&EPA
              United States
              Environmental Protection
              Agency
              Office of Research and
              Development
              Washington, DC 20460
EPA/625/R-94/003
September 1994
Manual
Alternative Methods for Fluid
Delivery and Recovery

-------
                                          EPA/625/R-94/003
                                            September 1994
              Manual

    Alternative  Methods for
 Fluid Delivery and Recovery
Center for Environmental Research Information
    Office of Research and Development
    U.S. Environmental Protection Agency
         Cincinnati, Ohio 45268

-------
                                       Notice
The information in this manual has been funded wholly or in  part by the  U.S. Environmental
Protection Agency (EPA). It has been subjected to the Agency's peer and administrative review and
approved for publication as an EPA document. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.

-------
                                               Contents
                                                                                                  Page

Chapter 1   Introduction	   1
            1.1   Role of Alternative Methods of Delivery and Recovery	   1
                 1.1.1   Containment	   1
                 1.1.2  Restoration	   2
                 1.1.3  Technical Impracticability	   2
            1.2   Overview of Contents	   2
            1.3   Summary of Important Issues	   2
                 1.3.1   Access	   4
                 1.3.2  Depth	   4
                 1.3.3  Recovered Phase	   4
                 1.3.4  Geology	   4
                 1.3.5  Availability  	   5
                 1.3.6  Current Experience	   5
                 1.3.7  Operating and Maintenance Costs	   5
                 1.3.8  Monitoring	   5
            1.4   References	   6

Chapter 2   Horizontal and Inclined Wells	   7
            2.1   Well Construction	   8
                 2.1.1   Directional Drilling  Components (or "Equipment")	   8
                 2.1.2  Drilling Fluids	 11
                 2.1.3  Inclined Wellbores Created With Sonic Drilling	 12
                 2.1.4  Well Installation and Completion	 14
                 2.1.5  Well Development	 17
                 2.1.6  Pumps	 17
            2.2   Design Considerations	 18
                 2.2.1   Pattern of Flow	 18
                 2.2.2  Wellbore Specifications	 18
                 2.2.3  Geologic Site Conditions	 21
                 2.2.4  Distribution of Contaminants	 24
                 2.2.5  Wellbore Hydraulics	 24
                 2.2.6  Site Conditions	 25
            2.3   Applications	 25
                 2.3.1   Pump and Treat	 25
                 2.3.2  Vapor Extraction	 26
                 2.3.3  Free-Product Recovery	 26
                 2.3.4  Sparging and Vapor Extraction	 27
                 2.3.5  Bioremediation	 27
                 2.3.6  Flushing	 27
                 2.3.7  Soil Monitoring and Sampling	 28

-------
                                        Contents (continued)

                                                                                                 Page

           2.4  Case Histories	 28
                2.4.1   Overview	 28
                2.4.2  Typical Costs of Completed Projects	 29
                2.4.3  Selected Examples	 29
           2.5   References	 36

Chapter 3  Induced Fractures	 39
           3.1  Methods of Inducing Fractures	 39
                3.1.1   General Considerations	 40
                3.1.2  Monitoring Fracture  Location	 41
                3.1.3  Equipment	 42
                3.1.4  Well  Completion	 43
           3.2  Design Considerations	 43
                3.2.1   Flow to a Gently Dipping Fracture	 45
                3.2.2  Forms of Fractures	 46
                3.2.3  Geologic Conditions	 47
                3.2.4  Fracture Permeability	 48
                3.2.5  Fracture Size	 49
                3.2.6  Well  Completion	 50
                3.2.7  Site Conditions	 50
           3.3  Applications	 50
                3.3.1   Vapor Extraction	 50
                3.3.2  LNAPL Recovery	 51
                3.3.3  Dual-Phase  Recovery	 52
                3.3.4  NAPL Recovery	 52
                3.3.5  DNAPL Recovery	 52
                3.3.6  Bioremediation Applications	 52
                3.3.7  Air Injection  	 53
                3.3.8  Steam Injection 	 53
                3.3.9  Electrokinetics	 53
                3.3.10 In Situ Treatment Zones	 54
                3.3.11  Barriers 	 54
                3.3.12 Monitoring  	 54
           3.4  Case Histories	 54
                3.4.1   Overview   	 54
                3.4.2  Selected Examples	 56
           3.5  References	 59

Chapter 4  Interceptor Trenches	 63
           4.1  Trench Construction	 63
                4.1.1   Conventional Methods	 63
                4.1.2  Specialized Methods	 64
                                                  IV

-------
                            Contents (continued)

                                                                                      Page

4.2  Design Considerations	  66
     4.2.1   Pattern of Flow to a Trench	  66
     4.2.2   Site Geology	  67
     4.2.3   Hydrologic Conditions	  68
     4.2.4   Trench Hydraulics	  71
     4.2.5   Contaminants	  72
     4.2.6   Site Conditions	  72
     4.2.7   Trenches and Horizontal Wells	  72
4.3  Applications	  73
     4.3.1   Hydrodynamic Control	  73
     4.3.2   Plume Interception	  73
     4.3.3   NAPL Recovery	  74
     4.3.4   Vapor Extraction	  74
     4.3.5   Other Applications	  75
4.4  Case Histories	  76
     4.4.1   Overview	  76
     4.4.2   Selected Examples	  76
4.5  References	  85

-------
                                                 Figures
Figure                                                                                               Page

2-1   Horizontal well used to intercept a plume	   7
2-2   Some applications of horizontal wells	   8
2-3   Directional drilling rig	   8
2-4   Directional drilling with a downhole motor	   9
2-5   Potential damaged zones around a horizontal wellbore	  11
2-6   Resonant stress wave providing energy to the drill bit	  12
2-7   Sonic drill head	  13
2-8   Creating a horizontal well with  the pull-back method	  15
2-9   Completing  a blind well using the washover pipe method	  15
2-10  Types of force normal to axes during installation of casing or well materials	  16
2-11  Flow paths to a horizontal well  in a thin aquifer with no regional flow	  18
2-12  Radial flow  in a vertical plane in the vicinity of a horizontal well	  18
2-13  Nomenclature describing  a horizontal well	  19
2-14  Orientations of horizontal  well relative to principal directions of permeability	  22
2-15  Drawdown resulting  from  a horizontal well as  a function of time and dimensionless length	  23
2-16  Discharge from horizontal wells of different lengths operated at constant drawdown	  24
2-17  Schematic of distribution of influx into a horizontal well	  24
2-18  Two approaches using horizontal wells to intercept contaminant plumes	  26
2-19  Schematic of flow patterns with horizontal wells	  26
2-20  Recovery of LNAPL with one horizontal well, and water control with another well below	  27
2-21  Inclined wellbore created  with resonant sonic  drilling, Sandia National Laboratories	  28
2-22  Distribution  of total lengths of horizontal wells drilled by summer  1993	  29
3-1   Injection pressure during creation of a fracture by injecting a sand-laden slurry	  41
3-2   Injected fluids leaking out of fracture during propagation	  41
3-3   Typical pattern of uplift over a shallow, gently  dipping hydraulic or pneumatic fracture	  42
3-4   Aboveground  equipment used for pneumatic fracturing	  42
3-5   Fracture created  by  isolating  a  zone with  a straddle packer	  43
3-6   Fractures created at the bottom of driven casing	  44
3-7   Methods of completing wells  with induced fractures	  44
3-8   Air discharge from a well  intersecting a flat-lying circular fracture  as a function of the permeability
      of the fracture	  45
3-9   Pressure, flux and travel time to a conventional well and a well intersecting a sand-filled fracture ...  46
3-10  Plan  and  section of a typical hydraulic fracture created in overconsolidated silty clay	  47
3-11  Locations of major areas  of soils related to glaciers  and vertisols	  48
3-12  The effective aperture of an open slot that is equivalent to that of a fracture filled with proppant. ...  49
3-13  Factors that affect how fractures should be propped	  49
3-14  Flux of air to three flat-lying fractures shown in cross section	  51
3-15  Well configuration to recover liquid and  vapor phases using suction at the  ground surface	  52

                                                    vi

-------
                                         Figures (continued)

Figure                                                                                           Page

3-16  Schematic application of steam injection into induced fractures	  53
3-17  Temperature as a function of depth and time	  54
3-18  Area of LNAPL recovery using fractured and conventional wells	  57
3-19  Methods of completing a well containing hydraulic fractures to recover LNAPL	  58
3-20  Discharges from C-10, C-12, and PW-1 as functions of time	  58
3-21  Cross-section along line of piezometers showing change in  head after five days of pumping	  59
4-1    Cross sections of a basic interceptor trench configuration showing  permeable backfill and
      perforated casing	  63
4-2   Schematic of continuous excavation and completion technique	  65
4-3   Schematic of deep trench excavation  equipment	  65
4-4   The pattern of flow to a trench with no regional  gradient	  66
4-5   Drawdown at the midpoint of a trench where discharge is constant	  68
4-6   Discharge from a trench held at constant drawdown	  69
4-7   Cross section of a trench partially penetrating an aquifer in  a regional flow, showing the
      vertical extent of aquifer captured by the trench	  69
4-8   Dimensionless arrival times as a function of location for various strengths and directions of
      regional flow	  70
4-9   Map of the distribution of influx and flow along an idealized trench	  71
4-10  Controlling the flow through a contaminated area by pumping from an upgradient trench  and
      injecting into a downgradient trench	  73
4-11  Interceptor trench coupled with a hydraulic barrier to minimize recharge from  an adjacent
      surface waterbody	  73
4-12  Trench used to capture plume and prevent migration to an offsite well	  73
4-13  Trench systems for DNAPL recovery	  74
4-14  Enhanced recovery of DNAPL by reintroducing water to increase the hydraulic gradient	  74
4-15  Excessive pumping rates  from the ground-water and DNAPL drains causing "pinch off' of the
      DNAPL plume	  75
4-16  Plan of the Westminster gasoline recovery site	  80
4-17  Plan of the Laramie DNAPL recovery  site	  80
4-18  Map of the northeastern Illinois vapor extraction site and detail of trench completion	  81
4-19  Typical subsurface profile	  82
4-20  Site map showing locations of monitoring wells, storage silos,  and  probable contaminant
      source area	  82
4-21  Extent of TCE plume in 1984 and 1986	  82
4-22  Model domain showing location of the interceptor trench, the 1986 TCE  plume limit, and extent
      of the capture zone	  83
4-23  TCE concentration distribution through time	  83
4-24  Site plan showing locations of initial interceptor trench segments	  84
4-25  Schematic cross section of trench construction	  84
4-26  Monthly extraction  from interceptor trench  sumps	  85
4-27  Observed and predicted discharges from trench system	  85
                                                  VII

-------
                                                Tables
Table                                                                                            Page

1-1    Issues Affecting Application of Alternative Methods for Delivery or Recovery	   3
2-1    Specifications of Directional Drilling Rigs	   9
2-2   Wellbore Curve Measurements	  19
2-3   Strengths of Prepacked Screens	 21
2-4   Tensile Strengths of Pipe	  21
2-5   Summary of Variables	 23
2-6   Summary of Applications of Horizontal Wells	  28
2-7   Sites Where Horizontal Wells Have Been Used for Ground-Water Recovery	  30
2-8   Sites Where Horizontal Wells Have Been Used for Vapor Extraction	  31
2-9   Sites Where Horizontal Wells Have Been Used for Assorted Environmental Applications	  32
2-10  Sites Where Horizontal Wells Have Been Used for Several Purposes	  32
2-11   Specifications of Horizontal Well,  Geismar, Louisiana	  34
3-1    Design Considerations for  Induced Fractures	 44
3-2   Summary of the  Effects of Induced Fractures Used To Improve the Recovery of Wells	  55
3-3   Specific Capacities Before and After Hydraulic Fracturing	  56
3-4   Specifications of Fractures Used  During the Pilot Test	  57
3-5   Average  Discharges and Ratios of Discharge	  58
4-1    Variables Used in Trench Analysis	 68
4-2   Summary of Applications of Trenches for Delivery or Recovery: Site  Characteristics	  77
4-3   Summary of Applications of Trenches for Delivery or Recovery: Trench Parameters	 78
4-4   Summary of Applications of Trenches for Delivery or Recovery: Product Recovery	  79
                                                  VIM

-------
                                 A cknowledgments
This manual was prepared under Contract No. 68-CO-0068 by Eastern Research Group, Inc. (ERG),
forthe U.S. Environmental Protection Agency's (EPA's) Office of Research and Development (ORD),
Center for Environmental Research Information  (CERI), Cincinnati, Ohio. Edwin Earth  of CERI
served as the Project Director and provided  technical direction and review. Heidi Schultz of ERG
directed the editing and production of this manual.

Principal authors were Larry Murdoch, Center for GeoEnvironmental Science and Technology,
University of Cincinnati, Engineering Research Division, Cincinnati, Ohio, and David Wilson, Inde-
pendent Environmental Consultants, Arvada, Colorado. Kevin Savage, Bill Slack, and Jim Uber of
the University of Cincinnati assisted in preparing sections of the manual. The following individuals
deserve special acknowledgment for their review: Steven Acree and Randall Ross, EPA ORD,
Robert S. Kerr Environmental Research Laboratory, Ada, Oklahoma; Randy Breeden, EPA Office
of Emergency and Remedial Response, Washington, DC; Subijoy  Dutta, EPA Office of Solid Waste,
Washington,  DC;  Rene  Fuentes, EPA Environmental  Services Division, Seattle,  Washington;
George Losonsky, International Technology  Corporation, Lake Charles, Louisiana;  and  Eugene
Donovan, Leslie Sparrow, and Bruce McClellan, HydroQual, Inc.,  Mahwah, New Jersey.
                                          IX

-------
                                             Chapter 1
                                            Introduction
Controlling subsurface fluids is among the highest pri-
orities  in  managing sites  with in situ contamination.
Some applications direct fluid movement continually in-
ward towards the site,  whereas others attempt to re-
cover contaminants and, ultimately, close the site. Both
these types  of  applications use  physical  methods of
delivery and  recovery, perhaps in conjunction with other
methods, to  meet their respective goals. Probably the
most common physical system used to recover fluids at
contaminated sites is a  vertical well with a submersible
water pump. However, this system is by no means the
most effective in every situation. Therefore, this manual
focuses on alternatives to conventional applications of
vertical wells—alternatives that improve performance or
reduce cost.

Developments in drilling technology over the past dec-
ade have  resulted  in considerable improvements.  For
example, technological  advancements now allow wells
to curve to a horizontal orientation, potentially placing
great lengths of screen in  a contaminated  zone and
offering several  additional advantages over their vertical
counterparts. Other developments involve methods of
fracturing  rock and soil  to improve the performance of
vertical and,  in some cases, horizontal wells. This tech-
nique is particularly suited  to increasing recovery from
low permeability formations, but it also has some spe-
cialized applications.  Trenches filled with gravel have
long been used for dewatering, and they can still be a
valuable option  at a contaminated site.

This manual presents these three alternative methods
of enhancing delivery and  recovery. The scope of this
document is confined to physical enhancement tech-
niques.  For this  reason, detailed discussion of chemical
methods (e.g., surfactant flushing) and thermal methods
(e.g., steam stripping) has been omitted. These chemi-
cal methods are mentioned, however, because the al-
ternative methods can  be  used in conjunction  with a
variety of other processes, including flushing and steam
stripping. In fact, nearly any in situ method of remedia-
tion that involves subsurface fluid flow either has been
or could be  used with  the methods  described herein.
Additional information on related technologies is avail-
able in the  following U.S. Environmental  Protection
Agency (EPA) documents:  Handbook on In Situ Treat-
ment of Hazardous Waste Contaminated Soils, Subsur-
face Contamination Reference Guide, Leachate Plume
Management, and Evaluation of Ground-Water Extrac-
tion Remedies,  Volumes 1-3 (1-4). Additional informa-
tion related to the existence, movement, and  transport
of ground water is described in Volumes 1  and 2 of the
EPA document Ground Water (5, 6).

1.1   Role of Alternative Methods  of
      Delivery and Recovery

Alternative  methods of delivery and  recovery are in-
tended to improve the performance of remedies ranging
from hydrodynamic containment, which arrests offsite
migration,  to restoration, which reduces  contaminant
concentrations.  The methods are applicable during in-
terim measures, in particular by improving the  effective-
ness of containment, and they also can augment the
performance of a variety of remedial actions selected as
possible long-term remedies.

1.1.1  Containment

Recent studies  (4) have evaluated the performance of
ground-water extraction  systems. These studies have
identified factors related to  hydrogeology,  contaminant
properties,  and  system  design that may  impede the
ability of those  systems to reduce concentrations  to
targeted values over the entire area of contamination.
Heterogeneities, such  as natural fractures, karstic fea-
tures, or variations in stratigraphy,  result in preferential
flow paths between wells, thus limiting recovery capa-
bilities. Moreover, nonaqueous-phase liquid (NAPL) is
present in the subsurface at many contaminated sites
(7, 8) and, with components  of the NAPL slowly dissolv-
ing in ground water, may act as a  persistent source of
contamination.

Those  issues have led to a higher appreciation of the
difficulty of remediation at some sites, and a recognition
of the current technical impracticability of complete re-
mediation at other sites (9).  Nevertheless,  the ability to
halt migration and prevent the spread  of contamination
is well within the scope of current technology. Thus,
emphasis  grows  on  expeditiously  implementing  sys-
tems to halt dissolved phase migration by controlling

-------
hydraulic gradients early in the investigation of a site.
Aggressively pursuing free-phase NAPL recovery also
is recommended at the earliest possible time to reduce
the potential for further contamination. Interim measures
must be  coordinated with final  remedies so that they
constitute the first phase of the overall remedial strategy.
Accordingly, containment and source recovery during
interim measures are a viable method of risk reduction
at many contaminated sites.

The effectiveness of interim measures and the extent to
which they reduce risks of exposure are based on their
performance. In many cases, systems of  alternative
methods  improve the ability to control hydraulic gradi-
ents beyond the capabilities of vertical wells. Central to
improved performance is the geometry of flow fields that
the systems produce. Whereas vertical wells are limited
to producing radial flow (in the absence of regional flow),
horizontal wells and trenches  produce linear flows in a
horizontal plane, and horizontal fractures produce verti-
cal flows. This expands the versatility of containment
systems, allowing them to be tailored  more closely to
the configuration of the site. An additional benefit of the
alternative  methods is the improvement in the amount
of fluid recovered per unit of power. Increased values of
specific capacities occur when using enhanced systems
in  productive aquifers.  Moreover,  in tight  formations
where  recovery typically occurs at fixed drawdown, the
enhanced systems are capable of increasing the dis-
charges  compared  with conventional  vertical  wells.
Where impenetrable surface structures cover NAPLs or
aqueous  plumes, horizontal wells offer the capability to
recover contaminants close to their source.

1.1.2  Restoration

Fluid   recovery  by  pumping  vapor,   aqueous,  or
nonaqueous phases may be sufficient to restore some
sites,  but heterogeneities, adsorption, low vapor pres-
sure, and other factors conspire to limit the effectiveness
of fluid recovery alone as a remedial tool. A wide range
of  other  methods has  been  and are  currently being
developed  to  accelerate  in  situ remediation.  Some
methods  involve injecting liquids or vapors that carry
nutrients  and electron  acceptors to stimulate in situ
organisms, thereby promoting  contaminant degrada-
tion.  Other methods involve degrading contaminants by
injecting an oxidant or mobilizing the contaminant with
a surfactant to enhance recovery. Another approach is
to heat the subsurface  by injecting steam or hot air to
vaporize organic compounds.

This diverse range of technologies shares the common
need to control the flow of subsurface fluids.  Where the
magnitude  or uniformity of fluid flow is impaired, the
performance of a conventional remedial method can  be
significantly reduced. Formations of low permeability or
those marked  by heterogeneities or other  hydrologic
complexities present particular difficulties. Stagnation
zones in the flow field resulting from well placement and
design further limit fluid flow. In view of these issues, the
alternative methods described  here can be a valuable
asset when used  in conjunction  with other remedial
methods  involving fluid flow. In this application, the pri-
mary purpose of the delivery and recovery techniques is
to improve the effectiveness of remedial methods involv-
ing  fluid flow. Improved effectiveness reduces costs or
may allow a technique to be successfully applied under
conditions where it would otherwise be unsuccessful.

1.1.3  Technical Impracticability

Currently, it is technically impracticable to restore some
sites to accepted standards due to the  presence of
immobile NAPL, strongly sorbed contaminants, forma-
tion heterogeneities, or other factors (9). As a result, the
methods described herein may be no better than vertical
wells in addressing limitations to restoration at some
sites. Alternative methods of delivery or recovery should
be implemented only when they provide a substantive
benefit to technically achievable goals.

1.2   Overview of Contents

The handbook contains chapters describing:

• Horizontal wells

• Induced fractures

• Interceptor trenches

Each chapter  contains a  description of  construction
methods, with particular emphasis on factors to consider
when selecting the method. The factors affecting design
are summarized, with particular emphasis on the pattern
of flow induced  by the method, the geologic and hydro-
logic site conditions that affect performance, and access
requirements for site implementation. In  addition, an
overview of applications suggests various possibilities,
although by no means have all the potential applications
been described. Each chapter closes with  several case
histories describing site conditions, system design, and
results. Cost information is  provided based  on published
or informally reported descriptions. The cost of a particu-
lar technique can vary greatly, however, and the reader
should use the  data given  here as a preliminary guide
to be supplemented by the cited  references and site-
specific vendor estimates.

