan (40 CFR 300.430 (a)(iii)). In this
case, engineering controls will be the preferred remedial option, and waste characterization will be
limited to determining gross landfill properties such as age, physical dimensions, and differential
settlement rates. However, if "hot spots" of concentrated contamination are identified within the
landfill, treatment of these areas may be required (6). This will call for more thorough waste
                  Establishing Data Needs  and Data Quality  Objectives
Overview of the DQO Process

       When environmental data are  collected for making  regulatory  decisions  concerning
hazardous waste sites, the decision makers must understand the level of assurance associated with
these data. To determine the level of assurance necessary to support the decision, an iterative
process should be used by decision makers and project planners.  This section describes the process
of establishing data quality  objectives and illustrates the uses of DQOs in the collection of waste
site data.  The DQO process  is by no  means the  only possible approach to planning a  waste
characterization study; however,  the rigorous, quantitative nature of this  process  renders it a
particularly powerful tool when applied  against the  variability inherent in heterogeneous wastes.
In addition, the process compels active participation by and communication among decision makers,
managers, field and laboratory personnel.  This team effort is  absolutely essential when  dealing with
these  complex materials.

       Data Quality Objectives (DQOs,  also called  Data Performance Criteria) are the full set of
constraints needed to design  a study, including a specification of the level of uncertainty that  a data
user is willing to accept in the decision. DQOs are developed using a process that encourages the
sequential consideration of  relevant issues.  Figure 3-2 shows  the principal stages  in the DQO
process.  Each of the stages results in  an important  criterion  (or  'product') for the study that

       the problem to be resolved at the site
       the decision needed  to resolve the problem
       the inputs to the decision
       the boundaries of the study
       the decision rule
       the uncertainty constraints

    Figure 3-2. Steps in the data quality objectives process.


These constraints or products are the DQOs that will be used to formulate a study design that
achieves the desired control  on uncertainty, allowing the decision to be made with acceptable
confidence. There are several benefits to  establishing DQOs

•      The data generated are of known quality.
•      DQOs help data users plan for  uncertainty.   All projects have some inherent degree  of
        uncertainty.  By establishing DQOs,  data users evaluate the  consequences of uncertainty and
        specify constraints on the amount of uncertainty they can tolerate in the expected study
        results. The likelihood of an incorrect decision is estimated a priori.
•      The DQO process facilitates communication among data users, data collectors, managers,
        and other technical staff before time and dollars are spent collecting data.
•      The DQO process provides a logical structure for study planning that is iterative and that
        encourages the data users to narrow many vague  objectives to one or a few critical
•      The structure of the process provides  a convenient way to document activities and decisions
        that can prove useful in litigation or  administrative procedures.
•      The process establishes quantitative criteria for knowing when to stop sampling.

        In establishing DQOs, it is important to follow the sequence of stages because the product
of one stage is often an input to later stages.  However, this process should be regarded as both
flexible and iterative;  as the study team sees the implications of different products, it should go back
as necessary and revise products of earlier stages to incorporate the new concerns.

        EPA's "DQO Training Software" provides an  interactive opportunity to practice the DQO
process. The software is available from the Quality Assurance Management Staff within EPA's
Office of Research and Development.
Heterogeneous Waste Example

       The following is an example of the DQO process applied to a heterogeneous waste problem.
This example was developed at the Heterogeneous Waste Characterization Workshop. To facilitate
the reader's understanding of the DQO process, this example has been integrated here into EPA's
draft DQO process guidance (8). What follows is an expanded description, based on the guidance,
of each of the seven stages in the DQO process with the relevant output from the workshop
example provided for each step.

       Note  that seven principal activities are described below and summarized in Figure 3-2 as the
components  of the DQO process, while Figure 3-1  shows other activities as part of the process.
Some of these additional activities are concurrent or ancillary steps, such as the establishment of
site-specific action levels for the contaminated waste.  Other activities, like evaluating alternative
sampling designs, logically follow  establishing the decision rule.  In reality, a rough  outline of the
sampling design is often developed in conjunction with establishing the quantitative uncertainty
statements. Establishing the uncertainty rules and statistical design is an iterative process,  rather
than a sequence of discrete tasks. While the draft EPA guidance treats the evaluation of statistical
study designs as part of the DQO  process, this chapter sets that activity forth separately so  that it
can be discussed in greater detail.

        The substantial information requirements prevent this example from addressing all the
relevant activities at each step of the process,  since the example represents a first attempt  at
developing DQOs for this problem. If more time had been available, the workshop  participants
would have been able to complete these other activities, refine and reconsider their output for each
step of the process, and include other relevant information such as site-specific  conditions, disposal
costs, disposal capacity, and potential adverse health effects associated  with waste mismanagement.
Defining DQOs for a real site would involve input and interaction from management and outside
interested parties such as the State, local government, and neighbors of the site.

       A total of more than 1.4 million drums containing radioactive and hazardous chemical wastes
are present at major DOE sites across the U.S.   The wastes are not well characterized and are
heterogeneous in terms of characteristics and distribution. At many sites, drums are stored in buildings,
stacked on the ground and/or covered with dirt.   At some sites, the drums have been placed in
approximate chronological order of waste generation and therefore, anecdotal site history and records
may provide insights as to their contents.

       Characterization of the drums is of primary importance since this information is needed to
determine processing/treatment requirements (remedial action).  In some cases, drums exhibit the
generator's label. DOE assumes all drum labels are accurate and will use them to the extent possible
to begin classifying drums  into categories (this assumption is not necessarily appropriate for sites not
regulated by DOE). In many cases drums are unlabeled and historical information is insufficient to
determine their contents. For these drums, new data will be needed to determine how to categorize the
drums. The type and variety of wastes we expect to find in drums depend on when the wastes were
generated at the sites,  waste parentage, and when the wastes were put in the drums. Opening each and
every drum for characterization is impractical and may well result in unacceptable worker exposure.
Therefore, non-intrusive investigative techniques are  preferred. The integrity of the drums is not always
clear and is  of concern, since some drums have been at sites for many years and some of the contents
may be reactive.

       The failure of past operations has diminished the public's confidence in management's ability
to safely control these wastes.   Given the potential magnitude of remediation costs, the resources
available for addressing the drum problem must  be used efficiently.   Interim measures such as
redrumming  may be considered to prevent a worsening situation due to drum deterioration.
Stage 1: State the Problem To Be Resolved

Product:      A description  of the  problem; initial thoughts on ways to resolve it; and any
              resource, time, or other practical constraints on the data collection.

Background:  The purpose of this step is to  describe  what is known or expected about the
              problem, to document the  general  approach that will  be used to address the
              problem, and to specify practical constraints on the approach to solving the problem.

 Activities:     State the problem as you currently understand it and sketch out initial ideas on
               possible ways to resolve this problem.  State any ideas or expectations about the
               problem that are apparent at this point in the process.

                      Summarize relevant information, including  preliminary  studies, and indicate
               the source and reliability of the information.

                   Develop  a  list of alternative  courses  of action (including  a  no-action
               alternative, if appropriate) that  may be  needed to address the problem. Identify
               existing policies of the regulatory agencies that may influence the alternative actions
               considered (e.g., agency emphasis on treatment rather than source containment).

                      Make an initial determination of whether new environmental data will be
               needed to decide among these alternatives. Determine the importance of social and
               political considerations to the problem.

                      If they have  not already been identified,  name the  members  of the study
               team including senior program staff, technical experts, and any senior managers or
               other representatives of the decision maker whose planning input will be needed
               during the process to ensure implementation of the study findings. A statistician
               should either be included in this team or be available to help with portions of this

                      As specifically as possible, describe the constraints  on the study that were
               identified during project scoping.  Specify all resource or time limitations for the
               study, including the  anticipated budget and the available manhours.  Identify any
               obvious practical constraints (such as the time  of year when data collection is not
               possible).  Identify health and safety considerations  that may  influence  the sampling
               design or sampling methods.

        The planning team, which was made up of representatives from several DOE sites, recognized
 that the following question has to be addressed prior to determining the appropriate remedial action:
 "How can the drums at each of these sites be categorized in order to facilitate determining how they
 should be handled? "

        Using the DQO framework, the planning team defined the following drum categories or drum
populations based on drum source, contents, and the remedial action appropriate for that category.

        1.     Non-radioactive,  non-hazardous
        2.      Transuranic (TRU)
        3.     Low-Level Radioactive Waste (LLRW)
        4.     Naturally Occurring Radioactive Waste (NORW)
        5.     High-Level Radioactive Waste (HLRW)
        6.     Hazardous non-radioactive
        1.     Hazardous TRU


       8.       Hazardous LLRW and
       9.       Hazardous NORW

 Types 7, 8, and 9 are mixed wastes, by definition.  Figure 3-3 illustrates the relationships among these
nine categories. Based on the above classification, the DQO team discussed the following project design
and drum classification strategy. Radioactive drums will be separated from non-radioactive drums based
either on labels or on a non-intrusive investigative technique. All radioactive drums that contain HLRW
wastes will be identified based on parentage, separated from all other radioactive drums, and classified
as HLRW. The remaining radioactive,  non-HLRW drums will be measured to determine if they contain
 TRU wastes. The first measurement will determine indirectly if the radionuclide specific activity exceeds
a specified level. A second test will be conducted on those drums that exceed the specific activity level
to determine if the drum contains alpha emitting radionuclides. If the drum satisfies the specific activity
and particle emission criteria,  the half-life will be measured to determine if it is greater than 20 years.
If the drum does not meet the specific activity, particle emission and half-life criteria, it will be placed
in a pile (either LLRW or NORW; an additional test may be needed if these two waste  types will be
disposed of differently) separate from the non-radioactive and HLRW piles. Another study will be
designed to determine if any drum designated as containing radio nuclide wastes also contains hazardous

       The process of identifying drum source and contents determines the category it belongs in and
implies the remedial action that is appropriate.   Since the consequences of mis characterizing drums
depend on the  contents, separate DQOs will be needed for each class of drums.  The rest of this example
will only address TRU drums. A similar approach would be used to address other classes of waste

       Since this is a general example of a DQO, site-specific conditions are not addressed here. In
addition, the consequences ofmischaracterizing drums are not specifically addressed because issues such
as the costs of disposal, disposal capacity, and health effects are site-specific.
Stage 2:  Identify the Decision/Question

Product:      A statement of the decision that will be made,  or the question that will be answered,
              using waste data. If possible, the decision should be expressed as a determination
              of whether to take one or more alternative courses of action.

Background: The study will usually produce data that are intended for use in making a decision
              (e.g., if the waste poses an unacceptable risk, then take remedial action). In a few
              cases, specific actions and action criteria cannot be identified, possibly because too
              little is understood about the problem or because this type of problem has not been
              identified as a priority  by the regulatory agency.  If specific  actions cannot be
              identified, then these studies fall  in the category of research. If they are able to
              conduct a site-specific research project, the study team should consult EPA's DQO
              guidance   for  environmental  research  (8)  to   establish DQOs  for  the
              study at this point.

6 "J
$ *
y ••
i *
1 	 J

Figure 3-3. Classes of drummed wastes found at DOE sites.

                      The decision maker (data user)  should be involved at this  stage  and is
               encouraged to provide general criteria for taking action.

                      The decision should be stated as narrowly and specifically as possible. If the
               problem is very complex, then the activities in this step will help the study team
               identify and organize elements of the larger problem.

Activities:      If several separate decisions must be made to address the problem, then begin by
               mapping out a decision or logic tree.  This exercise should reveal the relationship
               among decision(s). Try to determine the relative importance of each decision to the
               overall problem. Determine which decisions depend on other decisions. This  allows
               for  prioritizing  of the  decisions.  Determine which decisions  require  new
               environmental data and  the  importance of those data to the decision.  Use the
               DQO process for each of the decisions that require new data, starting with the most
               important decision.

                      Take the following steps to state the decision or question so that the  role of
               data in taking an  action  is clear.  Broad  statements of goals or objectives are not

               •       State the range of actions that might be based on the outcome  of the study.
               •       Specify the criteria for taking these actions, including specific "if..., then..."
                      scenarios when possible.   If the criteria  are not known at this  time,  then
                      specify how they will be established.
               •       Frame the decision as a hypothesis to  be tested. The hypothesis statement
                      should be qualitative at this point (e.g., null hypothesis: acceptable level of
                      contamination;  alternative hypothesis: unacceptable level of contamination);
                      a quantitative statistical formulation of the hypothesis will be developed  later
                      in the  planning process.


       Determine how to classify drums in a manner that facilitates disposition decision making. For
this specific example, the following is the statement of the decision with hypotheses for the decision:

       Determine whether to  classify a drum at a given site as containing TRU wastes.

       Hp Drum  does not contain TRU wastes
       Hf Drum  contains TRU wastes

      Alternative actions  include:

       •      dispose of drum containing TRU wastes according to special procedures,  or
       •      classify drum  as not containing TRU wastes,  then determine what other form of
              radionuclide waste  it contains, or
       •      if drum categorization is too hazardous, costly, or time-intensive,  treat all drums that are
              not HLRW as containing TRU wastes and omit the characterization study.

Stage 3: State the Inputs

The list of variables  or characteristics to be measured and other information needed
in order to make the decision.
Background: At this stage the study team should identify all variables or characteristics that may
              be relevant to the decision, and then focus on those that must be measured in order
              to have the information needed to make the decision.

                     The initial statement of project objectives may include only those of a short-
              term nature. However, it is more cost- and time-effective to address both immediate
              or short-term  information needs  and future or long-term  requirements concurrently.
              The short-term need may be only  to determine whether a waste is a source of
              radiation; however, upon reflection, long-term information needs, such as level of
              radiation,  source, whether the waste is also  RCRA hazardous,  etc.,  are  also
              identified. These concerns expand to storage and shipping regulations.

Activities:     Develop a list of variables/characteristics that may affect the decision and separate
              out those variables that need to  be measured in order to  make the decision (which
              action to take) or answer the question.

                     Select those measurements that together will provide sufficient information
              to make the decision.

                     Identify information from  other studies, regulations,  etc., that is needed to
              establish the criteria for taking action.

                     Confirm that each variable can be measured. If not, then determine if it is
              reasonable to  make assumptions about the  variable in order to draw conclusions
              without data.    If the  necessary  assumptions  cannot be made,  then select an
              alternative approach that involves  different variables.  If no practical approach can
              be developed, then consider shifting the effort to develop the research tools needed
              to address the problem.


       After consideration of factors affecting the decision,  the following informational inputs were
identified as necessary for making the decision on whether a given drum contains TRU wastes:

       (1)     Is the drum intact?
       (2)     Does the drum exhibit a radioactive label?
       (3)     Does the drum contain radioactive materials?
       (4)     Is the drum from a HLR W source ?
       (5)     Were TRU materials used at this site?
       (6)     What is the specific activity of the material in the drum?
       (7)     Is the radionuclide an  alpha emitter?
       (8)     What is the half-life of the material in  the drum?

Measurements will be needed for inputs 6,  7, 8, and possibly 3. Documentation as to parentage will be
used for inputs 3 and 4.  Visual observations will be used to address inputs 1 and 2.
Stage 4: Narrow the Boundaries of the Study

Product:      A description of the population(s) of interest,  including a description of the units
              that make up the population(s) of interest and the spatial and temporal boundaries.

Background: Spatial and temporal  boundaries are defined to incorporate all the units that make
              up the population(s) of interest.  In addition, criteria are established for identifying
              these units and distinguishing them from other similar units (e.g., to locate drums
              containing radioactive waste, the criteria will specify how to distinguish them from
              other drums).

                     The population(s) of interest can be defined as the population(s) of items for
              which the decision will be made or the question answered. On a heterogeneous
              waste  site,  a unit  of this population  may  be a container and its contents, an
              individual waste item, a volume of uncontainerized waste, or any of the items of a
              particular type or having certain contamination characteristics.

Activities:     Specify the population(s) of interest:

              •     the units of this population, including the criteria for identifying these units
              •     the temporal boundaries
              •     the spatial boundaries

                     Specify the smallest sub-population for which a separate summary statistic
              will  be calculated. If a separate  decision will be made for this group of units, then
              consider it as a population of interest.

                     Consider whether  the  population  that  is  accessible for measurement
              adequately represents the population(s) of interest. State any assumptions  that may
              be necessary to do so.

                     At this  point, an important step that is often neglected is the pilot study. For
              conventional environmental media such as soil and water, statistical methods for
              representing  contamination are  well  established.  The variability associated with
              sampling these media can often be estimated a priori.  This is not usually the case
              with heterogeneous wastes.   The purpose of  the pilot  survey is to estimate the
              inherent variability of contamination of the wastes and the measurement variability
              introduced by sampling and analysis.  This information is critical in developing the
              uncertainty levels that  the decision maker is willing to accept. If these values are not
              known, an unachievable level of uncertainty may be chosen in the DQO process.


       Measurements will be made on all intact drums at a site that has been determined to be
 radioactive, non-HLRW but without sufficient historical information to determine if TRU wastes are
present. For the half-life measurements, the definition of the boundaries of the survey should include
 the temporal component of the measurement: the time that will elapse between measurements to
 determine the half-life of the drums of interest. An individual drum is a unit of the  TRU waste drum

       Measurement of radioactivity will be done by Segmented Gamma Scanner and Passive Active
 Neutron  instruments. A pilot study may be required to collect analytical measurement performance
 information in order to estimate measurement error. If the results of the pilot indicate  that the precision
 and accuracy of the proposed analytical methods are inadequate or the methods are impractical, then
 another method will be needed.
 Stage 5:  Develop a Decision Rule

 Product:       A quantitative statement that defines how the data will be summarized and used to
               make the decision or to answer the question, including  quantitative  criteria for
               determining what action to take.

 Background: The purpose of this stage is twofold:

               •      to integrate the  decision and  relevant  variables (inputs)  into  a single
                      statement specifying how environmental data will be summarized and used
                      to make the decision or answer the question within  the study boundaries;
               •      to indicate what  actions will  be taken and the criteria for taking those

                      It is  important that someone with statistical expertise be involved at this
               stage to be  certain that the problem is framed in a statistically valid and efficient

                      The decision rule will be based on tentative or established action levels for
               the  contaminated waste:  either ARARs  or  site-specific  risk-based criteria. This
               decision rule will be reviewed and may require revision  after the uncertainty
               constraints for the decision have been established.

Activities:     Describe the summary statistic(s) and how they will be calculated (e.g., mean, range,
               maximum).  Develop a decision rule as an  "if.., then..." statement that incorporates
               the summary statistic, the action  criteria (based on the hypothesis test), and  the
               corresponding action(s) that will be taken under various possible scenarios (e.g., if
               TCLP extracts of piled manufacturing wastes exceed any of the regulatory criteria,
              the wastes will be treated in place or shipped to a hazardous-waste landfill).

                      Confirm that all the inputs are incorporated explicitly or implicitly in the
               decision rule. If not, then reassess the need for them in relation to the decision and
               either define a more narrowly-focused set of input variables or revise the decision
               rule to include the full set of variables, as appropriate.
                      Reconfirm the need for new or additional environmental data to make the
       If measurements show that an unlabeled drum exhibits a specific activity > 100 nanocuries per
gram, and is an alpha emitter and has a half-life > 20 years, then the drum will be classified as a TRU
drum and disposed of accordingly.
Drum does not contain TRU waste; specific activity
20 years or not an alpha emitter.
Drum contains TRU wastes.
                                                         100 nanocuries per gram or half-life
(Note: all drums labeled as containing TRU waste will be disposed of as TRU waste).
Stage 6: Develop Uncertainty Constraints

Product:       The decision maker's expressed desire to control decision errors, stated as limits on
               the  acceptable probability of making an  incorrect decision based on the study
               findings. These may be expressed as acceptable false positive and false negative
               error rates.

Background:  Environmental data collection always involves some error.  Therefore, some degree
               of uncertainty will exist in any decision based on environmental data. In this step,
               uncertainty constraints are established and  stated as acceptable probabilities of
               making incorrect decisions, i.e., acceptable decision error rates.  The uncertainty
               constraints are used  to  establish  quantitative  limits  on total study  error  and
               corresponding measurement and sampling error constraints.  These  are  used to
               finalize  the decision rule.

                      Uncertainty constraints should  be based on careful consideration of the
               consequences of incorrect conclusions. The  decision  maker will need to consider the
               political, social, and economic  consequences of decision errors when  setting
               uncertainty constraints.

                      The decision maker needs to be actively involved in the development of
               uncertainty constraints. In addition, the study team should work with a statistician,
               as  necessary,  during this  step to  ensure that the constraints  are  feasible  and
                      There are two types of decision errors for all studies that use a hypothesis

               •      False positive errors:   deciding to take an action when the environmental
                      data  incorrectly indicate that a problem  exists. In this case, the null
                      hypothesis is rejected when it should be accepted.
               •      False negative errors:  deciding not to take an action when environmental
                      data incorrectly indicate that a problem does not exist. In this case, the null
                      hypothesis is accepted when  it should be rejected.

               Uncertainty constraints for  these types of decisions can be expressed as limits on the
               acceptable rates of false positive and false negative errors.

 Activities:     A number of activities are involved in setting acceptable probabilities for decision

               •      Define false positive [f(+)] and false negative [f(-)] errors for the decision
                      and describe the  consequences of each type  of error.   Note that the
                      consequences can change depending on the size of the error, which may not
                      be well defined.
               •      Evaluate these consequences according to the relative  level of concern or
                      discomfort that they would cause.  Environmental,  public health,  economic,
                      social, and political consequences should all be considered. Differing relative
                      importance may be placed on these  consequences depending on the concern
                      of the decision maker, the degree of public interest in the study,  etc,
               •      Determine if false  positive or false  negative  errors are of greater concern.
               •      Establish, with statistical advice,  an acceptable probability for the occurrence
                      of each of these errors. Also, specify a "region of indifference," the area in
                      which one chooses not to control the probability of an incorrect outcome
                      because, under the stated conditions, either false positive or false negative
                      errors are acceptable.  This  region may be  narrow or  broad and must be
                      acceptable to the data users.
               •      Combine the probability statements into a formal statement of the levels of
                      uncertainty  that can be tolerated in the results. This formal statement may
                      take the form of a table or a  graph.
               •      Review the decision rule;  if necessary, revise or add quantitative measures
                      that will allow the decision uncertainty to be  evaluated.


       A false positive decision error is deciding that a drum contains TRU wastes when it does not.
   Consequences of False Positives:

    •          Limited disposal capacity designated for TRU wastes will be unwisely and unnecessarily
               used up
    •          Loss  of credibility if the error is discovered by subsequent analysis
    •          Unnecessary costs

A false negative decision error is deciding the drum does  not contain TRU wastes when it does.
 Consequences of False Negatives:

    •         Potential worker, nearby population, and environmental exposure if a drum of waste
    •         Facility management may be liable and may have to pay penalties
    •         Loss of credibility due to failure to prevent exposure and potential adverse health and
              environmental  effects
    •         Accidents may occur at facility locations where containment is lost
    •         Improper disposal may result in long term exposure.

Figure 3-4 is the  "discomfort curve " that was drawn for the first measurement (the specific activity of the
drum contents).  The study planners established that either decision is acceptable if the true specific
activity is between 80 and 100 nanocuries/gram (i. e., the region of indifference). If the true specific
activity is between 60  and 80 nanocuries/gram, it is acceptable to incorrectly judge the drum to exceed
the criterion 20% of the time. A 5% false positive rate is acceptable for drums exhibiting a lower specific
activity. The planners  were more concerned about false negative errors.  They specified that for drums
exhibiting a specific activity of 100-120 nanocuries/gram, the false  negative error rate must not exceed
10%; above 120 nanocuries/gram, it must not exceed 5%.

        If the measurement indicates that the drum has a specific activity > 100 nanocuries/gram, then
continue with tests for alpha emitter and half-life. Separate  discomfort curves would be drawn for each
of these measurements.   These curves would not necessarily be identical to that for specific activity.
 Stage 7: Optimize the Design (this stage may be considered part of the DQO process or it may be
 part of the Sampling and Analysis Design; the goal and output  is the same in either case. The
 detailed discussion of study design is given in the next section of this chapter.)

 Product:       The lowest cost design for the study (selected from a group of alternative designs)
               that is expected to achieve the desired constraints on uncertainty and any practical

 Background: In this step, statistical techniques are used to explore and evaluate various designs
               for the  study that meet  the constraints  specified  in the products from the DQO
               process. These designs should enable the decision to be made subject to error rates
               no greater  than  those  established  by  the  uncertainty  constraints within  the
               boundaries specified for this study.

Activities:      Assemble the information needed to develop alternative designs: the uncertainty
               constraints from the proceeding  step; any budget, safety or physical constraints; cost
               estimates for all study activities;   estimates of the natural  variability  for
               characteristics/variables to be measured; and estimates of the variability that will  be
               introduced by the sampling and  analysis  process.   If such  information  is not
               available, then  design pilot tests to generate it, make necessary assumptions,  or
               research improved analytical methods.

                      Have a statistician or someone with statistical expertise generate alternative
               designs  and estimate the  cost and anticipated error rates of each design. Select the
              most  cost-efficient design that has acceptable performance and meets all other needs
              of the decision maker including political  and social concerns.


              Likelihood of Concluding that
             Drum is Not a TRU Waste (%)

1 1

8 I
c ro
t_ C
•e .. o
c "5 E
- cr 5
0 C "°
c o w
O CO ^
'O) 'O 2=
cr Q

T —




CO .
o 52
-, CO
1— J
alpha emitter '
containing TR



! 1
0 0
CM •*

1 1
0 0





c_j | .





(%) 9}SEM
Figure 3-4. Uncertainty limits for specific activity in drummed wastes.

                      If there is a reasonable possibility that natural or analytical variability will be
               greater than estimated or assumed, evaluate the expected performance of this design
               under a range of alternative conditions.   Confirm that the design will yield useful
               results, even when conditions are more adverse than those expected or assumed.

                      If it appears that there is no design that will meet both the uncertainty and
               cost  constraints, then  determine whether  to  relax  the uncertainty or other
               constraints, or find additional funding to achieve  the desired uncertainty constraints
               within the specified boundaries for the study.


       In developing survey designs for a specific site, the planning team will need to refine and enhance
the constraints in the previous  steps with site-specific information to form the data quality objectives for
that site.   Once the decision maker and external data users  understand and accept these revised
constraints, the statistician will construct alternative designs that satisfy them.
                               Sampling and Analysis Design

Selecting the Strategy  and Methods

Formulating a Detailed Site Model

       Before  a sampling design can  be developed, the location and types of waste must be
determined. This may be done as a yes/no matrix involving the following:

       Areas of interest
       Hot spots versus average values of contamination
       Surficial versus deep contamination of the waste items and the ground
       RCRA versus non-RCRA contaminants
       Radioactive versus non-radioactive
       High level of contamination
       Low level, or trace, contamination

       Then the following questions are addressed:

       How is the waste emplaced?
       Are there patterns?
       Can it be segregated by type?
       What is the matrix?
       Is it layered?
       What degradation products could have been produced?

       As a result of these inquiries, either  a pattern can be identified upon  which to base a
verifiable model of the waste, or such a model cannot be developed. When such a model can be
created, it forms the basis for a sampling plan in which some "representative" portion of the waste


 is characterized. When it cannot be created, the achievement of a high degree of certainty in the
 waste characterization will require opening and sampling every waste container. A very closely
 spaced sampling grid will be needed for a waste pile, and the project will be very costly.

       An essential early part of the study will be grouping waste items into populations. A waste
 population may be defined on the basis of the contaminants present, the waste matrix, the level of
 contamination, or a process variable from the facility that generated the waste.  The criterion by
 which the waste is divided into populations is that, in the final disposition of the waste, all of the
 members of a population will  be treated alike.

       In the example above, each waste drum was treated as a unit of a population. The project
 planners were sufficiently confident of homogeneity within each drum that there was no need to
 investigate or distinguish among the contained  items. This is not always the case. Waste that has
 come to be co-located within a drum or a grouping of drums may be from similar populations or
 may be representative of many populations. The drum(s) may have been filled quickly or over a
 long period of time. The  materials  may be heterogeneous by process of generation but be from the
 same location.

       The assumptions  used to determine groupings (populations) must be well documented and
 tested against the needs  of the project.  As an example,  several bags of vials may be found within
 a drum and initially be grouped as a population.  However, the materials within the vials may be
 completely dissimilar chemically (by process of generation or by level of concentration), or they may
 all be similar by exhibiting gamma radiation but contain different nuclides. If these nuclides must
 be treated differently, the different bags actually represent different populations,

       The sheer magnitude of the project and/or economic concerns may suggest inappropriate
 groupings; i.e., that all  paper wastes, regardless of contamination,  be grouped together.  This
 approach should be taken only after serious  consideration of the sampling objectives,  addressing the

 •      What value is added by such a grouping?
 •      Can the items be mixed to produce a true population (that is,  can all the components be
       treated/disposed in the same way)?
 •      Is such a technique allowed by regulations (i.e., is this treatment by dilution)?
 •      Will a smaller number of samples be needed?
Aspects of a Preferred Sampling Design

       A sampling procedure must be practical and achievable. It is easy to set conditions on the
sampling  procedure which appear very  desirable and reasonable, yet  lead to  less efficient
contaminant concentration estimation and decision determination. Statistically, a good estimate
should have small bias and a minimum standard deviation of error.  Some statistical procedures
produce unbiased,  minimum variance  estimates  as  their goal (e.g.,  least-square regression).
However,  if the estimate is not required to be completely unbiased, it is often possible to obtain
estimates with a smaller root-mean-square (r.ms.)  error which have small but acceptable statistical
bias. With the best of intentions, a regulator might specify that the statistical sampling procedure
lead to unbiased estimates (certainly a commendable property) and inadvertently force large r.m.s.


 errors (relative to the given sample size) upon the estimates.   These correspond to a wide
 confidence interval around the estimated value and poor predictability in the measured parameter.

       Another desirable property of a statistical sampling plan is objectivity, as contrasted with
 subjectivity. An objective sampling plan obtains its estimates independently of the opinions and
judgments of the project personnel.  A completely objective procedure would require that prior
 knowledge and judgment not be used in the design of the sampling process but, instead, various
 random  number techniques  be  used to select locations for sampling  and other aspects of the
 process.   If the cost per sample  for collection and analysis is quite small (so that it is feasible to
 have large sample size), the very objective random sampling  procedures  are quite appropriate.
 Indeed, they are the popular and standard approach  in market surveys and many agricultural
 experimental designs.

       As the cost per sample for collection and analysis increases, it becomes desirable to allow
 some subjectivity in order to obtain the required estimate precision with a smaller sample size.  This
 may be achieved in classical sampling theory by using stratified random sampling and, perhaps,
 introducing other inferred structural properties of the population. Alternatively, one may shift to
 methodologies such  as Bayesian methods (which may require subjective estimates of prior
 probability functions)  or the random function approach of geostatistics (which builds on an inferred
 spatial correlation structure  within  the site).   Both methods use prudent, subjective judgment
 concerning the  statistical structure of the population and the uncertainties  relative to various aspects
 of the population to obtain greater precision from a given sample size.

       In the case of a very substantial cost per sample (often true for heterogeneous wastes), the
 need to  get  as  much information as  possible from a very limited number of samples  makes
 techniques such as sequential methods very appealing,  since they allow the information from data
 collection to be used as  soon  as available to optimize the selection of future samples in the process.
 They also permit  sample collection and analysis to terminate as soon as the required estimate
 precision or decision branch point is reached. This, in turn, minimizes the number of measurements
 required to attain the DQOs.

       All of the above sampling schemes  have  areas of appropriate application. It is not a
 question of which is  right or wrong, but rather what approach is best in a given situation. With
 heterogeneous wastes, the need to process many drums (or large unconfined waste piles) and the
 excessive cost  of direct analysis are dominant aspects of the problem.  An acceptable procedure
 has to be achievable within the limitations of money and time available, as well as being capable
 of implementation in such a way as to be safe  to the workers involved and the public at large. An
 unachievable sampling/decision procedure might result in greater danger to the public than a more
modest achievable procedure. This is one of the main reasons for involving technical, managerial,
 and administrative personnel in project scoping.