1.3   Summary of Important Issues

Horizontal  wells, induced fractures, and interceptor
trenches  present slightly different options for controlling
subsurface fluids, and each has slightly different advan-
tages and disadvantages among the issues facing ap-
plications at contaminated  sites. Table 1-1  summarizes
essential issues.

-------
Table 1-1.  Issues Affecting Application of Alternative Methods for Delivery or Recovery

Issue                   Horizontal Well                   Induced Fracture
                                                                                               Trench
Access

Fragile structures
over target

Poor access over
target

Depth

<6 m


6-20 m


>20 m


Recovered Phase

Aqueous

LNAPL
DNAPL



Vapor


Geology

Normally consolidated
clay

Swelling clay


Silty  clay till


Stratified sediment or
rock
Vertically fractured
sediment or rock
Coarse gravel


Thick sand




Rock



Availability
Current Experience
(Approximate)
•  Minimal surface disturbance    D  Evaluate effects of surface
                                     displacement
•  Standoff required
•  1  m minimum depth


•  Cost of guidance system
    increases at >6 m

•  No depth limit within
    environmental applications
O  Requires accurate drilling;
    best if water table
    fluctuations are minor

©  Requires accurate drilling
    and site characterization
•  Consider omitting gravel
    pack to save costs
©  Smearing of bore wall may
    reduce performance

©  Smearing of bore wall may
    reduce performance

©  Smearing of bore wall may
    reduce performance

O  Anisotropy may limit vertical
    influence of well
•  Orient well normal to
    fractures when possible
O  Possible problems with hole
    stability; penetrating cobbles

©  May be difficult to access
    top and bottom of
    formation;  hole stability
    problems

O  Feasible, but drilling costs
    more in rock than in
    sediment

10 to 20 companies with
capabilities; nationwide
coverage but may require
equipment mobilization

150 to 250 wells at 50 to
100 sites
O  Possible with horizontal well
•  1-2 m minimum depth
    No depth limit within
    environmental applications
©  Best with access to individual
    fractures


O  Caution; steeply dipping
    fractures may cause downward
    movement

•  Best with access to individual
    fractures
O  Induced fractures may be
    vertical and limited  in size

•  Relatively large, gently dipping
    fractures expected

•  Relatively large, gently dipping
    fractures expected

©  Stratification may limit upward
    propagation and  increase
    fracture size

©  Good where induced fractures
    cross-cut natural fractures
    (overconsolidated sediment and
    rock)

D  Permeability enhancement may
    be unnecessary

D  Permeability enhancement may
    be unnecessary
•  Widely used in oil, gas, and
    water wells drilled in rock
Several companies offer service;
nationwide coverage with equipment
mobilization
200 to 400 fractures at
20 to  40 sites
D  Excavation expected to be
    infeasible

D  Excavation expected to be
    infeasible
•  Installation with common
    equipment

©  Excavation costs increase with
    depth

O  Specialized excavation
    methods required
    Widely used to ensure capture;
    accommodates water table
    fluctuations

    Assuming mobile phase
    present and accurately located
©  Requires tight seal on top of
    trench
    Large discharge expected
    relative to alternatives

    Large discharge expected
    relative to alternatives

    Large discharge expected
    relative to alternatives

    Good way to access many thin
    beds or horizontal partings
©  Orient trench perpendicular to
    natural fractures when possible
•  Stability a concern during
    excavation

•  Stability a concern during
    excavation
D  Excavation difficult but blasting
    possible to make trench-like
    feature

Shallow trench (<6 m) installation
widely available from local
contractors; deep trench will require
mobilization

1,000+ trenches  at many hundreds
of sites
Key
• Good application
© Moderately good
O Fair, with possible technical difficulties
D Poor; not recommended using available methods

-------
1.3.1  Access

Access to the ground overlying the contaminated region
is commonly restricted, or access may be obstructed by
fragile structures such as product lines, tanks, and build-
ings. Horizontal wells are ideally suited to address ac-
cess problems because the drill rig can be located at
some distance from the obstructed area. Alternatively,
induced fractures created from vertical wells are limited
to areas where access allows creation of the well. Frac-
tures  have been  induced in the vicinity of horizontal
wells, although experience with this application for envi-
ronmental purposes  is currently limited.  Because  in-
duced  fractures  displace  the  ground  surface,  this
technique may be inappropriate where such displace-
ment  could damage overlying  structures.  Finally, the
excavation required to  create trenches  precludes their
application in many areas of limited access.

1.3.2  Depth

Depth is a minor factor for horizontal wells and induced
fractures; both can be created at depths far greater than
those that environmental applications require. Minimal
depths of 1 to 2 meters (3.3 to 6.6  feet) are commonly
required to contain induced  fractures within the subsur-
face. Drilling horizontal wells requires similar minimum
depths if high pressure jets are used to cut the bore. The
current  maximal limit of radio-beacon  guidance sys-
tems, which are the most economical method of guiding
horizontal wells, is 6 to 8 meters  (19.7 to  26.2 feet).
Therefore, creating a horizontal bore deeper than 6 to 8
meters (19.7 to 26.2 feet) requires a more sophisticated
and expensive guidance system. Trench depth greatly
influences cost, because deeper trenches increase both
the volume of contaminated soil that must be handled
(and possibly disposed of) and the cost of excavation.
The maximal limit of conventional excavation equipment
is approximately 4 to 6 meters  (13.1 to 19.7  feet), so
creating deeper trenches requires specialized  equip-
ment. Currently, trenchlike structures 100 meters (328
feet) or more  in depth can be created using specialized
excavators (Table 1-1).

1.3.3  Recovered Phase

All the alternative methods can recover aqueous phase
contaminants, but  they may differ in  their ability to  re-
cover NAPL.  It  is  feasible to recover free-phase light
(LNAPL) and dense (DNAPL) nonaqueous-phase liquid
using a horizontal well,  but this application  requires
particularly accurate depth control during drilling, as well
as a detailed understanding of  each site's subsurface
variability. Moreover,  even with  accurate placement of
the well, rises and falls  in the water table—and thus the
LNAPL  layer—can significantly hamper recovery of
LNAPL. Induced fractures can improve the recovery of
LNAPL from tight  formations. In addition, creating sev-
eral fractures over a range  of depth can  account for
water table fluctuations. Trenches  are widely used to
recover shallow LNAPL because they can intercept lat-
eral migration and accommodate changes  in the water
table.

Each  of the three methods  can be used  to increase
recovery of DNAPL. However, the benefits of recovering
some DNAPL must  be balanced  against the risk of
aggravating the recovery of liquid that  remains in the
ground. Addressing  this  issue  entails  locating the
DNAPL, which is a formidable task unto itself, and using
methods that limit the creation of vertical channels in the
area containing DNAPL. Horizontal wells offer the pos-
sibility of accessing a DNAPL layer without creating a
vertical conduit. Conversely, the location of induced frac-
tures  cannot be determined  precisely, so it is possible
that they will create pathways for downward migration.
Moreover, methods of sealing induced fractures remain
untested. Therefore,  the feasibility of abandoning the
fractures is currently unknown. Trenches  also  create
vertical channels,  but the trench location can be deter-
mined precisely, and methods of sealing and abandon-
ing trenches are available. Feasibility tests,  modeling,
and pilot studies are recommended before conducting
any DNAPL recovery project.

1.3.4   Geology

It  is  possible to use  horizontal  wells  in  fine-grained
sediments, although smearing along the wall of the bore
may require aggressive development. Anisotropy pro-
duced by stratigraphic layering limits the vertical influ-
ence  of a  horizontal  well. In  contrast,  anisotropy
produced  by  vertical fractures tends to enhance the
performance  of a  well oriented  perpendicular to the
fractures. Although drilling through coarse gravel and
rock is technically possible,  these formations are rela-
tively  difficult to penetrate and thus  increase the cost of
the well. In thick sand, a horizontal well may present
limited access to the upper or lower regions. Moreover,
steering a directional bore is difficult in soft sand, so
drilling accuracy may be compromised.

Induced fractures can increase the  recovery of subsur-
face fluids from fine-grained  sediments and particularly
from overconsolidated deposits, such as glacial drift or
swelling clay. However, induced fractures  may tend to
dip steeply in normally consolidated  deposits, which can
limit their size and effectiveness. Stratigraphic layering
can inhibit upward propagation of an induced fracture
and improve  performance,   even in normally consoli-
dated formations. However,  a  horizontal fracture may
offer little benefit in finely interbedded sands and clay.
Where vertical fractures are the dominant flowpaths,
induced fractures can be a benefit  if they are flat-lying
and cut across multiple natural fractures.  In contrast,
where induced fractures follow one vertical fracture,

-------
they may be of limited benefit. Both types of behavior
have been noted. Highly permeable formations, such as
sands and gravels, will benefit little from the permeability
enhancement  that  induced fractures  offer, although
there may applications where treatment materials are
injected into induced  fractures in sand. Induced frac-
tures are widely used to increase  the  discharge from
wells in rock formations.

Trenches are suited to applications in a wide range of
geologic formations. In fine-grained sediments, excava-
tion  is relatively easy, and trenches  offer significantly
greater discharge than vertical wells. In stratified sedi-
ments, they can cut across and access  multiple thin
permeable beds. Of course, excavating rock is generally
infeasible, although trenchlike features  can be created
using explosives placed  in a line of vertical boreholes.

1.3.5   Availability

Horizontal wells, induced fractures,  and trenches are all
methods available within the United  States and Canada.
As of the summer of 1993, approximately a dozen com-
panies had installed a horizontal well, with more enter-
ing  the market since  that time. Most companies that
create horizontal wells for environmental applications
are specialists in that market, although some are capa-
ble  of vertical drilling or  directional  drilling for  utility
installation. Depending on the location of the site, drilling
equipment may need to travel several hundred miles or
more.

At least two  companies specialize in designing and
creating induced fractures for environmental applica-
tions, and  several others  offer fracturing capabilities
along with other environmental services; these compa-
nies are capable of mobilizing within the United States
and Canada. Hydraulic fracturing of water wells is avail-
able from well drilling companies  in  many  locations,
although their experience with environmental applica-
tions typically is limited.

Widely  available excavation equipment  can  create
relatively shallow trenches  and should  require minimal
mobilization. Deeper trenches require the  more sophis-
ticated equipment and  engineering capabilities that spe-
cialist construction companies can offer.

1.3.6   Current Experience

Approximately 150 to  250  horizontal wells have  been
installed during the past 7 years for environmental ap-
plications at 50  to 100  sites. The drilling capabilities
used to install horizontal wells draw upon the experience
from directional boring of hundreds of kilometers to in-
stall utility conduits and oil wells. Specialized drilling and
completion methods designed to meet the challenges of
environmental applications continue to be developed.
From 200 to 400 fractures have been created at 20 to
40 sites. Fracturing capabilities draw upon the experi-
ence of thousands of applications in the petroleum and
water well industries and on the methods and effects of
injection grouting.  Because creating induced fractures
requires particular sensitivity  to  site conditions,  the
methods of creating fractures and anticipating their per-
formance  for environmental applications continues to
develop. This is the  newest of the three  alternative
techniques discussed  in this manual.

Trenches are widely used to recover contaminated fluids
from sites. Experience includes more than 1,000 appli-
cations  at hundreds of sites, making this the most ma-
ture of the techniques discussed here.

1.3.7   Operating and Maintenance Costs

Methods other than a  field of vertical wells may reduce
operating  and maintenance (O&M) costs in many ways,
such as by decreasing the number of pumps that must
be used,  reducing the time for routine  sampling and
measurement  of water levels, and  decreasing  power
consumption. Some maintenance aspects, such as re-
conditioning, may  be  more expensive for a horizontal
well or trench than for a vertical well, although the cost
on a per-well basis may be offset if the horizontal well
or trench  is equivalent to several  vertical wells.  One
comparative study  (10) found that the capital  cost of one
horizontal well was similar to five vertical wells, but the
O&M cost for the horizontal well was less than one third
that of the five vertical wells. In general, however, costs
are sensitive to site conditions and design quality, and
the relative economic advantages  of the  alternative
methods are difficult to generalize.

1.3.8   Monitoring

Induced fractures and horizontal wells in some circum-
stances can provide better monitoring capabilities than
vertical  wells. Small induced fractures can be used to
increase the area sampled and the fluid recovered by a
monitoring well. This application is particularly attractive
in fractured clay or rock, where it can be difficult to obtain
a water  sample of adequate volume from the formation.
Horizontal wells present  the  possibility of  monitoring
beneath potential sources, such as tanks or lagoons,
that cannot be penetrated  by a vertical well.

In situ monitoring of the distribution of head or pressure
when using alternative methods of delivery or recovery
can  be  accomplished using vertical piezometers  in
much the same manner as monitoring recovery by ver-
tical  wells. One exception is during the resolution of
vertical  head gradients; these gradients can be  impor-
tant in the vicinity of an induced fracture, horizontal well,
or trench  that partially penetrates  an aquifer but  are
often ignored when vertical wells are used. Head distri-
butions  caused by vertical wells are commonly moni-

-------
tored  using piezometers with long  screens, whereas
vertical gradients can only be measured with clusters of
multiple piezometers with  short screens  at different
depths or piezometers with  multiple ports separated by
packers (11).

Monitoring the rates and concentrations of recovered
fluids or the concentrations of compounds in the subsur-
face is independent of the method of recovery; similar
sampling and analytical techniques are used for alterna-
tive and conventional recovery methods.

1.4   References

 1. U.S. EPA. 1990. Handbook of in situ treatment of
    hazardous waste-contaminated  soils.  EPA/540/2-
    90/002.

 2. U.S. EPA. 1990. Subsurface contamination refer-
    ence guide. EPA/540/2-90/011.

 3. U.S. EPA.  1985. Leachate  plume  management.
    EPA/540/2-85/004.

 4. U.S. EPA. 1989. Evaluation of ground-water extrac-
    tion remedies, Vols. 1-3. EPA/540/2-89/054b,c.
 5.  U.S. EPA. 1990. Handbook:  Ground water, Vol. 1.
    Ground  water  and  contamination.  EPA/625/6-
    90/016a (September).

 6.  U.S. EPA. 1991. Handbook:  Ground water, Vol. 2.
    Methodology. EPA/625/6-90/016b (July).

 7.  U.S. EPA. 1993. Evaluation  of the likelihood  of
    DNAPL presence at NPL sites. EPA/540/R-93/073.

 8.  Cohen, R.M., J.W. Mercer, and J. Matthews. 1993.
    DNAPL site evaluation. U.S. Environmental Protec-
    tion Agency. Ada, OK.

 9.  U.S. EPA. 1993. Guidance for evaluating the tech-
    nical  impracticability of  ground-water restoration.
    EPA/540/R-93/080.

10.  Losonsky, G., and  M.S. Beljin. 1992.  Horizontal
    wells in subsurface  remediation. Hazardous Mate-
    rials Control Conference, New Orleans, LA (Febru-
    ary).

11.  Rehtlane, E.A., and F.D.  Patton. 1982. Multiple port
    piezometers vs.  standpipe piezometers:  An eco-
    nomic comparison. Second Aquifer Restoration and
    Ground  Water Conference, Columbus,  OH (May
    26-28), pp. 287-295

-------
                                               Chapter 2
                                  Horizontal and Inclined Wells
Recent advances in  directional  drilling  have forever
changed the well's image. No longer must a well be a
vertical cylinder; directional drilling methods can create
wellbores with almost any trajectory. Wells that curve to
a horizontal orientation are particularly suited for envi-
ronmental applications.

Horizontal wells are  not technically an innovation; the
water supply industry has used horizontal wells for many
years to collect water from beneath  rivers and  other
bodies of water (1). However, these  wells are drilled
radially from large caissons, making installation expen-
sive and rarely practical for environmental applications.

Directional drilling methods use specialized bits to curve
bores in a controlled arc; trajectory is monitored with
electronic sensors. This enables bores to be initiated at
a relatively shallow angle from the ground surface and
gradually to curve to horizontal (Figure 2-1). Blind well-
bores terminate in the subsurface. In some cases,  how-
ever, the well is turned upward and returns to the ground
surface, which makes it accessible from both ends. This
is called a continuous wellbore. Most horizontal  well-
bores are drilled  in roughly a straight line, but lateral
curves are certainly possible and may be important in
certain circumstances.
Figure 2-1.  Horizontal well used to intercept a plume. The solid
          line represents a blind well; the solid line plus the
          dashed line represents a continuous well.
Directional drilling can create two basic types of bores:

• Wellbores preserve the permeability of the host for-
  mation.

• Boreholes penetrate the subsurface without particular
  regard to the permeability of the host.
Creating wellbores requires special methods to avoid or
restore permeability damage, whereas methods used to
create boreholes are less demanding.

Inclined, or slant, wells are  installed in a straight well-
bore typically created with conventional drilling equip-
ment that is tilted. Inclined wells drilled with conventional
drilling equipment cannot be horizontal if they originate
from flat-lying ground.

In some cases, such  as when drilling into  a slope, an
inclined well can be horizontal or even slant  upward.
Directional drilling equipment can create wells that du-
plicate the geometry of inclined wells.

The  primary reason for including inclined wells in this
chapter is that they have recently received renewed
interest as a product of sonic drilling. This technique has
the remarkable ability to create wellbores without drilling
fluid  and extraneous drill cuttings; all of the material in
the path of the wellbore is collected as continuous core
sample. These attributes  make  sonic drilling ideal for
certain sampling and environmental drilling applications.
Thus, the  treatment of inclined wells in this chapter is
limited to applications involving sonic drilling.

Because of its orientation, a horizontal well is particu-
larly  suited to recovering  contaminants distributed as
broad, flat layers (Figure 2-2a). Such a distribution may
occur when LNAPL  floats on a water table or when
DNAPL accumulates on a low-permeability  bed. Direc-
tional drilling can place 100 meters (328 feet)  or more
of well  screen  in the layer, whereas the screen of a
vertical  well may  intersect  less than  a meter  of an
LNAPL  layer. Often, when contaminants have moved
with  the regional flow, the compounds distribute as  a
long, narrow, roughly horizontal plume. Asingle horizon-
tal well placed along the long axis of the plume (Figure
2-2b) offers an ideal geometry for recovering those con-
taminants.

In tight formations such as bedrock or till, where vertical
fractures provide the  primary  flow  paths, a horizontal
well can intersect many vertical fractures  (Figure 2-2c).
This  application facilitates access to the preferred flow
paths, increases well discharge,  and controls fluid flow
in the formation (2, 3).

-------
                                                       2.1   Well Construction
  a.
  b.
  c.
  d.
Figure 2-2. Some applications of horizontal wells: a) intersect-
          ing flat-lying layers, b) intercepting  plume elon-
          gated by regional gradient, c) intersecting vertical
          fractures, and d) access beneath structures.

Some advantages of a horizontal well  are unrelated to
hydrologic performance.  In many locations, such as be-
neath landfills, tanks, buildings, roads,  lagoons, or bod-
ies of water, access limitations prohibit entry by a drill
rig, prevent  penetration by a vertical hole, or restrict the
aboveground facilities necessary for  recovery  opera-
tions. As a result, recovery, sampling, or monitoring with
conventional drilling technology is difficult beneath many
structures that  may be sources of contaminants. Hori-
zontal and inclined wells, however, overcome those dif-
ficulties  by  allowing the rig to be adjacent  to  the
obstructing  structure, and the  wellbore to be  created
beneath  it (Figure 2-2d).

This chapter contains four  sections. The first section
describes current methods of constructing wellbores
and completing horizontal wells. The second  section
summarizes technical and economic factors that need
consideration when planning to use  horizontal wells at
a  contaminated site. Following this is a section  that
presents  some possible environmental  applications.
The chapter closes with an extensive summary of envi-
ronmental projects that have used horizontal wells and
a detailed description of four case histories.
Constructing a horizontal well entails directionally drill-
ing a wellbore and placing perforated tubing in the well-
bore to hold it open and provide access to subsurface
fluids. The well is developed to increase permeability in
the vicinity of the bore, and a pump is installed to recover
fluids. Details of this process follow.

2. 1. 1   Directional Drilling Components
        (or "Equipment")
Directional drilling uses three specialized components:
• A drilling  rig to power the system
• A bit to create  a curved hole
• A guidance system to locate and steer the bore

2.1.1.1   Directional Drilling
Directional drill rigs that currently create horizontal wells
for environmental applications typically consist of a car-
riage that slides on a frame and holds the drill rods at
an angle of 0 to 45 degrees (Figure 2-3). In  most cases,
hydraulic power energizes a motor on the carriage and
rotates the drill rods. A chain drive, rack and  pinion drive,
or hydraulic cylinder may push or pull the carriage to
advance or retract the  drill  string. A pump on the rig,
capable  of  handling slurries at 1 to 30 megapascals
(MPa)  (145 to 4,351 pounds per square inch [psi]), is
typically  used to inject drilling fluid.
  Chain
  Drive
                                   Drill Rod
                                               Bit
Figure 2-3. Directional drilling rig.

Drill rigs are available in a range of sizes and are distin-
guished chiefly by the torque and push/pull force they
provide  (Table 2-1).  Despite the range in sizes and
various details that manufacturers offer, the rigs share
some common features. The drill rig provides thrust to
the drilling tool and pull-back to the drill string. When
drilling a vertical well, the weight of the drill motor and
the drill string provide the downward force on the drill bit.
In contrast, when drilling a directional wellbore, the drill

-------
Table 2-1. Specifications of Directional Drilling Rigs (May
         1994)
              Mini Rigs
Midi Rigs
Maxi Rigs
Thrust/Pullback

Maximum
Torque
Drilling Speed
Carriage Speed
Carriage Drive
Drill Pipe
Length
Drilling
Distance3
Power Source
< 66.7 kN
(15,000 Ibs)
2.7 kN-m
(2,000 ft-lb)
>130 RPM
>30 m/min
Cable or
chain
1 .5-3 m
<200 m
<150 HP
66.7-444 kN
(15,000-
100,000 Ibs)
2.7 kN-m - 27
kN-m (2,000-
20,000 ft-lb)
130-100 RPM
28-30 m/min
Chain or rack
and pinion
3-9 m
200-600 m
150-250 HP
>444 kN
(>1 00,000 Ibs)
27 kN-m
(20,000 ft-lb)
<100 RPM
<28 m/min
Rack and
pinion
9-1 2m
>600 m
>250 HP
a Assumes nominal 12-inch wellbore, which is the effective maximum
 diameter wellbore for a mini-drill rig.
rig—typically a cable, chain,  or  rack  and pinion  sys-
tem—must provide the forward force on the drill string.
The rig must be anchored to provide a reaction against
which the mechanical system on the  rig  can operate.
Anchoring is typically accomplished by driving  stakes
through openings at the front of the rig (Figure 2-3) and
attaching the drill rig to a buried weight ("dead man") or
simply attaching it to a large, heavy piece  of equipment
on the surface. The drill rig must provide sufficient thrust
to advance the drill string the full length of the proposed
wellbore, and sufficient pulling force to retract  casing
into the completed wellbore. The relationship between
thrust and  drilling distance  depends  on  the formation
type and use of drilling fluids.

The drill rig  must also provide torque to the drill string.
Most drilling methods require that the drill  string rotates
while  it  advances into the wellbore in order to reduce
friction on the drill string. The drill rig must have sufficient
capacity to  overcome wellbore friction and supply the
necessary torque to the drill string throughout the pro-
posed length of the wellbore.
In addition, some rigs use pneumatic or hydraulic ham-
mers  to help advance the drill string,  a method rarely
used for environmental applications. In most cases, the
hammers are mounted on the drill rig, but several manu-
facturers also offer downhole hammers.

Some contractors have directional drill rigs of varying
sizes, whereas others may only have one size of drill rig.
In general, the midi-size rigs (Table 2-1)  are the most
versatile and can complete most projects.
The oil industry has  used other methods  of directional
drilling, with perhaps the most notable technique using
a high energy water jet to create the bore (4).  This
method employs a whipstock to bend continuous steel
tubing in a  tight arc (less than a 0.5-meter [1.6-foot]
radius of curvature) at the bottom of a vertical hole. The
result is a roughly straight horizontal hole extending
radially from a vertical access bore. This geometry pre-
sents some advantages over the inclined  entry rigs for
applications where access is extremely limited. Although
some have proposed environmental applications of this
method (5), apparently no demonstrations have been
conducted.

2.1.1.2   Creating and Steering the Wellbore

Directional drilling uses a downhole assembly that cre-
ates the wellbore and induces a curve in the trajectory.
Wellbores are created by cutting the formation  with a
rotating bit, water jet, hammer, or combination of these.
The trajectory is curved using a tool that is eccentric with
respect to the axis of the drill rod. In some cases, there
is a slight bend in the rod behind the bit, whereas in other
cases the bit itself has a beveled surface. Most com-
monly, when  creating directional wellbores,  bits are
driven by downhole  motors, jetting tools, and compac-
tion tools.
Downhole Motors. A downhole  motor uses pressurized
drilling fluid to rotate a cutting bit. The advantage  of
downhole mud motors is that they eliminate drill string
rotation and make it possible to drill a wellbore with a
short radius of curvature.  Downhole  motors are  gener-
ally preferred when drilling rock or resistant sediments,
where wellbore control is critical.
To curve the wellbore, a bent rod or "sub" is fixed behind
the motor. Rotating the entire drill string slightly offsets
the bent sub, causing the path of the wellbore to be
approximately straight. Pushing the  drill string without
rotation,  however, curves the wellbore by an amount
dictated by the angle of the  bent sub (Figure 2-4).
                                       Motor      Bit
                                                         Path With Push Only
                              Bent Sub
                          Figure 2-4.  Directional drilling with a downhole motor.

                          Jetting Tools. Jetting tools use hydraulic pressure either
                          to cut  the geologic formation or to assist rotary drilling
                          with bits. Water, mud, polymer, or other drilling fluid can
                          be used to form the jet. The hydraulic jet is directed from
                          either a bent housing or from a drilling fluid port on a drill
                          bit that is attached to a bent subassembly. A pump on the
                          drill rig controls the pressure of the hydraulic jet. To drill
                          a  curved section, the drill string  follows  the advancing

-------
bent subassembly along a curved path. To drill a straight
segment of the wellbore, the drill string is rotated (in one
direction or by alternating  directions/every 180-degree
turn); rotation prevents the hydraulic jet from having a
preferred orientation, and the drill string will not deviate
from the wellbore path.

Compaction Drilling  Tools. A compaction  drilling tool
creates a  wellbore by compacting sediments as  it ad-
vances into the  formation. The typical compaction bit
resembles a wood chisel,  with the  beveled part of the
chisel  providing the eccentricity to  curve the wellbore.
Just as with the bent subassembly, rotation  of a  beveled
bit results  in a straight wellbore, whereas pushing the bit
causes it  to turn. Accordingly, compaction drilling  re-
quires significant torque and thrust by the rig at the
ground surface. Compaction drilling is usually assisted
by cutting  the formation with a water jet or by advancing
the drilling tool with a hammer. Water is used to help
lubricate and  cool the drilling tool, but neither drill cut-
tings nor drilling fluid return to the surface.

Steering during compaction  drilling requires the  resis-
tance of the formation to be sufficient to bend  the drill
rods. In loosely consolidated sand,  it may be difficult to
steer using conventional beveled compaction  bits, al-
though some  bits with  enlarged cutting surfaces are
available to make steering in soft formations easier.

2.1.1.3  Location and Guidance  Systems

Typically, an electronics package placed behind the cut-
ting head:

• Locates the end  of the drill rod.

• Provides the azimuth and inclination of the  bottom-
  hole assembly.

• Provides the orientation of the drill face (bent sub-
  assembly).

This information, combined with the measured length of
drill  pipe,  can  be used to  calculate the position of the
bottomhole assembly. The most common guidance sys-
tems are magnetometer-accelerometer systems,  gyro-
scopes systems, and electronic beacons.

Magnetometer-Accelerometer  System.  This   system
uses three magnetometers to measure the position (azi-
muth) of the tool in the earth's magnetic field and  three
accelerometers to measure the  position (inclination) of
the tool in the earth's gravitational field.  The  system
sends information continuously to a surface computer
that calculates and displays the tool azimuth, inclination,
and drill-face orientation. An advantage of this system is
that the location of the bottomhole assembly is available
in real time.

If subsurface magnetic interference, for example from
metal  pipes  or tanks, is suspected of influencing the
azimuth readings, an electromagnetic secondary survey
system  may  be used in conjunction with a magnetic
guidance tool. The secondary system consists of a cable
laid out on the ground  in a rectangular configuration,
with the well  path running along the center of the  long
dimension of the rectangle. A direct current is applied to
the cable, inducing an electromagnetic field of known
intensity and size in the subsurface around the tool. The
induced magnetic field  generally  is strong enough  to
overcome any indigenous magnetic interference, allow-
ing the tool to display its location  in the induced mag-
netic field. The cable is unobtrusive and can be used at
most sites without fear that it will  inhibit surface activi-
ties. The combination of a magnetometer-accelerometer
tool and the secondary electromagnetic system  pro-
vides the most accurate (plus or minus 2 percent of the
vertical depth [VD]) guidance  system for shallow direc-
tional  drilling. The system loses  accuracy at depths
exceeding 35 meters (114.8 feet).

Gyroscope System. The gyroscope system is based on
the same navigational principles used in guided missiles
and airplanes. The system uses  three gyroscopes  to
measure azimuth and three accelerometers to measure
the inclination.  Before  the survey is made, the gyro-
scopes are aligned to true north at the ground surface.
The gyroscopes detect any deviation from true north
during the survey and relay the information to the  sur-
face, where a computer calculates the  azimuth, inclina-
tion,  and  drilling  tool  orientation.  This  system  is
unaffected by magnetic interference and therefore  may
be used in areas that do not allow the use of a magnetic
guidance system.

Electronic Beacon. The electronic beacon is a  battery-
operated sonde that sends a  radio signal from the  bot-
tomhole assembly. A hand-held surface unit then locates
the position  of the beacon,  calculates the depth  to
the beacon,  and  displays  the  drill-face orientation.
This method  is often called the "walkover" method be-
cause a technician carries the surface unit over the
beacon. The system cannot  be used in areas where
surface  obstructions prohibit access above the bottom
assembly.

Electronic beacons were developed to use with direc-
tional drilling for utility installation. They are widely avail-
able and relatively easy to use, with satisfactory results.
Widely available equipment can be used no deeperthan
approximately 8 meters (26.2 feet);  however, recent
advances in the technology provide walkover systems
that can be used to approximately  17 meters (55.8 feet)
in depth. Electronic beacons currently provide the least
expensive method of  locating and guiding directional
wellbores. Vertical accuracy  of current  systems is  at
least within 5 percent of the VD.
                                                   10

-------
2.1.2   Drilling Fluids

Drilling fluids have a variety of applications during direc-
tional drilling:

• Clean cuttings  from the bit and the end of the well-
  bore, and transport the cuttings to the surface.

• Provide wellbore stability.

• Control  subsurface pressures.

• Cool the drill bit and lubricate the drill string.

• Drive a  downhole drill motor, if one is used.

• Ensure  that formation information is obtained from
  cuttings, cores, or geophysical tools.

• Act as conduit for pressure pulses that communicate
  information from the guidance tool to the ground sur-
  face,  if such a  communication system is used.

Drilling fluids must provide these applications while mini-
mizing both fluid loss to the formation and reduction in
permeability around the  wellbore. No single fluid readily
meets these demands,  requiring  a tradeoff between
drilling needs and treatment objectives.

Drilling needs may constitute a higher priority than well
efficiency  for many horizontal wells used for environ-
mental applications. However, the drilling fluid should be
compatible with well materials and formation  soil and
water. Drilling fluid that penetrates the formation can be
difficult to  remove and, thus, can have  lasting negative
consequences (6). For instance, the fluid may change
the pH, redox potential, or ionic strength of the formation
waters. These changes  can cause minerals in the host
formation to dissolve or  precipitate. Clay-sized particles
in  some drilling fluids also  increase the surface area
available for precipitation of minerals.  Depending  on
formation and drilling fluid chemistry, the minerals most
likely to dissolve or precipitate are magnesium, ara-
gonite, oxides, hydroxides of iron, and possibly dolomite
(7). Drilling fluids that cause precipitation of minerals
reduce pore space and  permeability of the host forma-
tion (Figure 2-5).  The worst place for reduced  perme-
ability is the vicinity of the wellbore because all the fluid
that the well recovers must flow through this zone. Ac-
cordingly, it is critical to select drilling fluids and additives
that minimize damage to the formation  adjacent to the
wellbore.

In  addition to damaging  the formation,  it is possible for
drilling  fluids to  contain contaminants  that increase
disposal costs and  possibly contribute to migration or
sorption  of contaminants  during  drilling. Minimizing
or eliminating liquid drilling fluids is the ideal. As men-
tioned above, compaction drilling techniques require a
minimum  amount of drilling fluid. Compaction drilling,
however, is incompatible with some hydrogeologic con-
ditions. Therefore, many horizontal environmental wells
     Filter or Mud Cake
          0-1 inch
      Flushed Zone
       1-5 inches
    Transition Zone
      5-8 inches
  Axial View of Wellbore
Figure 2-5.  Potential damaged zones around a horizontal
          wellbore.
require fluids. Some experiments with air-based drilling
fluids have been conducted in environmental directional
drilling,  and use of air to contain cuttings transported
from the wellbore has recently been demonstrated (8).
Investigation of cryogenic drilling fluids also is presently
underway (9). Cryogenic drilling fluids have the advan-
tage of freezing the wellbore wall to provide stability and
improve sample recovery. In addition, they vaporize and
minimize cross contamination after use.

Bentonite  slurry and guar gum gel are the least expen-
sive and most widely used drilling fluids. Many drillers
are familiar with bentonite-based drilling fluid,  which
they can easily control to maintain a  stable wellbore and
efficiently  remove cuttings. The  major disadvantage is
that this type of drilling fluid creates  a mud cake that
must be removed during well development to restore the
original  permeability of the formation. A selected drilling
fluid should form an easily removed mud cake (less than
a few millimeters thick).

Guar gum is a cellulose-like polymer derived from the
guar bean and commonly used as a food additive. As a
drilling fluid, guar gum is mixed with water to form a gel
with the following strengths:

• It reduces friction on the drilling tool, thereby increas-
  ing drilling tool penetration rates.

• It lacks  fine-grained  clays that can clog formation
  pores.

• It either biodegrades naturally or  it  can be degraded
  with fluid additives.

• It is nontoxic.

Guar gum has several disadvantages, however:

• Biodegradation of guar gum can increase biomass in
  the vicinity of the wellbore; this can reduce  perme-
  ability, which must be addressed  during completion.

• Low gel strength can limit the  ability to transport
  cuttings.
                                                    11

-------
• Crosslinking to create a stiff fluid improves ability to
  transport cuttings but increases problems with circu-
  lation.

• Elevated temperatures accelerate  bioactivity, which
  could cause premature degradation in the mud  pit.

Guar gum is also used with the other alternative meth-
ods discussed  in this manual. Those applications are
described in detail in following chapters.

2.1.3   Inclined Wei I bores Created With Sonic
        Drilling

This section describes two sonic drilling methods with
respect to  inclined wellbores:  rotosonic and  resonant
sonic.  Both  methods use the same type of sonic drill
head and dual drill pipe or casing advance method, but
they differ in the capabilities to rotate  the casing during
drilling. These methods entail advancing a casing  into
the formation and removing drill cuttings from the inside
of the  casing via an inner drill  pipe/core pipe. In some
cases, the inner drill pipe advances first into the forma-
tion, and the outer casing follows. The major difference
between the two sonic drilling methods is the rotosonic
method relies on rotary  power and water  as a drilling
fluid to  aid in  drilling,  whereas  the resonant sonic
method uses rotary  power and drilling fluid only when
drilling rock. Following is a discussion of the principles
of sonic drilling and sonic drilling equipment, and  a
description of sampling methods used in sonic drilling.

2.1.3.1   Sonic Drilling Principles

The axial oscillations that the sonic drill head produces
induce a sinusoidal energy wave in the drill  pipe  (10).
Maximum energy transfers from the drill head, through
the drill pipe, and to the  formation when the sinusoidal
wave is in resonance with the drill string and a standing
wave is created in  the pipe (Figure 2-6). More than  one
frequency of axial oscillations can produce resonance
and a  standing wave. The fundamental resonance fre-
quency,  Rf,  relates to the length of the  pipe  by the
equation
Rf = v/(2L),
(2-1)
where vis the speed of sound through steel and L is the
length of the drill pipe.  For instance, for a steel drill pipe
30 meters (97.8 feet) long,
    = (5,000 m/sec)/(2 x 30 m) = 83 Hz.
(2-2)
Additional  resonance frequencies  are  whole  number
multiples (overtones) of the fundamental resonance fre-
quency. Higher resonance frequencies are used when
greater energy is required to advance the drill string
through more resistant formations or at greater depths.
The maximum sonic drilling depths  depend on the host
                                  Resonant Sonic
                                  Drill Head
                                  Antinode Location:
                                  Point of Maximum Strain
                                  (Compression or Expansion)
                                  in Molecular Structure

                                  Superimposed
                                  Induced Pressure Wave
                                  and Reflected Pressure Wave
                                  Expanding and Compressing Pipe
                                  Steel Drill Pipe

                                  Node Location:
                                  Point of Minimum Strain
                                  in Molecular Structure
                                  Standing Wave:
                                  Fundamental
                                  One-Half Wavelength
                                  Condition
                                  Amplitude of
                                  Wave Cycle
                                  Drill Bit

                                  Wave Variation With Time
Figure 2-6. Resonant stress wave providing energy to the drill
          bit (10).

formation. For example, for unconsolidated sand  (sedi-
ments most compatible with sonic drilling),  the  maxi-
mum drilling  depth is  approximately  200 meters (656
feet).

The drill pipe's  penetration rate  can be optimized by
controlling its frequency of oscillations between  reso-
nance overtones. As a result, the soil  particles adjacent
to the drill  pipe  cannot vibrate in unison with  the drill
pipe, and they begin to vibrate in random directions. This
random movement fluidizes the soil within approximately
0.6 centimeters  (1/4 inch)  of the drill pipe, thereby re-
ducing the friction on the drill pipe.


2.1.3.2   Sonic  Drilling Equipment

The sonic drilling method uses a combination  of tech-
niques to penetrate the subsurface:

• Thrust or pull-down

• Mechanically induced vibrations

• Rotary power

A hydraulic chain driven winch on the drill tower provides
the pull-down (and  hoist). The sonic drill head provides
the vibrations (axial oscillations)  and rotary power.
                                                    12

-------
2.1.3.3   Sonic Drill Head
2.1.3.4   Drill Pipe
The sonic drill head contains three  main components
within its housing (Figure 2-7): an oscillator that consists
of two out-of-balance counter rotating rollers, an air spring
isolator, and a rotational drive (10).

The out-of-balance counter rotating rollers are hydrauli-
cally driven. The rotation of the  rollers  is timed and
synchronized to provide a sinusoidal oscillation in the
drill string (Figure 2-7). The  energy from the rollers
transfers through the drill string to the cutting head and,
thus, to the formation. The oscillation can have a fre-
quency as high as 150 Hz.

The air spring isolator  is  a large-diameter piston set
within a closed cylinder. The piston connects to the
portion of the drill head that contains the counter-rotating
rollers, and the cylinder is attached to the drill tower. An
air  compressor supplies the volume of air on each side
of the piston; this air acts as a soft spring that  isolates
the compression waves  from the drill rig.

A hydraulically driven motor superimposes rotary move-
ment on the axial vibration of the  drill string. The  pipe
rotation is used to attach and remove pipe from the drill
string and to aid in drilling. The rotation of the pipe also
helps remove cuttings from the drill  bit  and keeps the
cutting edge of the bit against the formation.
The sonic method causes the drill pipe to, in part, be-
have like a spring (10). The axial oscillations that the drill
head creates cause the pipe to expand and contract (to
the extent allowed by its natural elasticity and inertial
properties). The alternating expansion and contraction
causes the pipe to dilate over its length, and the outside
diameter of the pipe to decrease cyclically and increase
with each  expansion and contraction. This  movement
decreases the friction on the outside of the pipe but also
induces stresses that common drill pipe does not nor-
mally experience. The stresses, which are concentrated
on  surfaces  found at the drill pipe  joints, can  cause
occasional failure  of the drill pipe. The rate of failure of
the drill pipe, however, can  be minimized by:

• Using threaded  drill pipe with hardened steel joints

• Avoiding use of welded drill pipe connections

• Periodically replacing  drill pipe

2.1.3.5  Drilling Tool

Sonic drilling penetrates a  formation  by displacement,
shearing,  or fracture by impact (11). Displacement drill-
ing tools are used to drill through sands and light grav-
els. Displacement occurs by fluidizing the soil particles
and causing  them to  move either into the formation or
into the center of  the drill pipe. The displacement drill
tool has no special shape or hardening  requirements,
                                       Upper and Lower
                                       Air Chambers
       Center Column
                               Air Spring
                               Piston
                          Oscillator
                          Housing
                        Force
                        Generating
                        Roller
                   Air Spring
                   Housing
                      Thrust
                      Bearing
                      Oscillator
                      Orbital Race
                 Drill Head  /
                 Outer Case
                                  Column to
                                  Drill Steel
                                  Adapter Flange
Figure 2-7. Sonic drill head (10).
                                                     13

-------
but a hard welded facing generally is used to  prevent
surface erosion.

The shearing drilling tool is used to drill clayey soils. The
clayey soils can only be sheared  if the axial oscillations
of the drill pipe overcome the elastic  nature of the ma-
terial.  Because clayey soils  have a  high dampening
effect on  the  drill  pipe, the tool  is designed with an
increased wall thickness at the cutting face. This pro-
vides clearance for the external  and  internal pipe sur-
faces that follow.

The fragmentation tool is used to penetrate lithified sedi-
ments and rock. The axial oscillation of the drill  pipe
causes the drilling tool to impact and  fracture the rock.
The continuous rotation of the drill pipe clears away rock
fragments from the  cutting surface. A flushing medium
often  is used to remove rock fragments from the well-
bore. A fragmentation tool generally has a hemispherical
array of tungsten carbide buttons on  a reinforced tool
surface. The tool is built with passages to  allow the
flushing medium to remove cuttings from the drill face.