       For characterizing heterogeneous wastes, maximum use of prior knowledge and non-invasive
measurements should be made.  If the decision can be made based on this information, further
sampling may be unnecessary. If more information is needed, sequential sampling, rather than a
fixed-sample-number approach, appears desirable since it allows knowledge gained at each stage to
be available in the  sampling at future stages.  This, in turn, permits the decision process to proceed
with the fewest samples and greatest dispatch  and to terminate as soon as a suitable basis for the
decision is obtained. The sequential approach is not popular with contracting departments because


 it appears much  simpler  to wrap up  a site study in one  stage. However, this  approach will
 frequently turn out to be optimum with respect to cost, worker safety, and time requirements.

        Statistical techniques potentially  applicable to heterogeneous wastes are listed and evaluated
 in Appendix B.  The statistical methodology needs to be selected carefully to obtain techniques
 which are most efficient in treating the special aspects of the site and the waste materials and the
 range of decisions/actions under consideration.  No single statistical methodology can serve as a
 "best procedure"  for all circumstances.  Instead, the project planners need to consider a variety of
 different methods, each of which might be optimal under particular circumstances.  There are also
 many circumstances which require some degree of further statistical research on existing methods
 to modify or adapt them to the special circumstances of particular waste sampling situations.  Thus,
 it appears that professional statisticians  should play an active role in  many aspects of project

        In  considering  cost-efficiency, the project planners must bear  in mind that waste
 characterization is only one of the  cost elements  in  dealing with the site. Implementing  the
 minimum acceptable sampling  plan may not minimize the overall site remediation cost. It may be
 appropriate instead to perform a more  rigorous waste characterization study so as to lessen the
 incidence of false positive results. This will result in classifying a smaller portion of the waste as
 exceeding the action level and requiring treatment.   Depending on the expense of treatment,  a
 substantial overall saving may be realized by expending more than the minimum required to
 characterize the waste.
Non-traditional Aspects of Statistical Sampling in Waste Characterization

       Traditional sampling theory is concerned with collecting data that are to be used to estimate
the average value of some variable of the population and to establish confidence intervals for that
estimate.  The characterization of heterogeneous wastes may involve additional considerations.
These include  the fact that the real purpose of the characterization  is to facilitate the decision
process in guaranteeing that relevant regulations pertaining to public safety are satisfied. Thus, the
estimation of mean values of the variables is only an intermediate step to ascertaining that the
proper actions are taken with a high probability.

       The presence of a single "hot spot" of contamination within the waste may be much more
relevant to taking the proper action  than the average value of the contamination, no matter how
accurately it is estimated. The location and measurement of a local maximum value within waste
material is not a conventional problem within statistical sampling theory. A reasonable body of
theory exists in studies of "search techniques" to locate objects of known size  and shape. However,
a combination of search theory with sampling theory may require further research and development
before it can be applied to waste characterization in all cases.
Selecting Sampling and Measurement Procedures

       Any measurement that can be made from outside the storage drum or unconfined waste pile
without introducing a physical presence within the waste can be considered to be "non-invasive" or
"non-intrusive."  There are numerous possibilities for such  measurements  (see  Chapter 5 for


 detailed discussion).  These include multispectral measures of reflectance (infrared, visual, and
 ultraviolet bands) from satellite and aircraft flights above the site; similar observations from ground
 level near the waste storage;  measurement of other physical properties such as temperature and
 radiation emitted from  the waste naturally or under excitation from  external sources;  X-ray
 photographs; and such activities as a careful visual observation of the drums for markings or stamps
 on the  drum, rust and other signs of age, and evidence of corrosion and leaking. Non-invasive
 techniques pose the least hazard to project workers, and they generally yield the least quantitative
 information about the waste. They may or may not be the least costly characterization methods.

        Semi-invasive (or "minimally-invasive")  techniques involve breaching the waste container or
 outer surface of the waste (as in a pile) but do not involve removing the waste for further testing.
 Therefore, these techniques usually develop information of a  semi-quantitative nature that are
 indicative of certain characteristics of the waste.  Such measurements are taken through  small
 punctures in the waste container or waste. For instance, instruments can be lowered or driven into
 the waste through the bung hole in the drum.

        Invasive measurements involve actually entering the waste, observing it, sorting, subsampling,
 and subjecting the  material to direct measurements.  Samples of the waste are taken and may be
 fully characterized by destructive means at field and permanent laboratories. During invasive
 testing,  samples can be taken  in  cross section, and materials may be sorted and segregated.
 Normally,  invasive sampling is the  only source of fully quantitative and representative data.
 However, this technique results in an increased potential for  accidental release of materials in
 transit and in laboratories, and may involve sample preparation processes that generate even more
 waste materials. It also entails the highest exposure to sampling personnel.

        Exposure concerns may, in fact, dictate the selection of sampling procedures, sample size,
 and project  phasing. If the risk is too great and cannot be mitigated by state-of-the-art technology,
 the project may have to be deferred until less hazardous methods have been developed, or a worst-
 case assumption requiring  a minimal characterization study may have to be adopted. The risk of
 worker exposure and potential for injury must be  weighted over  and  above cost and percent
 assurance of DQOs.

       In considering all of the available waste characterization methods, it is recommended that
 maximum use of prior waste  and site information be made.  If this information, combined with
various non-invasive measurements, is insufficient to answer the study questions, the investigators
 should next  examine semi-invasive techniques. If these will not suffice, fully invasive sampling will
be necessary. This should involve careful selection of the most cost-effective  statistical procedures,
preferably stepwise and sequential methods to minimize the required number of samples. The goal
is to maintain safety while optimizing the accuracy of the decision process  within the constraint
imposed by  the project budget.
Characterizing Landfills and Waste Piles

       Much of the discussion herein has focused on waste stored in sealed drums. However, there
are attributes  of landfills and unconfined piles of stored waste that require special note. The
general structure of the flowchart (Figure 3-1) and the comments in the list of statistical procedures
(Appendix B) all apply here.   However,  the examination of the waste  surface by non-invasive


 methods takes on a greater importance since the material is exposed, more or less, without the
 barrier of the metal walls of a drum. Thus, methods such as magnetic surveys, radar imagery, aerial
 photographs, satellite sensor imagery, electrical conductivity,  etc., can all  be quite appropriate,
 depending on  the  site  (see Chapter  5). Historical process information should be collected as
 completely as possible, and a site visit is vital.

       Regardless  of the nature of this information, it is imperative that the unconfined waste be
 considered in all three  dimensions, and not just over the exposed two-dimensional surface. The
 distribution of object sizes, types, materials, and hazard characteristics must somehow be inferred
 in three dimensions.

       Some  of the characterization decisions to  be faced are:   (1) What types of hazardous
 materials does the pile contain? (2)  What methods  of treatment  of the material are possible?
 (3) Can the material be so treated and/or sealed that the site can be converted to some beneficial
 use?  (4) If material can somehow leave the site, what are the possible pathways of migration
 (groundwater,  atmospheric dust and debris, surface runoff)? (5) Are "hot spots" of contamination
 present within  the unconfined waste?

       The techniques  of geostatistics, such as kriging and conditional simulation, and methods
 from biological spatial  sampling, such as transect sampling, appear to be particularly appropriate
 for landfills and waste piles. However, many other methods, such as stratified random sampling and
 systematic sampling, may also be appropriate if the cost for collection and analysis of each datum
 is reasonably small.

       One of the problems with a severely  heterogeneous waste pile is that the true diversity of
 the materials only emerges as the samples and information are assembled.  Thus, a stratification
 into semi-homogeneous categories is often not possible initially. Rather, it is necessary to keep a
 running waste  classification scheme going as information is obtained. Categories are added into
 consideration  as examples of those materials are  found.  Double sampling (see discussion in
 Appendix B) may be appropriate for these wastes. A large sample taken throughout the whole pile
 is collected in conjunction with a relatively inexpensive analysis scheme. This is used to establish
 significant categories of material and concerns.   Then a smaller sample  of carefully selected
 materials is collected and submitted to much  more detailed, and expensive, analyses.
                                 Finalizing the Project Plan

       When one or several alternative study designs have been developed, they are screened
according to the limitations of the project: its budget, schedule, health and safety requirements, and
DQOs. If there are no designs  that will be able to achieve the DQOs while satisfying the other
requirements, the project planners must return to an earlier stage, revising the study design or
loosening the DQOs (this must,  of course, be done in consultation with the decision-maker). When
more than one design meets the requirements, one design is chosen for  implementation.  This
choice is usually based on cost considerations,  but other factors may be of greater importance. For
example, there may be pressure to complete the study as quickly as possible, or the project team
may wish to  implement a statistical design they are familiar  with, rather than one that would be
experimental in their application.

        The final step is the preparation of the project planning documents. These generally include
 a sampling and analysis plan, a set of operating procedures, a QA project plan, a health and safety
 plan, and the overall project workplan. There may not be standard procedures for the non-invasive
 imaging techniques or the sample collection activities of the heterogeneous waste study. Innovative
 QA methods may be called for, and the  need for frequent on-site corrective actions can be
 anticipated. Consequently  the common practice of borrowing  boilerplate planning documents from
 previous studies is unlikely to be technically sound. Those preparing the planning documents and
 those who will be responsible for reviewing them should budget ample time for this stage.

        Recognizing the imperative need for careful concern for the health and safety of the public
 and the  workers  involved  in the waste characterization activities,  it  is  important  that
 characterization, remediation, and monitoring procedures be selected which are achievable with the
 available resources  of manpower and  budget.   It is also important to  evaluate the long-term
 consequences of "no action" along with the various possible remedial actions, since "no action" may
 have worse public safety effects than other feasible, but perhaps not ideal, solutions.

        A general course of action for heterogeneous waste characterization is recommended. This
 consists of three major stages: (1) The first stage entails  the project scoping, definition of data
 quality objectives and project  decision  choices, the collection of historical and non-invasive
 measurements, and a review to see if sufficient information is available for a decision at this point,
 (2)  If  not, the  project  team  proceeds to  examine  the relevant  and appropriate  statistical
 methodologies,  and to select the one or  several  choices  which are  both cost-effective and
 incorporate sufficient safety. (3) The last stage of the waste characterization study involves actual
 sampling (first, with cautious semi-invasive measurements, and then,  if that is not sufficient for a
 decision, with fully invasive sampling). In the actual sampling, a pilot  sample is recommended first
 as a guide for planning the future cycles of more complete  data measurements.  These cycles of
 sample collection and analysis are alternated with  review and  interpretation  until enough
 information is obtained for making the  appropriate action choice,

       There are two  actions that regulatory  agencies  could take to improve the quality of
environmental studies by influencing the process of study planning and DQO establishment. One
would target project managers, the other decision makers.

       Technical project managers are sorely in need of guidance for setting appropriate and
workable confidence intervals for environmental data. All of the sources of variability included in
population and measurement error need elucidation. DQOs are currently set according to rules of
thumb that vary across the country and among organizations.   To ground their studies on a firm
technical base, project managers need more than these arbitrary rules of thumb, and the regulated
community deserves more.  A detailed compilation of the courses taken in past studies could help
project planners select reasonable DQOs early in their own  projects, obviating the need for repeated
sampling and revision of goals.  Such a  compilation should be broken down according to the

medium to be characterized:   air, water, soil,  heterogeneous waste, and homogeneous waste
(industrial sludge, for example). For each medium, a number of past studies should be detailed.
The text should concentrate on the data quality objectives: what they were; on what basis they were
established; whether they were achieved; if not, why not; and the quality of data that was generated.
A simple synopsis of this type, detailing what has worked for different site conditions  and what has
not, would be a great help to study planners.  Unfortunately, it is likely that  little information is
available for studies characterizing heterogeneous wastes based on numerical  DQOs.

       A similar synopsis would be useful for decision makers. In this case, the description of each
past study would focus on the  uncertainty of the decision(s) that were to be made. The study
description would detail the uncertainty constraints associated with the  decision to  be made:  the
acceptable  frequencies of false positive and false negative decision errors  that were designated  and
the region  of indifference for the decision (see Figure 3-4). This document would be helpful in two

•     it would provide decision makers with information on what levels of certainty are achievable
       for various types of decisions concerning a range of media and contaminants, and
•     it would provide a national information base that could be used by  EPA  officials to establish
       uniform guidelines for making cleanup decisions.

       It must be emphasized that these documents would not alleviate the need for incisive, site-
specific planning  as each study is initiated.

       Project managers would benefit greatly from the development of analytical  and software
tools to aid in the planning process for characterization studies.   Optimizing the study design  and
scoping a study that is within budget yet achieves the DQOs is an iterative and frequently a very
tedious process. This process  could be expedited with the help of computer programs to allow quick
alterations of project design and comparisons among designs.  Such programs could perhaps be
based on existing software for accounting or cost-benefit analyses. Alternatively, new software could
be developed based on a gaming approach or a systems approach. A general, master program or
expert system could be adapted to be  quite specific to the needs and constraints of particular types
of projects. It must be  emphasized that no "cookbook" will ever be  available for this exercise:
difficult-to-quantify factors such as worker safety and professional judgment will always play an
important role. But it would be very worthwhile to adapt or develop tools to assess  and compare
the quantitative aspects of study design.

       Appendix B, which is a supplement to this chapter, summarizes a number  of statistical
methods  that might be useful in some of the circumstances of heterogeneous waste characterization
and gives a partial evaluation of the applicabilities of the methods. It  is  clear  from this summary
that there is a need for a very detailed handbook to  evaluate in depth  each of the methods for its
appropriateness in waste characterization, and to provide examples and guidance to users as to the
best choices in  particular  situations.    Enough experience has  been accumulated in waste
characterization and in the application of alternative  statistical techniques that an initial version of
such a handbook could be assembled now. The handbook would be expected to evolve over time
and would require the efforts of many contributors.


 1.      U.S. Environmental Protection Agency. 1986.  Test methods for evaluating solid waste. SW-
       846. 3rd edition and revisions. Office of Solid Waste and Emergency Response.

2.      U.S. Environmental Protection  Agency.    1988.   Guidance for conducting remedial
       investigations and feasibility studies under CERCLA. Interim Final.  EPA/540/6-89/004,
       OSWER Directive 9355.3-01.

3.      U.S. Environmental Protection Agency. 1987.  A compendium of Superfund field operations
       methods. EPA/540/P-87/001.

4.      U.S. Department  of Energy. Order 5400.1. General Environmental Protection Program.

5.      U.S. Department  of Energy.  Order  5400.5. Radiation Protection of the Public and the

6.      U.S. Environmental Protection Agency. 1991. Conducting remedial investigations/feasibility
       studies for CERCLA municipal landfill sites.  EPA/540/P-91/001. Office of Emergency and
       Remedial Response.

7.      U.S. Environmental Protection Agency.  1989. Methods for evaluating the attainment  of
       cleanup standards.  Volume 1.  Soils and solid media. EPA/230/02-89/042. Statistical
       Policy Branch.

8.      U.S. Environmental Protection Agency. 1991.  Data Quality Objectives Process for planning
       environmental  data collection activities. Draft, April 1991. Quality Assurance Management
       Staff, Office of Research and Development.

 Chapter 4

                 QA/QC  and Data  Quality Assessment

                             Jeff van Ee and Roy R. Jones, Sr.


       Quality assurance, quality assessment, and quality control are important for any study. The
 sampling of heterogeneous waste makes these study elements even more important because of the
 variability in the sampled material and the need to determine whether the data from a sampling
 effort accurately represent that material. Sources of variability in the measurement process, which
 can obscure the detection of natural spatial and temporal trends in the sampled material,  can be
 increased in the  sampling of heterogeneous material.  Those sources of variability have been
 identified previously in "A Rationale for the Assessment of Errors in the Sampling of Soils" (1), and
 they occur in the collection, transportation and handling, preparation, sub-sampling, and analysis
 of the sample. The challenge for the investigator is to determine whether the bias and variability
 introduced during those phases  of a study are sufficiently small, in relationship to the natural spatial
 and temporal variability of the sampled material, that they may  be neglected. Quality assurance,
 quality assessment, and quality control are meant to aid the investigator in meeting this challenge.

       The terms quality  assurance, quality assessment, and quality control are often misunderstood
 (see definitions, Chapter 2). Activities associated with these terms are often viewed as burdensome
 requirements that drain resources from an investigation and slow its progress. Some people view
 their completion of a Quality Assurance (QA) plan as representing their total commitment to QA
 when in fact it only represents the beginning. Definition of the data quality objectives (DQOs) may
 be attempted in a rigorous fashion, if it is done at all, but ascertainment of whether those objectives
 have been met usually proves to be difficult. All too often ascertainment of whether the  DQOs
 were met, or whether they were appropriately chosen, is never completed.

       It cannot be emphasized too strongly that the application of up-front, sound quality
 assurance procedures and documentation is the most important element for continuity, reliability,
 and confidence in the waste investigation project. Applied reiterative quality assurance provides
 for maximum utilization  of resources and ultimate economy of operations. It is the  best way to
 avoid having to repeat a job at an increased cost,

       There are two important conditions for successful application of QA to achieve project goals.
First, the support of QA  must come from the top down and permeate the way an agency or
program acts in its usual and accustomed way of doing business. QA must be supported throughout
the entire management and production structure of the organization generating the  data. QA
practices and procedures must start with the  inception of a problem-solving  design and be a
cooperative effort of all involved parties.  This is accomplished through established and accepted
guidance and standard operating procedures, implemented even in the initial scoping phase of


project definition.  Second, QA must be recognized and implemented as a reiterative and plastic
process.  By so doing, all parties to the study have an opportunity to apply their areas of specialty
or expertise. Especially as concerns heterogeneous  materials, QA must be an ongoing process,
evolving as problems are encountered and solved during the waste study.

       General guidance for QA/QC  in  hazardous-waste investigations is  provided by  EPA
(2,3,4,5).  This chapter attempts to lay out a basis for QA/QC in the sampling of heterogeneous
waste. The intention is not to duplicate standard guidance on the subject; rather, the purpose of
this chapter is to translate some of the standard QA/QC requirements and procedures into an area
where QA/QC can be most difficult to incorporate -  sampling of heterogeneous waste.

       Standard guidance and protocols for QA fall short when applied to heterogeneous materials.
In attempting to ensure and evaluate data quality, the study planner frequently discovers that:

•     there are no applicable standard reference materials for insertion into  the sample train;
•     there are no standard methods for necessary field and laboratory activities;
•     data  representativeness, precision,  and  accuracy  are difficult to assess with standard
•     those who audit project activities are unfamiliar with the procedures used and can neither
       evaluate the appropriateness of those procedures nor offer corrective guidance.

       Some of these issues, or portions of all of them, are beyond the current state of the art for
heterogeneous waste characterization procedures. For those that can be addressed, two scenarios
are  presented herein to focus discussion on the sampling of heterogeneous waste.  One  is the
sampling of waste piles with a variety of wastes, in a variety of forms, and in a variety of sizes. The
other scenario is the sampling of drums in which the materials were intended  to be isolated from
the environment and from human observation and contact. The application of QA/QC to both of
these scenarios  is important,  but  different  adaptations have to be made to reflect the different
nature of the problems.

       Two questions commonly asked in a QA/QC program are:

•     How many, and what type, of quality assessment samples are required to assess the quality
       of data in a field sampling  effort?
•     How can the information from the quality assessment samples be used to identify sources
       of error and uncertainties in the measurement process?

       Suggested methods to address these two questions are provided in EPA  (1),  particularly for
the sampling of soils.  Some adaptations may be made for  other media and other situations. The
following discussion draws upon the rationale in EPA (1) for the sampling of heterogeneous waste
and the assessment of data quality.

       The approach described herein  is  based on the  statistics  of a normal distribution.
Hazardous-chemical or radionuclide data from the sampling of heterogeneous wastes are unlikely
to follow a normal distribution. As discussed in the chapter on study planning, before using any
statistical model, the investigator must establish what the true distribution is. If a transformation
can bring the concentration data into a normal distribution, the approach described in EPA (1) may
be used to assess, and potentially control, sources of bias and variability throughout much of the


sample collection and measurement process.  Sampling of heterogeneous wastes will probably
require some modification of the approach, particularly when sampling involves remote sensing
methods or barrels of assorted waste. Suggestions are provided on how these situations may be
                        QA/QC in Sampling Heterogeneous Waste

       Sampling involves many different steps and techniques.  The general approach in QA/QC
is to examine those steps and techniques in a rigorous fashion to determine the bias and variability
and whether the data associated with the routine sampling meet the DQOs. A variety of tools are
available  including audits,  education, documentation, and  a  variety  of standard materials,
procedures, and methods. QA/QC for heterogeneous waste sampling is especially difficult because
of the paucity of standard materials,  procedures, and methods. Sound QA theory suggests that new
categorization and quantification techniques  should be developed for heterogeneous hazardous
waste sites.

       An initial question to consider is whether there are available standard materials, procedures,
and methods that  are appropriate for the questions that one seeks to answer  in the sampling
activity. The DQO process and the available sampling strategies and methods which may be
employed have been discussed in previous chapters.   Unfortunately, a poor sampling plan and
inappropriate  DQOs  can have  profound impacts  on a  QA/QC program.   For example, in
characterizing the contents  of a drum, many approaches may be taken depending on the overall
objectives of the characterization.   If the objective is to  determine whether the drum may be
transported by common carrier to another location, then  particular emphasis will be given to
determining whether  the contents  and packaging  are in conformity with  DOT regulations.
Alternatively,  if the objective is to determine whether the drum  may be stored in a particular
location for a long period of time, the form and substance  of the contents of the drum may have
to be determined, especially  if failure of the drum could  lead to significant risk to human health and
the environment. Measurements  of the drum  under those two scenarios would employ different
monitoring methods, have  different DQOs, and carry different levels of QA/QC. The level of
QA/QC  should be based upon the degree of confidence  required in knowing whether the data meet
the DQOs and in being able to  report the QA/QC data accurately.  Until the DQOs  have been
specified  and sampling  and analytical methods have been selected, a QA/QC program  for
heterogeneous materials cannot be prescribed, and major impediments to the  implementation of
the QA/QC program cannot be addressed.

       How  does  one  employ  QA/QC  in  the  characterization of waste  in a  drum? If a
conventional  method such as GC/MS is employed to  analyze samples from the drum, then the
analytical  QA/QC  may be conventional. QA/QC in the laboratory can be the same as that used
for  the sampling of other, more conventional media. Assessment of variability and bias in the steps
leading to the analytical step remains an issue. Alternatively, if a field physical method such as a
neutron assay method is  used, then new QA/QC methods may need  to be developed and validated.
In this instance, QA/QC for a remote sensing method would likely include the sampling and
analytical steps because they are more closely  intertwined. Common to both measurement methods
is the question of the number and timing of observations that are needed to assess the performance
of the measurement methods. Also common to  both methods is the question of whether the "truth"

can be ascertained. If the reported parameters are a function of the measurement method, there
can be problems in ascertaining the "truth."  Depending on the objectives of the study, it may not
be necessary to know the "truth," e.g., the absolute concentration of a substance in the drummed
wastes, to a high degree of accuracy and precision. If no standards are available for assessing bias
and variability at various steps in the sampling of heterogeneous materials, then the QA/QC can
be  complicated and  more severely restricted.   Application of QA/QC to characterization of
heterogeneous wastes requires  careful thought.
                            Assessment of Bias and Variability

       Systematic  errors,  termed  bias  (B), can be  introduced at many  points  in  the  waste
characterization process (Table 4-1). Bias causes the mean value of the sample data to be either
consistently higher or consistently lower than the "true" mean value. Bias may result  from faults in
sampling design, sampling procedure, sample  preparation, analytical procedure, contamination,
losses, interactions with containers, deteriorations, displacement of phase or chemical equilibria, and
inaccurate instrument calibrations. When the sampled material is heterogeneous, subsampling that
favors one type of item over another is an obvious potential source of bias. Laboratories usually
introduce various quality control samples into their sample load to detect possible bias. Bias in the
sampling of heterogeneous materials is difficult to detect. Components of bias can be discovered
by the technique described as standard additions or by using evaluation samples.  On the other
hand, it is difficult to demonstrate that bias is not present because an apparent lack of bias may be
the result of an inability to measure it  rather than its actual absence.

       Variability in the data can come from bias that changes over time, from natural variability,
differences between batches, or from imprecision in the collection, handling, transportation,
preparation, sub-sampling,  and analysis of the sample (Table 4-2). Also, variability occurs  in the
measurement process from the heterogeneity of the material and random  errors throughout the
measurement process. The  variability caused by any type of random error is frequently described
quantitatively by the variance, a2, of the random error, or by the positive square root, the standard
deviation, o, of the random error. The variances of independent random errors are additive in that
the variance of the sum of  errors is the sum of the variances of the individual errors (Table 4-2).
Other quantifications of variability do not have this useful, additive property.

       Separating the different components of  error requires proper selection, use, and timing of
QA/QC samples throughout  the  study. A rule of thumb is that the more observations that are
made, the greater the degree of confidence that may be placed in the observations.

 Table 4-1. Elements of measurement bias in environmental sampling
Bs =   Measurement bias introduced in sample collection not caused by contamination
Ex =  Measurement bias introduced in sample collection caused by contamination
Bh =   Measurement bias introduced in handling and preparation not caused by contamination
Bhc =  Measurement bias introduced in handling and preparation caused by contamination
Bss =   Measurement bias introduced in sub-sampling not caused by contamination
BSSC =  Measurement bias introduced in sub-sampling caused by contamination
Ba =   Measurement  bias  introduced in the  laboratory analytical process not  caused by
Bac = Measurement bias introduced in the laboratory analytical process caused by contamination
Bm =  Total measurement bias (sum of the above components)
       NOTE: Biases, other than contamination biases in the measurement of a sample, are often
dependent on the original concentration of the contaminant being measured and on the sample
matrix. Biases caused by contamination are listed separately because some QA samples, such as
rinsate samples, detect only contamination bias.
Table 4-2. Elements of variability in environmental data

           °«2= °m2+°P2

    where ot2 = total variability

          o^ =  measurement variability

          op2 = population variability

          °m2 = °s2 + oh2+oss2+oa2+ob2

    where os = sampling  variability  (standard  deviation)

           oh = handling, transportation and preparation variability

           0^ = sub-sampling variability

           oa = laboratory  analytical variability

           ob = between batch variability

        In the scenario of characterizing heterogeneous waste in a drum, analysis of the drum could
 involve the collection of one sample from the drum and its extraction and analysis by GC/MS.
 Experience would suggest initially that QA/QC for the sample in the laboratory would be relatively
 easy because of previous experiences, practices, and the ready-availability of standard materials for
 the GC/MS in the laboratory. More experience would suggest that the variability and biases in the
 laboratory measurement are likely to be minor in comparison to the variability  and biases involved
 in collection of the one sample. Conventional statistics suggest that one sampling of the drum of
 heterogeneous wastes would be inadequate and that some statistics-based approach  would be
 needed for increasing our confidence that our sample collection is representative of the contents
 of the drum. The bias and variability associated with the GC/MS analysis of the sample may be
 on the  order of 20%, while the act of collecting the sample could result in orders of magnitude
 differences  in the reported values depending on where and how the contaminant was physically
 sampled within the heterogeneous mixture. In many QA/QC programs, a great deal of resources
 and attention are devoted to the laboratory analysis when greater attention  should be given to
 earlier phases of the sample collection process.

        If a drum is  sampled by a remote  sensing  method,  assessment of bias and variability
 becomes more complicated.  Bias  may be represented in terms  that are related  to other remote
 sensing methods or to more conventional analytical methods.   The DQOs should be useful in
 determining how bias should be reported and assessed. As in the sampling of a drum for GC/MS
 analysis, the process of making the measurement can introduce more bias and variability than the
 analytical process itself. Variability is more readily assessed, but bias becomes more complicated
 when known concentrations  of contaminants in a sample matrix similar to  that of the routine
 samples may not exist.   Whenever possible, known materials  with known concentrations of
 contaminants that are indistinguishable to the operator of the analytical test method should be
 analyzed to help determine bias and variability of the test method. Duplicate  measurements may
 be made to assess variability, and calibration standards may be used to assess and control bias
 within  the measurement method, but double-blind materials are  preferred for assessment of
 cumulative bias and variability at the various stages of the measurement process.
                                     QA/QC  Samples

       One important tool, defined and described in more detail in EPA (1), is a suitable "reference
material" to assess bias and variability.  "Reference materials" may be described as materials having
a known  concentration,  or  property, that  is relatively stable over time,   "Standard  reference
materials" are defined as being materials produced by the National Institute of Standards and
Technology. Other terms may be used to define and describe reference materials. These materials
are most often thought of, in terms of laboratory measurements, as well-characterized,  relatively
homogeneous materials in small quantities that are usually used to assess bias and, alternatively,
precision  of a  method.   These  samples may  not  be appropriate for the  characterization of
heterogeneous waste. For less  orthodox, relatively new methods such as the physical imaging of
drum contents, there are apt to be few, if any, reference  materials that are suitable. When
inadequate reference materials exist, some consideration needs to be given to their manufacture.
The difficulty, cost, and time involved in making up the materials may not be  appropriate  for a
particular investigation. The decision to makeup  these materials depends, in part,  on the objectives
of the study and the degree of confidence that is needed in the results from the QA/QC  program.

       A relatively new practice, and an especially important one in assessing QA throughout the
 sample-collection and  measurement process,  is  the  preparation of site-specific performance
 evaluation (PE) materials.  These materials should be double-blind in that they cannot be readily
 recognized, and their established properties are unknown to the analyst. They offer the advantage
 of being able to accurately assess the performance of a measurement process through the use of
 natural materials that closely resemble native  materials under similar conditions. They may  be
 made from the same materials that are being routinely sampled. While site-specific performance
 evaluation materials  offer  opportunities  for obtaining unbiased estimates of bias and variability
 throughout the sample collection  and measurement process, they are not used often, in part because
 established procedures do not exist for their use and for their fabrication.

       Site-specific performance evaluation materials for characterizing heterogeneous materials
 may be made  by spiking uncontaminated waste  materials  with  a  known  concentration  of
 contaminant or surrogate. Site-specific performance  materials for the characterization of a drum
 by field physical methods  would at first appear to be less straight-forward, but may actually  be
 easier. Under the scenario of the drum that is being examined by a physical method, a site-specific
 performance evaluation material could be  made by packing a drum with items of known properties.
 For example, gloves, containers of liquid hazardous waste, and debris could be inventoried and
 packed into a drum for analysis by others. This drum  could be introduced into the routine sampling
 stream and could be classified as  a "double-blind." This "double-blind" sample would yield  valuable
 data on the performance of the measurement system if the drum were analyzed just once,  or if the
 drum were randomly reintroduced into the sampling stream several times over the course of the

       Other QA/QC samples, such as blanks, are described in EPA (1). While that document
 focused on soil sampling, these QA/QC samples may be used, with adaptation,  for other media and
 other situations.

       One of the most powerful tools available for QA/QC purposes is the concept of replicate,
 duplicate  and co-located  analyses.   They are  particularly important for  the  sampling  of
 heterogeneous waste because of the large spatial and temporal variability associated with these
 wastes.   Separation of natural  spatial and  temporal variability in  the sampled material from
 variability and bias in the measurement process may be achieved by the use of co-located, replicate
 and split samples at various points in the sample collection and measurement process. Creating a
 co-located sample in the sampling of a waste pile would entail the sampling personnel moving a very
 short distance from where a routine sample was  collected to collect another sample. The co-located
 sample would be identified and treated as a routine sample. (The co-located sample is sometimes
 known as a "field duplicate.") If the variation within the pile (the sample population) is very great,
 the decision  maker will be alerted to this by the  disparate results for the  routine and co-located
 samples. Prior to its receipt at the analytical instrument, a sample may be split. (The split samples
 may be known as "preparation splits.") At the instrument, replicate or duplicate  observations may
 be made to assess variability  in the use  and operation of the analytical instrument. All of these
 samples and procedures can assess variability from a variety  of sources at various times. The
 information  from these samples can  be  used  to  determine  whether the variability in the
 measurement process is so  large that it needs to be controlled or is sufficiently  small that it only
 needs to be reported. Figure 4-1  is a schematic of how QA/QC samples may be used to  identify
various sources of variability and bias.   Details on the  computation of variability from  the QA
 sample data are given by EPA (1).


                               QUALITY ASSESSMENT SAMPLES
                             DUPLICATES AND SPLITS
                                                              FES     FES     ELES   ELES     FRB

                                                                                           aa,ah,os aa,oh
Types of Samples
Field Evaluation (FES)
Low Level Field Evaluation (LLFES)
External  Laboratory  Evaluation

Field Matrix Spikes (FMS)
Field  Duplicates  (FD)
Preparation Splits  (PS)
Field  Rinsate Blanks (FRB)
Preparation  Rinsate Blank  (PRB)
Trip  Blanks (TB)
Samples may be submitted  non-blind,
definitions  of sources of error.
                                        Samples  of  known concentration are  introduced  in  the  field  as  early as
                                        possible to  check  for measurement bias and to estimate precision.

                                        Low concentration FES samples  check  for  contamination in sampling,
                                        transport, analysis, detection limit.

                                        Similar to  FES but without exposure in the field, ELES can  measure
                                        laboratory bias and, if used in duplicate, precision.

                                        Routine samples  spiked  with the  analytes of interest in the  field  check
                                        recovery  and  reproducibility  over batches.

                                        Second samples taken near routine samples check  for variability at  all  steps
                                        except  batch.

                                        Subsample splits are made  after homogenization  and  are used to  estimate
                                        error occuring in the subsampling and  analytical steps of the process,

                                        Samples obtained  by  rinsing the decontaminated  sampling equipment with
                                        deionized water to check for contamination.

                                        Samples obtained  by  rinsing the blank sample preparation  apparatus  with
                                        deionized water to check for contamination.

                                        Used  for Volatile Organic Compounds  (VOC), containers  filled  with
                                        American Society  for  Testing and  Materials  Type II water  are kept with
                                        routine samples through the sampling, shipment, and analysis phases.  Other
                                        types of trip blanks can be  created.

                                 single-blind, or double-blind,  as required for  data evaluation. See Table  4-2 for
                                Figure 4-1.  Quality  Assessment Samples


        Even when unconventional characterization methods are employed, the use of duplicate
 measurements is important. A drum could be analyzed by the same instrument and same operator
 consecutively or over a period of time.  The drum could be analyzed by the same instrument at
 different times  by different operators to examine the  variability associated with the  different
 operators. Alternatively, the drum  could be examined by two modifications of a basic physical
 method to determine the sensitivity of the reported data to  a particular method.  The choice and
 use of a particular duplicate or replicate sampling approach depends on the overall data quality
 objectives and the degree of confidence that is needed  in knowing the  source and magnitude of
 variability and bias throughout the sample collection and measurement process. Pertinent  estimates
 of variability may also be available  from previous studies.
                    How Many Observations or Samples are Needed?

        Several answers can be provided to this question. The precise answer depends on historical
 data from previous studies in which  QA/QC was employed to assess bias and variability throughout
 the sample collection and measurement process.  The precise answer also depends on the Data
 Quality Objectives and on the measurement quality objectives (MQOs). The distinction between
 DQOs and MQOs is important and requires explanation.

       Data Quality Objectives are intended to cover an entire study, but most often emphasis is
 given to the measurement phase of the investigation.   Precision, accuracy, representativeness,
 completeness, and comparability are  terms used in setting DQOs and are usually addressed in terms
 of the analytical portion of an investigation.  Decision-makers must be concerned with the larger
 aspects of these terms, however.  For example, a decision-maker may want to know whether the
 reported data are accurate to within 20% of the true value.  This QA/QC parameter might be
 important in considering the action  to be taken in relationship to an established action level.  The
 data quality objective for accuracy includes the method by which the sample was collected and  how
 accurately the collected sample measures the contaminant in an area.  A data quality objective for
 accuracy may not be appropriate for the analytical phases of an investigation, e.g., the MQO for
 accuracy, since biases and variabilities outside the analytical portion of an investigation are included.

       MQOs are meant to apply to the analytical phases of an investigation. Terms for precision,
 accuracy, representativeness, completeness, and comparability are more readily appreciated when
 applied to the analytical phase of an investigation. The distinction between DQOs and MQOs  is
 important because observations, i.e., samples, are taken in a QA/QC  effort to determine whether
 these objectives are being met.

       The question remains, then, how many samples are needed? If historical data indicate that
 inaccuracy or variability is increased in the preparation and handling of a sample, and this affects
 detrimentally the accuracy needed to meet the MQOs, then more frequent sampling of this portion
 of the measurement process is justified. As another example,  assume that the values reported are
 near the action level; bias is particularly important in meeting the DQO for accuracy, and the
 consequences in knowing whether a population of waste is above or below that action level are
 great, In this case, greater attention may need to be devoted to the sample-collection step and to
 the use of field duplicates to assess the variability. The number of samples required depends on
the available resources, the required degree of confidence, and the objectives of the study.

       The EPA (1) provides tables and a discussion for the number of samples that are required
to obtain a certain level of confidence when the data are normally distributed or can be transformed
to the normal distribution. The precision of an estimate, s2, of the "true" variance, o2, depends on
the number of degrees of freedom for the estimate which is  directly related to the number of quality
assessment samples.  Statistics texts contain tables that give the confidence intervals for  various
degrees of freedom, based  on an assumption that the data are,  or have been  transformed to,
normally distributed data.

       Equations are provided in EPA (1) for assessing the errors in the sampling of soils, and
these may be applied to more heterogeneous  materials.  The equations and tables in EPA (1) are
also incorporated into the public-domain computer program ASSESS (available from EMSL-LV).
This  program  resembles a computer-based spreadsheet, and computes  measurement  errors,
provided enough QA/QC  samples of the right type have been taken throughout the study. ASSESS
indicates when insufficient samples exist and certain variabilities cannot be computed.

       ASSESS can display graphically the degrees of confidence that exist for the measurement
of variability in the individual portions of a study. Certain  portions of the study receive more QA
and QC than other  phases.  For those portions of the study that are monitored to a high  degree,
the variability  may be low and ignored, or  the variability may be high and  may  need to be
addressed. For those portions of the study that are not monitored to  a high degree, the variability
may be low, but more QA/QC samples may be required, or the variability may be high and more
QA/QC samples may be required before potentially difficult and expensive corrections are taken.

       If the study is complete, the reported measurements from a QA/QC program can be termed
as quality assessments.  The quality assessment data are reported along with the routine  sample
data to provide the decision-maker with a measure  of the quality of data available for making the
decision. The form in which the QA results  are reported is crucial.  Tables of numbers buried
within voluminous data appendices are of little use  to decision-makers.  Instead, the results should
be presented in simple tabular form and interpreted  in the text of the report. What is the variability
of each stage of the sampling/analysis process, and what does it mean for interpretation of the
routine sample data? Based on this information, the decision-maker may choose to initiate another
phase of sampling and characterization.
                               Research Recommendations

       If quality assurance is the process of "demonstrating we are doing the right thing,"  and
quality control is the process of "demonstrating we are doing things right," then it becomes obvious
that existing QA/QC  procedures  must be  adapted  to deal  with unconventional waste
characterization  problems. Quality assurance, as discussed here, is a process that starts with the
initial planning of site operations.  The majority of problems  in both QA and  QC relate to the
question of representativeness of the sample data to the actual  conditions of the site; these
problems are especially acute with heterogeneous wastes.  For site managers who must have
confidence in the data  they use for decisions, the credibility  of the data is proportional to the degree
of representativeness that can be documented or demonstrated.

         The following QA/QC research needs address the concerns expressed by professionals
from a number of disciplines who are involved in operations at heterogeneous hazardous waste
sites. The first set of research needs is concerned with the Data Quality Objective process and its
dependence on statistical methods in formulating the DQO(s). When the generally used "status
quo" statistics are overwhelmed by the variety and  complexity  of a heterogeneous  waste site,
alternative planning tools are needed. These should include:

•      A model (or criteria for a model) to warn the project team that conventional  approaches
       are either failing or inflating the proposed project beyond reasonable or acceptable levels
       of time,  money,  or litigational potential.   This model would be one component of the
       broader project-planning tools recommended in the previous chapter.
•      An interdisciplinary  checklist of statistical terms  and definitions appropriate to DQO

       The second set of research needs  involves the quality control aspects of operations at
heterogeneous waste sites. Areas that need development include:

•      Field attainable quality control sampling methodologies. Methods should be defined and
       described to permit repeatable, practical samples to be taken with some confidence that the
       reported  data are representative and accurate for the sampled material. Standardization
       of sample collection methods (wipe  tests, physical  parameter measurement procedures, etc.)
       is essential for the larger scale debris and containerized waste problems.
•      Sampling methodologies that can deal with the problems  of particle size and  multiphase
       matrices. These may include some bulk extraction techniques that would effectively start
       the analytical process in the field.   These recommendations are expanded upon in  the
       following chapter; they are mentioned here because the availability of reliable methods for
       generating representative samples is very much a QA issue,
•      The techniques  of manipulation, preparation, and packaging of site  specific materials.  These
       are needed to provide reference materials for  performance evaluation.

       Many of these research or development efforts may be accelerated if they are conducted in
conjunction with ongoing site  investigations  rather than in isolated settings. The following measures
are strongly recommended:

•      Developers of new methodologies and procedures should document their applications as
       completely as needed to assure timely and  adequate training of end users of the products.
•      Practical trials, both  laboratory and  field,  should  be performed  in the presence  of
       experienced field personnel and active, qualified Quality Assurance staff that will  be
       expected to audit the actual use of the new method at different sites.
•      Parallel development of applied technologies should be encouraged and utilized wherever
       possible,  e.g., the development of  bulk quantities of reference materials for  treatability
       studies as a preparatory step in the  development of PE materials.


1.      U.S. Environmental Protection Agency. 1990. A rationale for the assessment of errors in
       the sampling of soils.   J.J. van Ee, LJ. Blume,  and T.H. Starks. EPA/600/4-90/013.
       Environmental Monitoring Systems Laboratory, Las Vegas, NV.

2.      U.S. Environmental Protection Agency.   1983. Interim guidelines and specifications for
       preparing quality  assurance project  plans.  EPA  600/4-83/004.  Quality Assurance
       Management Staff.

3.      U.S. Environmental Protection Agency.   1987. Data Quality  Objectives for remedial
       response  activities.  Development process. EPA/540/G-87/003. Office of Emergency and
       Remedial Response.

4.      U.S. Environmental Protection Agency. 1989. Preparing perfect project plans. EPA/600/9-
       89/087. Risk Reduction Engineering Laboratory, Cincinnati, OH.

5.      U.S. Environmental Protection Agency.  1986. Test methods for evaluating solid waste.
       SW-846.  3rd edition and revisions. Office of Solid Waste and Emergency Response.

 Chapter 5

                               Sample Acquisition

                       Janet Angert, Alan Crockett, and Timothy Lewis


       This chapter  presents procedures for sampling and characterizing heterogeneous hazardous
wastes.   The chapter deals with both uncontained wastes consisting of landfills and piles and
contained wastes such  as  drums and  boxes.   Unfortunately, there  is relatively little guidance
available for landfill  sampling beyond the standard site characterization and soil and waste sampling
techniques.    Landfilled  waste  is seldom characterized due  to  the difficulty of obtaining
representative samples and the hazards of invasive sampling.  The problems of sampling semi-
homogeneous waste in containers have been addressed by EPA in other documents (1,2,3). This
chapter concentrates on characterization of extremely heterogeneous contained wastes and relies
on the experience of DOE and its contractors for characterization of radioactive and mixed wastes.

       Sampling and  characterization of heterogeneous hazardous waste  pose several  major
problems, including  obtaining representative information and assuring personnel safety. For these
reasons, relatively few landfills have been intrusively sampled and very little literature exists on
managing the  unique problems of extremely heterogeneous waste.   Landfills  of relatively
homogeneous materials or of known content have been sampled using conventional technologies.
In other cases, unique  approaches have been  developed.  Current  EPA guidance on studying
municipal landfills that are NPL sites recommends against detailed waste characterization, unless
discrete, accessible "hot spots" of contamination have been identified  (4).

       Before embarking on the sampling of hazardous wastes in landfills, piles, or containers, it
is prudent to obtain all relevant information that may permit achieving the objective without actually
sampling (see Chapter 3). It is  also prudent to scrutinize the initially-defined project objectives and
approaches  critically on the basis of the sampling problems, hazards,  and  potential costs. For
example, it may be possible to consider a landfill as a black box, characterize the area around the
waste remotely, and sample contaminants leaving the site via air, leachate, groundwater, and biota,
rather than  intrusively sample the  waste.   The same approach holds for drum and container
sampling. Remote interrogation can provide some but not all information about the contained
waste. Constructive negotiation between the regulators and the regulated party should be part of
the DQO process.  This will allow the parties to avoid spending huge sums and committing large
staffs to sampling and characterization  projects that yield little benefit.

                  Characterization of Uncontained Heterogeneous Wastes
 General Considerations

        With the increasing number of landfill facilities  on the NPL, investigating  landfills for
 evaluation of source control  measures and removal alternatives is becoming increasingly important.
 Conditions at each landfill are unique and require detailed site-specific investigations into their
 structure and the  nature  of the  solid and  liquid  wastes contained within each system.  The
 hazardous and radioactive wastes present in the landfill can coexist in multiple phases (i.e., gas,
 liquid, pure solid,  or adsorbed to solids). The material  can be containerized or uncontained.

        Sample acquisition from a landfill can be very costly, labor intensive, and dangerous. The
 deposition  of debris  on the  site  poses difficult problems  for those  attempting to  collect
 "representative" samples of the site. The problems include, but are not limited to:

 •      What is representative? When faced with  a collection of materials of assorted sizes, types,
        and contaminants in  various phases, all arranged in unpredictable  distributions  over a large
        area (several acres), one can begin to realize the difficulty  of collecting a sample that is truly
        representative of the entire  site.

 •      How does  one characterize the health risks of  large entities  (e.g., a refrigerator)? Large
        objects  other  than white goods may be ground to manageable sizes.   However, the
        contamination may be associated with only the surface of the object. Grinding  would cause
        undue dilution  either by volatile  losses or by combining interior uncontaminated material
        with surface contaminated material. Wipe tests have been  used on surfaces, but currently
        no agreed-upon action limits in units of concentration per  surface area are available.

 •      Should  the objects  be sorted  into populations  by size, material type, or  suspected
        contamination? Without a priori knowledge of the contents  (e.g., percentages of paper,
        metallic objects, plastics,  etc.) and the contaminants present (e.g., heavy  metals,  volatile
        organics, semivolatile organics, etc.) in a given landfill, it is difficult to selectively sort the
        material to obtain  an estimate of the distribution of contaminants.

        A landfill can also be characterized by remote sensing coupled with indirect measurements,
 that is, sampling of environmental media rather than the waste itself. Landfill waste sampling is
 usually undertaken as a last resort and primarily  for remedial purposes. When waste sampling is
 called for, it is highly site-specific.   Indirect sampling is  preferred for several reasons, the most
 important being protection of worker health and safety.
Indirect Characterization Techniques

       In sampling heterogeneous wastes in a landfill, there are two requirements  that at first
glance appear to be diametrically opposed.   The first is  to obtain sufficiently large samples to
encompass all the types and sizes of materials representative of each portion of the entire site. The
second is to obtain  sufficiently small samples that they can  be handled by  the analytical laboratory,

 Theoretically, the largest sample that could be taken would be the entire landfill site. This, of
 course, is impracticable. However, a landfill can be approached as a black box. The reactions,
 behavior, and mobility of the contents of the black box can be ascertained by monitoring the
 effluents or emissions from the system.  Monitoring the input and output from the black box may
 serve as a means of obtaining the largest possible sample that is representative of the entire site.
 However, the investigator must keep in mind that landfills  are dynamic systems which can take
 more than 30 years after filling is completed to reach an equilibrium, where infiltration through the
 landfill cover equals leachate (5).
Leachate  Monitoring

       Monitoring of regulated landfills is required by Federal law.  Most states have their own
monitoring programs in place at hazardous landfills, and state requirements may often be more
rigorous than federal requirements.

       Monitoring programs are based  on knowledge  of the site. Historical records, aerial
photographs, and manifests aid in the design of an effective monitoring program.

       Monitoring wells are  used to detect the migration of hazardous wastes from the interior of
the landfill into adjacent ground water supplies.  A network  of monitoring wells around the
perimeter of the site can provide data to  aid  in the delineation of hazardous waste migration.
Monitoring wells and lysimeters can be installed  in the landfill interior to serve as an "early-warning"
predictor  of risk to adjacent ground water resources.  Internal monitoring wells are commonly
screened either at the bottom of the waste or at the top of the saturated zone  (the water table).

       Borings performed during well installation can be used to characterize the contents of the
landfill. Geophysical logging of the  borings can be used to obtain information on the subsurface
structure and  moisture regime within the  landfill.  Three types  of  geophysical techniques are
commonly  employed for borehole logging (6):

•      Natural Gamma Logging - will provide information on clay-rich (daily and  intermediate
       covers) and clay-poor materials (refuse)
•      Gamma-Gamma Logging - indication of changes in bulk density; cover material will have
       greater bulk density than surrounding refuse
•      Neutron Logging - indication of moisture content; the probe is also sensitive to hydrogen
       content of organic materials.

       The logging can be performed down the  borehole before or after the well casing is installed.
If the latter is  employed, the density  of the steel or the hydrogen content of the PVC pipe must be
taken into account. Calibration of the test method should be performed  on each site. Due to the
instability  of the refuse in landfills, open boreholes frequently collapse  around the probes, resulting
in an unanticipated excavation to retrieve the radioactive source.  NRC licensing is  required for
such logging procedures and calls for the service of a licensed subcontractor.

       The distribution of wells on the site  is ultimately dependent  upon the investigator's
confidence in  the uniformity  of the site.  If no information is available as to the contents of the

landfill or the leachate chemical characteristics, then at least one leachate well should be installed
for every 5 acres of fill area (7).

        Several precautions should be taken while installing  monitoring wells  within the interior of
a landfill. O'Hara et al. (8) discuss the seriousness of exploratory sampling  and monitoring well
installation at landfills containing buried drums.  If drums  are present on the site, they must be
located by non-intrusive,  remote interrogation procedures.   O'Hara et  al. (8) employed  a
magnetometer to indicate the possible presence of buried  metal (including drums). The boring
locations were based on the representative distribution within the reportedly deepest portions of
the pit and in areas having lower magnetometer readings than surrounding areas. The purpose of
the magnetometer screening was to minimize the potential risk of drilling  directly into a drum or
pocket  of drums,  particularly at shallower  depths.   A more detailed  discussion of remote
interrogation methodologies is presented in a later section of this chapter.

        If the  investigator has prior knowledge of exactly what radionuclides  or chemicals were
placed in the landfill,  the job of selecting indicator compounds or elements  for leachate monitoring
is fairly straightforward.   Volatile organic compounds are  present at most sites, and it has been
suggested that they be used as indicators for the presence of  other organic compounds (9,10). Thus,
without any  information on the contaminants that may be present, as a first  choice, volatile organic
compounds should be monitored as indicators of other organics. For inorganic contaminants, there
is no ubiquitous  element that can be used as an indicator. The  Superfund Public Health Assessment
Manual contains procedures for identifying the major contaminants (indicator chemicals) at a site
Soil-Gas Monitoring

        Soil may be a small percentage (< 10%) of the bulk material in a landfill. Thus, the term
"soil gas" may be a misnomer in the case of landfills, so the term "landfill gas" will be used instead.
Landfill-gas analysis can serve a variety of screening purposes, from initial site reconnaissance to
remedial monitoring efforts,  Obviously, landfill-gas sampling is intended for contaminants with
suitably high vapor pressures and, as such, is applicable to the volatile organic class of indicator
compounds.  Table 5-1 presents the types and  concentrations  of volatile  organic compounds
measured in landfill gas at 340 sites in California (12).

       Landfill-gas probes can be installed around the perimeter of the site and/or within the site
boundary. Samples can be collected in evacuated Summa canisters and the contents analyzed at an
off-site laboratory. Field portable instruments are gaining acceptance in the scientific community.
They allow immediate on-site analysis of solid, liquid, and gaseous samples.  Field analysis of landfill
gas is  usually by hand-held detectors, portable  GC or GC/MS, long-path Fourier transformed
infrared (FT-IR) detectors, ion mobility spectrometers, industrial hygiene detector tubes, and,
recently, fiber optic sensors.  Further information  on field screening methods is presented later in
this chapter.

 Table  5-1.     Concentration statistics  of specified contaminants  and methane  in  landfill-gas
methylene chloride
1,1,1 -trichloroethane
vinyl chloride
ethylene dichloride
carbon tetrachloride
ethylene dibromide
Number of Landfills
Contaminant Detected3
a       Landfill gas sampling was conducted at 340 landfills.
        Medians  and maxima of  the average sampling results  from individual  sites.
0       u =  Means  non-detected; the number shown is the detection limit.

Note:  Concentrations are in ppbv,  except methane  which  is in percent.  Taken from Baker  et  al.
Emission Rates  and Characteristics

        Vapors emitted from the surface  of landfills may pose health risks to adjacent  populations.
Typically, a  flux  chamber is used to estimate landfill  gas  emission rates (13).  The Air Resources
Board of the State of California recommends that the following compounds be analyzed (12):
methane, oxygen,  carbon dioxide, nitrogen,  vinyl chloride,  benzene, ethylene dibromide, methylene
chloride, perchloroethene,   carbon   tetrachloride,   1,1,1-trichloroethane,   trichloroethene,   and
chloroform.  The occurrence  of these compounds  emitted  from the  surface of landfills  is similar  to
the data shown in Table 5-1 for landfill gas.

        The  extent of gaseous emissions  from  the  surface of a landfill  can often be delineated by
observing the distribution of vegetation on the site. Dead vegetation or areas devoid of vegetation
can  give an indication of  areas of  gaseous  emissions.   Gaseous emissions  can kill plants by
displacement of oxygen from the  soil  or  by direct toxicity.

        Long-path  FT-IR  is  useful for the qualitative  and  quantitative measurement of VOCs and
low-boiling semivolatile organic compounds in the ambient air above the surface  of hazardous waste
sites. Though this technology is sensitive to  meteorological conditions such as  wind, particulate
matter,  humidity,  and rain,  most  of these affect  point  sampling  by canister or  direct  air sampling
methods as well.

 Geophysical Techniques

        Geophysical techniques are frequently applied to characterization of landfill sites. These
 techniques have been used successfully to delineate contaminant plumes, locate buried drums, map
 the location and extent of pits and trenches, and assess the general geologic setting of the site. The
 reader  is referred to  Telford et al. (14) and Sharma (15) for general works on geophysical
 techniques. An expert system to assist in selecting the appropriate geophysical techniques for a
 given study has been developed by EPA (16).

        Direct-current  and low-frequency alternating-current resistivity determinations usually are
 made by measuring the voltage drop between two electrodes placed in the ground after an electrical
 current has  been  induced  in the ground between two other electrodes. The most commonly applied
 electrode configurations are referred to as the Wenner and Schlumberger arrays.  Ultimately, the
 electrical  "soundings"  can be  related  to  the presence of  a contaminant plume.   Dissolved
 contaminants and buried wastes frequently cause ground water to have an abnormally low resistivity.
 Great caution should be used when interpreting the results of resistivity measurements. Improper
 selection of equipment and arrays, poor field procedures, improper interpretation,  and irregular
 terrain  all can  contribute to difficulties inherent in resistivity determinations.  Meaningful resistivity
 measurements cannot be achieved in landfills where  there is a highly random scattering of metallic
 and other conducting  materials present in the heterogeneous  matrix (17).

        Electromagnetic (EM) surveys are similar  to resistivity  surveys inasmuch as  they also
 measure electrical properties of the subsurface. EM does not require the implantation of electrodes
 in the ground, thus making it faster and easier than resistivity. It is also better suited for sites with
 a preponderance of asphalt or concrete slabs and for sites with highly conductive surface soils. EM
 is not as susceptible to metallic interferences as resistivity. However, EM is prone to interferences
 from spurious  electromagnetic fields such as power lines, underground cables, and  transformers,
 which are often present at landfill sites.

        Magnetometers, such as proton procession  devices, are very  useful for detecting buried
 ferromagnetic materials, especially buried drums.  A magnetometer can be used to  locate drums
 prior to well installation or removal activities.

        Ground-Penetrating Radar (GPR) is capable of detecting, in real  time, shallow earth profiles
 of dielectrical discontinuities related to subsurface conditions such as moisture content, lithology,
 bedding, voids, fractures, and man-made  objects. When conditions are optimal, which  seldom  is the
 case on a landfill, GPR profiles exhibit the greatest detail and resolution of any technique. Radar
 waves are dampened by moist, low-resistivity clays,  or other conductive material.
Tracer Studies

       Frequently used in hydrogeology, a tracer is matter or energy carried by ground water which
will give information concerning the direction of movement and/or velocity of the  water and
potential contaminants that may be transported by water (18). Tracer studies are not  commonly
performed on landfills. The physical and chemical behavior of the tracer should be similar to the
anticipated contaminant's behavior,

Aerial Photography

       Aerial photography and infrared imagery can be powerful tools for  defining sources of
contaminants on a much broader  scale,  i.e., the entire landfill site. Aerial reconnaissance to
determine the distribution of drums,  lagoons, discolored soils, and vegetation over the site  can
provide useful information about the extent and source of contamination. Subsequent sampling
efforts can draw on this source of information to focus on hotspots and problem areas.
 Waste Sampling

       As discussed earlier, intrusive sampling of landfills is costly, labor intensive, and dangerous.
 It is usually only performed when the final disposition of the site has been determined to be waste
 removal or treatment.

       Chapter 3 describes how data needs are defined and a sampling strategy is developed. No
 standardized approach is available for sample acquisition at landfill  sites. Each landfill waste
 characterization project is unique. Questions to be asked when collecting samples from a landfill
 for hazard characterization include:

 •      What part of the landfill should be sampled?
 •      Which items should be included as part of the sample?
 •      Should  samples of heterogeneous materials be composite? If so, how?
 •      How much of each type of material should be collected?
Sampling  Location

       The location from which to collect samples at a landfill site is determined from information
acquired by the application of many of the same nonintrusive techniques discussed previously:
photographs, historical records,  manifests, and  conversations  with people familiar  with site

       As much as possible should be learned about the site prior to sample collection. Review of
aerial photographs, historical records, and manifests can provide a great deal of information as to
the nature and location of wastes on the site.  Conversations with waste haulers who delivered
wastes to the site can provide insights not available  from historical records or manifests.

       A walk-over inspection of the site should be performed to gather information on a smaller
scale.  This approach  is semiquantitative.   Quadrants are laid  out over the landfill,  and field
personnel sketch or photograph debris on the surface of the site as they walk over each quadrant.
Debris is not handled or disturbed during the walk-over survey.  Photographs and sketches can be
later inspected off-site, thus greatly reducing the amount of time field crews are on the site.
Subsequent sampling operations can focus on areas with possible sources of hazardous materials
using evidence  gathered during the walk-over.

        If the landfill is inactive or abandoned, both historical and newly-acquired aerial photographs
 are invaluable in inventorying and evaluating the site. Historical aerial photographs are particularly
 useful in inactive  waste site investigations.  They are not only useful in compiling land use/cover,
 environmental, and other physical site-specific data, but also in directing ground investigation teams
 to exact locations for conducting drilling and leachate sampling operations.

        Various state and federal agencies have photographed most of the United States in recent
 years. As discussed by Erb et al. (19), at least one photograph exists for any land area within the
 continental United States. Most of the federal photographic data has been cataloged by the U.S.
 Geological Survey's National Cartographic  Information Center (507  National Center, 12201  Sunrise
 Valley  Drive, Reston,  VA 22092,  703-860-6045).
 Screening  Techniques

        Ground Penetrating Radar fGPR) and Magnetometers ~ On many landfills, particularly those
 located on industrial plant sites, drums are the most common solid waste. Concerns about public
 and occupational health and safety may make  investigators hesitant to directly drill or dig into
 buried drum disposal sites.  As described earlier,  GPR and magnetometry are two nonintrusive
 methods that can be used to locate pockets of drums or individual drums prior to sampling.

        Landfill  Gas and Emission  Gas ~ Landfill-gas and gaseous-emission surveys can be used to
 focus sampling efforts. Elevated levels of VOCs in samples collected from landfill-gas probes or
 flux chambers can  be indicative of subsurface zones contaminated with VOCs.  Sampling for
 isolation and remediation of these contaminated zones can be guided by the data collected during
 these surveys (12).
 Sample Collection

        Excavation—Once the area to be sampled has been identified by nonintrusive methods, a set
 of actual  representative samples of the debris present in the landfill is collected.  Common sense
 and experience are key ingredients in effectively and safely obtaining samples from landfills and
 waste piles. The hydrogeologic setting will also influence sampling device selection.

        Various  excavation procedures are used  to bring  subsurface refuse to  the surface for
 sampling.  Test pits give a more representative cross-sectional view of the landfill  structure than can
 be obtained from discrete samples or cuttings from borings.  The wider cross section exposed and
 the  greater amount of material available for  sampling can  give the site investigator greater
 confidence that the samples collected and analyzed are representative of the overall composition
 of the subsurface material.  Transect trenches can be dug to systematically  sample the site. Hill and
 Montgomery (7) have developed an  approach for installing test pits at hazardous waste landfill  sites.
Prior to test pit excavation, the cover material is removed and  segregated by  a  bulldozer.  This
material can later be used as a cap after the pit is backfilled.  Test pits are generally excavated with
 a backhoe, which limits the depth of investigation to approximately 20 feet. The  authors  have used
backhoes  equipped with custom-built arms or draglines for excavating below 20  feet, but find that
hourly  equipment costs increase substantially with depth. Hill and  Montgomery (7)  have  found that
placing excavated  wastes on plastic sheets to prevent migration of contaminants is awkward and


ineffective, so in their work the waste material is placed on stripped areas to avoid inadvertent
contamination of the surface material. Backhoes dump the waste out and the material is physically
characterized by field workers. The material is photographed at all phases of the operation. The
types of materials and their frequency distribution within each load is determined. The amount of
water (liquid) is determined.  Elaborate and complete documentation by the sampling personnel
is essential.

       Ideally, the location of drums on the site has been predetermined by GPR, magnetrometry,
or  other appropriate remote interrogation procedures.  However, unexpected encounters with
randomly scattered drums  will inevitably occur.  The backhoe operator should be experienced with
drum detection and retrieval. Overpacks for containing ruptured drums  and a  small pump and/or
adsorbent for recovering releases should be immediately available. Large-scale drum removal and
containment is best handled by qualified removal contractors as an emergency removal or as part
of site remediation activities. For further guidance on drum excavation and removal, the reader
is referred to a  separate document on drum-handling practices  at hazardous waste sites (1).

       Drilling —When refuse depths exceed 20 feet,  drilling is another technique for acquiring
subsurface refuse  samples.  Horizontal drilling has also been  used to  obtain material  from the
interior of waste piles.  Drilling can be safely performed  on drum disposal sites by the proper
utilization and interpretation of GPR or magnetic methods  (8).  Typically, a truck-mounted rotary
drill rig similar to those used for ground-water monitoring well installation is used for refuse drilling.
Air rotary equipment should be avoided, especially when VOCs are present, because it increases
the release of airborne contaminants.  Hollow-flight augers  are the most effective for refuse
sampling in landfills. A continuous  cased core is collected which  allows for identification of perched
liquids. When hollow-stem augers are used, the cores can be field-screened by XRF or hand-held
organic vapor detectors for the presence of heavy metals or VOCs, respectively.