2.1.3.6   Sampling Equipment and  Methods

Traditionally, sonic drilling has used three types  of sam-
pling methods: a core tray or plastic sleeve, a split tube,
and a core barrel liner (12). These methods can collect
core samples  at any angle.  The core tray or  plastic
sleeve is a commonly used core retrieval method. The
core sample is taken with the inner drill pipe either inside
the outer  casing or below the outer casing. The inner
drill pipe is removed from the wellbore. Then, the final
length of drill pipe that contains  the  core sample (still
attached to the drill head) is swiveled to about  55 de-
grees from vertical, and the drill  head vibrates  the drill
pipe. This causes the core sample to slide out of the pipe
and into either a plastic sleeve or onto  a core tray placed
at the open end of the pipe. The core tray  or plastic
sleeve method works well for lithologic characterization
and/or soil sampling. Because the samples are dis-
turbed when they slide out of the drill pipe, however, they
cannot be used to identify fine sedimentary structures or
to characterize hydraulic properties.

The split tube sampler is constructed of  split carbon
steel tubing, which has tapered threads and open caps
on each end. The tube attaches to the end of the inner
drill pipe and is driven into the formation. Then, the outer
casing is drilled over the split tube, and the  inner drill
pipe and tube are removed. After removing both the split
tube from the drill pipe and the open end caps, the tube
is split open. This sample method provides an undis-
turbed core sample that can be used for lithologic char-
acterization (including fine sediment structure analysis),
hydraulic properties characterization, and soil sampling.

A core barrel liner method was investigated at the De-
partment  of Energy's  (DOE's) Hanford, Washington,
site. The method uses a polycarbonate liner installed
inside the lowest piece of the inner drill pipe. The drill
pipe and liner are drilled into the  formation and then
removed. The liner and core sample are removed by
vibrating the drill pipe. The  core sample this technique
provides is equal in quality to a split tube core sample.

The temperatures of the core samples  collected using
all three methods described are generally higher than
the ambient subsurface temperatures. The sample tem-
peratures can be kept near ambient temperature (less
than 27°C [80.6°F]) by decreasing  the drilling rate and
cooling  the core sampler before it is  inserted in the
formation (12).

2.1.4   Well Installation  and Completion

Horizontal well construction can create two  types  of
wellbores: continuous wellbores and blind wellbores.

2.1.4.1   Continuous Wellbores

Many drilling  contractors are familiar with methods  of
completing continuous wellbores. These methods are
based on techniques developed in the trenchless tech-
nology industry for installing cable and pipe below roads
and rivers. The continuous wellbore is an advantage for
treatment systems that require access at both ends  of
the horizontal screen.

Some disadvantages of continuous wellbores are that
they require twice the surface access (exit and entrance
access) as a  blind wellbore. In addition, the  continuous
wellbore path is longer and  possibly more expensive to
drill. Also, well installation  in a  continuous  wellbore is
more stressful on the  well materials because of two
wellbore curves. Therefore, the strength requirements of
the well materials are greater  than those  in a  blind
wellbore.

The continuous wellbore generally has a 7- to 25-degree
approach angle and  a medium to long radius of curva-
ture (greater than 45 meters [147.6 feet]). To construct
a continuous wellbore, a pilot hole first is drilled from the
surface, through  the  curved section to the target depth.
The pilot hole then is drilled  along the horizontal section,
through another curved section, and back to  the ground
surface. The well materials,  such as casing, screen, and
filter packs, are assembled at the exit hole and attached
to the drill string in the wellbore. Generally a hole opener
or reamer is  placed at the end of the  drill  string, just
before the well materials, to facilitate their  installation.
The drill string, with the hole opener and well materials
in tow, is pulled  back through the wellbore. The hole
opener removes  any excess cuttings or formation mate-
rials that may have sloughed  into the wellbore  and
enlarges the pilot hole to the desired diameter. When all
the drill string has been removed from the wellbore, the
well materials are in place. Figure 2-8  illustrates this
                                                   14

-------
   a.
                                    Casing and Screen
Figure 2-8.  Creating  a horizontal  well with the pull-back
           method: a) drilling a continuous wellbore, b) ream-
           ing and pulling in casing, c) the finished well.

technique of installing casing, which is often referred to
as the pull-back method; the name is sometimes used
to refer to the entire task of installing a continuous well.

2.1.4.2  Blind Wellbores

Blind horizontal wellbores have an entrance hole and
terminate at the vertical depth of the target zone (Figure
2-8). This type of horizontal wellbore was first developed
in the  petroleum industry. The radius of curvature for a
blind wellbore can be short, medium, or long. An advan-
tage of a blind wellbore is that only one wellbore curve
needs to be negotiated during well installation. In addi-
tion, there is no need for site-access at an exit hole, and,
because a blind wellbore is shorter than a continuous
wellbore, it may be less expensive to create. However,
these  cost savings generally are offset because  the
drilling and well installation methods are  more complex
than for continuous wellbores. Drilling methods usually
require multiple trips in and out of the hole, as opposed
to a single round trip for continuous wells.

Blind  wellbores can be drilled using three techniques:
washover pipe, open wellbore, and pulling casing while
drilling. Washover pipe drilling entails initially drilling  a
prior hole (Figure 2-9). Then, the washover pipe is drilled
over the  pilot drill rods to  enlarge the wellbore and
provide a conduit for well installation. The washover pipe
installation is difficult because  it requires drilling two
pipes into the ground, thereby doubling the chances of
experiencing drilling problems. However, well installa-
tion is much easier in a washover pipe than in an open
wellbore.

Open wellbore installations have the advantage of being
the simplest and  easiest to complete but are limited to
sites where wellbore stability is not a problem. In this
method, the wellbore is drilled and cased to the end of
the curve. Then, a pilot hole is drilled in the horizontal
section, and a hole opener is used to make the horizon-
tal section the desired diameter. Finally, the well casing
is installed in the open wellbore. The obvious risk in this
method is that the wellbore may collapse before the well
casing  is completely installed.

The most complicated technique for completing a  blind
wellbore  involves pulling a casing as drilling proceeds.
A problem associated with this technique  is that casing
materials stick  because the  hole  cannot be  properly
cleaned as it is drilled (13). A modified method involves
drilling  a  pilot hole and installing the casing as the pilot
hole is  reamed to a larger diameter.

Either a washover casing or an open hole can be used to
install well materials into blind wellbores. The well materi-
als are assembled at the surface  and pushed into the
wellbore.  When well materials are in place, the washover
casing, if used,  is withdrawn the desired distance.

2.1.4.2   Stresses on the Casing During
          Installation

Friction against the wellbore  is the primary  cause of
stress on the casing during installation. Pushing or pull-
ing casing or well materials into a well requires force
greater than the friction generated along the surface of
contact between  the  hole and  the casing or screen.
Figure 2-9.  Completing a blind well using the washover pipe
          method.
                                                    15

-------
Friction force is characterized as the product of a friction
coefficient  and force normal  to the  surface  between
moving objects. Although neither of these two parame-
ters can be precisely characterized, identifying and con-
sidering several  influential  factors  should assist in
minimizing the force required to complete a horizontal
well.

Friction coefficients between solid objects at rest range
from 0.1 to 4. The exact value depends significantly on
the smoothness of the surface and nature of the mate-
rials, and especially on surface properties such as inter-
facial energy. Thus, steel and plastic casings will have
different coefficients of friction.  Likewise, coarse sand
will differ from clay. If solid surfaces  are separated by
small amounts of a  third material,  such as  a liquid,
effective friction coefficients can be markedly less than
dry surface friction coefficients. Small particles can also
lubricate surfaces by acting as ball bearings, thus yield-
ing lower effective friction coefficients.  Consequently,
friction  coefficients during  drilling operations can vary,
although Ta Inglis (14) recommends values of 0.2 to  0.4.

Several types of forces perpendicular to casing  or well
screen can exist simultaneously. Figure 2-10 illustrates
the origin of some of these forces, all of which can occur
in  blind and continuous wells. The long arc of the curve
section between the entrance angle  and the horizontal
section bends casing and well screen. The stiffness of
these materials requires that a force normal to the axes
be realized to affect  the  shape. This force should be
inversely proportional to the radius of curvature of the
hole. In the horizontal section, the normal force  can be
as small as  the weight of the casing or screen.  This
lower limit requires that the hole be open and straight,
so that it bears no other force. If the surrounding forma-
tion contacts  the well materials, then the in situ stresses
transmit as a  normal force. Formation materials also can
exert normal forces  through capillary  action;  for in-
stance, the stickiness of clay is, in large part, a manifes-
tation of the strong capillary action exhibited by clays.

Perfectly straight horizontal holes are difficult to drill.  The
highs and lows of the trajectory  deflect the pipe from a
straight path. Even though the  deflection may  not be
Figure 2-10.  Types of force normal to axes during installation
           of casing or well materials: A) flexural stress in
           long arcs, B) casing  weight  in straight  open
           sections,  C)  lateral stress created  by  tension
           through serpentines, D) pressure by surrounding
           formation.
sufficient to cause flexural reaction force in the pipe, an
additional  normal force  is generated when the force
along the trajectory is resolved into components parallel
and perpendicular to the general direction of the well. In
a mechanistic sense, the normal force components try
to straighten  the serpentine wellbore. This force varies
inversely with radius of curvature. The force is propor-
tional to the tension force that is placed on the casing or
well materials. This has the consequence of compound-
ing the friction force along the length of the well.

Friction is generated at all the points along the well,  and
the required force for placement can be represented by
the integral of friction force per unit length along the well.
Considering the types of normal forces described, an
expression of tractive force is

Fp = JV| F_L dx = JV| (Fflex + W + 2ji r N + a Fp) dx,   (2-3)

where Fp  is the force required for  placement, r| is an
effective coefficient of friction, and F± is the sum of
normal forces per unit length  of casing or well materials.
The normal force components are the reaction force  Fflex
of flexing the pipe normalized by length, the weight W
of pipe per unit length, stresses and pressure N exerted
by the formation, and  the perpendicular component of
the overall tractive force.  In this equation, ris  the radius
of casing  or well materials  and  a is a constant that
accounts for the geometry of serpentine sections.

For the pull-back  method of installing well materials  in a
continuous well,  the  pulling  force is transmitted as a
tensile stress in the casing material. It follows that the
tensile stress, o, in the casing is

        PFp       Ffiex + W+2KPN „,.„„„  ^
o =
    ji (2 r w - w2)   a n (2 r w - w2)
                                               (2-4)
where x is the length of casing in the ground, w is the
wall  thickness,  and p is a factor that accounts for the
loss  of material due to the slots. (A simple approximation
of p  is the  ratio of the total to the intact cross-sectional
area of the casing, while more complex analysis might
include  the stress intensity imposed by the sharp com-
ers of the slots.) The origin of coordinates is where the
casing enters the ground.  The casing breaks in tension
when o exceeds the tensile strength of the  material.
According  to Equation  2-3, stress increases  as more
casing  is  pulled  into  the  ground, and the increase in
stress is exponential if the well is serpentine. The maxi-
mum stress occurs where the combination of x and p is
the greatest. (The otherterms are constant.) This means
that  the casing will break either near the end that is
being pulled or at the first piece of screen; this prediction
is  consistent with observations of broken casing in the
field.
                                                    16

-------
Simple calculations using Equation 2-3 show that some
screens break if they are pulled into the ground unsup-
ported. As a result, a cable or rod is sometimes threaded
from the reamer through the casing and used as an
interior support. The interior support fastens to the end
of the casing  so that much of the load during pull-back
is supported by the pipe or cable rather than the casing.
This allows materials that are relatively weak in tension,
such as plastic casing with threaded or barbed joints, to
be pulled into a continuous wellbore. The elasticity of a
cable, however, must be sufficiently small for the cable
not to stretch  enough to allow the well materials to part.

Equation 2-3  can be similarly extended to assess the
compressive forces that occur while pushing casing and
well materials into a washover pipe  or open  hole. Wu
and Juvkan-Wold (15) present an analysis of compres-
sion stress that includes buckling of the well material.

2.1.4.4   Well Completion

Two methods of installing a gravel pack are available for
horizontal wells. One method places a tremie pipe along
the casing and injects the filter material into the annulus
between the casing and the wellbore. Placement of the
tremie pipe to the end of the well screen must be ac-
complished either during well material placement or as
part of a washover casing  system.  If the tremie pipe
clogs,  however,  it  can  be difficult to replace.  Placing
continuous sandpack around the well  screen  can be
particularly difficult because the rate required to flush the
sand through a tremie pipe over a long horizontal dis-
tance (greater than 90  meters [295.3 feet]) tends to
wash  out the wellbore during backfilling.

An  alternative method is to install a prepacked  screen,
which consists of an inner and outer screen with sand
in the annulus between them. Thus, the  filter  pack is
contained within the prepacked screen. Installation of a
prepacked screen requires a larger radius of curvature
for  installation than  standard wire wrapped screens.
Another alternative  is to wrap the well screen with a
geotextile filter membrane.

Another option is to use tremie pipe to install a bentonite
seal and grout in the annulus around the riser casing. A
grout  basket placed on the riser casing before installa-
tion keeps bentonite and grout from flowing into the well
screen. It is advisable to first install sand in the grout
basket to  ensure that the bentonite and  grout do not
seep around the edge of the basket.

A gravel pack may be unnecessary for some applica-
tions.  Where  horizontal wells are placed  in sands, re-
moving the fine-grained  sediments from the vicinity of
the well may not be possible during development. This
results in  a natural sand pack that  resembles similar
applications used for vertical wells  (16).  Moreover, a
gravel pack often is unnecessary for applications involv-
ing vapor extraction or bioventing. Eliminating a gravel
pack can significantly reduce the costs of a horizontal
well.

2.1.5   Well Development

Well development concerns and techniques for horizon-
tal wells are  similar to those for conventional vertical
wells; they depend on the hydrogeologic setting  and
remediation objective of the well. The drilling techniques
for installing a horizontal well in the vadose zone for the
purpose of vapor extraction or injection ideally will not
use a drilling  mud because of concerns of reducing the
gas permeability of the formation. However, if a drilling
mud is required, the well development technique would
be similarto that used for a well installed in the saturated
zone.

When drilling  a  well installed in the saturated zone and
using a bentonite- or  polymer-based drilling fluid, well
development  should begin as soon as possible after well
installation  and/or well completion. The drilling fluid and
mud cake  need to be removed to restore formation
permeability around the well screen and/or filter pack.
Well development may include volume flushing the an-
nular space with water, jetting water into the well screen
to remove drilling fluid, treatment with an acid or base
chemicals to  dissolve the drill fluid  or mud cake, or
treatment with a disinfectant to prohibit bacterial growth
in the case of a biodegradable drilling fluid. The devel-
opment fluid should be chemically compatible with the
formation water to avoid detrimental effects on the for-
mation.

2.1.6   Pumps

The pump  used in an inclined or a  horizontal ground-
water extraction well must be able to function continu-
ously in an inclined or a  horizontal  position. Currently
available well pumps are designed to operate vertically,
but no  studies  have  been published to  date on the
long-term performance of well pumps in nonvertical en-
vironmental wells.

Electrical centrifugal pumps can function in the horizon-
tal position, but the life of the motor and impeller bear-
ings, which were designed to operate vertically, may
limit their endurance.  Of the pumps that are currently
available, electrical centrifugal pumps are the most com-
monly used in horizontal wells because of their versatil-
ity and ability to be used at any depth.

Pneumatic  pumps can be  either diaphragm or bladder
actuated and typically make use of two check valves.
Usually the valves must be vertical to seat properly.
Apparently  some diaphragm  pumps function  horizon-
tally, because Wilson and Kaback (17) used diaphragm
pumps to recover  NAPLs  in shallow, low-volume hori-
zontal wells.
                                                   17

-------
2.2   Design Considerations

In horizontal well applications, it is important to consider
many factors that can affect the performance of a well
and the ability to drill and complete  the wellbore. The
pattern of flow in the vicinity of a well is a major factor
affecting  performance.  In addition, geologic and hydro-
logic conditions and contaminant distribution affect de-
cisions regarding:

• Drilling methods

• Horizontal wellbore specifications

• Length and elevation of the well

• Composition and design of well materials

2.2.7  Pattern of Flow

The  pattern of flow created by a horizontal well is a
fundamental aspect of  performance (18). Under ideal
conditions—a thin, confined aquifer of infinite extent with
no regional flow—radial flow adjacent to the  well  is
confined  to a small area and streamlines in the vicinity
of the well run roughly parallel and perpendicular to the
long  axis of the well (Figure 2-11). At great distances
from the well, streamlines run approximately radially, so
that in a  thin aquifer the pattern of flow to a horizontal
well resembles flow to  a trench (19). At steady state in
a thin aquifer, the  specific discharge from a horizontal
well with  no head loss along the casing approaches that
of a vertical well with a radius equivalent to one-fourth
the total  length of a horizontal well. Clearly, this can
represent a very large vertical well.

The pattern of flow changes as the formation becomes
thicker and the zone  of  radial flow around the well
becomes larger (18). In this case, the pattern is similar
to Figure 2-11 in plan but develops components of flow
in the third dimension (depth). Head losses associated
with the vertical component of radial flow adjacent to the
well (Figure 2-12) limit the advantages that a horizontal
well may offer over a fully penetrating vertical well, which
causes no vertical  flow. The effect is exacerbated by
anisotropic conditions, in which the ratio  of vertical to
horizontal hydraulic conductivity is less than  one (3),
such as in interbedded fine- and coarse-grained  sedi-
ments. Nevertheless, for many commonly encountered
geologic conditions, effects  of formation thickness and
anisotropy are small relative to improvements in specific
capacity resulting from a horizontal well's longer screen
length  (2).
Figure 2-12.
Radial flow in a vertical plane in the vicinity of a
horizontal well.
Figure 2-11.  Flow paths to a horizontal well in a thin aquifer
           with no regional flow. Lines are nearly straight and
           parallel within the patterned area.
A secondary benefit of a long screen is reduced velocity
of fluid in the vicinity of the well. This effect reduces the
possibility of inducing turbulence and associated non-
linear head losses (2, 20). Decreasing fluid velocities in
horizontal wells also should  reduce damage to  well
screens and filter packs due to migration and entrap-
ment of fine-grained sediment (2).

2.2.2   Wellbore Specifications

The well trajectory and materials of construction are the
principal wellbore specifications. To some extent, these
specifications are  interdependent. Installation and resi-
dence in  the subsurface put stress on the  wellbore
tubing. The magnitude of the stresses depends on the
method of installation, curvature and depth of the bore,
and other factors. To  ensure subsurface access, the
wellbore tubing selected must withstand these stresses.

2.2.2.1   Specification of Trajectory

Horizontal wells currently used for environmental appli-
cations typically enter the ground at a  shallow angle,
                                                    18

-------
gradually curve  until  they  are  horizontal,  proceed
through the contaminated interval, and then either ter-
minate or curve upward to the ground surface. Accord-
ingly, this trajectory  is  specified  by  the  following
parameters (Figure 2-13):
• Well head  location
• Entry angle
• Vertical depth to  target
• Measured  depth  to target
• Wellbore curve
• Step-off distance
• Plan path
    KOP
    EOC
    VDkop
    VDeoc
    VDtarget
    MDeoc
    MDkop
    MDtotal
    Heoc
Kickoff Point
End-of-Curve
Vertical Depth to KOP: Equation 2-5
Vertical Depth to EOC: Equation 2-6
Vertical Depth to Target: Given
Measured Depth to EOC: Equation 2-7
Measured Depth to KOP: Equation 2-8
Total Measured Depth: MDeoc + Horizontal Length
Horizontal Distance to EOC: Equation 2-9
Horizontal Distance to KOP: Equation 2-10
Total Horizontal Length: Heoc + Length of Horizontal
Section
Figure 2-13. Nomenclature describing a horizontal well (14).

Well Head Location. Generally the well head is in line
with the horizontal well at a practical distance from the
target zone.  It is possible,  however, to curve  laterally
before reaching the target zone to offset the well  head
from the major line of the well. Lateral curves stress the
drill rods, reduce the distance capabilities of the rig, and
increase stresses on the  well materials. The following
points  should be considered  when selecting the well
head location:

• Surface  obstructions may inhibit  the selection  of a
  well  head location. There must be sufficient  room at
  the well  head location for the  safe operation of the
  drilling equipment.

• The length  of surface piping to a treatment facility
  should be minimized.

• Multiple  horizontal wells may radiate from the same
  well head  location.  This enables well heads to  be
  located in a single well vault and one treatment sys-
  tem to be used for an extensive area.

Entry Angle. The entry angle (l|), sometimes referred to
as the approach  angle,  is the angle between the drill
stem and the ground surface at the entry hole. The entry
angle may be between 7 to 90 degrees from horizontal
(10 to 35 degrees is most common), depending on the
type of drill rig being used. Shallow horizontal well  in-
stallations  (less than 8 meters [26.2 feet] VD) generally
have a 10- to 15-degree entry angle, whereas deeper
horizontal well installations (greater than 8 meters [26.2
feet] VD) have a 15- to 35-degree entry angle.

Vertical Depth to Target. The VD to target is the differ-
ence in elevation between  the wellbore entry location
and the target zone.

Measured  Depth to Target. The measured depth (MD)
to the target is the distance along the wellbore path from
the wellbore entry to the target zone. The  MD is equal
to the length of drill rod in the ground.

Wellbore Curve. A radius of curvature defines the curved
portion. By industry convention, the radius of curvature
is measured  in feet, and a well is drilled with a buildup
rate  (BUR) measured  in  degrees of angle  per 30.5
meters (100 feet) of wellbore drilled. The curve begins
at the kickoff point (KOP) and ends at the end-of-curve
(EOC). The curve BUR that provides the desired radius
of curvature is given in Table 2-2.

Selection of a radius of curvature  depends on the target
zone location,  well materials,  and drilling equipment.