       Water does not necessarily have to be introduced into the borehole unless there is a concern
over the  release of potentially harmful contaminants into the  ambient air. In such an instance, clean
water can be introduced to prevent venting of hazardous landfill gases (it should be borne in mind
that this  can cause cross-contamination between cells of the landfill).

       Hill and Montgomery (7) estimate that up to 25% of drilling footage  in landfills will be
terminated  due to encounters with  impenetrable zones,  such  as  automobiles, refrigerators,
transformers,  and  the like. Appropriate adjustments in the projected cost of drilling should be

       Rathje (20,21) collects samples from municipal landfills for anthropological studies using
hollow-stem augers.  The contents  are unloaded onto sheets of plywood  for characterization. The
items on the plywood are placed into plastic bags for later sorting off-site. This may be acceptable
for municipal landfills that supposedly contain  no hazardous constituents. However,  Baker et al.
(12) have found no significant difference in the amount of hazardous VOCs present in landfill gases
from hazardous and reportedly nonhazardous landfills.

       Manual coring devices can  sometimes be used to collect  subsurface samples on landfills or
within waste piles. These devices, commonly used for soil sampling, include split-spoon samplers,
Shelby tubes, and zero contamination samplers. Dry solids are best sampled with a concentric tube
thief. These various devices are much smaller in  diameter than hollow-stem augers and will


 accordingly obtain smaller samples.  The smaller sample size may not accurately represent the
 diverse material present in the landfill. Only materials that fit inside the sampler will be retrieved.
 However, these devices may be suitable for sampling the landfill cover material, which usually
 consists of a fairly homogeneous layer of soil. For obtaining estimates of VOCs present in the near
 surface materials, intact subsamples may be removed from the coring devices with small-diameter
 hand-held coring devices. The contents of the hand-held corer are extruded directly into  40-mL
 glass volatile organic analysis (VOA) vials for subsequent purge-and-trap analysis (22). Several
 samples can be composited in a jar containing purge-and-trap grade methanol to obtain a more
 spatially representative estimate of VOC concentrations in the near surface matrix. Alternatively,
 samples from the split-spoon sampler can be sectioned and placed in wide mouth jars containing
 methanol (22,23). The  methanol  serves to preserve the sample by reducing losses of VOCs by
 volatilization (it should be noted that this is not currently a standard procedure; it is a draft ASTM
 procedure introduced as an alternative to Method 8240).

       Crude  Sampling —Finally, near-surface material can be exposed using  crude sampling
 implements such as shovels, picks, and trowels.  These procedures are suitable for metals and
 semivolatile compounds, but will cause losses of VOCs due to excessive disturbance.
Sample Handling

       Sieving—Investigators sampling debris at landfills frequently separate materials by  size,
although it should be borne in mind during project planning that the contaminants are not likely
to be uniformly distributed among the size fractions. The material exhumed by a backhoe, auger,
corer, or crude sampling device can be passed through a sieve. For estimation of bulk components,
the smaller size fraction may  be excluded  from the sample.   Wilson and Rathje (21),  in
characterizing  municipal landfills for anthropological studies, passed the material obtained using a
backhoe through a Vi-inch mesh. The material passing through the screen was discarded while the
retained portion was sorted and identified. Usually at hazardous waste landfills, information about
contamination associated with a range of size classes is needed for remedial planning. The smaller
items are more easily handled by the laboratory, are more amenable to  a variety of remedial
techniques,  and, due to their  greater  surface-to-volume  ratios,  can  potentially  leach more
contaminants  into the landfill. Thus, the finer material that passes through the mesh can  be
subsampled with a corer or  scooped into bags  or sample  containers for analysis.  The retained
fraction can then be placed in a sample container for subsequent leaching  or extraction procedures.
Items too large to fit into a container can be wipe tested. In a removal  program, sieving is generally
only conducted when preparing soil samples for x-ray fluorescence (XRF) field screening. A 20-
mesh screen size is recommended for this purpose. Samples  taken for VOA should not be sieved.

       Determining which items to exclude from the sample is a difficult issue to address. Not all
items contribute to the hazard posed by the landfill site. Inclusion  of all items may introduce error
in the sampling and analytical procedures. Examples  of extraneous material may include twigs,
leaves, or pieces of glass. In a very ad hoc manner, some investigators exclude all items larger  than
3 inches, i.e., "too big to fit in the jar," from the sample. Objects larger than % inches were excluded
from samples to be used for a bench-scale  stabilization treatability study (24). However, some large
items are essential contributors to the hazardous nature of a sample or site. For example, battery
casings may release considerable amounts of lead and should therefore be included as part of the

 sample.  Laboratory constraints must be  known before the  project planners specify  collecting
 samples  containing large items  for analysis.

       Comminution and Homogenization -Some investigators have obtained homogeneous samples
 from a mixed matrix of refuse by comminution and homogenization of the material. Commercially
 available shredders can be used for this purpose. The American Society for Testing and Methods
 (ASTM) has standard procedures for determining the efficiency of refuse size-reduction  equipment
 (25). However, no guidance is offered for exactly what items may be fed through such equipment.
 Obviously,  not all items are amenable to comminution, either because of their  size,  particular
 analytical concerns, or safety considerations. The drawbacks to comminuting heterogeneous wastes
 are enumerated below, under "Comminution of drum contents." If comminution will be  performed
 on the site, waste incompatibility should be determined first.  Hatayama et al. (26) have provided
 guidance on waste incompatibilities  that can be useful during  waste comminution processes. These
 authors have developed a "Hazardous Waste Compatibility Chart" that allows the user to evaluate
 potential adverse reactions for binary combinations of hazardous wastes. Binary waste combinations
 are evaluated in terms of the following reactions:   heat generation from a chemical reaction, fire,
 toxic gas generation, flammable gas  generation, explosion, violent polymerization, and solubilization
 of toxic substances.

        Sorting and Segregation —After refuse has been passed through a sieve, the retained items
 may be further separated by sorting or segregating by material type, size, or hazard class. Sieving
 may not necessarily precede sorting.  There may be a preponderance of large items that would
 make sieving impractical.

       The ultimate disposal or remediation approaches under consideration dictate how debris is
 sorted (see discussion on establishing waste populations, Chapter 3). The treatability of the waste
 should be determined early in the process and should be considered when sorting heterogeneous
 materials. Samples may be sorted to obtain representative material to determine the efficiency of
 incineration, for example. Treatment or disposal options for  specific waste types are covered in
 detail  in numerous reports.  The EPA (27) prepared a useful summary of major treatment and
 disposal options for various waste categories.

       The  presence  of hazardous  substances in the sorted  debris can be tested on site using
 numerous field screening methods  (28). These include XRF analysis for heavy metals, vapor
 monitoring or field GC analysis for VOCs, and high-vapor-pressure semivolatiles. Field test kits
 are available for screening-level analysis  of PCBs. However, the  EPA  requires that samples of
 record be laboratory  tested for PCBs before they  are combined  or  composited. Testing  to
 determine gross halogen content can sometimes be foregone if all insoluble  wastes  are to be
 incinerated at a facility capable of handling chlorinated organics.  However, testing for PCBs is
 required, regardless of the need for  testing other halogenated compounds.

       Containerized liquids within the sorted material can be tested for compatibility by mixing
 small samples and making visual observations for precipitation, temperature changes, or phase
 separations.  After  compatibility has been determined, if the debris is uncontained, it is often put
 into  drums for removal and subsequent disposal or treatment. The drums should be inventoried
 and labelled as they are filled in order to  avoid resorting the drums at the disposal or  treatment

        Large-scale Extraction --On-site batch extractions of bulky, heterogeneous wastes may give
 the most reasonable representations of the hazards posed by these materials. Such tests are not
 often performed. They are most appropriate when contaminant leaching from the surfaces of large
 objects is the major concern. Sorted and unsorted waste is weighed and placed in a 55-gallon drum.
 Water or another leaching fluid is added to the drum and the drum is placed on a drum roller. A
 batch leaching test of fixed duration is performed on the material in the drum. The leachate is then
 collected and aliquots are analyzed for the compounds of interest.  The remaining fluid may have
 to be drummed and treated as hazardous waste.  Such tests  are not standard, and their use must
 be negotiated with the regulatory agency during project planning.

       Wipe Tests —The hazardous chemicals associated with the exterior surfaces of large items
 are also commonly characterized by wipe tests.  In these tests, a unit surface area (usually 0.1 or
 0.5 m2) is wiped with a dry or solvent-moistened swab, and the swab is then extracted and analyzed.
 Wipe test results give no information on total contaminant level and are difficult to interpret with
 respect to health or environmental risk.  Wipe tests are frequently used on concrete surfaces.
 However, concrete is fairly porous, and contaminants  from the interior of the material can migrate
 outward after the wipe test has been performed. Some investigators no longer use wipe tests but
 rather disaggregate the concrete into small chunks with a jackhammer, and then perform a TCLP
 extraction on ground-up  chunks of concrete. Leaching tests of large items the size of transformers
 are being developed by the EPA. The efficacy of the washing  (leaching) procedure is evaluated by
 before-and-after wipe tests of surfaces on the debris (29). One of the problems with wipe tests at
 present is that there  are no rules of thumb for appropriate action limits in terms of concentration
 per unit surface area.

        Containerizing —If representative samples cannot be effectively screened in the field, then
 subsamples must be returned to the laboratory for analysis. A wide assortment of containers are
 commercially available for sample  collection. The type of container selected is dependent upon the
 size of the item(s) and the type of contaminants to be analyzed. Wide-mouth glass jars, typically
 125-mL  or 250-mL, are used for organic analytes. Immersion of solid samples into wide-mouth jars
 containing methanol has  been shown to reduce losses  of VOCs (22). Five-gallon ORME pails are
 commonly used to obtain treatability samples.

       Sample shipping  —samples collected in the field may be composed of hazardous material or
 have a component that is hazardous. The samples are  often transported to off-site laboratories for
 analysis.  Therefore,  there should always be a person in the field who has  experience with DOT
 regulations for shipping of hazardous  materials (30). Samples immersed in methanol are required
 to be shipped as a "Flammable Liquid." Documents accompanying the shipment must alert the
 laboratory to the anticipated hazard level of the samples.
                                 Field Screening Methods

Existing Methods

       Many field screening procedures are useful for determining the presence  of various
contaminants in landfills, waste pales, or drums. A review of all available procedures is  beyond the
scope of this document, but a brief description of a few procedures follows. For more information,

the proceedings from the first and second international symposiums on "Field Screening Methods
for Hazardous Wastes and Toxic Chemicals" (31,32) as well as the "Field  Screening Methods
Catalog: User's Guide" (28) should be consulted.  A new version of the latter document is  in
preparation and will be called the "Field Analytical Methods  Catalog"  (33). As  planned, the
document addresses analyses for volatile organics in  water, soil and sediment, soil-gas, and air;
semivolatile organics; pesticides; polychlorinated biphenyls; polynuclear aromatic hydrocarbons;
inorganic; and classical analyses.

       X-Rav Fluorescence —Toxic metals can be screened using a field portable XRF unit. The
XRF has a limited depth of penetration,  i.e., a few millimeters.  The depth  of penetration  is
dependent upon the matrix. Various components of a waste pile may be scanned to determine toxic
elemental concentrations  on the exterior of the pile. XRF  screening can be used in conjunction
with sorting to isolate components of a waste pile that exhibit elevated levels  of a particular toxic

       Organic Vapor Analyzers/Gas Chromatographs  -Organic vapor analyzers (OVAs) have been
used for many years in the industrial  workplace. OVAs are usually equipped with a photoionization
detector (PID). Organic chemicals whose volatility is sufficient to allow their partitioning into the
ambient air can be detected.  The PID  is somewhat limited as to the types of compounds it can
detect.   Most aromatic hydrocarbons  (e.g.,  benzene, ethyl benzene, toluene, and xylene) are
detectable using a PID. However, only certain halogenated VOCs, such  as  trichloroethene and
perchloroethene, can be measured.  Only total hydrocarbon concentrations are measured using the
OVA. No information is obtained  regarding the exact identity of compounds present. The level
of organic contaminants that  can be  detected varies with environmental conditions such as wind
speed and direction, humidity, rainfall, and temperature. OVAs can also be used to scan liquid and
solid samples  obtained from the interior of the pile by intrusive sampling methods.

       Field portable gas chromatography (GCs) are gaining recognition as a viable means  of
identifying  and quantifying VOCs and some semivolatile  organic compounds. A range of detectors
are available, allowing detection of a broad assortment of VOCs and semivolatiles. Liquid and solid
samples  can be screened by headspace  analysis (28). More recently, field portable GCs equipped
with mass spectrometers are being used on hazardous waste sites for confirmational analysis (34).

       Ion Mobility Spectrometers —The ion mobility spectrometer has been  used in the past for
detection of narcotics and  explosives. Recently, it has demonstrated its usefulness for screening  of
hazardous waste sites for volatile compounds. It is capable of detecting many different classes  of
compounds, ranging from  amines to acids. It can be used as a class detector (e.g., esters, ketones)
or to identify specific chemicals, The high sampling flow rate, usually measured at several liters per
minute, makes it ideal for "sniffing" around piles and dumps for the presence of a diversity  of
hazardous materials.

       Cone Penetrometer —Cone  penetrometers  were originally designed to  assess  soil strength
properties for railroad grades and unpaved roadways. Soil strength is determined by measuring the
resistance developed on the cone tip and on a fixed area (sleeve) of the rod behind the tip. The
tip resistance and sleeve friction vary with grain size of the soil and the degree of compaction or
cementation.   Penetrometers  are becoming increasingly popular as hazardous waste sampling
devices because they combine speed and versatility with a degree of safety  that is not available in
conventional drill-and-sample operations. Sensors  are now available for measurement of electrical


 conductivity of soil, natural radioactivity, and soil optical properties (fluorescence  and reflectance).
 The penetrometer, coupled with  the  appropriate sensors,  has  been used  to locate and  track
 fluorescent tracer dyes and waste oil and fuel in natural soils and landfills (35). The penetrometer
 can provide  information regarding the density  of the interior of the pile  and  the presence of
 hazardous  chemicals.

        Open-path  FT-IR —Open-path FT-IR  is useful  for the  qualitative and quantitative
 measurement of VOCs  and low-boiling semivolatile organic  compounds in the ambient air
 surrounding  the surface of  hazardous waste  piles.   Though  this technology  is sensitive to
 meteorological conditions such as wind, particulate matter, humidity, and rain, most of these also
 affect point sampling by canister methods as well.
New or Developing Technologies

       Fiber-Optic Chemical Sensors fFOCS) —These allow in situ monitoring of ground waters.
This  developing technology is based  on chemically selective  optical  fiber transducers.  The
chemically modified optical  sensors can be configured for either single or multiple component
characterization (36,37). Optrodes can provide rapid,  sensitive measurement of temperature, pH,
anions, cations, and organic pollutants. Sensors for pH, Ca(II), Pb(II), Se(IV), As(III), Hg(II), CN~,
O2, CO2, CHC13, CC14, C5H6> perchloroethene, trichloroethene, and benzo(a)pyrene have been or
are being  developed.    Miniature,  fully integrated  remote-sensing  modules  that incorporate
chemically selective, microporous membranes on optical fiber bundles with laser excitation sources,
fluorescence detection,  and a microprocessor transmitter/receiver for  telemetric  surveillance are
being developed. These systems  will provide accurate sampling with lower analysis costs and greatly
reduced sample handling requirements.  FOC sensors have the  same drawback as many  other
sampling/sensing techniques:  they yield point measurements, which must be appropriately
integrated to achieve representativeness.

       Raman Spectrometry—Surface enhanced Raman scattering (SERS) has shown improved
selectivity and sensitivity for detection and screening of trace level contaminants in ground water.
The narrower  Raman bands simplify identification of individual  components in complex mixtures.
The feasibility of utilizing SERS in harsh environments has been demonstrated, and new substrate
and optrode designs are being developed.

       Laser  Ablation  rLAVInductively  Coupled  Plasma (1CP) Atomic  Spectrometry —The
technique involves the generation of aerosols from solids and the  introduction of these aerosols into
a high temperature (~6,000-10,OOOK) argon inductively coupled plasma. At the high temperatures,
the sample is dissociated and atomized.  The free atoms  undergo excitation and ionization to
provide characteristic optical  and mass spectra of the  constituent atoms and/or ions. TCP-Atomic
Emission  Spectrometry (AES)  is  currently the most widely used analytical technique for the
detection of more than 70 elements. LA-ICP-AES provides a means of  analyzing solids directly by
sweeping the ablated particles into the TCP. The advantage of  laser ablation sampling is that it
eliminates  or minimizes the need for routine sampling, labelling, transportation, sample preparation,
analysis, and disposal of the  sampled material. It is also applicable to nonconductors as well as
conductive materials. LA-ICP-AES is  currently being developed as a method for remote  field
applications.  Small volumes  of liquids  can also be efficiently  sampled using a  direct injection

nebulizer,  and then analyzed  using  remote ICP-AES.   ICP-mass spectrometry (MS) provides
detection capabilities approximately three orders of magnitude better than ICP-AES.  Mass analysis
provides necessary elemental isotopic  determinations. ICP-MS instruments, which use laser ablation
for sample introduction, are commercially available, and field portable mass spectrometers are being
developed. ICP-Laser Excited Atomic Fluorescence Spectrometry (LEAFS)  can provide additional
selectivity that minimizes or eliminates spectral interferences for direct isotopic analyses in complex
sample matrices. Advances in diode lasers will make LA-ICP-LEAFS an attractive isotope specific
field  technique.

        Laser Induced Breakdown  Spectroscopy (IIBSV-Focusing a  high  intensity  laser pulse in
liquid, gas, or solid samples can produce a spark. The spark is a result of dielectric  breakdown of
the medium induced by the strong  electric fields of the pulse. The high temperature spark plasma
reduces the material to elemental  form and also excites the species. Emitting ions, atoms,  and
simple molecules are identified by  their characteristic spectral lines. LIBS has been  used to detect
atoms in vapors, aerosols, and particles. LIBS has also been used to determine uranium in solution
for possible application to process control in nuclear fuel reprocessing facilities, Real-time analysis
is possible because the laser spark both vaporizes and excites the sample.  The technique is  also
non-intrusive in the sense that only optical access to and from the sampled medium is required.

        Laser Induced Fluorescence  fLIF) -This technique uses relatively low power laser excitation
transmitted through fiber optic cables to samples, and the collection of the  resulting fluorescence
emitted by the species,  again through fiber  optics, for selective spectrochemical  analysis.  The
selectivity of the technique is  enhanced by the capability of generating excitation  and emission
spectra, as well as by utilizing time  resolution.   The technique is currently  targeted at  the
determination of polycyclic aromatic hydrocarbons (PAHs) in ground water,  gases, and soils. A
rugged, field transportable tunable  dye laser system has been tested. LIF has been tested for PAH
determinations using cone penetrometer technology.
                   Characterization of Contained Heterogeneous  Wastes
       In characterizing  contained wastes, the preferred approach is to minimize intrusion into the
waste and minimize waste handling, to curtail risks to personnel. For materials contaminated with
transuranic  wastes, the preferred hierarchy  of waste characterization approaches is to identify
container contents and contaminants from records and knowledge of the original waste stream; then
perform non-intrusive interrogation; next make a minimal intrusion for headspace sampling; and
finally, if absolutely essential, perform intrusive sampling. An alternate approach that is always
available is to minimize the waste characterization needs by specifying a very robust waste treatment
process that could handle all possible materials and contaminants.  This would eliminate the costs
and risks of a full characterization study. If intrusive sampling is required, it may be possible to sort
containers based on non-intrusive or minimally intrusive techniques and more fully  characterize
each population.  The statistical aspects of this approach are discussed in Chapter 3.

Non-Intrusive Techniques

Real Time Radiography

        Real time radiography permits an operator to remotely observe the contents of a container
using x-rays. Excellent resolution is obtained and it is possible to detect free and contained liquids,
metals,  individual objects, granular material, aerosol cans, gas cylinders,  wood, etc. The resolution
is often good enough to determine whether an aerosol can has been punctured and to detect a few
milliliters of liquid in a 55-gallon drum. The devices are commercially available and are estimated
to cost  $1.5- 1.7 M for a box (4x4x8 ft) size unit including training and installation. With digital
image  processing/x-ray  tomography, the  device  is better able to  distinguish between  metals
depending on their density. However, at high densities, differentiation becomes more difficult and
time consuming.

        Sonar  scans  are  frequently  used  to determine  drum integrity as part  of real  time
Radiation Measurement Instrumentation

       Measurements of gamma rays and neutron emissions enable non-intrusive determinations
of nuclear materials contained in drums. Radiation instrumentation is based on detection of alpha
and beta particles, gamma rays, and neutron emissions. Alpha and beta particles will not penetrate
the wall  of a drum and therefore have limited application in screening the contents within a drum.
Measurements of gamma and neutron radiation allow quick screening of the interior contents of
a drum.   Gamma and  neutron  measuring systems are  classified as passive or active,  Passive
measurements depend on measurement of intrinsic radiation emitted by decay and  interaction of
materials within a sample such as gamma emission following decay, spontaneous fission neutron
emission, and  (alpha, neutron) reactions. Active measurements induce a reaction in the sample that
produces measurable radiation such as using an x-ray generator to measure the edge  of a material,
and neutron source to produce fissions  in nuclear material or as elemental  analysis through neutron
activation analysis.

       Gamma-Ray Assay Methods —Gamma-ray assays rely on either low-resolution detectors such
as Nal that are unable to resolve individual gamma-ray peaks or high-resolution detectors such as
intrinsic  germanium that allow sufficient resolution to identify specific radionuclides.  Since gamma
rays are  directional and easy to collimate and filter,  they can be used to determine the precise
location  and distribution of radionuclides (38). Segmented gamma-ray scanners involve the use of
collimated detectors to examine multiple segments or pancake-shaped sections of a drum. They can
be used for screening or for quantitative detection and location of nuclear materials such as 235U
and 239Pu in containers up to 55 gal.

       Neutron Assay Methods—Neutron techniques  generally measure  large  samples,  in
comparison to gamma-ray methods, and have the advantage of good penetration and application
to a wide variety of samples (38). Both passive and active techniques exist and may use the same
detectors. Several neutron emission screening devices  exist including:

       Portable hand-held total neutron monitors which can be used to detect neutrons
       from spontaneous fissions (i.e., plutonium, californium, curium and other gamma
       and neutron emitters)

       Passive neutron coincidence counters which measure containers ranging in size up
       to 55 gal. drums to determine the presence of spontaneous fissions from nuclear
       materials such as plutonium

       Active well coincidence counters that use a random neutron source (i.e., Am, Li) to
       produce a fission  in  nuclear materials such  as  uranium for  measurement by
       coincidence  counting

       Combined passive and active coincidence counters which allow measurement of
       uranium and plutonium plus transuranic elements

       Active neutron interrogation which uses an external source such as 252Cf for sample
       irradiation and delayed neutron  counting or a deuterium-tritium (DT) neutron
       generator for differential die-away assay of nuclear materials and transuranics in a
       drum. These systems have sensitivities below 100 nCi/g of transuranics

       Passive coincidence counting and active measurements can be combined to enable
       one system to provide both passive and active assay,  such as the  active well
       coincidence counter, Shuffler, or DT system

       Neutron activation analysis (NAA)  is a non-invasive elemental analysis technique
       potentially suitable for materials contained in drums, including non-radioactive
       materials. It uses  high resolution gamma-ray measurements with active neutron
       systems  such  as the  Shuffler and Differential Die-away systems.    Additional
       development is  required to determine the overall capabilities for elemental detection
       and to determine what sensitivities are achievable. The prompt gamma neutron
       activation probe is very sensitive to chlorine, and may provide a non-intrusive
       technique to assay for chlorinated solvent vapors in the  headspace of drums.
       Chapter 6 describes the potential applicability of NAA techniques in more detail.
       Quantitative and semi-quantitative assays using radiation-based techniques are dependent
on  appropriate calibration,  development of reference  standards, and a measurement  control
program that includes operator training.

       Metal salts, powdered metals, the identity of liquids, and the presence of vapors within waste
containers cannot be detected remotely at this time.

Minimally Intrusive Techniques

       Headspace analysis is a  minimally intrusive technique that can be used to determine the
presence of volatile compounds and explosive conditions. Drums may have to be opened remotely
or using non-sparking tools  (1) if combustible mixtures may be present. The gas may then be
analyzed using a field portable explosimeter, organic vapor analyzer, vapor detection tubes, radiation
detection device, or gas chromatography. Alternatively,  headspace gas samples may be collected for
off-site analysis using canisters, gas bags, or sorbent traps.

       For drums containing radioactive  materials, self-sealing sampling ports are available to
minimize the potential loss of contents.  While radioactive mixed waste may be double and triple
bagged, it is expected that diffusion will occur through almost any container so that the presence
of volatile and some semivolatile compounds could  be determined.  In  some cases, it may be
possible to overpack a drum and, after some time, sample the headspace of the  overpack to detect
volatiles being emitted from the  drum of interest. Research on the feasibility of such an approach
is recommended.
Intrusive Container Sampling

Sort Drum Contents

       One method of handling containerized heterogeneous waste begins with sorting the contents
by material type. In other words, if the container holds clothing, rags, wood, plastic, metal, and
paper, the contents are removed and placed in various piles: clothing, rags, and paper in one pile,
wood in  a second pile, plastic in a third pile, and metal in a fourth pile. The next container is then
opened and the procedure repeated, adding to the piles of material from the first container.  The
process is repeated until all containers have been emptied and all the contents sorted.

       After sorting is complete, each  pile of material is weighed and  sampled. Wood can be
sampled  by sawing off small pieces from each item. Nonporous material can be rinsed with water
or solvent to extract surface contaminants. Materials that have only surface contamination can be
wipe sampled. Soft material can be  sampled in one of several ways:

•      Cut out pieces from each article  using scissors or shears until enough material is obtained
       for a sample.
•      Punch holes in the material using something similar to a large diameter paper punch.
•      Place the material back into one of the containers. Obtain a core sample using a hand-held
       coring device.

       The samples taken from each material  type are then analyzed for the contaminants of

       An advantage of this approach is that populations are established and contaminant levels
are determined for each material type,  not just drum  by drum.  A major  disadvantage is the
difficulty  of collecting "representative" samples, if that is a goal of the project. Within each material

type there is likely to be a wide range of contamination among individual items. Three possible
ways to deal with this variation are:

•      collect the entire pile of material, or a large portion of it, as the sample;
•      collect a number of equivalent samples from the pile to allow a statistical estimation of the
       range and mean of contaminant concentrations among items;
•      deliberately sample the most contaminated objects, if they are evident visually or by field
       screening.  This biased approach will yield an estimate  of the maximum contamination level
       for each material type.

Which of these (very different) approaches is appropriate is decided during project planning (see
Chapter 3).
Sort Drums Using Process Knowledge

       Another method of handling containers before sampling  is to sort them by hazardous
constituent based on knowledge of the process that generated the wastes. An  example of this
approach is  the system used at DOE's Fernald Environmental Management Project (FEMP) in
Ohio. There, all drums are labeled with a Lot Marking System Number (Lot No.). This number
is a 15-digit alphanumeric code that consists of five parts.  One of the parts  (Material  Type)
describes the material contained in the drum. Another part (Source Code) identifies the equipment
and process that produced the material. Together these codes are used to identify "waste streams."
A waste stream is contained in a group of drums that are labeled with the same Material Type and
Source Code. For  characterization purposes,  a waste stream constitutes  a single population of
interest. All  of the materials  in one waste stream are either RCRA hazardous for the  same
constituent(s) or non-hazardous.  Also,  all of the materials in a waste stream are  either high-level
radioactive waste,  low-level waste, or not radioactively contaminated.

       A waste stream that is identified using process knowledge and the container markings may
be split into several waste  streams  when  more information becomes available. For instance,
analytical data may indicate that a waste stream actually has more than one material. One may be
RCRA-hazardous for lead and another may be non-hazardous.  If new information indicates that
a waste stream actually consists of several different waste streams, then the newly identified waste
streams (populations) must be handled separately.

       Each  waste stream is statistically sampled  to reduce the total amount of sampling and
analysis required to identify the hazardous  constituents (39). At the FEMP, the approach taken
is biased toward  sampling maximally-contaminated items,  since the goal is to identify the
contaminants present and  ascertain whether any  of the drummed  objects  could be RCRA-
hazardous. Field personnel seek out items with stains or surface residues. The list below includes
common materials in a heterogeneous waste stream and sampling methods commonly used for these
materials at  the FEMP.

•      Clothing, paper, rags, plastic bags, plastic sheets - These materials are sampled using shears
       or scissors.  Sections of the matrix that appear to be contaminated or stained are cut or torn
       off in small pieces and placed in the sample container.

                           Treatment After Minimal Evaluation

       An alternative to intrusive sampling of heterogeneous material is to assume that the material
 is hazardous and treat it in a plasma arc furnace.  This furnace contains large crucibles that are
 heated by plasma jets. The crucibles are large enough to accommodate entire drums. Therefore,
 the drums do not have to be opened.  This reduces exposure of personnel to hazardous materials.

       Drums of waste are fed into the crucible where they are melted and mixed by the plasma
jets. This homogenizes the waste as well as treating it so that it will no longer be hazardous. The
 temperature is high enough to destroy organic contaminants.  If there is enough silica present, the
 resulting material is a very  stable glassy slag. Silica can be added to the melt if there is not enough
 in the waste. Once the glass is  solidified, any  hazardous  metals in the material are no longer

       The melt can be sampled before or after it solidifies.  It can be solidified in several forms,
 including pieces  small enough for the  TCLP analysis.  Since the material is now homogeneous, the
 problem of sampling and analyzing heterogeneous  material no longer exists.

       There are two limitations to this process. First, drums that are full of liquid cannot be put
 into the  furnace because of the  potential for steam explosions. This can  be prevented  through
 screening the drums  by  process knowledge or non-intrusive  techniques. Drums that are full of
 liquid can be treated by feeding the liquid in slowly instead of 55 gallons all at once.

       The second limitation is that there are currently only two plasma arc furnaces in operation
 that can handle hazardous, non-radioactive waste, and only one that  can  handle  hazardous,
 radioactive waste, Consequently, there is a limited  capacity  for this process. In addition, state
 governments require  that wastes crossing their  borders be fully characterized. This  defeats the
 purpose  of using this  method and leaves the problem of characterizing heterogeneous waste.

  1.    Additional plasma arc furnaces  are  needed.    In  addition, state  agencies  should be
       approached with information on the efficacy of the plasma arc furnace technology. Work
       needs to be done to change the restrictions on the shipment of waste to these facilities and
       to develop portable units.

  2.    Research  is needed to develop techniques to identify metals and organic  compounds
       remotely, without opening containers. In particular, a  testing program should be conducted
       to assay the ability of prompt gamma-ray analysis to  detect chlorinated organic compounds
       in drums non-intrusively.

  3.    Further development and commercialization of new field analytical and screening techniques
       is  recommended. These techniques have the potential to greatly diminish the need for
       sample handling and, to some extent, to circumvent the question, "What should constitute
       a sample?" that is so difficult with heterogeneous materials.

  4.    Clear guidance is needed for appropriate chemical parameters to be monitored at hazardous
       waste landfills. There is currently a considerable waste of resources in analyzing for the
       entire priority pollutant list.   Enough information exists to select appropriate indicator
       compounds for landfill sampling and monitoring.

  5.    Tracing contaminant migration at landfills by standard soil-gas monitoring techniques  is not
       an accurate technique, because the partitioning of VOCs between the sorbed and gas phases
       is different in debris than in soil. Bench-scale experiments are recommended for common
       types of  wastes, to define the partitioning parameters, and provide guidance on the
       appropriate use of soil-gas monitoring at landfills.

  6.    The  development of one or more standard, large-scale field extraction tests is recommended.
       The appropriate leaching solvent for these tests may be site surface or ground water.

  7.    In specialized coring applications, the construction industry is now using very large hollow-
       stem augers that create boreholes up to 8 feet in diameter. Large objects can be recovered
       from within the borehole using tongs. While they are quite  costly to mobilize/demobilize,
       these "super  augers" might have utility in some landfill studies and should be investigated
       for this use.


 1.     U.S. Environmental Protection Agency.  1986. Drum handling practices at hazardous  waste
       sites. EPA/600/2-86/013, U.S. EPA, Cincinnati, OH, 177 pp.

2.     U.S. Environmental Protection Agency. 1987.  A  compendium of Superfund field operation
       methods.  EPA/540/P-87/001.

3.     U.S. Environmental Protection Agency. 1985. Guidance document for cleanup of surface
       tank and  drum sites,  OSWER Directive 9380.0-3. Office of Emergency and Remedial

4.     U.S. Environmental  Protection Agency. 1991.  Conducting remedial  investigations/feasibility
       studies for CERCLA municipal landfill sites. EPA/540/P-91/001. Office of Emergency and
       Remedial Response.

5.     Rice, J. M., M.L. Voorhees, and A.C. Ohehe. 1985. Use of  cell model in predicting liquid
       movements  and  levels in a  landfill site. In: Proc.  Superfund  '85  - Management  of
       Uncontrolled  Hazardous  Waste Sites, Nov.  1985,  Washington, DC,  pp  182-188.

6.     Montgomery, R.J., D.A.  Wierman, R.W. Taylor, and H.A. Koch, 1985. Use of downhole
       geophysical methods in  determining the internal structures of a large landfill. In:  Proc.
       Eighth  Annual Madison  Waste Conf.  1985, pp 559-569.

7.     Hill, J.A., and RJ. Montgomery. 1986. Considerations in data collection for evaluation of
       source  control alternatives  at hazardous waste landfills.    In:  Proc.  Superfund  '86  -

       Management of Uncontrolled Hazardous Waste Sites. Dec. 1-3,  1986. Washington, DC,
       pp 292-296.

8.      O'Hara, P. F., KJ. Bird, and W.A. Baughman.   1986. Exploratory drilling into a buried
       uncontrolled drum disposal pit.  In: Proc. Superfund '86 - Management of Uncontrolled
       Hazardous Waste Sites. Dec. 1-3, 1986. Washington, DC, pp 126-131.

9.      Plumb, R.H. 1987. A practical alternative to the RCRA organic indicator parameters.  In:
       Proc. HAZMACON 87, April 21-24, Santa Clara, CA, pp 135-150.

10.     Plumb, R.H.  1991. The  occurrence of Appendix IX organic constituents in disposal site
       ground water. Ground Water Monit. Rev., Spring 1991, pp 157-164.

11.     U.S. Environmental Protection Agency.   1985.  Superfund Public  Health  Assessment
       Manual. Draft for Office  of Emergency and Remedial Response, Office of Solid Waste and
       Emergency Response, Washington, DC.

12.     Baker, L., R Capouya, C. Cenci, R Crooks, and R Hwang. 1990. The Landfill Testing
       Program: Data  Analysis and Evaluation Guidelines.   California Air Pollution Control
       Officers Association Technical Review Group, Landfill Gas Subcommittee, Air Resources
       Board of California Publication, September 1990, 29 pp.

13.     U.S. Environmental Protection Agency.  1986. Measurement of gaseous  emission rates from
       land surfaces using an emission isolation flux chamber - User's Guide.  EPA/600/8-86/008,
       U.S. EPA, February 1986.

14.     Telford, W. M., L.P. Geldart, RE. Sheriff, and D.A. Keys. 1986. Applied Geophysics.
       Cambridge University Press, New York, NY.  860 pp.

15.     Sharma, P.V. 1978. Geophysical Methods in Geology. Elsevier, New York, NY. 428 pp.

16.     Olhoeft, GR 1989. Geophysical Adviser Expert System, Version 1.0.  EPA/600/4-89/023.
       EMSL-Las Vegas.

17.     Johnson,  W.J.,  and D.W.  Johnson.    1986.    Pitfalls  of geophysics  in characterizing
       underground hazardous  waste, In: Proc. Superfund  '86 - Management  of Uncontrolled
       Hazardous Waste Sites. Dec. 1-3, 1986. Washington, DC, pp 227-232.

18.     Davis, S, N, D.J. Campbell, H.W. Bentley, and T.J.  Flynn. 1985. Ground-water Tracers.
       National Water Well Association, Worthington,  OH.

19.     Erb, T.L., W.R.  Phillipson, W.L. Teng, and T. Liang. 1981. Analysis of historic airphotos.
       Photogrammetric Eng. Remote Sensing 47(9): 1-12.

20.     Rathje, W.L. 1991. Once and future landfills. Nat. Geog., May  1991, pp. 117-134.

21.     Wilson, D. C., and W.L.  Rathje.  1990. Modern middens. Natural History.  May  1990,
       pp. 54-58.


22.     Lewis, T.E., A.B. Crockett, R.L. Siegrist, and K. Zarrabi. 1991. Soil sampling and analysis
       for volatile organic compounds.  EPA/540/4-91/001,  Ground-Water  Issue,  U.S. EPA,
       EMSL-Las Vegas, 24 pp.

23.     Siegrist, R. L., and P.D. Jennsen.  1990. Evaluation of sampling method effects on volatile
       organic compound concentrations in contaminated soils. Env. Sci. Tech. 24:1387-1392.

24.     Rupp,  G.L. 1989. Bench scale fixation of soils from the Tacoma Tar Pits Superfund Site.
       Final Report. EPA/600/8-89/069, U.S. EPA, Las Vegas, NV.

25.     ASTM. 1988. Standard Test Method for Characterizing the Performance of Refuse Size-
       Reduction Equipment, Method E959-83, Vol. 11.04, Ann. Book of ASTM Standards, Sect.
       11- Water and Environmental Technology, Philadelphia, PA, pp 516-524.

26.     Hatayama, H.  K., E.R de Vera, B.P.  Simmons, RD.  Stephens, and D.L. Storm. 1980.
       Hazardous waste compatibility. In: Proc.  Sixth Ann. Research Symposium on Disposal of
       Hazardous Waste, EPA/600/9-80-010, U.S.  EPA, Cincinnati, OH, pp 21-28.

27.     U.S. Environmental Protection Agency.  1981. Draft. Design report for surficial cleanup
       and  disposal  of chemical waste at the pollution abatement services site, Oswego, NY,
       Prepared by Camp Dresser and McKee, Inc.

28.     U.S. Environmental Protection Agency. 1988. Field Screening Methods Catalog - User's
       Guide. EPA/540/2-88/005. Office of Emergency and Remedial Response, Washington,
       DC, 65 pp.

29.     Taylor,  M.L.  1990.  Assessment of chemical and physical methods for decontaminating
       buildings and debris at Superfund  sites.  In. Proc. 15th Ann. Research Symposium on
       Remedial Action,  Treatment,  and Disposal of  Hazardous Waste, EPA/600/9-90/006,
       Cincinnati, OH.

30.     49 CFR 1982. Code of Federal Regulations, 49,  Parts 100 to 177,  October 1982, 231 pp.

31.     U.S. Environmental  Protection Agency.  1988.  Field Screening Methods for  Hazardous
       Waste Site Investigations. Proceedings,  First International Symposium, Las Vegas, October
       11-13,  1988. Environmental  Monitoring Systems Laboratory-Las  Vegas.

32.     U.S. Environmental  Protection Agency.  1991.  Field Screening Methods for  Hazardous
       Waste  Site Investigations.    Proceedings, Second International Symposium, Las Vegas,
       February 12-14, 1991, Environmental Monitoring Systems Laboratory-Las Vegas.

33.     Fribush,  H.  1991.  Field  Analytical Methods Catalog. U.S. Environmental  Protection
       Agency, Office of Solid Waste and Emergency Response, in preparation.

34.     Robbat, A.,  and G Xyrafas.  1988. Evaluation of a field-based, mobile gas chromatograph-
       mass spectrometer for the identification and quantification of volatile organic compounds
       on EPA's hazardous substances list. In:  Proc. 1st Intern. Sym. Field Screening Methods for
       Hazardous Waste Site Investigations, October 11-13, 1988, Las Vegas, NV, pp  343-348.

35.     Lurk,  P.  W., S.S.  Cooper,  P.O.  Malone, and S.H. Lieberman.  1990.  Development  of
       innovative penetrometer systems for the detection and delineation of contaminated ground
       water and soil. In: Proc. Superfund '90, November 26-28, 1990, Washington, DC, pp 297-

36.     Janata, J. 1990.  Chemical Sensors. Anal. Chem.  62:33R-44R.

37.     Murphy, E. M, and D.D. Hostetler. 1989. Evaluation  of chemical sensors for in situ ground-
       water monitoring at the Hanford Site. PNL-6854, Battelle Pacific Northwest Lab, Richland,
       WA, March 1989. 70pp.

38.     Eccleston, G. W., M.P. Baker, W.R. Hansen, M.C.  Lucas, J.T. Markm,  and J.R. Philips.
       1990.  Application of Safeguards Technology in DOE's Environmental Restoration Program.
       Los Alamos National Laboratory, New Mexico, LA-UR-90-2410.

39.     Westinghouse Materials Company of Ohio.   1989. Sampling Plan for Drummed Waste at
       the FMPC. FMPC-2185. Procedures Dept, Westinghouse Environmental  Management Co.
       of Ohio, P.O. Box 398704, Cincinnati, OH, 45239-8704.

 Chapter 6
                  Analytical Laboratory Requirements

                     Clare Gerlach, Wayne McMahon, and James Poppiti

       The topic of the chemical analysis of heterogeneous wastes can be subdivided variously to
address any of several situations arising from the sampling of a complex debris site.  If the
heterogeneity is dealt with by the field samplers (i.e., physical segregation is done on site), the
samples received by the analytical laboratory may be homogeneous though they are substituents of
a vastly heterogeneous population.  Alternatively, physically diverse materials may be packaged
together  and submitted as  a unit to the laboratory.  The responsibility then falls on laboratory
personnel to decide whether to analyze the whole sample or to  separate or homogenize it before
performing the analysis,  Thus there are three basic  options for the laboratory in dealing  with
heterogeneous waste samples.

       Within these options additional levels of complexity are possible. Sometimes a multiphase
liquid sample is received, and obtaining a representative aliquot presents a problem.  Sometimes
a  solid precipitate introduces a  complication in  an otherwise homogeneous medium.   And
sometimes a large solid  object is submitted that may have small areas of highly concentrated
contaminants but, if pulverized and prepared as a  unit, weight concentration would yield data that
did not properly represent the  nature of the problem.

       There is need for thoughtful evaluation of samples and  their physical characteristics. An
important function of the  personnel engaged in sample  receipt, recording,  and handling is the extra
effort and careful attention to detail including documentation at this stage that can prove important
when data use questions arise later.

       Sample preparation procedures can range from simple extractions to complex methods for
separation and emulsification.  Novel procedures  for sample preparation are being developed but
are yet untested.  Homogenization of samples containing volatile organic compounds (VOCs) is not
recommended because of the high potential  for VOC loss during sample manipulation. Carefully
controlled headspace methods are often the best alternative for the analysis of VOCs.

       The presence of radioactivity in a sample imposes unique requirements on the analytical
laboratory. Whether or not radioactive contamination is present  is key knowledge needed for the
subsequent treatment of the sample and for the characterization and remediation of the site itself.
Information on sample radioactivity is also needed to define safe handling procedures and to
protect laboratory workers and the facilities from radioactive contamination and exposure. This is
part of DOE's ALARA principle, to keep radiation exposure "as low as reasonably achievable. "
Though samples should be screened for the presence of radiation before being  submitted  to a

 laboratory, there is widespread experience that often this is not done. A simple, quick radiation
 screening procedure would be very useful to the analytical lab. Presently, four procedures must be
 done to ensure that samples do not contain radiation: gross alpha, gross beta, liquid scintillation,
 and gamma scan. (This characterization is in addition to the health physics monitoring by probe
 and smear for alpha and beta/gamma activity normally done on radioactive samples upon receipt
 at a radiochemical laboratory.) This rigorous pre-testing is expensive and time-consuming.  If a
priori site knowledge gives the analyst reason to suspect radiation, safeguards must be taken,  If it
 is feasible to perform only one test, the recommended method is liquid scintillation.

        When sample preparation must be  customized for particular  sample types, as with
 heterogeneous samples, it  is critical that careful record  keeping be done to document the condition
 of the sample-as-received and to detail all procedures performed upon the sample.  The positions
 of sample  receivers, handlers,  and preparation personnel should be staffed with highly qualified
 personnel; if possible, these positions should be upgraded to reflect the  importance of these jobs.
                                     Project  Planning

 General Requirements

        The  effective and appropriate  planning  and management  of a  study to  analyze
 heterogeneous hazardous wastes demands a cooperative effort between project management, field
 scientists, and the  laboratory project manager.   This cooperation should begin with a carefully
 focused experimental design that is based upon realistic and agreed-upon DQOs, as described in
 Chapter 3.

        The need for close communication between client and laboratory personnel is especially
 pronounced  when the job entails heterogeneous  hazardous waste  characterization. Unconventional
 sample handling and preparation techniques  may  be required; thus a  laboratory representative
 should be a part of the project planning team.  Close attention to formulating achievable DQOs will
 save later trouble by clearly defining the specific needs of the particular analytical  regimen. It is
 essential that DQOs be  tailored to the project objectives; establishing DQOs that are too rigid
 merely  places an extra onus of responsibility on  the laboratory personnel. Communication with the
 project manager should not end at the planning stage. Indeed, continued contact and consultation
 is key if the proper decisions are to be made while allowing the most flexibility to  the analytical

        It is important that the analysis request be complete. The request should contain:

 •     A clear, complete description of the sample preparation, extraction, and  analysis procedures
       that  will be required, including detailed performance specifications.
 •     Documented reporting requirements.
 •     A listing of the required reference and performance  evaluation materials.
 •     Mechanisms for the laboratory to obtain technical assistance from the EPA in  implementing
       or gaining approval of method modifications or performing non-routine methods.

       When routine methods do not exist for a particular characterization, it is usually because
 lower detection limits are required, an unusual combination of chemicals is present, the sample
 matrix is complex, or the sample/analyte type is unique to a particular site.

       Analytical laboratories that wish to undertake the complex preparation and analysis  of
 heterogeneous hazardous wastes may need to procure special equipment ranging from  fusion
 crucibles to large volume centrifuges. Special mechanical reduction mills will often be needed, and
 they must be located in a different part of the laboratory from the routine sample preparation area
 (because of added dust  and vibration). For non-routine analyses, the  laboratory should have highly
 trained personnel and instrumentation that is  not dedicated to production work, especially if new
 methods or untested modifications are requested.
 Quality Assurance and Quality Control

       For the purpose of heterogeneous waste characterization, project specific QC protocols
 should be developed. Project planning should address the homogenization of samples in the field,
 if possible. If this is precluded by field conditions or sample complexity, the laboratory may attempt
 to  homogenize  the sample  to  obtain  a representative aliquot. Once  the  material  has been
 homogenized,   existing  laboratory  QA/QC  protocols  are readily  amenable  to sample

       There are certain minimum  criteria that must be met by  analytical laboratories charged with
 the  responsibility of generating environmental data to  support waste management and
 environmental investigations. Such criteria have been well documented for the analysis of waste.
 National consensus standards for environmental data operations have been developed by multiple
 organizations. For example,  the DOE has adopted  the  ANSI/ASME NQA-1 standard. EPA's
 Quality Assurance Management Staff (QAMS) has developed QA/QC program standards for its
 regional  offices  and its  research and  development  laboratories.   The Nuclear Regulatory
 Commission has  issued QA guidance for low-level waste disposal  sites, and the Department of
 Defense has developed different approaches for implementing QA/QC among the different services.
 In addition, the ASTM committee D-34 on Waste Disposal is developing a QA/QC standard. The
 intent of each program is to  assure the technical integrity of the data collection process and to
 assure all data are  of known quality. Presently  the American Society for Quality  Control is working
 with representatives of the various agencies that have developed QA/QC standards to meld key
 elements and concepts of NQA-1 and EPA QAMS  guidance so that conflicts between the various
 requirements can be resolved.

       The total  error associated with  waste  characterization is  the sum  of  population and
 measurement  errors.   The measurement  error associated with any data set  is  a composite of
 collection,  handling, transportation, subsampling, and  analytical variabilities (see Chapter  4,
 especially Table 4-2). Due to the complexity of defining and collecting a representative sample from
 a heterogeneous population, the sources of error associated with the sample collection process are
likely to be the major factor. In comparison, error associated with analytical measurements would
generally be expected to be much smaller.

                        Sample Receipt,  Handling, and Preparation

General Considerations

       As previously  discussed, to meet sampling and analytical program objectives, there must be
iterative communication between sampler,  analyst, and  data user. Standard operating procedures
(SOPs) for collecting,  handling, and subsampling of samples will be determined by the project's data
quality objectives.

       Unless information indicates otherwise, the analyst assumes that field sampling was done
properly and that the sample received at the laboratory is representative of the material of interest.
The assumption that the total sample is representative of the waste of interest is not necessarily
transferable to subsequent aliquots used for analysis unless consideration is given to subsampling
error potential. Though it is generally accepted that the greatest source of measurement error is
the field sampling step, the responsible laboratory manager must take care to mitigate small but
additive errors in the subsampling and analytical steps.

       When a sample is received at the laboratory, a stepwise characterization procedure begins.
Figure 6-1 is a flow chart of the analytical plan.  Clearly,  the nature of heterogeneous hazardous
waste dictates that  strict safety procedures coupled with good laboratory practice must  be observed
throughout the process.

       The  first step upon sample receipt is the proper inspection and  safety screening  of the
sample. The laboratory chain of custody begins here, and there must be careful documentation of
all operations  performed on the sample.  Any gross abnormalities  or contradictions of the field
sample documentation should be brought to the attention of the laboratory manager  (1).

        Because sample matrices and types of contamination are widely variable, many treatment
and analytical judgments must be made. It  is advisable that experienced analysts, in consultation
with sampling  personnel,  decide which  subsampling  strategies  should be used.

       Several steps may be  involved in the progressive analysis of heterogeneous samples. It is
important to consider a priori knowledge in deciding upon specific analytical procedures because
each step taken can potentially change the composition of the remaining sample.

       In all cases, a documentation of appearance is necessary. When the presence  of radiation
is suspected, it is necessary to screen for radioactivity. For multiphase samples, it is important to
observe and record an estimation of the phase ratio. If a  visual inspection determines that a sample
contains  more than one phase, the project  manager should be consulted to discuss the merits of
homogenization versus individual  phase analysis.  Each  phase of a multiphase sample may be of
particular interest.








Sample Preparation

       Samples composed of several phases (each of which is homogeneous) can be separated and
analyzed separately for metals and organic constituents.  Another possibility is to mix the phases
to  achieve a homogeneous sample and remove  aliquots for metals and organics analysis.
Heterogeneous  solids can be  ground up to achieve a homogeneous sample  for  metals  and
semivolatile organic  analyses.  The entire sample of heterogeneous material can be analyzed for
volatile constituents, or the sample may be extracted with methanol or polyethylene glycol (PEG)
and an aliquot analyzed.  If there is superficial contamination of large particles, leaching tests can
be  performed.

        One basic option available to the laboratory in dealing with heterogeneous waste is to
 separate out discrete sample elements.  This approach works primarily for liquids or mixtures of
 liquids  and solids. Consider a sample that contains three discrete layers: an organic  layer on top,
 a water layer in the middle, and a solid at the bottom. Often the sample will have separated during
 shipment, and the analyst can take an aliquot of each layer for analysis. The analytical results are
 then combined mathematically using the weight and the concentration found in each layer.

        Sometimes an individual layer can be further separated.  Samples from oil refineries often
 contain a "rag" layer which consists of an emulsion of oil, water, and solids. This rag layer has a
 density greater than oil but less than water.  This layer is homogeneous when sampled; however,
 upon  addition of solvent (e.g., hexane and/or  acetone), the solid material drops out of the emulsion.
 In this case, the layer can be separately analyzed for the material dissolved in the solvent and the

       The second basic option available to the laboratory is homogenization of heterogeneous
samples. In this case the sample is cut, ground, shredded, etc., to achieve a homogeneous material
that may be sampled. This technique is restricted primarily to solid samples that will be analyzed
for metals and non-volatile organic compounds. There are several types of laboratory grinders on
the market that can be used to homogenize soils and debris.

       Liquids can be homogenized to some extent; however, sample integrity will be compromised
if the phases begin to separate during aliquot removal. An ultrasonic bath or horn  can be used to
aid in dispersing separate layers long  enough to get a representative aliquot of the sample. Another
option is to add an emulsifying agent or surfactant to the sample followed by ultrasonic mixing.
Non-ionic surfactants, such as polyethoxynonylphenol can be used to disperse oil into water. Use
of emulsifying  agents, however, introduces the possibility of adding contaminants which are of
interest in the  sample, thus raising detection limits.   They may also  cause interferences in
subsequent analysis, particularly if they coextract with organic contaminants of interest. Use of ionic
surfactants may avoid some of these problems.  For routine analytical work, emulsifying agents or

surfactants are not generally recommended because of these problems. Furthermore, it is not
possible to know a priori whether a particular emulsifier/surfactant will work on a given sample.
Whole-Sample  Analysis

       One way to avoid the issue of heterogeneity is to analyze the entire sample provided. If the
entire sample  is used, the issue  of heterogeneity is  not germane  as  far as the laboratory  is
concerned. Heterogeneity may still represent a problem from a field sample standpoint since the
lab can only provide results as good as the sample itself permits.

       This approach is really only feasible for volatile  constituents due to sample size constraints;
however, it represents a real possibility for heterogeneous solids. Solid samples may be taken  in
a 40-mL vial with septum seal.   At the laboratory the vial  is weighed. The vial may then be
submitted to either a heated headspace purge or water (5 mL or more) may be introduced and the
vial coupled directly to a purge unit.  The purged gas is collected on the normal volatile trap for
subsequent analysis by gas chromatography or combined gas chromatography-mass spectrometry.
Several manufacturers make heated head space purge units capable of analyzing the entire sample
in a 40-mL VOA vial. After  analysis, the vial  is emptied and  weighed to  determine the exact
sample size.

       Another approach  which represents  analysis  of the  entire  sample  for  volatile  organic
constituents is  to extract the sample using methanol or PEG. The  entire sample  (in a jar or  in a
40-mL VOA vial) can be extracted with methanol and a portion of the methanol introduced  into
the purge chamber of a purge-and-trap device. The assumption is that  all  volatile components from
the sample are extracted into  the  methanol (or PEG)  and an aliquot is analyzed.  Normally the
amount of methanol  introduced depends on whether the project requires quantifying the more
volatile constituents (i.e., the gases). Typically 5 to 25  jiL of methanol is used. If higher volumes
are used, the early part of the chromatogram is obscured and the detection  limits for the more
volatile constituents are increased.
Sample Digestion. Extraction, or Fusion

       Once the sample aliquot is selected, the sample is further prepared for analysis. For metals
this normally involves digestion with strong mineral acids.  Extraction with methylene chloride,
hexane, or acetone is used for semivolatile organics, and purge-and-trap or direct injection may be
used for volatile organics.

       Heterogeneous aliquots can be digested for metals analysis. One technique for achieving
the digestion is  to use a microwave-assisted digestion method  or a bomb technique. The microwave
digestion works well for aqueous and  organic liquids and solids; however, the sample size is limited
depending on the amount of organic  material in the sample. For pure organic materials (i.e., oils)
only about 0.25 gm is recommended due to pressure buildup in the Teflon digestion vessel. Newer
digestion vessels are capable of higher presures and slightly more sample can be used; however, one
should select sample size carefully.  Organic solvents should never be digested due to rapid increase
in pressure during heating,  Sample bombs (Teflon lined) may also be used; however, they are
likewise limited in sample  volume.


        Organic liquids and solvents may be analyzed directly for most metals by dilution with an
 appropriate solvent (e.g., kerosene, toluene, methyl isobutyl ketone) and direct introduction into the
 ICP or Flame AA. If using this approach, an internal standard is highly recommended to account
 for changes in sample viscosity.  The internal standard is preferable to the more tedious method
 of standard additions.

        Samples that require organic analysis will be extracted using one or more organic solvents.
 The  resulting extract is  homogeneous and  can be diluted or concentrated  to  the desired
 concentration  range. Volatile organics are analyzed  using the purge-and-trap technique. Purge-and-
 trap can be used on large  sample sizes and thus can be used to advantage with heterogeneous
 samples since the analytes end up in a homogeneous (gas) phase.

        Samples contaminated with volatile  organic  compounds must be subsampled, handled, and
 treated by special procedures due to the potential for  volatile  loss during handling. Participation
 of laboratory personnel in the planning of field sampling for VOCs (and all other contaminants) can
 help in the selection of proper sample collection containers and will facilitate subsampling in the
 laboratory. The three drawings of Figure 6-2 illustrate various innovative sampling and shipment
 vessels. Two of these are ready for purge-and-trap analysis without analyst intervention. The other
 is an engineer's drawing of an "ideal" shipping container, complete with septum port and an insert
 sleeve for various indicators (2).

        A sample preparation technique that is primarily used for the analysis of metals in solid
 matrices is high temperature fusion (see Chapter 5  for field applications). Sample preparation by
 high temperature fusion was originally  developed for X-ray  fluorescence (XRF) and arc/spark
 emission spectroscopy. X-ray fluorescence is particularly  susceptible to the effects of sample
 heterogeneity (particle size, orientation and surface roughness). By use of fusion, a homogeneous
 glass disc is prepared, and analytical errors due to particle physical properties  are eliminated.
 However, interelement effects, which may be severe in XRF, need to be compensated for in order
 to obtain accurate results.  Recently, sample preparation using fusion has been  applied to ICP

        Fusion is accomplished by mixing a suitable flux with the sample in an appropriate ratio
 (between 1:1 and 10:1, flux: sample) followed by heating the mixture in a suitable crucible in a muffle
 furnace up to  a  maximum temperature  of 1200 Deg. C.  Once the flux becomes  molten, it is
 maintained at temperature for 15 to 20 minutes so  that the sample matrix is dissolved. The melt
 can be poured into a dilute acid solution, either hydrochloric, nitric, or aqua regia, or allowed to
 solidify in the crucible. If the latter procedure is followed, the solidified disc is removed and placed
 in a Teflon or plastic beaker, diluted acid solution is added, and the disc is leached until complete
 dissolution occurs.

       Fusion methods have some limitations.  In many cases fluxes which have the same purity
 as an ultra-pure acid are not available. Thus high blanks may adversely affect detection limits. Also
 if the flux and/or temperature used to prepare a sample  is inappropriate, losses of the more volatile
 elements such as tin, antimony, and arsenic may occur.

       Fusion methods have many advantages. In many  cases it is difficult or impossible to dissolve
matrices, even with the use of hydrofluoric or perchloric acid. Difficult matrices such as silicon
carbide  or refractory brick are examples. Fusion methods may also be quicker than standard digest-


                          40 ml VOA BOTTLE
                          W/ SEPTUM
                                              STEEL BODY
                          STAINLESS STEEL NEEDLE
                          W/ FEMALE LUER
                          STAINLESS STEEL MALE
                          LUER TO WDIATUBE
                          ZHE VALVE
                                                                         CLAMPING DEVICE ON
                                                                         TECHMAR'S PURGE
                                                                         AND TRAP SYSTEM

                                 TEFLON BALL
                                 (SHOWN REMOVED)
                                                                              40 cc VIAL
                                                       PURGE AND TRAP DEVICE
          40 ml VOA BOTTLE
                         LOCK RING
                        1/2" DIA SEPTUM
         LOCK RING
                                   AN IDEAL SAMPLE CONTAINER
Figure 6-2. Samplers that allow the entire sample to be analyzed for volatile constituents.


 ions using acids.   Finally, fusions allow for the use of much larger sample sizes in the sample
 preparation stage. Both microwave and oxygen bomb techniques are limited to samples weighing
 a few hundred milligrams. Fusions do not have this restriction. Thus there is a greater chance of
 employing a more representative sample in a fusion technique.

 The Effect of Particle Size

        Generally,  the  larger the sample  the better when dealing with heterogeneous materials.
 Large samples,  however, may not always be  feasible  when performing micro analyses. One
 approach to circumventing this difficulty is to follow the guidelines set forth in Pierre Gy's sampling
 theory (3; see also the discussion in Appendix B). To minimize subsampling  error, Gy's theory
 suggests a minimum sample size that increases as the size of the largest particle increases. These
 "minimum"  sample  sizes are  larger than the normal samples used in environmental analysis.
 Sometimes the analytical subsample size can be increased or the size of the subsample that is
 digested or extracted  can  be increased. Then a  subsample of the well-mixed  digestate or extract can
 be  analyzed. Another  approach involves  aliquoting a sample of the required size into several
 smaller subsamples  that can be digested or  extracted and then analyzed individually or as a
 recomposited sample.

        Table 6-1 gives the maximum particle  size allowable for normal sample size, according to
 Gy's theory. If the maximum particle size is greater than that listed in the table,  particle size
 reduction (PSR) is suggested.   Fortunately, if a priori knowledge exists about the sample or the
 mechanism of contamination, the use of large  sample sizes or PSR may be avoidable.

 Table 6-1.     The Relationship of Sample Size to Maximum Particle Size

                     SUBSAMPLE SIZE               MAXIMUM PARTICLE SIZEa
                        (grams)                               (centimeters)

                           1                                          .1
                           2                                          .13
                           3                                          .14
                           4                                          .16
                           5                                          .17
                          10                                          .21
                         20                                          .27
                         30                                          .31
                         40                                          .34
                         50                                          .37
                         75                                          .42
                        100                                          .46b

a The maximum particle  size allowed can be  approximated for other sample sizes by using this
    equation: Maximum particle size = the cube root of (sample size in grams x .001).
b The Toxicity Extraction Procedure and the  Toxicity Characteristic Leaching Procedure allow
    samples to contain particles as large as 0.95 centimeters.

References:  Maney (4); can  also be calculated from (3).


Leaching  Tests

       For heterogeneous wastes, leaching test results may better represent potential hazards to
humans and the environment than would mass-concentration data for particular analytes.  With
large items  contamination is likely to be superficial and mass data of little relevance to hazard.
Also, standard laboratory protocols for the digestion  of environmental  samples preparatory to
analysis do not always work well for larger particles or synthetic materials. Of course, the choice
between whole-sample analysis and a leaching test is made during project planning, not in the
laboratory upon receipt of samples.

       Often, the leaching test that is mandated is the Toxicity Characteristic Leaching Procedure
(TCLP). The TCLP is required whenever RCRA applies; that is, if it is to be ascertained whether
the waste is  a characteristic hazardous waste, or if the  success of waste treatment  before land
disposal is to be assessed. The nominal sample size for the  TCLP is 100 grams.

       With wastes containing large  items, the project planners must decide whether to subsample
and leach portions of the  items that  appear to be "hot spots,"  or to  ascertain  representative
contaminant values by extracting random subsamples from a  number of different items. If the latter
course is chosen, a subsampling scheme must  be devised.   When the items  are of varied
composition, another order of complexity is introduced. In this case, the investigator may choose
to  leach  subsamples of the  different types  of items  separately, then  composite the results
mathematically on  a weight or volume basis.

       Innovative leaching tests  are  not  performed  with any frequency in  hazardous  waste
characterization. However, in some  cases they may be eminently appropriate.  For "worst-case"
leaching scenarios, standard batch leaching protocols can be altered to use  harsher  solvents, or
aqueous eluants of high or low pH. For leaching large items, standard tests can be done with
specially-constructed large leaching apparatuses. If particle size reduction to a standard, fine grain
is not performed, longer leaching times may be appropriate.  The disadvantage of a large-scale test
is the volume of waste  solvent generated.  However, this may be offset by the need for far fewer
leaching tests to attain representativeness. The  application  of innovative leaching tests (or any non-
standard procedures) must be  cleared by the regulatory agency during  project planning.
Innovative Techniques

Radiation Screening Parameters

       It is important that laboratories recognize the need for proper screening of samples that are
suspected of containing radioactive materials, and it is critical to recognize the limitations and
strengths of various screening devices.   The laboratory's license for dealing with radioactive
materials will dictate its need for and use of screening equipment. (It must be emphasized that, for
sampling and sample shipment of radioactive materials to be in compliance with regulation, a
thorough radiation screen must have been performed in the field.)