Table 2-2. Wellbore Curve Measurements
                                                         Radius of Curvature (RoC)
                                                                  (feet)
                                                                             Buildup Rate (BUR)
                                                                             (degrees/100 feet)
                                                                             RoC x BUR = 5,730
50
100
150
200
250
300
400
500
750
1,000
114.6
57.3
38.2
28.7
22.9
19.1
14.3
11.5
7.6
5.7
                                                     19

-------
Wellbore curves have been classified based on their
radii of curvature  as short-,  medium-,  or  long-radius
wells: a short radius of curvature is less than 45 meters
(147.6 feet); a medium radius of curvature is 45 to 250
meters (141.6 to 820 feet); and a long radius of curva-
ture is greater than 250 meters (820 feet).

As  the radius of curvature tightens,  both the bending
stresses on the drill rods and the frictional resistance
between the rods  and the formation increase.  These
stresses conspire to effectively reduce the  total length
that can be achieved by a particular combination of drill
rods and boring machine.  Horizontal sections are limited
to approximately 100  meters (328 feet)  in short radius
of curvature boreholes (13, 14). The  diameter and ma-
terial of the casing  also affect the wellbore radius. A rule
of thumb in the river crossing industry is that the radius
of curvature, measured in feet, should be 100 times the
diameter of the installed  pipe measured in  inches. For
example, if diameter of the well casing is 4 inches, then
the  radius  of  curvature should be approximately 120
meters (394 feet).

The radius  of curvature of a borehole should be as long
as practical within the constraints of the other borehole
criteria. A shallow entry angle and large radius of curva-
ture increase the borehole length and reduce the possi-
ble  locations for the well  head. The increase in  drilling
cost associated with drilling a longer borehole must  be
weighed against the benefit of reduced stress on well
materials.

Step-Off Distance. The step-off distance (Heoc) is the
horizontal distance between the entry hole and the be-
ginning of  the horizontal section  or the EOC  of the
wellbore. The step-off distance may  be  determined  by
site-specific conditions, such  as the available surface
area for the drilling equipment, well head(s), and asso-
ciated treatment systems.

Plan Path.  Normally a horizontal well is planned to  be
straight from entry  point to wellbore termination or exit.
The size and  shape of the plume and subsurface ob-
structions, however, may  require the well path design to
include lateral curves. A horizontal wellbore  can  be
drilled with a lateral curve, but the radius of curvature
should be large to  reduce stress on well materials.

2.2.2.2   Estimating Wellbore Specifications

A detailed  determination  of the wellbore trajectory is
required during drilling. This allows the expected  loca-
tion to be compared with information from the electronic
systems that locate the bore in the subsurface. Typically,
designs of  wellbore trajectories are developed and re-
vised  graphically in conjunction  with site  conditions.
After establishing the trajectory, either graphical or geo-
metric methods should quantify it. A few equations that
may be helpful follow.
For the VD of the KOP (VDkop),

VDkop = VD, - [k-,(cos If - cos 1^1 BUR,
 (2-5)
where VD, is the VD to the target zone, k1 is a constant
that  equals 5,730 when BUR is  in degrees per 30.5
meters (100 feet), lf is the final wellbore angle (0 degrees
for a horizontal well), and /, is the entry angle. The VD
to the EOC is  equal to the VD to the target zone if the
surface elevation at the  entry point is equal to the ele-
vation of the point above the target zone. If those eleva-
tions are unequal, then  the VD to the EOC  relative to
the entry point is given by
 (2-6)
      = MDkop + [(nkJ/180] [(I, - IJ/BUR],       (2-7)

where MDkop is the MD to the KOP and is given by

MDkop = VDkop/sin(lj.                           (2-8)

If the step-off distance is not fixed, then it can be deter-
mined by the entry angle (/,) and the radius of curvature
of the borehole (or BUR) as follows:
VDeoc = VDkop - Ik, (cos lf - cos

The MD to the EOC (MDeoc) is given by
     = Hkop +
 (2-9)
where the horizontal distance to the kickoff point (Hkop)
is given by
Hkop =
2.2.2.3   Casing and Screen
(2-10)
Casing and screen in a horizontal well must be stronger
than in vertical applications. Horizontal wells require the
extra strength  to withstand tensile  or compressive
stresses along  the axis of the casing that result from
pulling or pushing the casing  into the bore. Furthermore,
the weight of the overburden produces stresses that
may collapse a horizontal casing.

Strength of casing  depends on the  casing  material,
diameter, and wall thickness. In general, stainless and
carbon steel  casings are stronger  than  plastic casing.
However, plastic casing, particularly high-density poly-
ethylene (HOPE), can bend in a tighter radius than steel
casing.  The collapse strength, which is the  maximum
allowable force  applied perpendicular to the casing, de-
creases  with increasing diameter (Table 2-3). Con-
versely, the tensile strength, which is  the  maximum
allowable axial  force, increases with increasing diame-
ter. A typical formation  exerts approximately 0.02 MPa
per meter depth (0.8 psi/foot), so the collapse strength
is related to the allowable depth at which the casing can
                                                   20

-------
Table 2-3.  Strengths of Prepacked Screens (According to
          Johnson Filtration Systems, Minneapolis,
          Minnesota)
Nominal Size
   Collapse
Strength (MPa)
   Tensile
Strength (kN)
Prepacked 304 Stainless Steel
    2 inch                10.3
    4 inch                4.14
    6 inch                1.38

Prepack Schedule 40 PVC
    2 inch                1.72
    4 inch                1.03
                         24
                         58
                        170
                          4.9
                          7.1
be used. Tensile strength is important when pulling cas-
ing into a horizontal bore, so the allowable stress has
been converted to allowable pulling force in Table 2-3.

To estimate allowable pulling force for casings of other
sizes and materials, multiply the effective  pulling area
times the strength of the tubing (Table 2-4). The effective
pulling  area is  the continuous cross-sectional  area,
which for solid casing is rcdw/2  (with d the outer casing
diameter and wthe wall thickness). Most casings that
fail in tension crack at either the slots in the screen or
the joints between  individual pieces of casing. To esti-
mate the upper  limit of the strength of a screen made
from  plastic pipe with slots cut in it, use the approach
outlined above, but reduce the  effective pulling area to
account only for the solid material in between the slots.
The strength  of joints  between casing sections can
range widely, from  essentially the same as the casing
itself for welded joints to small values for threaded joints
designed for vertical wells.
Table 2-4. Tensile Strengths of Pipe

Pipe
           Tensile Strength (MPa)
PVC

HOPE

Fiberglass
                   9.4

                   6.5

                 1.3-6.9
Casing and screen in a horizontal well used for environ-
mental applications must also be resistant to whatever
chemicals  exist at a particular site. In this respect, the
requirements for horizontal wells are the same as those
for vertical wells.

New materials  and joining capabilities are being devel-
oped  for casing  and screen for environmental wells.
Distributors of  well supplies should be able to  supply
information about the capabilities of these new products.

2.2.2.4   Filter Pack

Filter packing around a well screen is recommended for
horizontal  wells that produce liquids from  formations
containing fine-grained sediments, just as it is for verti-
cal wells in similar sediments. To select a size gradation
for a granular filter pack, use the same approach as for
a vertical well (16), although keep in mind that entrance
velocities into horizontal wells can be considerably less
than those into vertical wells.

Several methods exist to place filter packing into the
annulus around a casing or screen, most of which in-
volve injecting the particles of the filter pack in a  liquid
or air stream (21). Moving fluid  more readily transports
HOPE particles, but these methods are more expensive
than sand. Quality assurance is  problematic in the injec-
tion of filter  pack materials; it is difficult to ensure that
the packing  is continuous around the entire casing.

An alternative to injecting filter pack materials is to use
a prepacked screen, which consists of an inner screen,
an annulus  filled with sand, and an outer screen. The
prepacked screen currently is the most popular method
of installing  a gravel pack in horizontal wells because it
ensures that the pack is continuous and uniform. Pre-
packed screens are  available in most sizes  and casing
materials.

Unlike most vertical  wells, filter packing in a horizontal
well  can  be  a significant component of the  cost of the
well. Therefore,  careful examination of each application
should establish the need for a filter pack and determine
whether  the cost of the pack will  result in  a valuable
increase in the performance of the well. Many wells used
for vapor extraction, for example, may require only a well
screen.  Likewise, liquid-phase recovery  wells  com-
pleted in clean  sands formations  may have  naturally
developed filter packs (16), thereby avoiding the cost of
the filter pack.

2.2.3  Geologic Site Conditions

Geologic conditions,  ranging from composition  of the
formation to the heterogeneities caused  by depositional
environment, affect details related to  drilling and com-
pletion and the use of horizontal wells.

2.2.3.1   Formation Composition

The  composition of the formation  affects the  methods
used to drill  a directional wellbore. Highly resistant for-
mations, such as siliceous sandstones and metamor-
phic or  igneous rocks, are difficult to cut. Loosely
consolidated formations that are cobble-rich can be dif-
ficult to penetrate because the  cobbles rotate with the
drill bit. Generally, loosely  consolidated granular sedi-
ments present steering problems because the bit cannot
exert enough lateral force on the formation to deflect the
drill rods. Specialized technology can  address all these
issues, although it also may increase the cost of creating
the wellbore.
                                                    21

-------
2.2.3.2   Stratification

Drilling in a stratified formation may present problems in
some cases, particularly when intersecting  a resistant
unit at a  shallow angle. In this case, the wellbore tends
to ride along the top of the resistant formation. If recog-
nized, however, this problem can be addressed.

Formation stratification largely affects the application of
horizontal wells,  particularly when the distribution  of
contaminants follows a sequence of alternating fine- and
coarse-grained sediments (impermeable and perme-
able layers). In this case, recovery would be primarily
from thick, high permeability strata in the vicinity of the
well, with recovery from strata overlying or underlying
the well less than anticipated. The degree to which this
occurs depends on the thickness and permeability of the
stratigraphic layers.

In general, contaminants  flow in a horizontal  plain un-
less hydraulic, chemical, or gravitational forces provide
an impetus for flow in the vertical direction. A horizontal
well should be  placed as near the contaminant as pos-
sible to decrease the likelihood of an impermeable layer
impeding the flow between the well and the contaminant.


2.2.3.3   Heterogeneities in the Horizontal Plane

Horizontal wells are suited to accessing formations that
are heterogeneous in a horizontal plane. Formations
whose major component of permeability comes from
vertical fractures are excellent candidates. In addition,
horizontal wells can  access braided  stream deposits
with laterally discontinuous  sand bodies.  Many glacial
sediments  contain  both vertical fractures and  laterally
discontinuous sand bodies, and many near-surface rock
formations contain vertical fractures.
2.2.3.4   Aquifer Thickness

Horizontal wells can be effective in aquifers of any thick-
ness. However, the efficiency of a horizontal well relative
to a fully penetrating vertical well increases as the aqui-
fer  becomes thinner  (2). This is because  flow must
converge from increasing distances both above and
below the well as the aquifer thickens (Figure 2-12). A
vertical  well that fully penetrates an aquifer, however,
induces  primarily horizontal flow and  does not suffer
from the effects of vertical  flow (assuming  that draw-
down is modest). The specific capacity of both horizontal
and  vertical wells increases as  the  aquifer thickens.
Therefore, the performance (measured by specific ca-
pacity) of a horizontal well in a thick aquifer is better than
the same well in a thin aquifer. This occurs because less
interaction with the upper and lower no-flow boundaries
occurs in a thick aquifer than in a thin layer.
2.2.3.5   Anisotropy

The permeability of most formations is anisotropic, al-
though the principal directions depend on the geologic
conditions. Interbedded  sediments, for example,  typi-
cally are more permeable in a horizontal than in a verti-
cal plane, and even in the horizontal plane permeability
is greater along one direction than other directions. Ac-
cordingly, the directions  of greatest, intermediate, and
least permeability are mutually perpendicular in an or-
thogonal set. In other geologic formations, the principal
directions of permeability may differ significantly. In any
case, these directions should be considered when locat-
ing horizontal wells.

For situations with distinction only between vertical and
lateral  permeability, the effects of anisotropy resemble
the effects of aquifer thickness; an anisotropic aquifer of
a certain thickness behaves the same as  an isotropic
aquifer  of different thickness (20, 22).  The  effective
thickness he of an anisotropic aquifer of actual thickness
h is given  by he  = h /ch//cv, where /ch and  kv are the
horizontal and vertical  permeabilities, respectively. Ac-
cordingly, an  anisotropic  aquifer, for which /ch//cv =  100,
that is 5 meters (16.4 feet) thick behaves like a homo-
geneous aquifer 50 meters (164 feet) thick with k = kh.
Similarly, if kv is greater than /ch, as is possible where
vertical fractures  are common, then the effective thick-
ness of the aquifer is less than the actual thickness.

In general, a  horizontal well drains the largest volume if
placed normal to the plane that contains the axes  of
maximum and intermediate permeability (Figure 2-14a).
This only can be possible if the direction of greatest
permeability is vertical. Where the greatest permeability
is horizontal, a horizontal well placed perpendicular to
the direction of intermediate permeability drains the  larg-
est volume (Figure 2-14b).

  a. Ideal Orientation   b. Fair Orientation   c. Poor Orientation


I/.
' ^max
^•v min
K
intermediate



<*""

	 ->*" 	
Figure 2-14.  Orientations of horizontal well (dashed line) rela-
           tive to principal directions of permeability.

When locating the well, other factors to consider along with
the direction  of anisotropy include surface obstructions,
direction of regional flow, and location of contaminants.

Well tests  are available to estimate formation anisot-
ropy (23).

2.2.3.6   Drawdown and Discharge

The drawdown and discharge of a horizontal well depends
on aquifer and well characteristics. Aquifer characteristics
                                                    22

-------
include permeability, storage coefficient, and thickness
and boundary conditions. Relevant well characteristics
are length, height in the aquifer, and details of the hy-
draulics  resulting  from flow through  the slots  in the
screen and along the axis of the wellbore. Most of these
factors also are common to analyses of vertical wells,
with  the exception of wellbore  hydraulics,  which are
commonly ignored for vertical wells that have fairly short
screens. The next section includes a discussion on well-
bore hydraulics; therefore,  this  section examines the
effects of the other factors.

Horizontal wells placed in productive aquifers commonly
use a pump operating at constant discharge. The pump
produces a drawdown that increases with time, just as
with a vertical well, and the magnitude of the drawdown
depends on the factors listed above. As  an example,
consider a horizontal  well of length L at midheight in a
confined aquifer of thickness h and infinite lateral extent.
The dimensionless drawdown Pd as a  function  of time
depends on the ratio of well length to aquifer thickness,
and on the ratio of vertical to horizontal permeability
(Figure 2-15). This is consistent with the conclusions
presented in the qualitative discussion above.  At any
given  time,  the  drawdown  increases  as  the well
becomes shorter, the  aquifer becomes thicker,  or the
ratio of vertical to horizontal hydraulic  conductivity be-
comes smaller—all of which contribute to decreasing Ld
(Figure 2-15). It is noteworthy, however, that these vari-
ables cause the most severe differences early  on, but
the differences diminish and approach the  results for
Ld = oo. The interpretation of Ld = oo is that the well cuts
the full thickness of the aquifer, which is impractical for
most horizontal wells  but is fairly typical of many instal-
lations of trenches. Thus,  based on Figure 2-15, the
drawdown  in a horizontal well apparently approaches
that of a trench with  increasing time. The abscissa of
Figure 2-15 is dimensionless time, which must be trans-
lated to real time using the aquifer properties and well
length  (18).

The results for a well operating at constant discharge in
a confined aquifer are by  no means applicable to all
horizontal wells.  For example, results  from cases in
which horizontal wells extract vapor from areas where
the ground  surface is open to the atmosphere differ
significantly from results when the ground  surface is
sealed. Moreover, it would be inappropriate to use Fig-
ure 2-15 to analyze cases in which  horizontal wells are
used in confined aquifers of relatively low permeability;
the difficulty in this case results from the requirement of
constant discharge. In many applications in low  perme-
ability formations,  horizontal wells operate at constant
drawdown  using a water level sensor to control the
pump.  This results in a well with  a roughly constant
drawdown  but  with a discharge that  decreases with
time.
      10
     10-1
     10-2
     10-3
            I    I    I     I    I    I    I    I
                Trench Solution
                 I    I    I
       10-6
10-3
                                               103
Figure 2-15.  Drawdown resulting from a horizontal well as a
           function of time and dimensionless length. Draw-
           down and time are dimensionless.
For example, consider a well at the top of a confined
aquifer, where the aquifer thickness is h = 20 (variables
are defined in Table 2-5). The dimensionless discharge
decreases as a function of time and also as a function
of well length (Figure 2-16). The effect of well length on
dimensionless discharge occurs because the discharge
per unit length of the well (Q/L), not the total discharge,
decreases with increasing well length. For example, at
fd = 10,000, the dimensionless discharge is 0.078 when
L = 10, whereas  it is 0.055 for L = 20. As a result, the
actual discharge, obtained using the expression in Fig-
ure 2-16, is proportional to 10 x 0.078 = 0.78 for L = 10,
whereas it is proportional to 20 x 0.055 = 1.1 forL = 20.
Therefore, as the well  length increases from 10 to 20,
the actual discharge increases 40 percent. By present-
ing the results in dimensionless form, estimates can be
obtained of the actual discharge for any combination of
aquifer conditions. Details of this analysis are presented
in Murdoch and Franco (24).

2.2.3.7  Vertical Fractures

The  presence of naturally occurring vertical fractures
enhances the performance of a horizontal well accord-
ing to the principals outlined above,  but such fractures

Table 2-5.   Summary of Variables

Ky         Vertical Hydraulic Conductivity
Kh         Horizontal Hydraulic Conductivity
K         Average  Hydraulic Conductivity
h         Aquifer Thickness
S         Storage Coefficient
L         Well Length
t          Time
rw         Well Radius
AP,,,       Drawdown at the Well
                                                    23

-------
     0.12
     0.10
     0.08 -
     0.06 -
     0.04 -
         0     2,000    4,000    6,000    8,000    10,000
                            td

Figure 2-16.  Discharge from horizontal wells of different lengths
           operated at constant drawdown (24).

are absent in many formations. Vertical fractures can be
induced from horizontal wells using methods developed
for vertical wells, described in Chapter 3. Such induced
fractures can significantly reduce limitations related to
formation thickness, anisotropy, and layering. The petro-
leum  industry routinely induces  hydraulic fracturing of
horizontal wells. In  addition, techniques developed  for
soil (25)  have been used to create fractures from hori-
zontal wells  used for environmental applications.

2.2.3.8   Regional Gradient

Horizontal wells can provide hydrodynamic control in
aquifers with a regional gradient. Special considerations
are necessary, however, if the wells are  intended to
intercept contaminated plumes. In some cases, horizon-
tal wells  can be placed in the direction of the regional
gradient, particularly to  intercept an elongated plume
that is moving relatively slowly.

In other cases, horizontal wells can be placed perpen-
dicular to a regional gradient to intercept ground water.
This application is best suited to  relatively thin aquifers
and relatively minor regional flow rates. This is because
the certainty of all the water being  intercepted decreases
as the aquifer becomes thicker or the regional flow rate
increases. Other aquifers above  or below the well also
may affect interception. Monitoring  piezometric heads
with multilevel vertical piezometers and estimating gra-
dients at various depths in the vicinity of the horizontal
well are recommended to ensure that all water is  being
intercepted.

2.2.4  Distribution of Contaminants

Once  the distribution of contaminants is known, or as it
becomes resolved in  more detail,  locating  horizontal
wells ensures hydrodynamic containment before recov-
ery begins.  Details of the  plume geometry  also  may
affect the design of a horizontal well.
2.2.4.1   Elongated Plume

Elongated plumes can resemble the shape of the cap-
ture zone of a horizontal well, which also  is elongated
(2). Therefore,  elongated plumes are ideally suited to
this application. A regional gradient,  however,  affects
the capture zone of a horizontal well with travel time
contours forming teardrop shapes. The  narrow  end of
these teardrops is on the downgradient side of the well.
Inasmuch as many plumes are elongated in the direction
of regional flow,  regional effects must  be considered
when designing the horizontal well. Using a few vertical
wells  downgradient from the horizontal well may  be
advisable to ensure complete capture.

An elongated plume may be wide enough to warrant
placing a horizontal well perpendicular to its long axis to
intercept migration with the regional flow.
Some elongated plumes result from elongated sources
and can be independent of regional flows (e.g., contami-
nants beneath a leaking pipe). Horizontal wells are ide-
ally suited to this application.

2.2.4.2   Equidimensional Plume

Horizontal wells that  access equidimensional  plumes
provide the best performance when located perpendicu-
lar to the direction of maximum permeability.

2.2.5   Wei I bore Hydraulics

A notion carried over from experience with vertical wells
is  that a horizontal well recovers water uniformly along
its length. This is true with wells that have a  large  screen
diameter (greater than 15 centimeters [6 inches]) and
are installed in  low-permeability formations. Head loss
in  large  diameter screens is negligible when compared
with the head loss in the formation next to the screen.
However,  when  small  diameter  wells are  installed in
highly permeable formations, the head  loss along the
screen can cause the rate of inflow into the well to vary
with distance from the pump (Figure 2-17).