       Radioactive material is the physical material which, by the process of nuclear decay, emits
some form or forms of radiation. For the purpose of discussing the screening of samples, only the

three principal types of radiation need be considered: alpha, beta, and gamma radiation. Because
of the different absorption characteristics  for each type of radiation, different detection systems  are
required. Both alpha and beta emitters may also release gamma  radiation as part of their decay,
but no single screening method is adequate to detect all forms of radiation being emitted from a

       The screening of samples that may contain radioactive materials  serves many purposes. This

•      provides health and safety protection for laboratory personnel
•      provides guidance on correct shielding, clothing, and monitoring to be used
•      directs the sample through an authorized and appropriate subsequent analysis
•      provides information  for a radioactive materials  inventory
•      fulfills administrative and licensing requirements pertinent to compliance parameters

       In deciding on appropriate radioactive screening protocols it is essential to understand the
reasons for performing the planned screening.  In some cases, a  simple screening with a Geiger-
Miiller probe  or a Nal  probe is  adequate.  But, if more specific data are needed to protect
personnel from inhalation and ingestion hazards, a series of scans for alpha, beta, and gamma rays
may be necessary (5, 6).

       Several detectors are  available for gamma or X-ray emissions, alpha emissions, and beta
emissions. Each has both advantages and limitations, as shown in Table 6-2.

Table 6-2. A Comparison of Several Radiation Screening Devices
Emission Type
I. Gamma,

II. Alpha

III. Alpha,
IV. Beta

Pressurized ion
Thin window probe
Alpha scintillation
Sealed-gas pancake
Beta-gamma pancake
Beta scintillation
Tritium detectors

Measures bulk of
High resolution
Integrates sample
Measures low energy
Many probe
geometries available
Higher counting
Sensitive to both
Sensitive to beta
Sensitive to 3H
X-rays may be absorbed
No information about
isotopes emitting alpha,
beta rays
Must be kept at liquid
nitrogen temperature
X-rays may be absorbed

Only good for surface or
near-surface sources
Fragile window
Alpha sensitivity much less
than beta sensitivity
Not good for alpha scan
Lower counting efficiency
Smaller area covered
Must be close to source
Not sensitive to 3H or
Need large surface area

       One of the most versatile detection systems for use in screening is the liquid scintillation
counter (LSC). In this system a small amount of sample is placed in a counting vial, and a liquid
scintillation cocktail is added. The total volume of the vial is usually 5 to 20 milliliters. Any alpha
or beta radiation is absorbed by the cocktail and the energy is emitted as light.  The light is counted
by photomultiplier tubes.  Advantages of this system are high counting efficiencies and sensitivity
to forms of radiation least likely to be  detected by gamma systems. Thus  use of an LSC system and
removal of  small amounts  of sample provides the broadest coverage of potential radioactive

 isotopes. Although LSC is not particularly  sensitive to high energy gamma emissions, gamma rays
 are usually emitted in concert with a mode of decay (e.g., alpha or beta), which is easily detectable
 by LSC. By the use of LSC, it is possible to do an excellent broad screen for radioactivity, although
 the use of an unprocessed sample may preclude precise quantitative results due to quenching and
 other effects. This method is particularly  applicable to liquids. It is also applicable to solids if small
 amounts of the sample are placed in the  cocktail or if smears of the sample are placed in the
 cocktail (a smear is a small piece of filter paper, cloth, polystyrene peanut, or foam that has been
 used to collect small amounts of sample by rubbing a surface). A less expensive system than the
 LSC can be implemented by using a hand-held device such as a pancake style alpha/beta detector
 or a scintillation detector fixed to a simple planchet holder device. One disadvantage of this system
 is the relatively low counting  efficiency for some isotopes, in particular, tritium.

       A laboratory which has concern for all types of radioactive materials might be best served
 by setting up an LSC system.   A laboratory which prefers not to set up an LSC system can utilize
 multiple monitoring systems to provide coverage for the various types of radiation potentially
 present in the samples. Such systems can  be set up relatively inexpensively and provide adequate
 information for personnel protection.  Ancillary elements of the screening system, in addition to
 procuring the correct hardware, will include appropriate calibration of the equipment, routine (at
 least daily) performance  checks for  equipment, and training of screening technicians in  the
 operation, application, and limitations of all equipment used.

       A disadvantage common to all screening systems for alpha and beta activity is the necessity
 of opening the sample containers. Although this  requires breaching sample integrity,  it is the only
 way to obtain valid screening data for pure alpha  or beta emitters. In laboratories where this is not
 appropriate in the sample receiving area, it may be necessary to screen for dose rate in one area
 and alpha/beta in a second, sample processing area.
 Special Requirements for Analysis of Mixed Waste Samples

        Support for DOE's Environmental Restoration and Waste Management program presents
 a challenging opportunity to the analytical laboratory. Past operations of the DOE facilities have
 left a legacy of both homogeneous and heterogeneous mixed waste requiring physical and chemical
 characterization to determine appropriate disposition according to treatability  and regulatory

        The presence of  radioactivity in waste  samples collected  for RCRA hazardous waste
 characterization mandates special  handling in the analytical laboratory (7). Correct guidelines for
 safe handling of radioactive samples should be established by the laboratory. For example, health
 physics guidelines at the Oak Ridge National Laboratory (ORNL) limit the total activity of "very
 high" radioactivity isotopes  (such as  90Sr) to 0.1  microcuries for  monitored benchtop sample
 preparation,  10 microcuries for radiochemical  hood work,  and 10  millicuries  for  glove  box
 operations (8).

       Researchers at DOE National Laboratories are experimenting  with analytical techniques for
 the analysis of radioactive  wastes and mixed wastes to meet the objectives of RCRA  SW-846 testing
 (9). Because of the lack  of standard regulatory  methods for the RCRA analysis of radioactive
wastes, researchers are exploring sample preparation techniques  and modifications of existing EPA


 protocols (SW-846 methods and Contract Laboratory [CLP] methods) to address this immediate

        At ORNL research and development efforts have focused on the analysis of volatile organic
 and semivolatile organic  contaminants in radioactive waste  samples.  Tomkins et al.  (5)  have
 described the adaptation of SW-846 method 5030 for glove box use.  In this manner, the volatile
 organic analysis may be completed in  a remote non-radioactive laboratory using conventional EPA
 GC/MS techniques.

        The analysis of semivolatile organics in mixed waste via current regulatory protocols may
 present an unacceptable safety hazard due to the level of activity present and the large volume of
 material to be extracted (typically a 1 L aqueous  sample) (6). Reduction  of sample size in the
 extraction step is an alternative but at the  expense of analytical  sensitivity. Continuous liquid-liquid
 extraction is preferable  to the conventional separator funnel technique since  direct operator contact
 with the radioactive samples is minimized. In another useful technique, Tomkins and Caton (7)
 have demonstrated the use of commercially available solid-phase extraction  cartridges to  minimize
 the analyst's exposure to the radioactive sample and to effectively separate  the radioactivity from
 the analytes of concern. These also generate minimum waste from the extraction step.
Neutron Activation Analysis

        The technique of neutron activation analysis (NAA) is a powerful elemental method capable
of measuring many elements in a variety of matrices.  In general, it is very sensitive, with low
detection limits for environmentally important elements such as arsenic, selenium, the halogens, and
many metals which exceed the capabilities of other commonly available  analytical techniques.  The
method lacks sensitivity for a few key elements of particular environmental concern.  These include
lead, copper, and  cadmium.  When properly utilized for the determination of those elements for
which it is most suited, NAA can contribute to the overall characterization of hazardous materials.

        Because the sensitivity is very good for many elements, the sample sizes generally used for
NAA are very small; less than one gram and often only milligrams. In most applications, the ability
to use a small amount of the material to be analyzed  is a distinct advantage.  In addition, the
requirement for sample  handling  and preparation for analysis is greatly reduced compared with
other techniques which require solubilization.  Samples can be analyzed as solids, liquids, or even
gases, and the  elemental determinations are not dependent on the chemical state.

       A major advantage  NAA brings  to  characterization of heterogeneous materials  is its
application to macrosamples, that is, the direct analysis of very large samples, perhaps kilograms
or even eventually a 55-gallon drum (10). Both the excitation source (neutrons) and the analytical
probe (gamma-rays) are very penetrating. Given the proper facilities for irradiation and counting,
there is no reason  large samples cannot be  assayed for many analytes. While analytical  imprecision
will arise because  of the uncertainty of the location of the analyte within the sample volume, the
level of uncertainty likely will accommodate the data  quality objectives for gross  analysis or
screening. The construction of such a facility at a large site would minimize the required handling
and  transport of hazardous  materials.   The cost of this specialized technique  may  be at least
partially offset by the elimination of the need for sample homogenization, segregration, etc., required
by any other technique.


       At the  Center for  Chemical Characterization  and Analysis at Texas A&M University,
preliminary studies on the feasibility of developing a "large sample" NAA facility are underway. The
program is still in its infancy, but some progress has been made.  One initial finding was that the
element cadmium,  which is  of great environmental  concern,  was not  determinable  using
conventional NAA at the expected levels of interest.  A variation of NAA called prompt gamma
activation  analysis (PGAA), however, is  very sensitive for  cadmium (11).  A separate PGAA
procedure  for cadmium is under development. The PGAA method using isotopic neutron sources
for activation is also likely to be field portable, possibly allowing field screening of wind rows or
landfills directly without excavation or sample removal (see Chapter 5 for further discussion of field

       A variation of the  NAA technique which has  been highly developed for certain mining
applications and for bore-hole logging is fast neutron activation analysis  (FNAA). Although the
sensitivity is not particularly good for certain elements of concern, such as lead, FNAA may be
appropriate for use as a screening tool (12).
                         Waste Disposal in the Analytical Laboratory

       In the process of performing an analysis, laboratories generate waste. Laboratories are
subject to the same RCRA regulations as are industrial production operations. However, wastes
generated by the laboratory can differ significantly from wastes generated at industrial production
operations. The American Chemical Society's Task Force on RCRA has prepared a handbook to
aid generators of laboratory waste in determining the appropriate responsibilities for proper waste
disposal (13).

       Satellite waste collection containers at the analyst's work station may pose a safety hazard
if not properly vented, because all laboratory salvage must remain in a sealed container except when
waste is added or  removed. Safety concerns such as  pressure build-up from vapors or potential
chemical reactions from the  mixing of waste forms  must be addressed in the laboratory waste
management plan.   Modifying the waste collection vessel by adding a vent tube with appropriate
absorbent, such as charcoal for organics and carbonates for acids, is a possible solution (14).

       When processing  heterogeneous wastes, laboratories may end up with large volumes of
excess sample and  extract. Little guidance is available to laboratories regarding disposition of excess
sample. Two general options are available for  sample disposal: return unused and unmodified
sample to the sampler (generator) for disposal or arrange disposal by the analytical laboratory.
While disposal options are straightforward, the actual  process can be complex if heterogeneous or
radioactive samples are involved. Each option for disposal has advantages and limitations. These
are noted in Table  6-3. 40 CFR 261.4(d)(l) allows exception  to the RCRA regulations for handling
and storage of samples for analysis and the return of excess  sample to the generator.  Sampling of
heterogeneous waste requires close coordination with the laboratory to assure sufficient sample size
is submitted for analysis yet assure minimal sample remains  for disposition by the laboratory.

        Most waste disposal procedures are specified in  regulations. The mechanism of excess
sample disposition should be specified in the project work plan.  The project planners should be
responsible for determining potential hazards and  safety considerations for sample disposition. Such
information should accompany the samples to the laboratory.


                Reporting Requirements for Analysis of Heterogeneous Waste

    Reporting the results from the analysis of heterogeneous hazardous waste entails an expansion
of the usual reporting requirements. The report should include all pertinent information about the
condition and appearance of the sample-as-received (15).   Other report  highlights that are
particularly informative in the case of heterogeneous samples are:  detailed  descriptions  of any
sample screening analysis that preceded the requested analysis; an outline of the sample preparation
steps and any observed phenomena that occurred; and a reconstitution of the sample if a series of
analyses were performed on segregated subsamples from the same sample.
Table 6-3.  Sample Disposal Options

Option A - Laboratory disposes of all sample excesses
    Pros:      (1)     Most convenient for the project as a whole
              (2)     Site manager avoids storage and disposal problems
              (3)     Trained chemical personnel handle samples
              (4)     Eliminates unnecessary transport of materials

    Cons:      (1)     Potential safety hazard to laboratory personnel
              (2)     Insufficient  information may  accompany  sample  to determine  proper
              (3)     Adds to overhead cost of laboratory operations
Option B - All sample excesses returned to waste site

   Pros:       (1)     Eliminates potential hazards from sample mixing
              (2)     Avoids storage problems in laboratory
              (3)     Responsibility and cost of disposal handled as with remaining on-site waste

   Cons:      (1)     Storage, handling problems for site manager
              (2)     Additional transport of samples
              (3)     Return of sample to site may not be possible
              (4)     May require reanalysis for waste acceptance criteria
              (5)     Potential for RCRA or CERCLA liability if altered sample is returned to

   Reporting requirements should be  specified when requesting all  chemical data and should
include the following.

•      Summary of analytical results for each sample
•      Results from QC samples such as blanks, spikes, calibrations


 •     Reference to standard methods or detailed description of analytical procedures
 •     Raw data printouts for comparison with summaries

       Most standard methods specify  some set of reporting procedures. Analytical services
 programs such as EPA's Contract Laboratory Program present a set of specific reporting procedures
 as a contractual requirement (16,17). This has led to the designation "CLP" reporting package, a
 term that is correct in reference to CLP Routine Analytical Services methods.

       It is critical  that the laboratory keep clear  and specific records of the physical characteristics
 of the sample  on receipt, because a single heterogeneous sample may  require segregation of
 components before  analysis.  The records should include size, weight, visible components, and  any
 information that can be derived  from visual inspection. If sample components are segregated before
 analysis, descriptions of each component, including weight/volume measurements or percent of total
 sample, should be provided in the report.  It is frequently recommended that photographs or even
 video documentation be used.

       The laboratory should describe the sample preparation procedures used in the analysis in
 enough detail that the data user can understand how the sample was manipulated during analysis
 because this may affect the usability of the data.  For example, if the sample was ground, some
 inorganic analyses may be biased depending  on whether  the grinder  had  metal or ceramic
 components. If the sample was extracted with an organic solvent, components that do not leach
 from the sample under normal conditions  may be  detected (e.g., decomposition of rubber gloves in
 organic solvents may cause false positive response).  In many of these cases, once the sample is
 prepared, it may be analyzed in the same manner as a homogeneous sample.

       The project manager should consider requesting that the results of analysis  of individual
 components be reported, in addition to a total sample content, in cases where the waste consists
 of discrete  elements that were segregated before analysis.  That will allow recalculation of  site
 concentrations  if one component is  heavily  contaminated or  if individual components  are
 contaminated with different analytes. It will also  facilitate development of alternative strategies  for
 remediation, including treatment of only  the most contaminated fractions.

       If an innovative procedure or a new application of a known procedure is  used for sample
 preparation or analysis, a detailed description of the procedure should be provided. The description
 should include  the measurement principle as well as examples of expected interferents, dynamic
 range,  and other pertinent factors.  Although laboratory and field chemists may be familiar with
 these methods, they may  be unknown to the data user and errors in interpretation of data may

                            Conclusions and Recommendations
       Many research efforts can be described that will lead to improvements in the analytical
procedures used for heterogeneous hazardous waste.   Basically, they fall into three categories:
sample preparation, instrumentation, and documentation.  The preparation steps range from the
design of a clear decision path to the development of special sample containers that will allow ease
of handling while maintaining sample integrity. Instrumentation needs should include the evaluation

of various methods for sample emulsification,  fusion techniques, and continuing refinement of
screening equipment.  Documentation is important for the establishment of new procedures and
their implementation in the laboratory. It is crucial, too, for the concise description of the sample-

       The unusual nature of heterogeneous hazardous waste presents a challenge to researchers
in all areas, from sample handling to data interpretation. Careful planning of this research will pay
off in simplified and focused analysis in the decades ahead.

 1.     Keith, L.H. 1988. Principles of Environmental Sampling.  American Chemical  Society,
       Washington,  D.C.

 2.     Manufacturer of Environmental Sampling and Sample Preparation Equipment, Catalog 104,
       Associated Design and Manufacturing Company, Alexandria, VA.

 3.     Pitard,  F.F.  1989.    Pierre Gy's  Sampling Theory and Sampling Practice. Volume 1:
       Heterogeneity and Sampling, Volume 2: Sampling Correctness  and Sampling Practice, CRC

 4.     Maney, J. 1990. Subsampling  in the environmental laboratory. ENSECO  Corporation,
       internal document.

 5.     Tomkins, B.A., I.E. Caton, M.D. Edwards, M.E. Garcia, R.L.  Schenely, LJ. Wachter, and
       W.H. Griest. 1989. Determination  of regulatory organic compounds in radioactive waste
       samples, volatile organics in aqueous liquids. Anal. Chem. 61:2751-2756.

 6.     Tomkins, B.A.,  I.E. Caton, G.S. Fleming, M.E.  Garcia, S.H. Harmon, R.L. Schenley,
       C.A. Treese, and W.H. Griest.  1990. Determination of regulatory organic compounds in
       radioactive waste samples, semivolatile organics in aqueous liquids, Anal. Chem. 62:253-257.

 7.     Tomkins, B.A., and I.E. Caton. 1990. Preparation of radioactive 'mixed' waste samples for
       measurement of RCRA organic compounds. In. Waste Testing and Quality Assurance,
       Volume  2, STP 1062, D. Friedman, Ed., American Society for Testing  and Materials,
       Philadelphia,  PA (1990) pp. 351-364.

 8.     Martin Marietta Energy Systems.    1990.  Guidelines  for  Radiochemical Laboratories.
       Procedure RP-2.16 (2/9/90) in Health Physics Procedure. Oak Ridge National Laboratory,
       Oak Ridge, TN.

 9.     Griest, W. H., R.L. Schenely, B.A. Thomkms, I.E. Caton, Jr., G.S. Glemmg, LJ. Wachter,
       S.H. Harmon, M.D. Edwards, and M.E. Garcia.  1990. Adaptation of SW-846 methodology
       for the organic analysis of radioactive mixed wastes. In Proceedings of the Sixth Annual
       Waste Testing and Quality Assurance Symposium. Volume II, American Chemical  Society,
       Washington, D.C. (July 16-20), pp. II-106-II-116.

10.     Grazman, B.L., and  E.A.  Schweikert.    1991.   A  non-destructive  approach for the
       determination of cadmium in large heterogeneous samples. J. Nucl. & Radioanalyt. Chem,
       in press.

11.     Grazman, B. L., and E.A. Schweikert.   1991.  A  brief review of the determination of
       cadmium by prompt gamma-ray neutron activation analysis.   Texas A&M University,
       Internal Document.

12.     Grazman, B. L., and E.A. Schweikert. 1991. On  the non-destructive analysis of a municipal
       solid waste compost, J. Nucl. and Radioanalyt. Chem., in press.

13.     ACS Task Force on RCRA Report. 1986. RCRA and laboratories. American Chemical
       Society, Washington, DC.

14.     Emergency  Standard Practice for Generation of Environmental Data Related to Waste
       Management Activities. American Society for Testing and Materials, Philadelphia, PA.

15.     U.S. Environmental  Protection  Agency.   1990.  Guidance for data usability in risk
       assessment.    EPA/540/G-90/008.    Office of Emergency  and  Remedial  Response,
       Washington,  DC.

16.     U.S. Environmental Protection Agency.  1991. Contract Laboratory Program, Statement of
       Work for Organics Analysis. Document OLM 01.8.

17.     U.S. Environmental Protection Agency.  1991. Contract Laboratory Program, Statement of
       Work for Inorganic Analysis.  Document OLM  01.0.

 Chapter  7
                            The Larger Perspective

                                     Roy R. Jones, Sr.

       The  preceding chapters demonstrate  that  the  application of  conventional  sampling
approaches to heterogeneous waste is very difficult and frequently provides unsatisfactory results.
The problems presented by heterogeneous wastes and debris are complex and interdisciplinary.
When the question of available technology to deal  with such materials is considered, another order
of magnitude of complexity is introduced.

       Among the principal conclusions to be drawn:

•      Because  the methods for heterogeneous waste  characterization  are not as well developed
       as those for  conventional environmental media, a common nomenclature has not been
       adopted by those working in this field.  The confusion of terms  (and,  hence, expression of
       concepts) impedes communication among professionals and hinders progress in this area.

•      The potential pitfalls in study planning, as encountered during the DQO process, are more
       acute for heterogeneous waste studies than for other environmental studies. This makes it
       imperative that reasonable, site-specific study  goals be emphasized, which in turn may lead
       to devoting more project resources to study planning.

•      In designing the heterogeneous waste  characterization study, investigators have a  very wide
       variety of statistical models to choose from.  Insofar as possible, they should make use of
       historical and process data, and data obtained by non-intrusive methods. Study design may
       be seriously constrained by worker health and safety considerations. A design with several
       stages of data collection and evaluation, rather than one massive effort, may be very cost-

•      Existing  QA/QC methods are seriously inadequate for heterogeneous waste studies. A new
       set of modified methods must be developed. In this effort, the development of site-specific
       comparison or reference materials to  improve both  QA and QC should take priority.

•      An enormous suite of field methods is already used with heterogeneous waste. In addition,
       there  is a vigorous research effort directed towards developing field instrumentation.  A
       strong trend towards field measurement and analyses is evident.

•      Laboratory personnel must know what to expect in samples from the heterogeneous waste
       site. An  analyst must be part of the study  planning team from its inception, and the

       laboratory may need special equipment to handle unusual samples. Laboratory logistical,
       safety, and waste disposal problems may be greatly exacerbated by heterogeneous samples,
       especially if they contain radioactivity.

Heterogeneous waste sites can be placed in three general categories:

•      those sites with problems amenable to current methods of waste characterization

•      sites where current strategies or technologies  of waste characterization are unsatisfactory,
       but methods now under development (or methods used in other industries)  offer great

•      sites for which current approaches, even augmented by new methods,  probably cannot yield
       "representative" waste  characterization

       For these latter sites, we must change the way we go about the characterization studies and
probably change the questions we ask.  The discussion that follows summarizes these three types
of situations and recommends general measures to expedite the characterization of heterogeneous
hazardous wastes.
                             Successful  Waste Characterization

       Most large waste sites, to one degree or another, may be considered heterogeneous. As long
as the question of "Hazardous Wastes" is not a concern, the materials may be handled with relative
ease and dispatch by conventional  segregation, reduction, and disposal techniques.   Materials
representing an economic benefit may be salvaged and recycled, modified  by reduction  (shredding,
grinding, compacting, etc.)  and/or composting, and the residues disposed of by incineration, burial,
etc. Dangerous materials may be dealt with by the safest expedient means, particularly by re- or
over-pack and shipment to a hazardous waste facility.

       Where hazardous chemicals or radionuclides  are of concern, there are several factors that
may enhance the likelihood of successful  characterization.  Obviously, conventional methods work
best when the wastes are not very heterogeneous. There are standard methods for sampling liquid-
filled drums. A single kind of production waste, with perhaps a second phase that can be separated
from  it,  is amenable to characterization (see Appendix A).  Piled or  containerized  wastes  the
investigator can be confident are only superficially contaminated may  be handled by wipe tests
followed by some type of washing.

       We also  succeed when we ask less difficult questions. Studies aimed at simply  identifying
the contaminants present, or confirming/denying the presence of one particular contaminant, have
a better chance of attaining  their DQOs than  those requiring rigorous quantitative, "representative"
sampling.  These types  of questions are asked when,  for  example,  contaminated industrial
equipment is to be sold for re-use in the same industry, or is to be treated as scrap and melted in
a blast furnace.

        Process knowledge concerning the creation of the heterogeneous wastes aids in successful
 waste characterization. When the contaminants of concern can be narrowed to a small number a
priori, project resources can be allocated towards more samples and fewer per-sample analyses. For
 sites that originally operated as dumps or waste reprocessing facilities, it is of great benefit to
 establish early that some or most of the wastes present could not be contaminated; they can then
 be dealt with as ordinary debris.   Process knowledge  is of most use for DOE sites and those
 regulated under RCRA. It is often unavailable for CERCLA sites.

        Finally, investigators can sometimes look for contamination solely in the environmental
 media  surrounding the waste, and not  attempt to  characterize the  waste itself. Conventional
 monitoring techniques can be used, and the need for waste handling is  obviated. This strategy may
 be applied to some DOE and RCRA sites, and it is an option for CERCLA sites that are municipal

                                  Methods Development
       As described in the preceding chapters, there are many promising methods whose further
development would aid in the characterization of heterogeneous hazardous wastes. These include
field analytical methods, statistical project design tools, sample handling techniques, and QA/QC
methods. In addition, there  are undoubtedly study design and implementation tools currently  being
used in other technical disciplines that could be adapted to heterogeneous waste characterization.

       Communication among professionals will be the key to enlarging the suite of methods
brought to bear on heterogeneous waste sites.   As one example, there is a need for a  new and
continuing effort  within EPA and DOE and  jointly staffed by  both. What is needed  is the
formation, support, and use of an "Applied Technology" Group, Office,  or Committee. Broadly
speaking, this group of technical specialists would continue to work in the spirit and intent of the
Heterogeneous Waste Characterization Workshop. It would serve as a focal point for:

•      The continued preparation and dissemination of an accepted glossary of words and phrases
       needed to assure clear communications  between agencies and disciplines. The definitions
       herein were included only for the  purposes of this discussion; they can in no way be
       considered  as "official" definitions of any agency, unless they were drawn  from regulations
       published in the Federal Register.

•      Inter- and intra-agency innovative technology  projects,  maintaining lists of availability and
       modifications of methods.

•      Investigations of state-of-the-art extramural technologies, with an eye toward acquisition and
       incorporation of technologies from other industry disciplines into  hazardous waste site

•      Development  and maintenance of an Applied Technology "bulletin  board" posting
       information accessible to all EPA regional offices and DOE facilities.

There is no need to create an entirely new organization. The organizational elements already exist
within the different agencies and the private sector, and, in some cases, they are duplicating effort.
For example,  the  Department of Energy has  an Office of Technology  Development. The
Department of Defense maintains programs and facilities in each department. In the U.S. Army,
there is the Corps of Engineers; the Navy has the Naval  Facilities Engineering  Command,
Environmental Quality Division.  The Air Force program is the  Engineering Services Center
Environics Division. Because of defined security responsibilities and classification systems within
these agencies, some technologies and processes  will always be justifiably restricted from public
access. However, those that are not classified would share in the technology  evolution with the rest
of  the hazardous waste handling community, and the  classified sector could benefit from  the
developments of the proposed consortium.

       EPA has both a Technology Transfer Program and a Technical Support Program. They are
developing capabilities  with contractors, consultants, and national trade associations such as ASTM.
Based on this beginning, EPA, DOE, and DOD could initiate  a federal consortium charged with
creating improved communications and interchange of ideas in the non-classified areas of hazardous
waste problem solving.

       Other Federal agencies and departments have long and productive histories of developing
and applying scientific/engineering technologies to the solution  of  problems.  If the initial
consortium can successfully deal with "turf and "not-invented-here" syndromes that are the main
impediments to interagency cooperation, other agencies and departments should be anxious to join
in the efforts.
                                 A Changing Perspective

       The intractable problems posed by heterogeneous materials will not be solved simply by
better communication among hazardous waste professionals.   Nor should we anticipate rescue
through the application of ever-more-sophisticated technologies to  ever-smaller  samples.  There are
instances where the waste characterization questions  now being asked cannot be answered through
any effort of reasonable magnitude. The drum of assorted laboratory wastes (Figure 1-1) is such
an instance. The question Is any item in the drum contaminated to a level above the action level? can
be answered when there are only a few items per drum and a few  drums on the site. For practical
reasons the question is unanswerable when there are many drums or many items per drum, and no
technology now on the horizon will change this. For  such situations, fundamentally new approaches
to waste characterization are needed.

       There are many sites that are not amenable to the classical approaches for the selection of
samples to be considered "representative" of the site.   If,  early in the site investigation and/or
subsequent site characterization phases, a site or critical portions  of the site could be designated
as "Heterogeneous Hazardous Wastes Areas," they possibly could be exempted or temporarily
waived  for an additional defined time period from some of the restrictive regulations that now
confound site studies. This site/portion would be the  last  area to be approached as a  unit and
would serve to temporarily  store identified, characterized hazardous  wastes that  were not amenable
to the characterization and treatment techniques applied to  the  rest of the site. These would still
be heterogeneous, but they would be characterized  and quantified amounts of waste that could then

be manipulated, treated, or over-packed for ultimate disposal as required to prevent their escape
into the environment.  The final regulatory goals would have to be satisfied eventually, but the
heterogeneity of the materials would not impede the conventional handling of those other portions
of the site wastes more easily dealt with. An economy of scale or at least a reduction in problems
introduced by waste variability  could be achieved.

        By such controlled manipulation and on-site accumulation and temporary storage, not only
would some segregation and volume reduction be accomplished, but the remaining heterogeneous
hazardous material might then be more amenable to statistical treatment. If the manipulation could
include some unit processing operations such as sorting, grinding, or homogenization, the final
resultant products could be a  quantity of homogeneous waste, either hazardous, dangerous, or
acceptable, and another quantity of very highly contaminated hazardous waste. This approach is
not possible under current restrictions placed on the handling of waste by "treatment" regulations
or interpretations of technically weak or  loose  regulations.  This makes applications of new or
innovative technology to problem solving very difficult or impossible to implement.

        Ultimately, the actual solution to these waste accumulations may lie in approaching them
from a different viewpoint using methods imported from  other industries. One possibly fruitful
approach is that used in the mining industry. There is an analogy between a heterogeneous waste
site and an ore body.  Both may contain deposits of valuable material. While the economic benefits
of an ore accrue to the producers of the material recovered, the value of treating a heterogeneous
waste site is not strictly a function of the value of the components recovered.  However,  any
recovered component of value reduces the cost of removing the environmental threat posed by the
original site.

        Furthermore, the "waste" residues from the "mining" of a waste site are an identified and
characterized waste  stream  that  may  be sequentially treated  to prevent  or  reduce future
environmental problems with the waste. By establishing a specific series of unit operations as a
processing scheme, the waste site may be dealt with in a manner that demonstrates actual progress
almost from the start of operations rather than an indeterminate period of study and then re-study
of the study.  If initial activities  can  include tangible benefits for the  surrounding community, either
aesthetic or economic,  public awareness expressed as negative concerns may be modified into
positive community activities.

        Sampling has evolved along  defined paths that  reflect the matrices perceived to provide the
most likely route to affect an organism exposed to that matrix. The most obvious routes, inhalation,
ingestion, and, to a lesser degree, adsorption and absorption, focus attention to air, water,  and
actual contaminants in sediment, soil, or product residues. In most of these cases, the question of
whether a sample is "representative" of a given area or volume is amenable to statistical analyses.
As long as wastes are present in relatively discrete and definable areas or volumes, ranges of
particle sizes, and numbers of phases present, the conventional methods of taking a representative
sample work.

       If there is to be acceptable progress in dealing with the heterogeneous areas of hazardous
waste accumulations or sites, sampling technology must evolve actively into the acquisition of data
from site "hot spots."   Just as  new characterizations and  identifications of waste produce new
problems and lead to new technologies  of manipulation and treatment, so must new approaches be
developed dealing with the identification and characterization of the wastes. Characterization and


treatment of heterogeneous waste accumulations must be recognized as an iterative design and
implementation process. In some cases, radically different approaches may be required. Sampling
may become an element in a series of unit processes that actually alters the physical structure and
distribution of the components of the waste.

       Initial clean-up and removal activities provide the best opportunity for identifying and
characterizing the waste materials on the site. They can also include preliminary manipulation and
segregation of the hazardous materials remaining for remedial activities. While this may be deemed
"treatment" and cause some regulatory problems, it would provide the opportunity to expedite the
final resolution of the site problems more rapidly and at less total expense to the taxpayer and the
responsible  party.

       This chapter and the preceding chapters developed from an interdisciplinary scientific and
technological  viewpoint.  It was  not the purpose of the authors to enter the area of policy and
planning for either EPA or DOE. However, numerous technical approaches that were advanced
by workshop  participants had to be qualified "but the regulations don't allow that." This problem
of overly-rigid regulations for waste sites has been addressed in detail by the National Advisory
Council for Environmental Policy and Technology (NACEPT) (1). In its  report, NACEPT has
documented the regulatory impediments to innovative site study  and waste transport and treatment
techniques.  Additional institutional barriers exist where there are multiple agency jurisdictions and
conflicting regulations. This has been recognized by EPA and DOE, and there has been a growing
cooperation between  the two agencies as  expressed in  the activities of  the "Harmonization"
committees and other interagency groups active for the last few years. It is  a sincere hope that as
problems evolve, so do solutions, and that this report  will be a contribution to more effective
solutions to some of the problems.

       NACEPT.  1991.  Permitting  and  compliance policy:  Barriers  to  U.S.  environmental
       technology  innovation.  Report and recommendations of the Technology Innovation and
       Economics  Committee,  National Advisory Council for Environmental Policy  and
       Technology. EPA 101/N-91/001. Washington, D.C.


Appendix  A
                         Hypothetical Case History

                           Drum Characterization

                              Tom Starks and Gretchen Rupp


       In 1983 the new owner-operator of XYZ Specialty Metals came under pressure from the
 state RCRA agency to clean up the wastes stored on the plant property. Of principal concern were
 two unlined lagoons, each containing more than 2000 cubic yards of chromium sludge. After a brief
 engineering study, it was decided to solidify the sludge chemically and store the product on site in
 drums. A small concrete batch plant was set up next to the lagoons. The sludge was  excavated with
 a backhoe,  mixed with water, fly ash, and Portland cement, and decanted into 55-gallon steel drums.
 The drums were labeled by lagoon, batch number, and within-batch drum number,  and a drum
 inventory was recorded. Over a four-month period, all of the sludges were solidified. The resulting
 drums were stored on an asphalt pad on the plant  site.

       In 1990 the plant operator decided to build an addition on the plant. This required cleaning
 up those areas of the plant property where wastes were stored. Integral to the cleanup would be
 establishing that the drummed solidified sludges were not RCRA-hazardous  and disposing of them
 at an off-site landfill. The  consultant brought in to do  the necessary studies established that these
 were not RCRA-listed wastes, but could not ascertain from process information whether they would
 qualify as characteristic wastes.  The consultant made a walk-through examination of the drum
 storage area.  At that time it was discovered that many drums were corroded, and liquid had
 apparently leaked from a number of these. Apparently solidification had been incomplete in some
 batches, and there was an alkaline supernatant several inches deep atop the concreted sludge in the
 drums filled from these batches. This had induced corrosion of the drums at the liquid-concrete
 interface. This situation was confirmed by a preliminary study that involved opening  selected drums
 from a number of different batches. It appeared that  about five percent of the drums contained
 free liquid.

       The consultant investigated the original uses of  the two lagoons and the details of the 1983
 sludge stabilization. Lagoon A had been a flow-through settling basin.  Lagoon B had been the
 disposal site for sludges dredged from Lagoon A. At the time of the sludge fixation, the sludge in
 Lagoon B had been fairly dry. The  sludge was dry at one end of Lagoon A, but  it  graded to a
 saturated condition  at the other end, In excavating the sludge, the operation began at one end of
 each lagoon and worked gradually to the other end. Excavated sludge was mixed with water, then
 blended with fly ash and cement. A single batch filled  10-15  drums. When the last of a batch was
 decanted, it was sometimes necessary to hose out the equipment to clean out the  last  of the
 mixture; thus the last drum might contain extra water. There were changes in process variables and
 equipment over the course  of the operation. As the  solidification operation was getting underway,


the blend formula was altered several times. Different mixers and mixing tanks were used during
the four-month period. There was a major equipment change between the processing of the A and
B lagoon sludges.
                                 Initial Project Planning

       The consultant and the owner discussed the decisions that would have to be made before
off-site drum disposition could proceed. The state regulatory agency would have to decide:

•      whether a drum contained a hazardous waste.  If so, that drum would have to  go to a
       hazardous-waste disposal facility. If not, it could go to a municipal landfill (the consultant
       had first established that a local landfill would accept non-hazardous drums).

•      how the corrosive supernatant in some drums would be treated as the drums were handled.

The owner would have to decide, on the basis of cost factors:

•      whether it would be less expensive to dispose  of the drums as a hazardous waste or to
       undertake an expensive characterization study  in the hope of demonstrating their non-
       hazardous nature; and

•      whether to go forward with site cleanup once the costs of the characterization study and
       drum disposition were known.

The owner did not set a maximum project budget, but indicated to the consultant that inordinate
costs would compel him to postpone the project. No time constraints were identified for the drum
characterization study.

       At this point a RCRA specialist from the state regulatory agency was brought into the
planning process.  He and the consultant first discussed how to determine if the concrete was a
hazardous waste.  It was agreed that the characterization was to be done by vertically coring the full
depth of concrete in selected drums, pulverizing the  core, and performing a TCLP test on a
subsample of the (well-mixed)  powder.  The TCLP extract would be analyzed for total chromium.
If it exceeded the regulatory criterion (5 mg/1), the sample would be judged hazardous.

       The handling  of the liquid supernatant was discussed next. The consultant proposed a
cursory study to  establish average supernatant pH, RCRA corrosivity,  and heavy metal
concentrations. If these were within regulatory limits, any supernatant encountered as a drum was
handled would be  dumped out on  the ground.   The state RCRA  specialist insisted  that a
considerable range in these properties could be anticipated, and therefore the supernatant must be
separately characterized each time it was  encountered.  This would have entailed extraordinary
analytical costs, so the following plan was agreed on. As each drum was handled, any supernatant
would be siphoned out into a general holding tank. The collected,  well-mixed supernatant would
be characterized for those constituents regulated in the facility's NPDES permit. If it were within
the specified discharge limits,  it would be piped to the  plant's surface water discharge; if not, it
would be used up gradually within the plant as process feed water.

       The consultant and his statistician now began planning the drum characterization study. They
considered it important to take into account:

•     differences between sludges from the two lagoons, especially since they were processed using
       different equipment;

•     the changing nature of the  wet mixture over time, as reflected in differences among batches;

•     the relative uniformity of the mixture within batches, except for the occasional "wet" drum
       at the end of a run.

In addition, the state RCRA specialist pointed out that drums on the outside of the stacks had likely
had much more exposure to the elements, and the concreted sludge might be different from that
in the interior of the blocks of drums. The 6120 drums of interest were stacked in eight separate
blocks that were four drums across, three drums high, and 64 drums  long. Since the interior drums
in each block could not easily be surveyed, it was decided that the drums in the two exterior rows
would be surveyed and  disposed of first, and then the interior  drums  would be surveyed and
disposed  of. It was expected that the information  obtained in the  survey of the exterior drums
would be  useful in evaluating the population of interior drums, since the drums had similar origin.

       Four of the blocks contained all  3050 drums,  from 254 batches, filled with concreted sludge
from  Lagoon A, and the other four blocks contained the 3070 drums, from 256 batches, filled with
concreted sludge from Lagoon B.   There were 1514 drums in the two exterior rows of the four
blocks from Lagoon A  and 1534 drums in the two exterior rows of the  four blocks from Lagoon B.

       An exploratory survey of the exterior drums  was performed to find the precision and
accuracy  of  the proposed measurement process  and to estimate  the  within-population and
within-batch variance of log-transformed measurements of chromium concentration in the TCLP
analysis of the concrete in the drums.  The exterior drums were divided into four populations:

•     Population 1 - all exterior drums from Lagoon A that were not the last drum of their batch
•     Population 2- all exterior drums from Lagoon B that were not the last drums of their batch
•     Population 3- all  exterior drums from  Lagoon A that were the last drums of their batch
•     Population 4- all  exterior drums from  Lagoon B that were the last drums of their batch

From the  drum inventory, it was  determined  that Population  1 contained 1404  drums  from 250
batches; Population 2 contained 1411 drums from 255 batches;  Population 3 contained 110 drums;
and Population 4 contained 123 drums.  If selected drums contained  liquid, the liquid was drained
into the holding tank prior to sampling the concrete in the drum. The measurement process for
the concreted sludge required that a 2 cm diameter hole be drilled vertically through the concrete
in each selected drum from a randomly selected starting point on the  upper surface of the  concrete.
The material obtained from the  hole was mixed and then subsampled  to obtain the 100 g laboratory
sample upon which the TCLP analysis was performed.  Each drum selected in the sample was
weighed after removal  of any liquid that might be present.

       In  the exploratory survey of the exterior drums, ten batches were  randomly selected from
each of populations 1 and 2 (with probability proportional to batch size,  and with replacement) from


 among the batches that had at least two drums in the exterior rows. Two drums from each selected
 batch were randomly chosen.  Of this pair, one drum was randomly selected and one sample was
 obtained from it by the method discussed above; three samples were obtained from the other drum.
 The three samples taken from  one drum were obtained by drilling a hole in the concrete and then
 taking two subsamples ("splits") after mixing the material from the hole, and by then drilling a
 second hole (a "field duplicate" or "colocated" sample) and taking one subsample from the material
 obtained. With data from these samples, the consultant could  estimate the between-batch variance,
 the within-batch variance, the  within-drum variance, and the  between-subsample variance (which
 also includes variances owing to handling and  to analytical error).  The sum of the within-drum and
 between-subsample  variances  represented the total measurement error  variance for a  single
 subsample taken from the material obtained  from a single drill hole  (employed here are  some
 quality assessment procedures discussed in the chapter  entitled, "QA/QC and Data Quality
 Assessment"). In the exploratory study, a sample also was taken from each of ten randomly selected
 (simple random  sampling) drums  in each  of populations  3 and 4.  These samples provided
 information about within-population variances for the end-of-batch drums.

       The results of the exploratory study were that the measurements from all drums sampled
 were below the TCLP action level for chromium (i.e., 5 mg/1) with values ranging from 1.1 to 4.6
 mg/1.  The weights of the selected drums from populations 1  and 2  were very similar with a
 coefficient of variation of only 3 percent.  The weights of the drums in populations 3 and 4 were
 quite variable (CV = 75 percent), and the mean weight of the  drums in these populations was only
 60 percent of that in the first two populations.   The variance estimates for TCLP chromium
 concentrations obtained from Population 1  were sufficiently  similar to those obtained  from
 Population 2 data  that the decision was made  to  pool  the variance estimates. Of the total
 measurement-error  variance  for  the log-transformed  data,  90 percent was  attributable to
 subsampling and analytical error. The total measurement-error variance was only 6 percent of the
 within-population variance and therefore was considered to be sufficiently small by the usual rules
 of thumb (i.e., error variance should be less than either one-tenth or one-sixteenth of population
 variance. The idea  behind these rules of  thumb is that further reduction  in measurement error
 variance will not have sufficient impact on total within-population variance to make the attempted
 reduction worthwhile.). The estimated within-batch variance was found to be  only 10 percent of the
 between-batch variance,  which  showed  that  relative  to  the  variation  between  batch-mean
 concentrations, individual measurements were typically close to both the mean for the drum and
 the mean of the batch from which the sample  was taken.

       At this point the consultant and the state RCRA specialist explored the issue of how much
 confidence the state needed in its drum disposition  decision and the sampling design that would be
 needed to assure this confidence. The specialist wanted to be 95 percent confident that all drums
 individually fell below the specified action level for the TCLP  analysis. On this basis the consultant
 determined that one of two approaches might be followed. One approach would be to sample at
 least 95 percent of the drums.  This would  in essence imply that all drums be sampled, and those
 found,  if any, that did not meet requirements  would be sent to the hazardous waste facility. The
 second approach would be to randomly sample one drum from the "not last" drums in every batch
 and test whether the sample TCLP measurement was significantly below the 5 mg/1 action level.
 If such a sample measurement from a batch was not significantly below the action level, they could
then either sample all other "not last" drums in the batch to decide which needed to be sent to the
hazardous waste facility, or send all "not last" drums in the batch to the facility without further
sampling. All the drums that were the last  drums of their batches would have to be sampled and


 tested as in the first approach. The second procedure, sampling two drums per batch, would require
 considerably less sampling, but the criterion that the sample measurements would have to meet to
 establish a significantly-below-action-level condition would be more stringent, and the procedure
 would be based on  distribution assumptions that would be  difficult to justify. The  consultant
 calculated that the first approach would require approximately 6000 routine samples, while the
 second approach would require about 1000 routine samples.

       The XYZ owner-operator could not justify proceeding with the plant  expansion if it required
 1000 or more TCLP tests of concrete cores. The consultant formulated an alternative, less-intense
 sampling  strategy to support a completely different decision rule, and the state RCRA specialist
 agreed to this alternative.   In this new plan, the state would only require that,  for each drum
 population, the leachate from the aggregate of all drums of concreted waste must meet the TCLP
 action-level requirement. This would be demonstrated by showing that the leachate sample mean
 was significantly below the action level. This could be accomplished by taking a sample of drums
 that was much smaller than would be needed in the two scenarios discussed above.

       The new sampling scheme is as follows.  Within  populations 1 and 2 (the "not last" drums),
 batches will serve  as primary sampling units  and be selected with probability proportional to batch
 size minus one (the excluded last drum) and with replacement. Batches that were selected in the
 exploratory study will be  included in the  population  to be sampled, and, if selected, the data
 obtained from the first  drum of the previous sampling will be used rather than taking new
 measurements. After batches are selected, one drum within each selected batch will be  chosen by
 simple random sampling. In populations 3 and 4,  sampling of drums will be accomplished by simple
 random sampling.  Drums selected in the exploratory study will not be eligible for selection in this
 sample as  results  from the two studies will be combined to estimate mean concentrations for
 populations 3 and 4.   The decision to include batches that had previously been sampled from
 populations 1 and 2, and to exclude drums previously sampled  in populations 3  and 4,  was to
 simplify  the estimation  procedures  for population means and  standard  errors of means. The
 sampling  with replacement of the batches  recommended for populations 1 and 2 is  somewhat
 unusual, but, in this case, while  it will cause some loss in efficiency, it will greatly simplify the
 estimation of the standard error  of the mean relative to  the  sampling  without replacement

       Acceptable false negative  and false positive errors were established as follows. The regulator
 specified that the state would only be willing to allow the concreted sludge to be sent to the sanitary
 landfill if the mean TCLP were significantly less than the 5 mg/1 chromium action level when using
 a one-sided 5  percent significance level test.  Put another way, there would be a likelihood of 5
 percent that any drum population having a mean TCLP chromium > 5 mg/1 would be judged non-
 hazardous. The facility owner-operator, his  consultant, and the consultant's  statistician worked out
 the acceptable false  positive error rate.  They  knew from the preliminary study that the mean
 concentrations for  all the populations were probably about 3 mg/1. They also knew that, as 5 mg/1
 was approached, the sampling requirements for a given degree of certainty would increase greatly.
 Therefore, they specified a 95 percent probability that  a drum population would be judged non-
 hazardous if its true mean TCLP chromium level were 3 mg/1.  That is, in terms of tests of
hypotheses, the significance level of the test was specified as  5 percent, and the power (i.e.,
probability of accepting the alternative hypothesis, \i < 5) of the test if the true value of n is 3 was
 specified as 95 percent.

                                       Study Design

       Given these sampling goals and uncertainty constraints, the statistician proceeded to design
the sampling scheme. He reasoned as follows:

       For populations 1 and 2, a measurement of TCLP concentration of chromium would be
obtained from a sample from one randomly selected drum from each batch selected in the batch
selection process.  If n is the number  of batches to be selected in the sampling of a population h
then the sample mean is
                              x  =     x   n            h=1'2
where x1 is the measurement on the chosen drum from batch 1. The formula for the estimated
variance of the estimator of the mean is (from reference 1, equation 6.21).

                                         x.-F]2 /  [n  - 1]) =
The estimate of the standard error of the mean is just the positive square root of V (xh).

       The Central Limit Theorem comes into play here. Standardized sample means tend to the
standard normal distribution as sample size increases.   The confidence statements  and tests of
hypotheses employed here are based on the idea that the sample mean is normally distributed. To
be significantly less than the action level, the sample mean must be less than the action level by 1.65
standard errors  of  the  mean.  The  regulator's  decision rule,  therefore,  is If the mean  TCLP
chromium for population i is less than 5 minus 1.65 standard errors of the mean, the population
will be judged non-hazardous, and all drums in that population can go to the sanitary landfill. If
it exceeds this value, that population is a hazardous waste. If one lets C represent the critical value
that is the action level minus 1.65 standard errors, then the owner wants to have a 95 percent
probability that the sample mean will be less than C when the true mean is 3 mg/1. To achieve this
probability, the value 3 must be 1.65 standard errors of the mean below C, or in other words 3.30
standard errors of the mean  below the action level. Hence for each population, one  should have
for standard error of the mean the value (5 - 3)/ 3.3 = 0.61 mg/1, or less. As  observed in  the
preceding paragraph,  the standard error of a  population mean is a function of the within-population
variance, sj;, and  of the sample size for the population. Therefore, since the data from populations
1 and 2 in exploratory study provided a pooled estimate, sj; = 16.0, the suggested sample sizes for
sampling populations 1 and 2 are the same and that common sample size is the smallest integer n
such that

                                    [16.0 / n] °-5 <;  0.61,

which is n  = 43.  It should be pointed out here  that in  determining sample means and standard
errors, we are working with actual data, not log-transformed data. The log-transformations were


necessary to stabilize variance so that a variance components analysis could be performed to
determine the proportions of variance  arising from various sources. But transformations are not
used here because we may appeal to the Central Limit Theorem and because we are interested in
investigating the means rather than variances.

       For populations  3 and 4 (the "last-of-batch" drums), the sample mean must be a weighted
mean since there are different amounts  of concrete in each selected drum and we wish to estimate
the average TCLP chromium concentration over the aggregate of all concrete in the population of
last-of-batch drums from a lagoon. That is, if W; is the true weight of the concreted sludge in last-
of-batch drum i and T, is the true mean TCLP chromium concentration for that drum, then we are
interested in the value  of (£WjTj / £W}) where the  sum is over all last-of-batch drums in the
population (3 or 4). Thus, the sample estimate of this value is
                                        N         N
                                   r = E
where wf is the weight of the i th selected drum minus a typical empty drum weight w0 in the sample
from a population. The estimate is in fact a ratio estimate where both numerator and denominator
are determined from sample results. Therefore, a formula for the variance of a ratio of random
variables must be employed before an estimate of the standard error of the weighted mean can be
obtained.   Let w  be the arithmetic mean of the weights of all the drums (after removal of
supernatant) minus w0 in the sample from population.  The estimated variance of the weighted
mean is (from reference 2, Equation 4-21.4)
                 s2 = (1 - f)(l/n)[£ wfo-rfl / [(n -
where f is the sampling fraction n/N. The estimates of s2 obtained for populations 3 and 4 in the
exploratory study were so similar that they were pooled to obtain the value 25.0. Unfortunately,
the Central Limit Theorem does not apply  to ratio  estimates and so  we must appeal  to the
Chebychev Inequality which states that the probability that an estimate will differ from its expected
value by more than k standard deviations is no more than k"2.  Because of the conservative nature
of the inequality and the fact that it is two-sided while we are interested only in one-sided deviates,
the regulator may be satisfied that the observed mean TCLP leachate concentration is significantly
less than the action level if the observed value of r is four standard deviations less than the action
level  (i.e., confidence greater than 94 percent), and the factory owner will be satisfied if the value
3 mg/1 is four standard errors less than the action level. Hence, the owner may want to choose a
sample size n such that the standard deviation will be no greater than (5 - 3)/4 = 0.5. In this case,
for the sample from population 3 which  contains n =  1 1 0 drums, the owner will want n to be the
smallest integer for which

                               (1 - n/100) (1/n) (25.0) z (0.5)2.

The sample size required  for Population 3 is n = 54.  For Population 4, the population size is
n = 123, and a sample of size n = 56 is needed.


       It should be noted that the standard errors of the means or ratios that will be used in the
 significance tests will be estimates that employ the data from the final sampling study, not those
 from the exploratory study. Hence, the actual estimates employed may be smaller or larger than
 the target values (0.61 and 0.5).

       The use of field duplicates and splits would continue in this study as in the exploratory study,
 but they would be used at a reduced frequency relative to the routine samples. However, at least
 20 field duplicates and 20 split samples should be obtained.  The number 20 is chosen because it
 provides estimates of variances that we can say are within a  factor of 2 of the true value with 95
 percent confidence.  These QA samples should be uniformly spread through the routine sample
 stream. In addition, in populations 1 and 2, the  statistician recommended taking a sample  from one
 additional drum in ten of the chosen batches in each population to provide  more  information about
 within-batch variability.

       The approximate minimum number of samples that would be obtained and TCLP tests that
 would be performed in this design would be 236. The consultant now compiled cost information
 to allow the owner-operator to decide on a course of action. The owner needed to consider
 whether the cost of the sampling and chemical analyses would be less than the savings in sending
 the four populations of drums to  the sanitary landfill as opposed to sending  the  drums to  the
 hazardous waste  facility. If it were not substantially less than those  savings,  then it would be
 reasonable for the owner to declare the drums to be hazardous and  not perform  the  sampling
 studies. If the cost of sampling and chemical analysis were considerably less than the savings and
 if the result of the investigation were that the sample mean, or weighted mean, were less (say,  3.6
 mg/1) than the action level, but not significantly less than the action level, the  consultant would
 recommend that the owner and regulator consider taking additional samples from the population
 to see if the additional data would provide the confidence required that the population mean were
 below the action level.

       Once the exterior drums have been studied and disposed, the owner must then address the
 four corresponding populations of interior drums. If the sample mean or weighted  sample mean
 for an exterior drum population is above the action level and there is no reason  to expect that the
 corresponding  population of interior drums will  be different from the exterior  drums,  then it would
 be reasonable for the owner to declare that interior drum population hazardous  and save the cost
 of sampling it. If the sample mean or weighted mean of an exterior population  were found to be
 significantly less than the action level, then the owner should first try to show that the  corresponding
 interior  population has  similar  or better TCLP characteristics than the exterior population. For
populations 1 and 2, this might be done by randomly picking 20 batches that were sampled in the
 exterior population and that also have drums in the interior population, randomly picking an
interior drum from these batches and then performing a paired samples one-sided nonparametric
test. This procedure cannot be applied to the two populations  of last drums in batches. Also, since
there were some batches that had no exterior drums in populations 1 and 2,  the regulator considers
it prudent to require that a sample be taken from one randomly chosen drum from each of these
batches to ascertain that these batches do not yield TCLP chromium concentrations above the range
observed in samples of batches that had exterior drums present.

      If the nonparametric tests indicate the need for additional investigation of a population of
interior  drums from  either lagoon A  or  B that do not contain the last  drums in batches, the
investigation  of the  population would  follow a  similar procedure  to that employed for the


corresponding population of exterior drums, but the additional information obtained in the sampling
of exterior drums should be taken into account in estimating standard deviations and sample sizes.
The two populations of interior last drums of batches from lagoons A and B will be sampled in a
similar manner to the procedure employed for populations 3 and 4 of exterior last  drums.  The
consultant will use information obtained in the sampling of populations 3 and 4 to determine
appropriate sample sizes.

1.      Raj, Des. 1968. Sampling Theory. McGraw-Hill, New York, 302 pp.

2.      Hansen, M.H., W.H. Hurwitz,  and W.G.  Madow. 1953. Sample  Survey Methods  and
       Theory, Vol. 1. John Wiley, New York, 638 pp.


Appendix  B
            A  Survey  of Available Statistical  Techniques

                                      Leon Borgman

       The Waste Characterization Strategies work group included both professional statisticians
 and individuals with substantial experience in using and interpreting statistics in environmental
 applications.  The group compiled the following list of statistical categories and subdiscipline that
 have been used or appear to have potential usefulness for heterogeneous waste characterization.
 The tabulation is quite impromptu and represents those topics which arose during a "brainstorming"
 session. Generally, the more important and useful methods are toward the beginning of the list
 since they occurred to the participants first. References to literature and software are given for the
 most-used techniques. This list is far from complete as a guidance to project planners, but it could
 serve as a basis for the development of such guidance.

       To place the methods within the context of heterogeneous waste characterization, many (but
 not all) have been classified within a multiple category framework. This consists of the following

              Completeness of the statistical theory
       A.     The method is well established
       B.      Some questions remain
       C.     Many questions remain
              Applicability to heterogeneous waste problems
       I.      The method would be applicable to heterogeneous drums
       II.     Homogeneous  drums
       III.     Unconfined waste
       a.      Potentially a very useful method
       b.      Moderately useful
       c.      Not particularly useful
              Data requirements
       i.      Considerable amounts of data needed
       ii.      Moderate amounts of data needed
       iii.     Useful with sparse amounts of data

       The methods are classified into these categories on a preliminary basis; it may well be that
with more investigation, the  listed choices  would be changed.  Nonetheless, this listing should have
some value as an initial examination of possible statistical techniques in heterogeneous waste

                                      List of Methods

(1)    Equiprobable pure random sampling

              The population being sampled is regarded as a finite array of possible samples. No
       assumptions  as to  the probability law,  statistical stationarity of the  elements of the
       population, etc., are made. What is "out there" is just a collection of possible samples. The
       randomness is introduced by requiring that every possible sample have an equal likelihood
       of being included in the data to be collected. This is achieved by using a random number
       table or computer program to select which of the samples are going to  be collected into the
       data set.   One characterization  of the  technique is to say  that  "The population is
       deterministic; the sample is random" (1). The method is very objective and is particularly
       desirable if sample collection and analysis is relatively cheap so that very large data sets are
       feasible. The method is directed toward the estimation of the mean, variance, and similar
       parameters of the finite population of possible samples.
                                                                              A; I,II,III; b;  i
       References: (2, Chapter 2;  3; 4, Chapter 4; 5; 6)

(2)    Weighted pure  random sampling

              This is similar to (1) above, except that the possible samples are  not collected in such
       a way  as to make each equally likely to be included in the  data. Instead, some rule is
       devised so that each sample has its own probability of being included in  the data set. The
       weights can be based on the area of the strata from which the sample was selected, the size
       of the group from which the sample was taken, or the probability that an object of that size
       is encountered  in the sampling procedure.   This modification of random sampling is
       sometimes  called "probability sampling."
                                                                              A; I,II,III; a;  i
       References: (3,  Chapter 2; 7, page 26)
(3)     Stratified random sampling as a modification to items Tl) and (2] above

              The population  of possible samples are divided into  subpopulations, usually  so that
       each subpopulation has less internal variation.  Then the methods of (1)  or (2) above are
       applied separately to each subpopulation.  Some of the objectivity of method (1) is lost
       because the classification into  subpopulations  usually requires subjective judgments. A more
       even coverage of the population of possible samples (perhaps for contouring purposes)  can
       be obtained with stratification than with method (1).
                                                                              A; I,II,III; a; i
       References: (2, Chapter 5; 4,  Chapter 5; 5; 6)

(4)     Systematic random sampling and/or cluster  sampling
              A rectangular grid, or some similar regular geometry, of locations to be sampled is
       laid out over the population  of possible samples. The equal likelihood for inclusion of each
       possible sample into the data  set is maintained by using some random mechanism to
       establish the placement and orientation of the grid  in its exact position upon the population,
       For example, random numbers can be used to establish where the center of the grid will be
       placed, and then another random number can be picked to make a random rotation of the
       grid. The methods of (1) can then be used on the data. Some of the advantages of the
       method are that a very even coverage  of the region for contouring or search can be
       achieved,  and the logistics of sampling can be simplified (since the grid can be surveyed and
       staked previously and the samples can be collected in a rapid sweep over the grid).

                                                                             A; I,II,III; a; i
       References: (2,  Chapter 8; 4, Chapter 8;  5; 6)

(5)     Gy's methods for sampling homogeneous mixtures of granular materials

              These techniques were developed in the context of the mining industry to sample
       crushed ore. The ore is thought of as consisting of an inert substance (gangue) and a
       valuable material (some metal or metals),  which are intimately intermixed. Even though the
       ore is crushed to some degree of fineness, some of the value is hidden from the assay by a
       covering or armor of the gangue.  The smaller the diameter of the particles, the more the
       value is released to be measured in the assay. The  basic question, then, is  "what sample
       volume and reduction of particle  size is required to provide an estimate of the total value
       of the ore with a pre-specified accuracy  or confidence interval?"  The procedures depend
       on the characteristics of the ore and a substantial amount of calibration against practical
       experience.  The best derivation of the basic mathematics of the method are given by Gy
       (8) in a paper in French.   A recent exposition in English of Gy's derivation  has been
       presented in a report by Borgman (9). The method appears to have applicability to waste
       characterization problems, although  some trial and error will probably be  required. The
       above discussion illustrates how Gy's sampling procedure extends  from sampling statistics
       into sample collection and preparation. Thus, it seeks to provide an overall framework for
       sampling, whereas the traditional random sampling schemes address  only  the statistical
       aspects of data computation and  leave questions of mesh size, volume of sample, etc., to the
       chemist. From this perspective, Gy's methods consider and involve many other practical
       sources of estimation error not included in the simple example used above to introduce the
                                                                                  B; III; 11
                                                                           tailing piles: a
                                                   extreme intercorrelated heterogeneity:  c

       References: (8, 9, 10, 11, 12)

(6)     Correlated spatial, temporal, intervariable geostatistical procedures  (variogram.  kriging
       conditional  simulation)

              This is the  general body  of  statistical techniques that have come to be called
       "geostatistics," although the methods are finding applications in many areas besides  geology
       and mining. A more descriptive name is random field theory or random function  theory.

              In contrast to the various types of random sampling above, the assumption  in
       geostatistics is that the population may be treated as though it is a realization of a  random
       process; i.e., there  is a  probability law and various  statistical  assumptions concerning
       "stationarity"  and "spatial intercorrelation" which  model our uncertainty concerning the
       population. In contrast to the random sample model where "the sample is random  and the
       population is  deterministic" in the geostatistics frame of  reference, "the population is
       random and the sample is deterministic."  This is interpreted to mean that the locations to
       be sampled can be selected deterministically (i.e., by subjective choice rather than by
       random numbers).

              In most of the methods, the first step is to select the appropriate assumptions  to
       mathematically model the population and to estimate from data or  past experience a
       function that describes the spatial intercorrelation within the population (the variogram, the
       covariance function, etc.). Fortunately, many of the techniques are very robust relative to
       errors in these preliminary estimates;  i.e.,  about the same answer is obtained even if the
       covariance function picked is somewhat wrong. Because of this, the covariance function (or
       variogram)  is often picked  subjectively  in situations where  there  is  very little initial
       information. As data are  obtained, these estimates can be updated so that the final answers
       can be based on much better covariance estimates.

              Kriging gives an unbiased estimate with minimum expected-square-error for the
       average of some attribute of the population over a specified spatial region of the population.
       There is a variety  of different types of Kriging methods  depending  on the assumed statistical
       structure of the population and the one or more attributes to be estimated.