Although a detailed analysis  of wellbore hydraulics is
beyond the scope of this manual, the effects of various
parameters  can be anticipated. Consider a very long,
small-diameter well with a small pump operating at one
          Pump

                             Influx
Figure 2-17.  Schematic of distribution of influx into a horizontal
           well.
                                                   24

-------
end. Fluid flows into the well adjacent to the pump at
some rate per unit length  of well screen. That rate of
influx must decrease with increasing distance along the
wellbore  due  to head  losses  associated  with  flow
through the  casing. It follows that the inflow rate, and
thus the  effect  of the well, may be negligible beyond
some critical distance where the sum of all the influx is
equal to  the discharge of the pump (26). This effect
would be most severe where long, narrow wellbores are
installed in highly permeable formations (as mentioned
above). In some applications, however, the influx would
decrease  along  the  wellbore but then increase as it
approaches  the  far end of the well (27). This occurs
because  water converges  from beyond the end of the
well (as in Figure 2-17). This effect from the flow pattern
in the aquifer offsets the effects of wellbore hydraulics.

Hydraulics cause variations in inflow along the length of
the wellbore, although the  magnitude of this effect dur-
ing field applications has yet to be quantified. Formation
heterogeneities  (e.g., vertical fracture) that  are  inter-
sected by the wellbore also cause variations in the rate
of inflow  and may dominate the effects of wellbore hy-
draulics.  Logging techniques that use a borehole flow-
meter (16) to determine the inflow into a vertical well are
available, and similar methods could be applied where
the distribution of inflow into a horizontal well is critical.

2.2.6  Site Conditions

The  major site conditions that affect well design  are
related to access and obstructions of the well trajectory.
A horizontal  or inclined well typically is selected when
surface structures restrict subsurface access. In some
cases, such as  at refineries or other large industrial
complexes, surface structures exist a considerable dis-
tance from the contaminated zone and restrict access
to  even a directional wellbore. In other cases, sites are
simply too small to permit sufficient step-off distance.
The  entry location for a horizontal  well must provide
sufficient space to set up the drill rig and related equip-
ment. The required  area for this activity currently  is at
least 5 by 10 meters (16.4 by 32.8 feet),  and can be
considerably more depending on the type of drill rig.

The  following are additional considerations  related to
site conditions:

• Elevation  differences in  excess of 16 meters  (52.5
  feet) between the entrance hole and the exit  hole
  should be avoided  because they may cause prob-
  lems with  the  drilling fluid system.

• The wellbore path  should avoid existing  piles, moni-
  toring wells, metal footings and pipelines. Subsurface
  metallic objects decrease the accuracy of a primary
  magnetometer-type steering  tool. Drilling projects
  that use a magnetometer-type steering tool and re-
  quire wellbore accuracy in an  area that has subsur-
  face magnetic interference need a secondary subsur-
  face survey system.

• Drilling should  not occur  in areas where overhead
  structures or wires may limit the use of construction
  equipment.

• A continuous wellbore design should provide at least
  a 12-meter (39.4-foot) wide area for the exit hole and
  an available area for laying out the well string  during
  pull-back installation.

2.3   Applications

Horizontal or inclined wells have been used for most of
the same purposes as  vertical wells (28): air sparging,
soil vapor extraction, ground-water extraction, and injec-
tion.  The following sections discuss only the effects of
horizontal  and inclined wells on the efficacy of these
methods.

2.3.1  Pump and Treat

Recovery  and treatment of  ground water, "pump and
treat," is perhaps the most common method of address-
ing ground-water contamination, accounting for approxi-
mately one-quarter of horizontal well applications (28).
Horizontal  wells  can  recover aqueous-phase  com-
pounds at reduced energy costs by using fewer pumps.
It  is  important to  recognize that processes  such as
sorption, diffusion, and  dissolution from NAPL sources
are unaffected by well geometry. Accordingly, limitations
of pump and treat systems now widely recognized for
vertical wells (29) also apply to horizontal wells.

Horizontal well design approaches depend  on specific
site conditions, but the following general approaches are
applicable to many field cases:

• Use horizontal wells to increase the rate of recovery
  from low-permeability formations. This is  particularly
  applicable to  relatively massive formations  with ver-
  tical fractures. The well should be placed  in the area
  of greatest contamination, and perpendicular to ver-
  tical fractures. To pump, maintain constant drawdown
  to  maximize discharge.

• Place a horizontal well along the axis of an elongated
  plume to recover contaminated  water and arrest mi-
  gration (Figure  2-18a). Because the shape of many
  plumes  resembles the elongated capture zone  of a
  horizontal well,  an excellent opportunity exists to re-
  cover a  majority of the plume with one well. Keep in
  mind, however, that an elongated plume  probably is
  moving, and  that regional gradients can skew  the
  capture  zone of the well. Well placement should  an-
  ticipate the effect of the  regional  gradient. For in-
  stance,  it  may  be appropriate  to extend  the  well
  further downgradient than the current extent of con-
  tamination. Pumping  should begin  immediately after
                                                   25

-------
         Horizontal Well
    Capture Zone
                                         Source
Figure 2-18.  Two approaches using horizontal wells to inter-
           cept contaminant plumes.

  creating such a well because, if neglected, the well
  offers a path along which contaminants can migrate
  under the natural gradient. Horizontal wells also may
  combine with vertical wells to contain downgradient
  migration.

• Place a horizontal well perpendicular to the direction
  of migration of a  plume  (Figure  2-18b). This ap-
  proach,  which  resembles that of  an  interceptor
  trench, is most applicable when a relatively thin stra-
  tigraphic unit contains a broad plume. Because some
  of the plume may travel above or below the horizontal
  well, this approach may be appropriate when trench-
  ing is  not an option  because of access, depth, or
  other factors.

All of the applications  described above would benefit
from modeling studies to improve the chances of plume
capture. This effort requires that site hydrogeology and
distribution of contaminants are well known. Moreover,
the pattern of flow that a long horizontal well induces
may take some time to become  steady, so transient
analyses are recommended during the design process
to avoid using the overly optimistic conditions seen at
steady state.

2.3.2  Vapor Extraction

Vapor  extraction  projects account for  approximately
one-quarter of the applications of horizontal  wells. This
is because of the relatively large area  covered by the
well and the ability to  access beneath structures and
place recovery wells close to the source of volatile or-
ganic compounds  (VOC). The  source  of air is a key
factor during the design of vapor extraction systems that
use horizontal wells. One horizontal well  beneath an
open surface  causes flow that is vertically downward
and converges toward  the screen (Figure 2-19a).  This
focuses the extraction on the ground overlying the well,
although the lateral extent may be limited. It is possible
to increase the lateral extent affected  by the well by
placing a layer, or cap, with low pneumatic conductivity
over the  well (Figure 2-19b). In this case, air flows into
the subsurface beyond the covering layer and expands
the size  of  the area affected by the  well (30).  This
increased area, however, comes at the expense of re-
ducing the magnitude  of air flux through the ground
directly above the well. In some cases, the region above
the well may become stagnant and receive little reme-
dial effect. Installation  of air inlets in the cap  may be
advisable so that the benefits of both covered  and un-
covered sites may be realized. Other horizontal wells or
vertical wells may also be used to provide air  inlets in
the subsurface.
Figure 2-19.  Schematic of flow patterns with horizontal wells:
           a) without cap and b) with cap.

2.3.3  Free-Product Recovery

A horizontal well can present an ideal geometry to re-
cover layers of free-phase LNAPL or DNAPL, although
complicating factors can make this application difficult to
execute effectively.  Free-phase LNAPL, and  in some
cases  DNAPL, commonly occurs as  broad,  flat-lying
layers and can be recovered by a horizontal well placed
within or slightly below the layers. This, of course, re-
quires that the NAPL location is known and well  place-
ment is accurate. However, accurate determination of
LNAPL location can be difficult,  and even approximate
                                                   26

-------
determination of DNAPL location can be difficult (31).
Assuming that the NAPL can be identified, to optimize
recovery, well placement must have a relatively narrow
window of depth. For example, consider a LNAPL layer
0.3 meters (1  foot) thick.  The best placement  of the
recovery well  would  be  at  or  slightly  below the
NAPL/water interface. If the well is a few decimeters too
shallow, it would overlay the NAPL, and recovery would
be minimal. (Suction  would have to be used to recover
any liquid.) If it is a few decimeters too  deep, water
recovery must  precede NAPL  recovery. Lowering the
water table significantly may cause the newly created
vadose zone to trap some of the NAPL (32). For depths
within 10 meters (32.8 feet) of the ground surface, cur-
rent methods can  drill accurately to within a few vertical
decimeters (less than a foot)  (accuracy is better than 5
percent of the depth).

Natural fluctuations in the water table may change the
depth of the interface between NAPL and  water. This
can be particularly problematic if fluctuations cause the
NAPL to move significantly above or below the horizon-
tal well. The severity of this problem can be assessed
by monitoring water levels  for at least 1 year and, pref-
erably, by accessing  historical records. Using an inter-
ceptor trench may be preferable at NAPL sites with large
seasonal fluctuations in water level.

NAPL recovery deflects the interface with water so that
it may be advantageous to simultaneously recoverwater
using another pump either  below (for LNAPL) or above
(for DNAPL) the interval of  contaminant recovery. Using
two pumps is a common method of recovering LNAPL
with vertical wells (32),  and this method has been used
to improve the recovery of DNAPL (33). A similar geome-
try is feasible with a horizontal well, except an additional
well must be created either above or below the contami-
nant recovery well (Figure 2-20). Alternatively, it may be
cost  effective to control water levels with vertical wells
while a horizontal well recovers free product.
At least eight  horizontal wells  have been  installed to
recover free product. One of those wells was designed
to recover  machining oil  beneath a  building  and,  al-
though specific data are  unavailable, it is reportedly
producing more oil than expected (17). At least seven
wells were installed for free-product recovery and hydro-
dynamic control to halt the migration of  petroleum-
contaminated ground water into a residential area (34).

2.3.4  Sparging and Vapor Extraction

A technique that couples sparging with vapor extraction
has been used with horizontal wells at a minimum of five
locations since the DOE Savannah River site in Aiken,
South Carolina, first  demonstrated this approach (35).
During an initial 139-day demonstration, air was injected
into one well below the water table, and air and  VOCs
were recovered from another well above the water table.
This  approach  recovered  7,200  kilograms  (15,873
pounds) of trichloroethylene (TCE), which was five times
more than was  estimated for recovery using vertical
wells (35).

This application  consists of two horizontal wells,  one
located over the other as shown in Figure 2-20.  The
primary advantage of using horizontal wells in this ap-
plication is that they encourage the injection of air along
the length of the well, thus lengthening the area covered
compared with sparging using vertical wells. This can
represent a significant improvement because in the ab-
sence of confining layers, air channels  from a vertical
well may rise rapidly, remain close to the well, and cover
a small portion of the aquifer. A horizontal well used for
the upper well  in this application is less critical and can
be  replaced with several vertical wells to recover in-
jected air and  stripped VOCs. As of 1993, at  least 19
horizontal wells  have  been  installed  in  integrated
sparging  and vapor recovery systems at five sites.
Figure 2-20.  Recovery of LNAPL with one horizontal well, and
           water control with another well below.
2.3.5  Bioremediation

Horizontal wells can be used to enhance bioremedia-
tion, either by injecting liquids,  such as nutrients and
hydrogen peroxide, or by  injecting gases, such as air or
oxygen. The design considerations and approaches are
similar to applications for pump and treat or vapor ex-
traction.
2.3.6  Flushing

Uniform infiltration from a horizontal well can be tapped
to enhance soil washing or flushing. Twenty-six horizon-
tal wells  have been installed as surfactant infiltration
galleries, and two wells  as recovery wells. Information
on their  performance is  unavailable,  but Wilson  and
Kaback (27) provide some information  resources.
                                                  27

-------
2.3.7   Soil Monitoring and Sampling
Another potentially beneficial application of horizontal
wells involves collecting samples and/or monitoring be-
neath structures. Two known cases of horizontal wells
used for monitoring or sampling exist. A slotted pipe was
installed beneath a mixed-waste landfill at Sandia, New
Mexico, to obtain soil vapor samples  (36).  At another
site, soil samples were taken while drilling a horizontal
well. The  sampler was a hydraulically powered coring
tool with an inner  barrel actuated by the drilling fluid
pump pressure  (37).  The inner barrel was hydraulically
pushed into the formation,  forcing the  sample into an
enclosed plastic cylinder. After withdrawing the cylinder
into an  outer housing, a valve closed to protect it from
contamination  while  the  drill string and  core sample
were withdrawn from the wellbore. This method allowed
access  to  high-quality samples  during drilling,  but it
required removal of the drill  bit, insertion and removal of
the core barrel,  and reinsertion of the drill bit. Each drill
rod required handling four times  during this operation,
which illustrates the excessive labor involved in recov-
ering one  sample from a long wellbore. Moreover, this
sampling method requires that the wellbore remain open
while the drill rods are removed, which either requires
specialized drilling techniques or  a stable formation.
Methods of obtaining soil samples without removing drill
rods are currently  unavailable. Such a method would
markedly  increase the value of  horizontal drilling and
thus, is currently being investigated. Lower drilling costs
and improved methods should increase horizontal well
applications for sampling and monitoring.

Inclined wells created by sonic drilling greatly comple-
ment horizontal wells because they promise to markedly
improve sampling capabilities. Sonic drilling uses high-
energy  vibrations to resonate a drill  string of heavy
casing.  The vibrations enhance  penetration  and force
formation  material  into the  casing. A sample retrieval
system  clears the casing and brings the sample to the
surface. Thus,  sample recovery  is integrated with the
drilling process. The quality  of the samples that this
technique  obtains can be excellent because  no fluids
are required during drilling.
Sonic drilling has  created  approximately 12 inclined
wellbores  and slant wells. The Sandia National Labora-
tory in 1993 obtained soil samples beneath a mixed-
waste landfill using two wellbores  inclined at 15 degrees
from horizontal  and 43 meters (141.1 feet) long (Figure
2-21) (38). Soil samples were taken ahead of the drill bit
to reduce sample temperature and analyzed in the field
for metals. (Chromium was the primary contaminant.) A
pipe 15 centimeters (6 inches) in diameter was driven to
create the wellbore. Then, a  10-centimeter (4-inch) poly-
vinyl chloride (PVC) casing  and screen was inserted as
the pipe was jacked out. An inverted plastic sleeve was
inflated  in  the screen to obtain vapor samples.
                         Chemical Waste Landfill  /
Figure 2-21.  Inclined wellbore created with resonant sonic drill-
           ing, Sandia National Laboratories (38).

Similar studies have  been conducted at the DOE Han-
ford site, where approximately five wellbores were cre-
ated  for sampling  and  one inclined wellbore  was
completed for soil vapor extraction  by 1993 (12). The
"Case Histories" section that follows describes the soil
vapor extraction well. Cobble-rich sediments at Hanford
present extremely challenging conditions for drilling and
sampling, and the success of sonic drilling at this loca-
tion is a powerful testament to  the  capabilities of this
application.

Using sonic techniques to drill three  inclined wellbores,
continuous cores were recovered at Newark Air Force
Base  (AFB),  Newark, Ohio  (39).  The wellbores were
drilled under a building at 45 degrees to a VD of approxi-
mately 5.5 meters (18 feet). The wellbores readily pene-
trated a reinforced concrete beam.

2.4   Case Histories

This section overviews known applications,  discusses
average costs of previous projects, and describes four
selected applications of horizontal or inclined wells.

2.4.1   Overview

As of the summer of 1993, there  were more than 100
applications of horizontal wells at 30  contaminated sites
(17). Approximately one-quarter of the wells strictly ex-
tracted ground water, one-quarter extracted vapor, and
the remaining half was used for a  variety of techniques
including free-product recovery,  sparging, sampling,  or
combinations of two  recovery techniques (Table 2-6).
The majority of the wells were created  at depths of less
than 8  meters (26.2 feet).  This  partially reflects the
tendency to find more contaminants at shallow depths
than at great depths.  The depths of the majority of these

Table 2-6. Summary of  Applications of Horizontal Wells (28)

Vertical
Depths
<8 m
8-30 m
>30 m
Totals
Ground-
Water
Extraction
Wells
13
9
4
26
Soil
Vapor
Extraction
Wells
18
4
3
25

Other
Wells
52
1
2
55


Totals
83
14
9
106
                                                   28

-------
wells  also reflect the lower limit of electronic beacon
locating systems. The electronic beacon, or "walkover,"
method currently is significantly less expensive than any
alternative. Therefore, the cost of creating  wells in-
creases abruptly if they are deeper than 8 meters (26.2
feet).
The measured depth, or total length, of horizontal wells
created for environmental  purposes generally ranges
from  10 to 670 meters (32.8 to  2,197 feet), but the
majority of the wells are in the range of 30 to 150 meters
(98.4  to 492 feet)  (Figure 2-22). Like the  range of
depths, the range of well lengths probably also results
from  both technical  and economic  factors. Wellbores
150 meters (492 feet) long  can be drilled with  relatively
small, compaction-type rigs and can be completed with
pull-back techniques. Accordingly, this type of well is
typically  less expensive than those  created with more
sophisticated rigs and completion methods.
  20
   15
   10
 
-------
Table 2-7.  Sites Where Horizontal Wells Have Been Used for Ground-Water Recovery
Reported
Contaminant
Type/Phase
TCE/APL and
DNAPL
TCE/APL and
DNAPL
Halogenated
hydrocarbons/
APL and DNAPL
TCE/APL and
DNAPL
TCE/APL and
DNAPL
TCE/APL and
DNAPL
Petroleum
hydrocarbons/
APL and LNAPL
Petroleum
hydrocarbons/
APL and LNAPL
Petroleum
hydrocarbons/NR
Petroleum
hydrocarbons/NR
JP-4jetfuel/APL
and LNAPL
Leaking
underground
storage tank/APL
and LNAPL
Petroleum
hydrocarbons/NR
Petroleum
hydrocarbons/NR
Lime Lake
leachate/APL
TCE/NR
TCE/NR
Petroleum
hydrocarbons/NR
Petroleum
hydrocarbons/NR
Site
Taylor, Ml
Savannah River
Site, Aiken, SC
Geismar, LA
Union City, CA
San Francisco
Bay area, CA
Tinker AFB, OK
Sacramento, CA
Carson City, NV
Fullerton, CA
Beaumont, TX
Williams AFB,
AZ
Houston, TX
Geismar, LA
Geismar, LA
Barberton, OH
Savannah River
Site, Aiken, SC
Savannah River
Site, Aiken, SC
Los Angeles,
CA
Norwalk, CA
Date
1987
June 1991
August
1992
July 1992
1992
November
1992 to
February
1993
December
1991
April 1992
April 1992
July 1992
July 1992
September
1992
November
1992
February
1993
February
1993
May 1991
November
1992
February
to March
1992
August to
September
1992
Number
and
Type of
Wells
2
1
2
1
1
5
1
1
1
2
2
1
1
2
9
1
2
2
2
GWE
GWE
GWE
GWE
GWE
GWE
GWE
GWE
GWE
GWE
GWE
GWE
GWE
GWE
GWE
SVE
SVE
SVE
SVE
Site Geology
Unconsolidated
sediments
Fine silty sand,
silt, and clay
Sand and clay
Fine silty sand
and clay
Sand, sand
and gravel,
intermittent clay
Fine sand, silt,
and clay with
shale and
sandstone units
Sandy clay
Unconsolidated
channel sands
Silt, clay, and
intermittent fine
sand
Silty sand with
clay stringers
Fine sand and
silt clay with
gravel and
cobbles
Clayey soil
Sand and clay
Clay and sand
Clay and sand-
size tailings
Fine silty sand,
silt, and clay
Fine sand, silt,
and clay
Unconsolidated
silty sand
Silty sand and
sandy silt
Type of
Wellbore
Blind
Blind
Blind
Blind
Blind
Blind
Blind
Blind
Blind
Blind
Continuous
Blind
Blind
Blind
Blind
Blind
Blind
Blind
Blind
Measured
Depth/
Horizontal
Length
(Feet)
550/—
488/258
435/363
469/400
1 00/70
200/80
920/200
362/100
250/1 00
460/325
— /300
2,200/500
320/—
185/—
295/—
320/—
750/—
300/1 52
725/420
720/420
600/400
360/—
320/—
Vertical
Depth
(Feet) Well Materials
<25
152
14
16
15
25
110
150
3x35
50
25
110
25
235
35
18
36
20
40
105
110
45
85
25
7 in. ID
6 in. ID
6 in. ID,
HOPE;
4 in. ID,
SS filter
Sin. ID,
PVC
Sin. ID,
SS
6 in. ID,
HOPE
4 in. ID,
SS filter
3 in. ID,
PVC
Sin. ID,
PVC
6 in. ID,
HOPE
SS
HOPE
20-slot
12-slot
pack
1 0-slot
1 0-slot
20-slot
12-slot
pack
20-slot
20-slot
20-slot
6 in. ID 20-slot
HOPE;
4 in. ID HOPE
filter pack
6 in. ID
prepack
4 in. ID,
PVC
6 in. ID,
HOPE
SS
1 0-slot
20-slot
6 in. ID
Enviroscreen
4 in. ID,
HOPE
6 in. ID
Sin. ID,
PVC
Sin. ID,
PVC
4 in. ID,
HOPE
20-slot
HOPE
1 0-slot
1 0-slot
20-slot
Drilling
Contractor
Drilex
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Michels
Environmental
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
                                                        30

-------
Table 2-7.  Sites Where Horizontal Wells Have Been Used for Ground-Water Recovery (continued)
Reported
Contaminant
Type/Phase       Site
Date
Number
and
Type of                Type of
Wells    Site Geology   Wellbore
Measured
Depth/
Horizontal  Vertical
Length     Depth                Drilling
(Feet)      (Feet)  Well Materials Contractor
Petroleum
hydrocarbons/NR
Petroleum
hydrocarbons/NR
Hydrocarbons/NR
Hydrocarbons/NR
Hydrocarbons/NR
Orcutt, CA