              Conditional simulation is a body of techniques that produce by computer a  detailed
       version of what the population might be, consistent with the measured data and the assumed
       statistical  structure  and  spatial intercorrelations of the population.  A number  of such
       simulations  can be produced,  each of which agrees  with  the data  and has the proper
       population statistical structure. These can be useful to characterize the degree of variability
       remaining after a suite of samples (data) have been collected from the population, either
       to  help in finding future locations  for sampling or to study the variation in possible
       responses to remediation  action  consistent with the amount of information embodied in the
       current data. If there is too much variation, more data can be collected.

              Because of the subjectivity involved, the geostatistical procedures are much more  in
       the spirit of engineering design. Much judgment is used, and experience, skill, and art are
       important  factors.  However, if properly used, the methods  make a maximum use of the
       available data and are optimally cost-effective.  Thus,  they are appropriate for situations
       where sample  collection  and analysis  are very costly  and it is desirable to get as much
       information as possible from each datum.


              The random sampling model and the geostatistical model are very different frames
       of reference. Both  are valid and appropriate in various situations.  There are also various
       elaborations of the  random sampling model which take into account many of the structural
       features  incorporated  into the geostatistical model.  The situation is  analogous to the
       apparent conflict in physics between particle theory and wave theory. A fairly elaborate
       examination of the  various considerations favoring one model over the  other in a particular
       application is given by Borgman and Quimby (13).
                                                                particularly suitable for III; a
                                                                     may be useful for 1,11; b

       The amount of data needed depends  on the balance permitted between objectivity and
       subjectivity  allowed in obtaining the correlation/variogram functions.

                                                              if great objectivity is required: i
                                                      if substantial  subjectivity is permitted: iii

       References:  (1,  13, 14,  15, 16, 17, 18, 19, 20, 21, 22)

(7)    Double sampling

              The procedure proceeds in two phases or stages. A first data  set is collected and
       analyzed. The information from this first set is then used  to plan and perform the collection
       and analysis of a second set of samples. Within the random sampling models, the first data
       set is often used to  obtain estimates of the population variance, and this provides guidance
       for picking  the best  sample  sizes  for the subsequent set  of  measurements. In the
       geostatistical frame of reference, the first data set is often picked to estimate or verify the
       variogram function selected and other structural  assumptions made about the population,
       while the second data  set provides the basis for a least-cost computation of the estimates
       with the target precision.

              Another common application of the method uses  a large first screening sample  with
       inexpensive  analysis procedures, followed by a small second sample, carefully selected on the
       basis of the first sample results and analyzed with more precise, costly methods.

              Procedurally and  logistically, the  double sampling process has much to recommend
       it in many situations. It is a very practical and reasonable tool for characterization of waste
       materials. A natural extension is to use more than two stages in the process (see items 8
       and 9 below).
                                                                              A; I,II,III; a; n
       References: (2, Chapter 13; 4, Chapter 6)

(8)     Formal fWald)  sequential testing

              These procedures are developed in the context of quality control.  A sample of data
       is collected and used in a formal test procedure with three possible actions: (1) accept the
       null hypothesis, (2)  reject the null hypothesis, or (3) take another sample  of data. A formal

       stopping rule, derived under some choice of optimality conditions, is used to determine
       which of the three actions is most appropriate.

              The advantage of the sequential methods is that the action level is reached with the
       smallest possible number of data measurements for a given power of the test. This is
       contrasted with fixed sample size procedures where the sample size is selected initially,  and
       then the data are collected and processed. Usually, in order to guarantee the desired  test
       power, a larger size of sample than needed is chosen as a conservative measure.

              In contrast to the sequential, phased data collection discussed in the next section, the
       formal procedures here are very restricted as to the decisions which are made. Essentially
       one  may  accept, reject, or continue  sampling relative  to a formal  test  of hypothesis
       concerning some population parameter (e.g., mean or variance), usually with a likelihood
       ratio test.  In the next section, the  decision framework is  much more loose and practical,
       allowing the introduction of intangibles, alternative courses of action, value judgments,  etc.
       The  formal Wald scheme may be a part of that, but it probably will only be used as a
       contributing source of information for the actual  practical decision process. Thus,  the
       method here is good for  waste characterization if it is subsidiary  to other  methods.
       However,  care  must  be  exercised to make  sure  the underlying assumptions of  the
       methodology are satisfied in the waste storage circumstances.
                                                                              A; I,II,III; b; 11

       References: (23, 24)

(9)     Sequential, phased data collection and analysis (a more informal, "ad hoc" type of sequential

              This is an approach that is potentially very important for waste characterization,
       remediation, and monitoring. It is a natural extension of double sampling, can incorporate
       aspects of the sequential testing and procedures, can be  used with either random sampling
       or geostatistical techniques, and generally allows the user to "proceed up the learning curve"
       in a way so as to make the required decisions with as  little  data  as necessary. The
       sequential  method is intentionally a loose,  informal procedure which proceeds in alternating
       stages of data collection and judgmental interpretation.  The project team must carefully
       define action  alternatives, types of data needed at each stage,  and  so forth, but  the
       sequential  accumulation of information  is not restricted  to the  formal Wald process
       discussed above.  A sequential approach often allows the project team to avoid the  over-
       expenditure usually associated with a single-stage process consisting  of (1) plan sample,  (2)
       collect sample, and (3) make decision.  In order to guarantee that  enough information is
       available for a safe decision, the number of samples collected is typically set  conservatively
       large in the single-stage scheme.   If the individual samples are expensive to collect and
       analyze, the cost of this conservatism  can be substantial.  The phased sampling approach
       avoids this by undersampling initially, then adding additional samples as needed to minimize
       the number of samples ultimately collected. There may, of course, be additional costs in
       placing contracts for each stage.  These must be factored into the overall cost.

                                                                             A; I,II,III; a; 11
       References: (24, Chapter 8; 4,  Chapters 6  and 7)


(10) Compositing Tmixing) of diverse samples

              Compositing is not a complete statistical technique in itself, but it can be used with
       many different techniques. Compositing is particularly useful where collection of samples
       is cheap and easy but analysis is costly, where relatively few (less than 20 percent) of the
       results are expected to be above the blank values,  and where the goal is  to estimate a
       population average.  The samples are separated into groups, and the samples in a given
       group are mixed together (composite) to obtain a single group sample for  analysis.  The
       number of samples per group and the number  of groups  are determined by the
       requirements of the estimate accuracy and analysis accuracy. A type of composite kriging
       is presented and studied by Weight (25).

       References: (4, Chapter 7; 25; 26; 27; 28; 29)

(11)    Bayesian methods

              At one time there was  a very fierce debate among statisticians as to whether classical
       or Bayesian statistics are most desirable for use. Classical statistics were based on the use
       of prior information to design the mathematical model and assumptions for the statistical
       population but required that only newly collected data with proper randomization be used
       in making the inferences.  Bayesian methods (based on a famous relation called Bayes
       theorem) required the subjective estimation of a priori probability laws be made by the user,
       then data collected and processed with the a priori probabilities to get a posteriori probability
       law. This  was then  used for making  the statistical inferences. The subjective a priori
       probabilities are strongly involved in the results of the inference and, thus, two investigators
       can reach different conclusions from the same data if their initial  opinions differed. Some
       statisticians found this very repugnant, since they wanted objective procedures that would
       lead all investigators to the same results. The Bayesian enthusiasts argued that in the real
       world, prior information is important and should be used to get as good an inference as
       possible with available data. It did not bother them that investigators could get different
       answers since this was not any different from other fields  of engineering and  science where
      judgment is an important component of any study.

              Many of the same considerations arise  in waste characterization. For compliance
       with regulation and for questions  of litigation, highly  objective procedures are  advantageous.
       This suggests classical statistics and random sampling  methods. However, where cost
       effectiveness  is important (samples and analysis very costly), it is useful to introduce more
       subjective judgment to enable the  utilization of smaller data sets. The Bayesian methods
       are  one scheme for doing this  in a systematic, formal way. The formal procedures may bury
       the  investigator in mathematical difficulties, but an informal Bayesian approach (engineering
       and scientific judgment at each step of a sequential, phased approach) can be very useful.
       The previously mentioned methods of geostatistics, double sampling, phased  and  multistep
       sampling, and sequential testing  and estimation are all methods which also attempt to allow
       the  user to  input prior information to arrive at greater efficiency and cost effectiveness.

                                                                        A,B; I,II,III; a,b; 1,11

       Reference:  (30)


(12) Importance   sampling

              This is a new proposed procedure which tries to sample sequentially so that each
       new observation is  collected at a location that is most important or significant to the
       decision or action being considered.   The method needs more  work for validation in
       practice, but it appears very promising.
       Reference: (31)

(13)    Bootstrap resampling/simulation for bias and confidence intervals with minimal data

              There  is considerable interest in obtaining confidence intervals for estimates and
       procedures which are so complex statistically that it is almost impossible to derive the
       intervals from mathematical manipulations. In this case, one answer would be to perform
       multiple sets of experiments and get a confidence interval from the value computed from
       each set. The cost of the multiple sets of experiments usually make this approach infeasible
       from budgetary concerns.  A body of techniques that have come to be called "bootstrap
       methods" provides a sort of "poor man's confidence interval" that can be derived from a
       single set of measurements.

              In this procedure, a single sample of n observations is used to provide an empirical
       distribution function for the population; then  synthetic samples are developed by Monte
       Carlo procedures which imitate what one would have obtained with additional sets of n
       observations. The statistic or quantity of interest is then computed for each of many such
       additional data sets,  and the histogram of the  statistic or quantity  is processed to provide
       a "rough" estimate of the confidence interval for the quantity. There is no intent to argue
       that this confidence interval will  be as good as a confidence interval based on the full
       measurement sets;  rather, it provides  a rational  procedure that can be introduced when time
       and cost prevent more elaborate measurement programs.

              Various extensions of these procedures have been developed at the U.S. Army Corps
       of Engineer Waterways Experiment Station,  Vicksburg, Mississippi,  in the context of coastal
       engineering, and the results of these studies  are just now  beginning to appear in the journals
       (32,33,34). More applications are  in press.
                                                                           B; I,II,III; a; 11,111
       References: (35, 36, 37,  38, 39)

(14)    Artificial intelligence, expert opinions,  subjective probabilities  futility  theory)

              Sometimes useful where there is little  or no data available, yet some planning or
       other action is necessary.

       References: (40, 41, 42,  43)

(15)   "Hot spot" analysis and search theory

              The search for targets with a given shape (elliptical, square, etc.) with an appropriate
       grid which may be redefined as the effort proceeds.
                                                                            A,B; I,II,III; a; i

       References: (4, Chapter 10; 44, p289ff; 45)

(16)   Extremal Analysis

              Developed in engineering fields  as a way to  extrapolate to maxima or extremes.
       Should be useful for "hot spot" problem.
                                                                            A,B; I,II,III; a; i

       References: (46, 47, 48, 49)

(17)   Censored or truncated data analysis

              These methods  are useful where the analysis method is incapable of detecting values
       beyond a particular threshold magnitude, or when data  sets contain many "less-than" values.
       Only values above or below some limit can be measured. The data have  been censored at
       some cutoff. How can one  proceed with such data so they are properly included  in
       calculations with other methods?  Some research has been done on these problems, and the
       listed references will prove an entry into the literature.

       References: (50, 51, 52, 53)

(18)   Nonparametric ^distribution-free) methods

              This group of techniques  uses rank order methods to obtain statistical procedures
       that are valid regardless of what the probability law for the parent population is. They are
       especially useful in waste characterization problems when little is known about the parent
       population and data are sparse.

              There is a variety of nonparametric techniques covered by a very extensive body of
       literature. The methods often  involve simple computational requirements and sacrifice  very
       little "power" or decision accuracy as compared to parametric methods. The latter have
       more rigid assumptions that must be satisfied.

       References: (54, 55, 56)

(19)   Attribute (yes/no)  testing, rather  than numerical testing

              The real question in  many  studies is "which action should be  taken?"  Various
       numerical measurements and their confidence intervals are only intermediary to the actual
       remedial action that is to be selected. However, there are methods in decision theory and
       other statistical fields that relate directly to the discrete selection process.  This includes the

       area of cost-benefit analysis (57),  which has considerable importance in most decision

       References: (57, 58, 59, 60, 61)

              Although reference 58 is illustrated with applications from the petroleum industry,
       the methodology is useful in many contexts, including management of environmental waste.

       The rest of the list is a tabulation of various other bodies of statistical techniques that may
be relevant to waste characterization studies. There is no claim for completeness in this list, and
almost certainly other topics could be included following further deliberation and discussion.  The
methods are listed here with appropriate references, but with only  brief discussion.

(20) Proportion testing

              Where the problems are concerned with  the fraction or proportion of a population
       with a specified attribute, one can often use  techniques, usually constructed around the
       binomial  probability  law, which produce estimates and/or make tests of hypothesis
       concerning  proportions.

       References: (2, 4)

(21) Control  chart  methodology  and quality  control

              Methods developed  in industrial quality control may find applicability in waste
       management practices.  Reference 62 is  a very old, but classical and very readable,
       introduction to traditional techniques. In the last several decades, however, there has been
       an explosion of new, and often somewhat controversial, procedures in this field associated
       with the names "Deming" (63) and "Taguchi" (64).  Each of these "schools" of methods has
       its advocates and detractors. The new procedures are distinguished from the older type of
       quality control by their involvement with the whole process of production which leads to
       quality, and  particularly the human element.  There is less emphasis on just meeting contract
       specifications  and more emphasis on  striving to increase  quality (or  even achieve zero
       defects) by whatever ideas or methods can be found.

       References: (62, 63, 64, 65, 66, 67)

(22) Compliance   testing

              This is really a subtopic within quality control which is concerned with testing to
       guarantee that quality specifications are met. It usually functions in a monitoring mode.

       References:  same as previous topic, plus (68)

(23)  General multivariate methods (cluster,  discriminate, classification, principal component, and
       factor analysis)

              Most environmental data are actually multivariate in nature. That is, more than one
       quantity is required to characterize the situation and associated risks. Consequently, it is
       often artificial to pick one (perhaps dominant)  property and make tests of hypothesis solely
       on that scalar quantity. There is a rich body of techniques that deal directly with the total
       vector   of  properties.    These methods  will probably  find  much greater use in  the
       environmental context in the future.

       References: (69,  70, 71, 72, 73)

(24)   Source search strategy, as in locating source of a plume of contaminants in ground water

              This can be viewed as a  parallel  to the "hot spot"  search strategy  of topic (15).
       Actually, both are subcategories  of the field  of pattern recognition and characterization.
       Much work has  been done in recent years on pattern recognition relative to the analysis of
       remote  sensing data from satellite measurements.  References  74 and 75 give discussions
       of the topic from this more general perspective.

       References: (4, Chapter 10; 44, p.289ff; 45, 74, 75)

(25) Decision  theory  techniques

              In a general sense, this can be thought  of as including all of the other topics in this
       discussion,  as  they influence decision-making relative to environmental concerns.  However,
       the terminology is usually not used in that  broad sense.  Instead, it is generally taken to
       refer to a body  of  statistical  procedures often related  to economics  and  business
       management.  These methods  include decision trees and various simulation  schemes.

       References: (58, 76)

(26)    System reliability theory for components and total system

              Complex systems can sometimes  be subdivided into separate components which
       behave independently,  or  act according to joint  probability laws with other components. The
       study of how the reliability of the total system is related to the separate reliability of each
       component (or subgroup of components) constitutes  system reliability theory. It may be
       thought of as a subfield of operations research.

        Reference: (77)

(27)   Risk analysis ^related to decision theory)

              Just as with decision theory, the area of risk analysis summarizes many aspects of
       the various topics in this list.  It  is usually  restricted to decision-theory concerns. These
       often relate to economics, business management, and psychology.

        References:  (78, 79, 80, 81, 61, 82)

 (28)  Response  surface techniques
              This is a recently developed group of techniques which  organize, usually  with
        multiple regression methods, the search for maxima, minima, or other critical values of a
        variable or vector defined over a multidimensional space. Thus, it might have applicability
        to topic (24) in searching for the source of a plume of contaminants in ground water, or
        moving toward a maximum "hot spot" concentration in topic (15).  However, the methods
        are much more general than those applications and seek to produce a multiple regression
        model of the variations of the variable(s) in the space of definition.

        References:  (83, 84, 85)

        No attempt has been made to  place the later topics (numbers 20-28)  within the classification
 scheme used previously, since most are specialized techniques or broad  general classes of methods
 that do not allow a general statement of appropriateness  to  waste  characterization.    For
 completeness, however, they  are included.

        It is clear that a large  number of statistical methods are available. As a first step,  the listing
 provides a set of possible approaches for characterizing a new situation for which the appropriate
 method is not obvious.  It also underscores and emphasizes that the statistical methods for waste
 characterization are not "cut and dried" procedures laid out in handbooks,  particularly when there
 is a large premium placed on cost-effectiveness.  In large waste characterization and remediation
 projects, substantial  savings  may be possible with suitable initial  statistical investigation of the  most
 appropriate methods.

        Clearly there is a need for some type of handbook, giving examples, evaluations, and
 comparisons of these various methods as applied to waste characterization. Such a book should
 carefully avoid being "captured" by any particular school of thought within statistics, but should
 include all possible ways to approach the problems in waste characterization.

1.      Borgman, L.E.   1988.  New advances in methodology  for  statistical tests  useful  in
       geostatistical  studies.  Mathematical  Geology  20(4):383-403.

2.      Cochran, W.G.  1977. Sampling Techniques, 3rd Ed. John Wiley and Sons, Inc., New York.

3.      Sukhatme, P.V., and B.V.  Sukhatme.  1970.  Sampling Theory of Surveys with Applications,
       3rd Ed. Iowa State University Press,  Ames, Iowa, 452 pp.

4.      Gilbert, R.O.  1987. Statistical Methods for Environmental Pollution Monitoring.  Van
       Nostrand Reinhold Co., New York, 320 pp.

5.      Gaugush,  R.F.    1987.  Sampling  design  for  reservoir  water  quality measurements.
       Department of the Army,  Environmental and  Water  Quality  Operational  Studies.
       Instruction Report E-87-1.


6.      Gaugush, R.F. 1990. Sampling Design Software, version 2.0. Environmental Laboratory,
       U.S.A.E. Waterways Experiment Station, P.O. Box 631, Vicksburg, MS 39180.

7.      Raj, Des. 1968. Sampling Theory. McGraw Hill, New York, 302 pp.

8.      Gy, P. 1967. L'Enchantillonnage des Minerais en Vrac: Tome I: Theorie Generate, Revue
       de L'Industrie Minerale. Also published in Memoires du Bureau de Recherches Geologique
       et Minieres, No. 56, 1967,  186 pp. (Chapter 4, Theorie de L'Enchantillonnage Equiprobable,
       give an excellent presentation of the theoretical foundation for Gy's method).

9.      Borgman, L.E. 1991. An exposition of Gy's derivation in L'Enchantillonnage des Minerais
       en Vrac, Theorie Generale. Statistics Laboratory Report No. 91-2, Department  of Statistics,
       University of Wyoming, Laramie, Wyoming.

10.     Gy, P.M. 1982. Sampling of Particulate Materials, Theory and Practice. Elsevier Scientific
       Pub., New York, 431 p.

11.     Pitard,  F.F.   1989.    Pierre Gy's  Sampling  Theory and Sampling  Practice,  Vol.  I:
       Heterogeneity and Sampling. CRC Press, Boca Raton, Florida, 214 pp.

12.     Pitard, F.F. Undated.  Software  Programs: QE1,  VARIO, and  COAL. Francis F. Pitard,
       14710 Tejon Street, Broomfield,  CO   80020.

13.     Borgman, L.E., and W.F. Quimby. 1988. Sampling for tests of hypothesis when data are
       correlated in space and time. In Principles of Environmental Sampling  (Keith, Lawrence
       H, ed.), p. 25-43, The American Chemical Society.

14.     Journel, A.G., and C.J. Huijbregts. 1978. Mining Geostatistics. Academic  Press, New York.

15.     David, M. 1977. Geostatistical Ore Reserve Estimation. Elsevier, New  York, 600 pp.

16.     Clark, L. 1978. Practical Geostatistics. Elsevier, London, 125 pp.

17.     Isaaks, E. H., and R.M. Srivastava. 1989. An Introduction to Applied Geostatistics. Oxford
       University Press, Oxford, England, 561 pp.

18.     Borgman, L.E., M. Taheri, and  R. Hagan.  1984. Three-dimensional, frequency  domain
       simulation of geological variables.  Geostatistics for Natural Resources  Characterization,
       Part 1, NATO ASI Series C, Mathematical and Physical Sciences, p. 517-541.

19.     Borgman, L.E.    1987. New advances in methodology  for  statistical tests useful in
       geostatistical studies.  Presented at the National Meeting of the Mathematical Geologists
       of the United States, Redwood City, CA, April 1987.

20.     Easley, D.H., L.E.  Borgman, and P.N.  Shive. 1990.  Geostatistical simulation for geophysical
       applications, Part 1: Simulation.  Geophysics 55:1435-1440.

 21.    Flatman, G.T., EJ. Englund, and A.A. Yfantis.  1988. Geostatistical approaches to the
       design of sampling regimes.  In Principles of Environmental Sampling, L.H. Keith, ed.,
       American Chemical Society, Washington, D. C., pp. 73-84.

 22.    Englund, Evan.  Undated.  Software Program:   GEO-EAS.  U.S.  EPA,  EMSL-LV, Las
       Vegas, NV  89193-3478.

 23.    Siegmund, D. 1985, Sequential Analysis, Tests and Confidence Intervals. Springer-Verlag,
       New York, 272 pp.

 24.    Wetherill, G.B., and K.D.  Glazebrook.  1986. Sequential Methods in Statistics, 3rd Ed.
       Chapman and Hall, New York, 264 pp.

 25.    Weight, W.D. 1989. Knowledge-Based  Computer Systems for Two Topics in  Geology:
       Micron Gold  Deposits and Planning  Geological  Sampling Procedures. Ph.D. Thesis,
       Department  of Geology and Geophysics, University of Wyoming, Laramie, Wyoming.

 26.    Garner, F. C.,  M.A.  Stapanian, and L.R. Williams.   1988.   Composite sampling for
       environmental monitoring.   In. Principles of Environmental Sampling, L.H. Keith, ed.,
       American Chemical Society, Washington, D. C., pp. 363-374.

 27.    Rajagopal, R. 1990. Economics of screening in the detection of organics in ground water.
       Chemometrics  and Intelligent Laboratory Systems 9:261-272.

 28.    Rajagopal, R., and L.R. Williams. 1989. Economics of sample compositing as a  screening
       tool in ground water quality monitoring. Ground Water Monitoring Review, Winter 1989,
       pp. 186-192.

 29.    Garner, F.C.,  M.A.  Stapanian, E.A.  Yfantis,  and L.R. Williams.  1989.  Probability
       estimation with sample compositing techniques, Journal of Official  Statistics 5(4):365-374.

 30.    Box, G.E.P., and G.C. Tiao. 1973. Bayesian Inference in Statistical Analysis.  Addison-
       Wesley, Reading, Massachusetts,  518 pp.

 31.    Johns, M.V.  1988. Importance sampling for bootstrap confidence intervals. Journal of the
       American  Statistical  Association 83(403):709-714.

 32.    Prater, M.D., T.A.  Hardy,  H.L. Butler, and L.E.  Borgman.  1985. Estimating error of
       coastal stage frequency curves.   Proceedings 19th International Conference on Coastal
       Engineering,  Sept. 3-7,  1984, Houston, Texas.

33.    Walton, T., and L. Borgman. 1990. Simulation of nonstationary non-Gaussian water levels
       on Great Lakes. Jour. Waterway,  Port, Coastal, and Ocean  Engineering, ASCE 116(6):664-

34.     Miller, M.C., W.E. Roper, L.E. Borgman, and J.J. Westerink. 1991.  Development of water
       level and wave height design data. Proceedings, Panel on Wind and Seismic Effects, UJNR,
       23rd Meeting, May 14-23, 1991,  Tsukuba, Japan.


35.     Efron,  B.  1982. The Jackknife, the Bootstrap,  and Other Resampling  Plans. Regional
       Conference Series in Applied Mathematics No. 38, Society  for Industrial and Applied
       Mathematics, Philadelphia,  Pennsylvania, 92  pp.

36.     Efron,  B., and R. Tibshirani.  1986. Bootstrap methods for standard  errors, confidence
       intervals, and other measures of statistical accuracy.  Statistical Science  l(l):54-77.

37.     Efron, B. 1990. More efficient bootstrap computations. Journal of the American Statistical
       Association 85(409):79-89.

38.     Diaconis,  P.,  and B. Efron. 1983. Computer-intensive methods in statistics.  Scientific
       American  248(5): 116-129.

39.     Larson, K.M. Undated.  Software Program: Bootprog. SRA Technologies Inc., 4700 King
       Street,  Suite 300, Alexandria, VA 22302.

40.     Plansky, L.E. 1989. On the application of artificial intelligence to geology. The Compass,
       p. 165-169.

41.     Weight, W.D., L.E. Borgman, and W.F. Quimby. 1989. GEOSAM, A knowledge-based
       computer system for planning geological sampling procedures. The Compass, p. 171-184.

42.     Waterman, D.A.  1986.  A  Guide to Expert Systems. Addison-Wesley Pub., Reading,
       Massachusetts,  419 pp.

43.     Van Ee, J. Undated. Software Program: Environmental Sampling Expert System (ESES),
       U.S. EPA, EMSL-LV, Las Vegas, NV 89193-3478.

44.     Davis,  J.C. 1973. Statistics and Data Analysis in Geology, 2nd Ed. John Wiley and Sons,
       New York, 646 pp.

45.     McCammon, R.B. 1977. Target intersection probabilities for parallel-line and continuous-
       grid types of search. Mathematical Geology 9(4):369-382.

46.     Gumbel, E.J. 1958. Statistics of Extremes. Columbia University Press, New York, 375 pp.

47.     Gumbel, E.J. 1954. Statistical theory of extreme values and some practical applications.
       National  Bureau of Standards Applied Mathematics Series  No.  33,  U.S. Government
       Printing Office, Washington, D.C.

48.     Borgman,  L.E.   1961.  The exact  frequency  distribution  of near  extremes. Journal
       Geophysical Research 66:3295-3307.

49.     Galambos, J. 1978. The Asymptotic Theory  of Extreme Order Statistics.  John Wiley and
       Sons, New York, 352 pp.

50.     Schneider, H.  1986. Truncated and Censored Samples from Normal Populations. Marcel
       Dekker, Inc., New York, 273 pp.


51.    Newman, M.C., et al. 1990. Software Program: Uncensor V2.0m. Savannah River Ecology
       Laboratory, Aiken, SC  29801.

52.    Helsel, D.R. 1990. Less than obvious: Statistical treatment of data below the  detection
       limit. Envir. Sci. & Tech. 24:1766-1774.

53.    Lambert, D., B. Peterson, and I. Terpenning. 1991. Nondetects, detection limits, and the
       probability of detection. Jour. Amer. Stat. Assoc. 86:266-277.

54.    Gibbons, J.D.  1971. Nonparametric  Statistical Inference.  McGraw-Hill Co., New York,
       306 pp.

55.    Lehmann, E.L., and HJ.M. D'Abrera. 1975. Nonparametrics: Statistical Methods Based
       on Ranks. Holden-Day Pub., San Francisco,  457 pp.

56.    Hettmansperger, T.P. 1984. Statistical Inference Based on Ranks.  John Wiley and Sons,
       New York, 323 pp.

57.    Mishan, E.J. 1976. Cost-Benefit Analysis. Praeger Publishers, New York, 454 pp.

58.    Newendorp, P.D. 1975. Decision Analysis for Petroleum Exploration. The Petroleum Pub.
       Co., Tulsa, Oklahoma, 668  pp.

59.    Brown, R. V., A.S. Kahr, and C. Peterson. 1974. Decision Analysis for the Manager. Holt,
       Rinehart, and Winston, New York, 618 pp.

60.    Fabrycky, W.J., and G.J. Thuesen.  1974. Economic Decision Analysis. Prentice-Hall, Inc.,
       Englewood Cliffs, New Jersey, 390  pp.

61.    Singleton, W.T., and J. Hovden (editors).  1987. Risk and Decisions. John Wiley and Sons,
       New York, 232 pp.

62.    Shewhart, W.A.   1939. Statistical Method From The Viewpoint of  Quality Control.
       Department of Agriculture Graduate School,  Washington, DC, 155 pp.

63.    Walton, M. 1986. The Deming Management Method. Dodd, Mead, and Company,  New
       York, 262 pp.

64.    Taguchi,  G. 1981.  On-Line Quality Control During Production. Japanese  Standards
       Association, Tokyo, 154 pp.

65.    Barker, T.B.  1986. Simulation by experimental design ~ A Taguchi concept. 1986 ASQC
       Quality  Congress  Transaction, Anaheim, CA, American  Society for  Quality Control,
       Milwaukee, WI.

66.    Box, G.E.P., and S. Bisgaard. 1987. The scientific context of quality improvement. Quality
       Progress, vol. 20.

67.     Juran, J. M., P.M. Gryna, Jr., and R.S. Bingham. 1979. Quality Control Handbook, 3rd Ed.
       ASQC Quality Press, Milwaukee, WI.

68.     Schilling, E.G. 1982. Acceptance Sampling in Quality Control. Marcel Dekker, New York,

69.     Eaton, M.L.  1983. Multivariate Statistics, A Vector Space Approach. John Wiley and Sons,
       New York, 512 pp.

70.     Tabachnick, E.G., and L.S. Fidell.  1983. Using Multivariate Statistics.  Harper and Row,
       Pub., New York,  509 pp.

71.     Press, S.J. 1972. Applied Multivariate Analysis.  Holt, Rinehart, and Winston, Inc., New
       York, 521 pp.

72.     Flury, B. 1988. Common Principal Components  and Related Multivariate Models. John
       Wiley and Sons, New York, 258  pp.

73.     Romesburg, H.C. 1990. Cluster Analysis for Researchers.  Robert E. Krieger Publishing
       Company, Malabar, Florida, 334 pp.

74.     Ahuja, N., and B. Schachter. 1983. Pattern Models. Wiley Interscience, New York, 309 pp.

75.     Andrews, H.C. 1972. Introduction  to Mathematical Techniques in Pattern Recognition.
       Wiley Interscience, New York, 242 pp.

76.     Ferguson, T.S. 1967. Mathematical Statistics, A Decision Theoretic Approach. Academic
       Press, New York, 396 pp.

77.     Barlow, RE., and F. Proschan. 1975. Statistical  Theory of Reliability and Life Testing
       Probability Models. Holt, Rinehart, and Winston,  Inc., New York, 290 pp.

78.     Borgman, L.E. 1972. Risk evaluation in engineering. In  McGraw-Hill Yearbook of Science
       and Technology. New York.

79.     Megill, R.E.  1977.  An  Introduction to Risk Analysis.  Petroleum Publishing Co., Tulsa,
       Oklahoma, 197 pp.

80.     Borgman, L.E. 1963. Risk criteria. Jour. ASCE, Waterways  and Harbors Div., Vol. WW3,
       pp. 1-35.

81.     Kogan, N., and M.A. Wallach. 1964. Risk Taking.  Holt,  Rinehart, and Winston, New York,
       278 pp.

82.     Sjoberg, L. (editor). 1987.  Risk  and Society. Allen and Unwin, London, 245 pp.

83.     Box, G. E. P., and N.R. Draper.   1987, Empirical Model-Building and Response Surfaces.
       John Wiley and Sons, New York, 669 pp.


84.      Deming, S. N., et al.  1991. Sequential  Simplex Optimization. CRC Press, Inc., Boca Raton,

85.     Cornell, J.A. 1984. Volume 8: How to Apply Response Surface Methodology. American
       Society for Quality Control, Milwaukee, WI.
 «U.S GOVERNMENT PRINTING OFFICE: 1992. eue.OO^it 1617    144

                                                                               > me
                                                                               CO 3 3
                                                                               o> < »•
                                                                               3 ^' CD
                                                                                  3 ^
                                                                                  o> w

                                                                              b13 m
                                                                              i   <


                                                                            —. ~, rn i