Charlotte, NC
Las Vegas,
Las Vegas,
Cherry Hill,
NV
NV
NJ
October
1992
June 1993
1993
1993
1993
4SVE
2SVE
1 SVE
1 SVE
2 SVE
Sandy
Sandy
sandy
NR
NR
NR
soil
silt and
clay



Continuous
Continuous
Continuous
Continuous
Continuous
1 x 1207—
3 x 707—
150/100
120/110
4267—
8407—
— /140
1 x 4
3x5
8
7
<5
<5
<5
2 in. ID, 10-slot
PVC
2 in. ID, 10-slot
PVC
4 in. ID HOPE
4 in. ID HOPE
2 in. ID HOPE
UTILX
UTILX
UTILX
UTILX
UTILX
APL = aqueous phase liquid; DNAPL = dense nonaqueous phase liquid; GWE = ground-water extraction;
LNAPL = light nonaqueous phase liquid; NR = not reported; SS = stainless steel; SVE = soil vapor extraction; TCE = trichloroethylene
Table 2-8.  Sites Where Horizontal Wells Have Been Used for Vapor Extraction
Reported
Contaminant
Type
TCE
TCE
Petroleum
hydrocarbons
Petroleum
hydrocarbons
Petroleum
hydrocarbons
Petroleum
hydrocarbons
Hydrocarbons
Hydrocarbons
Hydrocarbons
Site
Savannah
River Site,
Aiken, SC
Savannah
River Site,
Aiken, SC
Los Angeles,
CA
Norwalk, CA
Orcutt, CA
Charlotte, NC
Las Vegas,
NV
Las Vegas,
NV
Cherry Hill,
NJ
Date
May 1991
November
1992
February
to March
1992
August to
September
1992
October
1992
June 1993
1993
1993
1993
Number
and
Type of
Wells
1 SVE
2 SVE
2 SVE
2 SVE
4 SVE
2 SVE
1 SVE
1 SVE
2 SVE
Site Geology
Fine silty sand,
silt, and clay
Fine sand, silt,
and clay
Unconsolidated
silty sand
Silty sand, and
sandy silt
Sandy soil
Sandy silt and
sandy clay
NR
NR
NR
Measured
Depth/
Horizontal
Type of Length
Wellbore (Feet)
Blind 300/152
Blind 725/420
720/420
Blind 600/400
Blind 360/—
320/—
Continuous 1 x 120/ —
3 x 707—
Continuous 150/100
120/110
Continuous 426/ —
Continuous 840/ —
Continuous — /140
Vertical
Depth
(Feet) Well Materials
105
110
45
85
25
1 x 4
3x5
8
7
<5
<5
<5
6 in. ID HOPE
Sin. ID,
10-slot PVC
3 in. ID,
10-slot PVC
4 in. ID,
20-slot HOPE
2 in. ID,
10-slot PVC
2 in. ID,
1 0-slot PVC
4 in. ID HOPE
4 in. ID HOPE
2 in. ID HOPE
Drilling
Contractor
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
Eastman
Cherrington
UTILX
UTILX
UTILX
UTILX
UTILX
APL = aqueous phase liquid; DNAPL = dense
LNAPL = light nonaqueous phase liquid; NR =
        nonaqueous phase liquid; GWE = ground-water extraction;
         not reported; SVE = soil vapor extraction; TCE = trichloroethylene
                                                           31

-------
Table 2-9.  Sites Where Horizontal Wells Have Been Used for Assorted Environmental Applications

                                                                         Measured
                                                                         Depth/
                                                                         Horizontal  Vertical
Purpose Site
Characterization, Kirtland
monitoring at AFB, NM
landfill
Bioventing-diesel Cincinnati,
OH
Ground water Seattle, WA
and petroleum
recovery-LNAPL
Product recovery- Ponca City,
petro-hydro OK
Date
May 1993
July 1993
June 1993
1988
Number
of Wells Site Geology
1 Sand, gravel,
and cobbles
2 Fluvial
sediments
1 Medium to
coarse sand
7 Fine sand, silt,
and clay
Type of
Wellbore
Continuous
Continuous
Continuous
Continuous
Length
(Feet)
400/—
1 90/1 40
320/1 80
980/—
Depth
(Feet)
40
10
13
17
Well Materials
None
2 in. ID,
20-slot PVC
4 in. ID,
20-slot
prepack PVC
4 in. ID
slotted PVC
Drilling
Contractor
Charles
Machine
Works— Ditch
Witch
University of
Cincinnati
UTILX
UTILX
APL = aqueous phase liquid; DNAPL = dense nonaqueous phase liquid; GWE = ground-water extraction;
LNAPL = light nonaqueous phase liquid; SVE = soil vapor extraction; TCE = trichloroethylene
Table 2-10. Sites Where Horizontal Wells Have Been Used for Several Purposes
Measured

Purposes
In situ stripping of
TCE -contam-
inated soil and
ground water
Air injection, SVE,
bioactive barrier
hydrocarbon
Dual well system
for SVE and GWE
with TCE in soil
and ground water
Radio frequency
heating/SVE of
TCE in soil
Dual well system
for SVE and GWE
of TCE-contam-
inated soil and
ground water
Washing soils
contaminated with
gasoline


Air injection, SVE,
and gasoline
remediation
system for
underground
storage tank

Site
Savannah
River Site,
Aiken, SC
Ponca
City, OK
Tinker
AFB, OK

Savannah
River Site,
Aiken, SC
South San
Francisco
Bay Area,
CA

Minden, NV


Del City,
OK

Date
October
1988
September
1988

November
1991

September
1992
October
1992

October
1992


October
1992
Number
and Type
of Wells
1 SVE
1 injection
3 SVE/
bioinjection
1 SVE
1 GWE
and control
1 SVE
1 SVE
1 GWE

22
infiltration
galleries
2 recovery
4 recovery
3 GWE/
SVE/
injection
Depth/
Horizontal Vertical
Site Geology
Fine silty sand,
silt, and clay
Unsaturated
fine silty sand,
silt, and clay
Fine silty sand,
silt, and clay

Fine silty sand,
silt, and clay
Sand, gravel,
and
intermittent clay

Sandy clay
with gravel


Fine silty sand,
silt, and clay
Type of
Wellbore
Blind
Blind
Continuous
Blind
Blind

Continuous
Blind
Blind

Blind
Blind
Blind

Continuous
Length
(Feet)
290/200
485/300
—MOO
270/70
270/70

570/230
210/110
210/110

20/—
35 and
65/—
135/—

300/—
Depth
(Feet)
75
155
20
15
25

40
18
20

3
3
3

12
Well Materials
4 in. ID,
10-slot SS
screen;
2 in. ID steel
perforated pipe
2-4 in. ID,
20-slot PVC
6 in. ID
prepack SS
6 in. ID
prepack SS
4 in. ID,
10-slot
fiberglass
3 in. ID,
20-slot HOPE;
Sin. ID,
20-slot SS

2 in. ID PVC
4 in. ID PVC
2 in. ID PVC

2 in. ID,
20-slot PVC
Drilling
Contractor
Eastman
Cherrington
S&S Harris
Michels
Environmental

Charles
Machine
Works
Eastman
Cherrington

UTILX


S&S Harris
                                                          32

-------
Table 2-10.  Sites Where Horizontal Wells Have Been Used for Several Purposes (continued)
Purposes Site
Dual well system Cincinnati,
for diesel recovery OH
and GWE and
ground-water
control
Number
and Type
Date of Wells Site Geology
December 1 product Fluvial
1992 recovery sediments
1 GWE
and control
Measured
Depth/
Horizontal
Type of Length
Wellbore (Feet)
Continuous 297/ —
Continuous 325/ —
Vertical
Depth
(Feet)
19
21
Drilling
Well Materials Contractor
4 in. ID PVC Underground
Research, Inc.
4 in. ID PVC
In situ stripping of  Fallen, NV  December  3 SVE
soil and ground              1992
water contam-                       4 air
inated by                           injection
underground
storage tank
Silty sand      Continuous 170/—     4      2 in. ID,
                      1707—     5      10-slot PVC
                      2807—     4

             Continuous 1507—     14     2 in. ID,
                      1807—            10-slot PVC
                      2407—
                      2807—
                                                                                              UTILX
Air injection, SVE, Moore, OK March
and gasoline 1993
remediation
system for
underground
storage tank
Dual wall system Portland, April 1993
for SVE and GWE OR
of contaminated
soil and ground
water by
underground
storage tank
4 GWE/ Fine silty sand, Continuous 5007— 12
SVE/ silt, and clay
injection



1 SVE Silt Continuous 475/300 6.5

1 GWE Continuous 475/300 23
and control



4 in. ID,
20-slot PVC




4 in. ID,
20-slot PVC

4 in. ID,
20-slot PVC
prepack
screen
S&S Harris





UTILX






APL = aqueous phase liquid; DNAPL = dense nonaqueous phase liquid; GWE = ground-water extraction;
LNAPL = light nonaqueous phase liquid; SS = stainless steel; SVE = soil vapor extraction; TCE = trichloroethylene
significantly less  costly than the ones  detailed here,
were unavailable at the time of writing. Each discussion
includes the following information:
• Location of site, date of study, contact for the con-
  tracting agency, and contact for the drilling contractor.
• Remediation objective.

• Geology of site as related to drilling operations.
• Design,  description  of well trajectory,  completion
  technique,  intended purpose, and well performance
  (if available).
• Results, discharge, contaminant  concentration with
  time,  area  where drawdown is affected, comparison
  to model, and cost of project (if available).

2.4.3.1   Ground-Water Recovery, Geismar,
         Louisiana
Location.  Two ground-water extraction wells were in-
stalled in Geismar during August 1992 at an abandoned
herbicide manufacturing plant (37).
Objectives. The wells were installed as components of
a pump and treat ground-water extraction system. The
system  was designed to remediate  ground water con-
           taminated with ethylene dichloride  (EDC) and mono-
           chlorobenzene (MCB)  beneath the  plant. There were
           two ground-water contaminant plumes: the south plume
           had  an area  of  approximately  1,486  square meters
           (16,000 square feet), with MCB concentrations of 100 to
           200 ppm and EDC concentrations of 10 to 100 ppm; the
           north plume had an area of approximately 557 square
           meters (6,000 square feet), with MCB concentrations of
           100 to 370 ppm and EDC concentrations of 10 to 70
           ppm.  The conceptual plan was to install two horizontal
           wells with screens 122  meters (400 feet) long. The goal
           was to install each horizontal well on top of a clay layer
           along the base of the affected zone, parallel and through
           the longitudinal axis of  each plume.

           Site Conditions. The two wells,  H-50 and H-51, were
           installed beneath an abandoned wastewater pipeline
           (south plume) and a product loading area (north plume),
           respectively.  The  contaminated  portion of the aquifer
           consisted of Holocene silty clay.

           The wellbore path for H-50 was designed to run west to
           east through the  principal area of contamination. The
           wellbore path had to run through a narrow corridor of
           vertical subsurface drill  shafts and pilings that supported
           an  overhead  steel  superstructure.  The  corridor was
                                                   33

-------
6.1  meters (20 feet) wide, allowing a maximum of 2.4
meters (8 feet) clearance to the south and 3.7 meters
(12 feet) clearance to the north. The abandoned waste-
water pipe ran about 1.5 meters (5 feet)  above the
proposed well path.

The wellbore path for H-51 was  adjacent to an active
railcar loading area and beneath a concrete  road. The
H-51 wellbore was allowed a 0.6-meter (2-foot) vertical
and a 2-degree horizontal deviation.

Well Geometry. The well geometry for the two horizontal
wells is presented in Table 2-11.
Table 2-11.  Specifications of Horizontal Wells, Geismar,
          Louisiana
                 H-50
                                   H-51
Entry Angle
Kickoff Point
End-of-Curve
Horizontal Length
15° to Horizontal
19ftMD/5ft VD
59ftMD/12ft VD
459 ft
15° to Horizontal
19ftMD/5ft VD
67ftMD/14ftVD
467 ft
Drilling Method. The drill rig used for this job was rated
to 610 meters (2,000 feet) for vertical drilling,  with a
hydraulic drill stem hoist that had a 311-kilonewton (kN)
(70,000-pound) hoist and a 133-kN (30,000-pound) pull-
back capacity. The drill rig mast could incline from verti-
cal to 15 degrees to horizontal.

This application  used two different downhole assem-
blies. In the curved section of the wellbore, the downhole
assembly consisted of a 30.5-meter  (100-foot) radius,
and a 17.1-centimeter (6.75-inch) diameter RDM mud
motor with a 31.1-centimeter (12.25-inch)  expandable
bit. Wellbore surveying was done with a tool face indi-
cator, which provided  inclination and  tool face orienta-
tion but did not provide wellbore azimuth. Data from the
surveying system were communicated to  the ground
surface by creating pressure pulses in the drilling mud.
When drilling neared the end of the curve, a film-based
multishot magnetic survey tool confirmed the tool face
indicator readings and the wellbore location. (The tool
face indicator and the multishot magnetic survey tools
are not commonly used in  environmental directional
drilling.) In the horizontal section  of  the wellbore, the
downhole assembly consisted  of a  91.4-meter (300-
foot) radius and a 12.1-centimeter (4.75-inch) diameter
RDM  mud motor with a  22.2-centimeter  (8.75-inch)
expandable  bit. Here,  wellbore surveying was accom-
plished with  a wireline  magnetometer-accelerometer lo-
cator and an electromagnetic secondary survey system.

The wellbores were drilled and the curved section of the
wellbore was cased with a 10-inch HOPE  casing. The
horizontal section of the wellbore was left open  for the
well  installation.  The open hole method was  possible
because the formation had a high clay content. In addi-
tion, a mixed metal hydroxide (MMH) additive was used
with bentonite in the water-based drilling fluid. MMH is
an insoluble crystalline inorganic compound containing
two or more metals in  a unique hydroxide lattice. The
bentonite/MMH mixture extended the gel strength of the
drilling fluid and  allowed suspension of soil cuttings
indefinitely when the drilling fluid circulation was halted.

Well Materials. The wells were constructed of  6-inch
internal diameter (ID), 0.5-millimeter (0.02-inch)  slotted
HOPE with a 4-inch ID  stainless steel prepack inside.

Well Installation. The well materials were pushed into
the open wellbore. The horizontal section of the wellbore
was relatively clear of cuttings, so the well materials did
not experience extra friction during well installation.
Each well was wash developed before the drill rig was
removed from the drill site. Drilling fluids were displaced
with potable water and  pumped into tanks for treatment
and disposal. The drill crew washed the HOPE liner with
a wash tool to remove the filter cake and the fine-grained
sediments. Water discharging  from  the wells became
relatively free of sediments after 6 to  8 hours of flushing
with water.

Field Effort. Well H-50 required 17 days of construction,
11 days longer than anticipated. A mechanical problem
with the steering mechanism  on the downhole mud
motor resulted in  two aborted curved section attempts
before the problem was  recognized and  corrected.
There also was one aborted attempt to drill the horizon-
tal section due to overcompensating while controlling
the elevation  of the wellbore. The well was completed
with a  110-meter (363-foot)  horizontal well screen (as
opposed to the 122-meter  (400-foot) planned length).
Most of the wellbore was steered along the specified
trajectory, except a 12.5 meter (41-foot)  section that fell
0.3 meters (1  foot)  below the specified trajectory. The
wellbore was successfully steered through the narrow
corridor and away from the  vertical pilings and monitor
well. Soil cores were also retrieved successfully.

H-51 was constructed in 6 days. The completed well had
122  meters (400 feet) of well screen, as planned. Most
of the wellbore was  placed  as  specified, although
a 21.6-meter (71-foot)  section lay 0.3 meters (1 foot)
below  the maximum  specified depth  of 4.9  meters
(16  feet), according to the surveys performed after
completion.

Results of Tests Performed on Wells. Three steps were
required to determine the pumping performance of each
well:

1. The well was pumped for 8  hours to  evaluate pump
   size and flow rates.
                                                   34

-------
2. The well pump control was manually operated to
   obtain the optimum settings for the continuous auto-
   matic pumping operation.
3. The well was placed in continuous automatic opera-
   tion.
Each well  elicited different pumping results because of
the heterogeneity of the sediments' hydraulic properties.
The  results for H-50 were:
• A  long-term  pumping  rate of 4.9 liters (1.3 gallons)
  per minute.
• An area of influence that spanned 48.8 meters (160
  feet) on either side of the well.
• An estimated  one pore  volume of the well  capture
  zone that  pumped once  every 149  days.
The  results for H-51 were:
1. A long-term pumping rate of 22.3 liters (5.9 gallons)
   per minute.
2. An area of  influence  that spanned 21.3 meters (70
   feet)  on either side of the well.
3. An estimated one pore volume of the well  capture
   zone that was pumped  once every 15 days.
Current Status.  The wells  are pumping  as part of a
full-scale remediation system.
Cosf.The wells cost approximately $400,000 each.

2.4.3.2  Ground-Water Recovery, Barberton, Ohio
The  horizontal well project in Barberton exemplifies a
multiple-well installation. A pilot well was installed to
characterize drilling conditions, determine sediment hy-
draulic properties,  and test the well  screen and filter
pack design. The pilot well installation provided impor-
tant  information that was used to optimize the design of
the remaining eight horizontal wells (40).
Location. One  pilot and  eight production ground-water
extraction wells were installed in Barberton during 1993.
Objectives. The wells were installed to collect leachate
seeping from the base  of two  Solvay  Process waste
impoundments (lime lakes) into adjacent streams. The
leachate had a pH of 12 to  13 and  contained  various
hazardous constituents,  including chlorinated hydrocar-
bons, asbestos, and lead (40).
Site  Conditions. The lime lakes, Lime Lake 1 and 2, are
located  within  a large chemical manufacturing facility
that includes several portions of major waterways. There
are two streams that flow along  the north, east, and
south boundaries of Lime Lake 1 and the north and
northeast  boundaries  of Lime Lake 2. The lakes are
about 12.2 meters (40 feet) above local grade and con-
tain clay- to sand-sized lime spoils that are 12.2 to 15.2
meters  (40  to 50  feet) thick and rest  upon a  hard,
chemically altered  layer of native soils. The hard layer
acts as  an aquitard and causes the leachate to form a
mound within the lime lakes. A thick slaker sand and
cinder cover was spread across the surface of the lakes
to prevent surface erosion (40).

Well Geometry. The pilot well was constructed in a blind
wellbore with  a  17-degree to horizontal entry  angle,
drilled with a 183-meter (600-foot) radius of curvature,
231-meter (757-foot) MD, 152-meter (500-foot) horizon-
tal length, and 11.6-meter (38-foot) VD. The geometry
of the eight horizontal wells resembled that of the pilot
well. Two wells had the horizontal section shortened to
91.4 meters (300 feet) due to  a  rapid change  in the
elevation of the hard layer (40).

Drilling Method. The pilot well was drilled with a medium-
sized utility-type drilling rig equipped  with a hydraulic
spud jet drilling tool, a mud motor to drill through hard
layers,  and a  magnetometer-accelerometer steering
tool. A washover pipe enlarged the wellbore. A guar
gum,  water-based drilling fluid was used during drilling.
The eight production wells were drilled  in a similar man-
ner, except for modifications based on the experience
gained  from drilling the  pilot well. The  modifications
included additives mixed to the guar gum drilling fluid to
improve  its ability to clean  the  wellbore and maintain
wellbore integrity and a reduced rate of penetration
during drilling.

Well Materials. The pilot well was constructed  of 10-
centimeter (4-inch) ID, 0.5-millimeter (0.02-inch) slotted
HOPE casing, and a 20/40 sand filter pack. The other
eight  wells were constructed of 15-centimeter (6-inch)
ID HOPE, 0.5-millimeter (0.02-inch) slotted HOPE cas-
ing, and no filter pack.  The decision not to include a filter
pack  in  the  eight production wells was  based on the
assumption that it would become encrusted and clogged
with carbonate minerals. The fine materials that  did
enter the pilot well were attributed to over-pumping and
not the well design. The decision to omit the filter pack
in the production wells  proved correct because fine-
grained  materials did not enter the wells during well
development and testing. An electrical centrifugal pump
was used in each well.

Well Installation Method. The well materials for the pilot
well were pushed into the washover pipe. The filter pack
was installed around the well screen with a "sand shoe"
as the washover pipe was withdrawn from the wellbore.
Installation  of  the remaining wells was similar except
that they did not have a filter pack.

Field Effort.  The pilot well installation  required two at-
tempts.  Some problems occurred while constructing the
pilot well:

• Drilling fluid  and  lime spoils fractured to the surface.
                                                   35

-------
• The  hydraulic spud jet met refusal at several thin,
  hard layers.

• The washover pipe became stuck twice and was re-
  trieved by pulling on  it with bulldozers.

These problems were overcome by changing the drilling
fluid rheology, slowing the rate of penetration of the drill
stem, and using a downhole  mud motor to drill through
the hard layers.

Results of Tests Performed on Well. The pilot well pro-
duced 6.1  liters (1.6 gallons) per minute after 45 days of
pumping. The eight  production wells produced 15.1 li-
ters (4 gallons) per minute on an intermittent pumping
schedule that was determined for each well. The rate of
15.1 liters (4 gallons) per minute was selected to provide
a flow velocity that would keep the pump discharge line
clear of solids and minimize the infiltration of fine mate-
rials into the wells.

Current Status. The wells currently are being used to
collect the lime lake  leachate.

Cost. The total cost of the nine-well system has not been
released.

2.4.3.3   High-Angle Inclined Boreholes, Newark
        AFB,  Ohio

Location. Three high-angle  boreholes were drilled  at
Newark AFB in Newark during 1993.

Objectives. The purpose of the boreholes was to obtain
continuous core samples beneath Building 4.  The soil
beneath the building was believed to be contaminated
with chlorofluorocarbons.

Site Conditions. The soil was silty clay with some sand
and gravel.

Well Geometry. Three boreholes were drilled, each at a
45-degree angle. The boreholes reached vertical depths
of 5.2 to 5.5 meters (17 to 18 feet).

Drilling Method. The wellbores were drilled using the
rotosonic technique  described earlier with  the drill rig
mast at a 45-degree angle.

Field Effort. The wellbores successfully penetrated  a
reinforced concrete beam.

Cost. The three boreholes were drilled at a total cost of
$7,530.

2.4.3.4   Inclined Well for Soil Vapor Extraction
        Well,  Hanford Site, Washington

Location. The DOE (through the Westinghouse Hanford
Company and Pacific Northwest Laboratory) and Water
Development Corporation  formed  a Cooperative Re-
search and Development Agreement (CRADA) for the
purpose of developing a resonant sonic drilling method
that can meet the rigorous drilling conditions found at
the DOE's Hanford site in southeast Washington. Phase
I of the research program included installing a 45-degree
angle soil vapor extraction well at the 200 West Area
Carbon Tetrachloride Expedited Response Reaction site
(200 West area) at the Hanford site (12).

Objectives. The purpose of the inclined well was to
demonstrate the ability of the resonant sonic drill rig to
create an angled wellbore, and to  install a soil vapor
extraction well at the 200 West area.

Site Conditions. The upper geologic unit of the Hanford
formation beneath the 200 West area consists of two
fades: coarse-grained  sand and  granule-to-boulder
gravel from which matrix is commonly lacking; and fine-
to coarse-grained sand and silt that commonly display
normally  graded rhythmites from an  inch  to several
inches. In general, the coarse fades is 5-percent boul-
der and ranges in thickness from 6.1 meters (20 feet) to
greater than 61 meters (200 feet). The underlying  fine
fades consists of 1.5 to 18.3 meters (5 to 60 feet) of silts
and fine sands, which in turn overlay sediments of Plio-
Pleistocene Ringold formation (12).

Well Geometry. The wellbore was drilled at a 45-degree
angle to a  36.3-meter (119-foot) VD and a 51.2-meter
(168-foot) MD.

Drilling Method. The wellbore was drilled using the reso-
nant sonic technique with a 300-horsepower drill head
and the drill rig mast at a 45-degree angle.

Well  Installation  Method.  The  well  materials  were
pushed into the wellbore.

Well Materials. The well materials consisted of a  7.6-
centimeter  (3-inch) ID stainless steel screen and riser.

Field Effort. The wellbore was drilled and the well was
installed in  9 days.

Results of Tests Performed on Well. The inclined well
yielded satisfactory flow rates compared to an adjacent
soil vapor extraction well drilled with a cable tool drill rig.

Current Status. The well will be used to remediate the
vadose zone at the 200 West area.

Cost. No  cost analysis  was performed on this well.
Phase II of the CRADA will include cost analysis of the
resonant sonic drilling method.

2.5   References

When an  NTIS  number is cited in a reference,  that
document is available from:
  National Technical Information Service
  5285 Port Royal Road
  Springfield, VA 22161
  703-487-4650
                                                  36

-------
 1.  Hantush, M.S., and I.S. Papadapulous. 1962. Flow
    of  ground  water to  collector wells.  Proceedings
    of  the American  Society  of  Civil  Engineers.
    88(HY5):221-244.

 2.  Losonsky,  G.,  and M.S. Beljin.  1992.  Horizontal
    wells  in   subsurface  remediation.   Hazardous
    Materials  Control  Conference, New  Orleans, LA
    (February).

 3.  Mukherjee, H., and M.J. Economides. 1991. A para-
    metric comparison of horizontal  and  vertical well
    performance.  Society  of  Petroleum  Engineers
    (June), pp. 209-216.

 4.  Dickinson,  W.W., R.W. Dickinson, J. Nees,  and E.
    Dickinson.  1991. Field production results with the
    ultrashort radius radial  system in unconsolidated
    sandstone  formations. Presented at the 5th UNI-
    TARJUNDP International  Conference on  Heavy
    Crude  and  Tar  Sands,  Caracas,  Venezuela
    (August).

 5.  Dickinson,  W.W., R.W. Dickinson, PA. Mote, and
    J.S. Nelson. 1987. Horizontal radials for geophysics
    and hazardous waste remediation. In: Superfund
    '87. Silver  Spring, MD.  Hazardous  Waste Control
    Research Institute, pp. 371-375.

 6.  Howsam, P., and R. Hollamby. 1990. Drilling fluid
    invasion  and permeability impairment in granular
    formations. Quarterly J.  Engin. Geol. (London) 23:
    161-168.

 7.  Hayatdavoudi, A.Z. 1994. Personal communication
    from drilling fluid specialist, Eastman Cherrington
    Environmental  Corporation,  to  D.  Wilson, Inde-
    pendent  Environmental  Consultants,  LLC (Febru-
    ary).

 8.  Westmoreland, J. 1994.  Evaluation of an air drilling
    cutting   containment  system.   Sandia   Report
    SAND94-0214  UC-721  (April). Albuquerque, NM:
    Sandia National Laboratories.

 9.  Simon, R., and G. Cooper. 1994.  Use of cryogenic
    fluids for environmental drilling in unconsolidated
    formations. In: Vozniak, J.P, ed. Drilling technology.
    New York, NY: American Society of Mechanical En-
    gineers. PD-Vol. 56: pp.  199-207.

10.  Barrow,  J.C.  1994.  The  resonant sonic  drilling
    method: An innovative technology  for environmental
    restoration  programs.  Ground  Water Monitor.  Rev.
    14(2):153-161.

11.  Dance, D.R., and N.W Beattie. 1981. Sonic drilling.
    Northwest Mining Association Fall Short Course (no
    address available).
12. Westinghouse Hanford Co. 1994. Phase  I: Reso-
    nant sonic CRADA report. WHC-SD-EN-TRP-007.
    Richland, WA: Westinghouse Hanford Company.

13. Westinghouse Savannah  River  Co.  1993.  Sum-
    mary report of the drilling technologies tested at the
    integrated demonstration site. WSRC-TR-93-565.
    Aiken, SC: Westinghouse Savannah River Co.

14. Ta Inglis, T.A. 1987.  Directional drilling, petroleum
    engineering, and development studies, Vol. 2. Bos-
    ton, MA: Graham and Trotman.

15. Wu, J., and H. Juvkan-Wold. 1994. The  effect of
    wellbore curvature on tubular buckling and lockup,
    drilling technology. In: Vozniak, J.P, ed. New York,
    NY: American Society of Mechanical Engineers.
    PD-Vol. 56. pp. 17-25.

16. Driscoll, F.G. 1986. Ground water and wells, 2nd
    ed. St. Paul, MN: Johnson Division Publisher.

17. Wilson, D.D., and D.S. Kaback. 1993. Industry sur-
    vey for horizontal wells. Final report. WSRC-TR-93-
    511.  Aiken, SC:  Westinghouse  Savannah  River
    Company.

18. Joshi, S.A. 1991. Horizontal well technology. Tulsa,
    OK: PennWell Books.

19. Murdoch, L.C. 1994. Transient analysis of an ide-
    alized interceptor trench. Water Resour.  Res.  (in
    press).

20. Bear, J.  1979. Hydraulics  of ground  water.  New
    York, NY: McGraw-Hill.

21. Karlsson,  H.,  G.  Losonsky,  and G.E. Jacques.
    1992. Horizontal wellbore completions for aquifer
    restoration and their economics. Proceedings of the
    10th  Annual Hazardous Materials  and  Environ-
    mental  Management Conference, Atlantic City, NJ
    (June),  pp. 154-181.

22. Muskat, M. 1937.  The flow of homogenous fluids
    through  porous  media.  Ann Arbor,  Ml: J.W Ed-
    wards, Inc.

23. Streltsova, T.D. 1988. Well testing in heterogene-
    ous formations.  New York, NY: John Wiley and
    Sons.

24. Murdoch, L.C., and J. Franco. 1994. The  analysis
    of constant drawdown wells  using instantaneous
    source  functions. Water  Resour.  Res. 30(1):117-
    124.

25. U.S. EPA. 1991. The feasibility of hydraulic fractur-
    ing of soil to improve remedial actions.  Final report.
    EPA/600/2-91/012 (NTIS PB91-181818).

26. Dikken, B.J. 1990. Pressure drop in horizontal wells
    and  its  effect  on production  performance.  JPT
    Nov:1,426-1,433.
                                                 37

-------
27. Tarshish, M. 1992. Combined mathematical model
    of flow in an aquifer-horizontal well system. Ground
    Water 30(6):931-935.

28. Wilson, D.D., and D.S. Kaback. 1993. Results of a
    survey  of horizontal  environmental well installa-
    tions. In: Vozniak, J.P., ed. Drilling technology 1994.
    New York, NY: American Society of Mechanical En-
    gineers. PD-Vol. 56. pp. 193-199.

29. U.S. EPA. 1993. Guidance for evaluating the tech-
    nical impracticability of ground-water  restoration.
    EPA/540/R-93/080.

30. U.S. EPA. 1991. Soil vapor extraction technology
    handbook. EPA/540/2-92/003.

31. Cohen, R.M., J.W Mercer, and J. Matthews. 1993.
    DNAPL  site  evaluation. Ada, OK: U.S. Environ-
    mental Protection Agency

32. American Petroleum Institute. 1989. A guide to the
    assessment and remediation of underground petro-
    leum releases. API Publication 1628. Washington,
    DC: American Petroleum Institute.

33. Sale, T.C., and D. Applegate. 1994. Oil recovery at
    a former wood treating facility. Proceedings of the
    Innovative Solutions for Contaminated Site Man-
    agement, Water Environment Federation,  Miami,
    FL.

34. Downs,  C.  1993. Personal communication  from
    Conoco Corporation to D. Wilson, Independent En-
    vironmental Consultants, LLC (June).
35. Looney, B.B., T. Hazen, D. Kaback, and C. Eddy.
    1991. Full-scale field test of the in situ air stripping
    process at the Savannah River Integrated Demon-
    stration  Test Site.  WSRC-RD-91-22. Aiken, SC:
    Westinghouse Savannah River Co.

36. Wemple, R.  1994.  Personal communication from
    drilling specialist, Sandia National Laboratories, to
    D. Wilson, Independent Environmental Consult-
    ants, LLC (April).

37. Conger, R.M., and K. Trichel. 1993. Aground-water
    pumping application for remediation of a chlorinated
    hydrocarbon  plume  with horizontal well technology.
    In: Proceedings of  the Seventh  National Outdoor
    Action Conference and Exposition, Las Vegas, NV.
    Dublin, OH: National Ground Water Association, pp.
    47-61.

38. Sandia National Laboratories.  1993. Sonic drilling
    rig hits heavy metal at closed landfill.  Lab News
    45(10). Albuquerque, NM: Sandia National Labora-
    tories.

39. Clough, R. 1994. Personal communication from Al-
    liance Environmental,  Inc.,  to D. Wilson, Inde-
    pendent Environmental Consultants, LLC (March).

40. Gaillot, G. 1994. Design and installation of an in situ
    leachate collection system utilizing horizontal wells
    within two former waste impoundments. Proceed-
    ings of the Eighth National Outdoor Action Confer-
    ence and Exposition, Minneapolis, MN (May 23-25).
    Dublin, OH: National Ground Water Association (in
    press)
                                                 38

-------
                                              Chapter 3
                                        Induced Fractures
More than  50 years ago,  petroleum engineers recog-
nized that induced fractures could increase the produc-
tivity of oil wells. Within the past 5 years, EPA research
has shown  that similar techniques can improve the per-
formance of environmental wells (1, 2). This chapter
focuses on the methods  currently  used for environ-
mental applications that  primarily involve  injection of
fluids to drive fracture.

Several fluid  injection methods induce fractures using
fluid pressure to  dilate a wellbore  and open  nearby
cracks. Once fluid pressure exceeds a critical value, a
fracture begins to  propagate and continues to grow until
injection ceases, the fracture intersects a barrier (or the
ground surface), or  the injected fluid leaks  out through
the fracture walls. After injection is complete, fractures
are held open either by asperities on the fracture walls
(naturally propped fractures) or by permeable material,
or proppant, injected during propagation. The result is a
layer designed to be more permeable than the adjacent
formation.

The  maximum dimension of fractures created by inject-
ing fluids is limited by either a tendency for the fracture
to climb and intersect the ground surface or by the loss
of fluid through the fracture walls.  Therefore, all  else
being equal,  the  maximum  dimension  increases  with
increasing depth and decreasing permeability of forma-
tion. At a depth range of 1.5 to 5 meters (4.9 to 16.4 feet)
in overconsolidated  silty clay, the typical maximum di-
mension of a fracture is approximately three times its
depth.

The  primary application of induced  fractures is to in-
crease the  discharge of and the area affected by a well.
Field demonstrations  conducted in  rock or silty  clay
have shown that, compared with a conventional well,
induced fractures  can increase well discharge 10 to 50
times and increase  the distance for detecting pressure
effects 10 times. According to theoretical modeling stud-
ies,  these  results  are consistent with  the expected
change in performance when adding a permeable layer
to the vicinity of a well.

Even though the improvement in performance is strik-
ing,  induced fractures do not solve  all the problems of
remediation in tight formations. The relative  increase in
performance is greatest in the tightest formations, where
the performance of conventional wells is poorest. Ac-
cordingly,  it may be possible to  improve the rate  of
recovery of a contaminant by  an order of magnitude  or
more and  still have a rate that is less than desired for
timely  closure.  In  tight  formations,  induced fractures
may improve the performance of hydraulic control and
containment at the site,  or they can be combined with
other enhancement technologies, such as hot air injec-
tion (3), to accelerate recovery.

A secondary  application of fracturing techniques is  to
deliver solid compounds to the subsurface.  This appli-
cation  fills fractures with granular compounds that im-
prove  the  remedial process in  various ways. For
example, injecting solid nutrients or slowly dissolving
oxygen sources  can improve  bioremediation, or inject-
ing electrically conductive compounds (e.g., graphite)
can improve electrokinetics.

Hydraulic (injecting air) and pneumatic (injecting liquid)
fracturing both are recognized  methods of inducing frac-
tures for environmental applications. Both the mecha-
nisms  and the  results  of the  methods share some
striking similarities (3, 4),  yet they each have distinc-
tions. Capabilities are evolving so rapidly, however, that
a direct comparison of hydraulic and pneumatic fractur-
ing methods is unnecessary. For this reason, this chap-
ter focuses on the characteristics and factors affecting
performance of induced fractures, and avoids direct ref-
erence to hydraulic or pneumatic fracturing except in the
description of techniques.

Most induced fractures that have been used for environ-
mental applications to date are shaped like  gently dip-
ping disks, or in some cases they  are  slightly bowl
shaped. Vertical  fractures have been used infrequently.
Therefore, this chapter focuses on gently dipping frac-
tures. Applications  of induced vertical fractures  hold
promise, however, particularly to improve the perform-
ance of horizontal wells.

3.1   Methods of Inducing Fractures

Methods used to induce fractures and improve the per-
formance of wells range from  injecting fluid to detonat-
ing explosives. Many of the environmental applications
                                                   39

-------
involve fluid injection, although a few cases have involved
use of explosives to enhance permeability of bedrock and
improve contaminant recovery (5,6). The following section
focuses on methods that use fluid injection.

3.1.1   General Considerations
The fundamental process of inducing fractures by inject-
ing  a  fluid is straightforward: the fluid is injected into a
borehole until the pressure exceeds some critical value
and a fracture nucleates. This method can create frac-
tures  in most naturally occurring materials, from rock to
unlithified sediment  or soil. Once the fracture nucleates,
fluid continues to  be injected, driving  propagation away
from the borehole.

3.1.1.1   Injection Fluids

Three fluids have  been used to create fractures for most
environmental  applications:  gas, water, and  gelled
water. The major issues affecting choice of injection fluid
are:
• The equipment required  for injection.
• Safety concerns.
• The potential to mobilize contaminants or other injec-
  tion restrictions.
• The ability to transport solid grains into the fracture.
Air is the primary gas used to create fractures, although
other  gases may  have applications in specialized con-
ditions, such as when anaerobic conditions are required.
Air injection requires relatively simple equipment. Large
injection rates, however, demand special safety precau-
tions.  There is relatively little possibility of mobilizing
liquid  phases, although  there is a strong  possibility of
mobilizing vapor phases. Local regulations may govern
injection of air into the subsurface.
Fine-grained particles or powders can readily be trans-
ported into fractures by injecting  air (7). The ability to
transport particles decreases with increasing grain size
and density, however,  limiting capabilities of  injecting
significant volumes  of coarse-grained sand. This is a
topic of current research, and the development of spe-
cialized equipment and proppants promises to markedly
improve the transport of proppant during air injection.
Injecting water can  create fractures in low-permeability
rock formations. Relatively simple pumps and packers
are used, although  pressures in excess of 5 MPa may
be required to  initiate the fracture. Safety precautions
relate to potentially high pressures. Injection of water is
restricted by  regulations in some locations. Injecting
water will have limited effect on mobilization of vapors,
although it may  mobilize liquids. In  most cases, the
injected water and  any  fluids mobilized as a result of
injection should be readily recovered through the result-
ing fracture.  Water can transport solid  grains  into a
fracture,  although the best results are achieved using
plastic particles that have a density similar to water (8).

Guargum gel is a viscous fluid commonly used to create
fractures. Guar gum is a food additive derived from the
guar bean. Mixed with water, guar gum forms a short-
chain polymer with the consistency of mineral oil. Adding
a crosslinker causes the polymer chains to link and form
a thick gel capable of suspending high concentrations
of coarse-grain sand. This property makes guar gum gel
ideal for  filling fractures  with solid material. An enzyme
added to the gel breaks the polymer chains, allowing
recovery of the thinned fluid from the fracture.

Fracturing with guar gum gel requires several special-
ized pieces of equipment. A mixer is required to blend
the gel,  crosslinker, and enzyme, as well  as  sand or
other solids. This method also requires a pump capable
of handling a slurry containing a high concentration of
sand grains. The safety  precautions are similar to those
for pressurized water. Injecting guar gum gel negligibly
effects mobilization of vapors, although it may slightly
mobilize  liquids after the gel breaks down. The fracture
confines  the gel during  injection, however,  and  prompt
recovery of the gel should eliminate interaction with pore
fluids. Areas that regulate subsurface injection may  re-
strict injection of guar gum  gel. Because in situ organ-
isms metabolize the organic components of guar gum
gel, its use is commonly avoided when fractures are
created to enhance discharge from a drinking water well.
The major benefit to using guar gum gel is the ability to
suspend  a high concentration of coarse-grained sand
(1.2 to 1.8 kilograms  of sand per liter of gel [10 to  15
pounds of sand per gallon of gel]) as a slurry in the gel.

3.1.1.2   Injection Pressure and  Rate

The pressure required to initiate a fracture in a borehole
depends on several factors, including confining stress,
toughness of the enveloping formation,  initial  rate of
injection, size of incipient fractures, and pores or defects
in the borehole wall. In  general, the injection pressure
increases with increasing depth,  injection  rate, and fluid
viscosity. For instance, propagating a fracture by inject-
ing liquid into soil at 75  liters (19.8 gallons) per minute
and at a  2-meter (6.6-foot) depth requires 60 to 85 kPa
(8.7 to 12.3 psi) of pressure, which increases approxi-
mately 20 kPa per meter (0.9 psi per foot)  of depth. In
contrast,  the  pressure required to  create a fracture  by
injecting air, with injection rates of 20,000 to 30,000 liters
(706 to 1,059 cubic feet) per minute, is in the range of
500 to 1,000  kPa (73 to  145 psi).

The  pressure during  propagation decreases  in most
operations, but the details  of the pressure history de-
pend  on   a variety of factors. For example,  slight  in-
creases in pressure occur when sand concentration in
slurry increases (Figure  3-1).
                                                    40

-------
  400
  ,300
  200
  100
                   Propagation Starts
Sand Arrives

 Sand Content Increases

              Pump Off
      01     234567
                         Time (min)

Figure 3-1. Injection pressure during creation of a fracture by
          injecting a sand-laden slurry.
3.1.1.3   Leakoff

Propagation could continue indefinitely if the fracture is
created in  infinitely impermeable  material,  but in real
materials several factors  limit the size  of the fracture.
Some of the injected fluid  flows out through the walls of
the fracture and into the pores of enveloping soil or rock
(Figure  3-2). Workers in  the oil industry dubbed this
process "leakoff." The rate  of leakoff increases as the
fracture grows and offers more surface area  through
which the injected fluid can flow. Other factors that affect
the leakoff rate include the relative permeability of the
fractured formation and the viscosity and pressure of the
fluid. Accordingly, the rate of fracture propagation de-
creases as the rate of leakoff increases, and propaga-
tion ceases entirely when the leakoff rate equals the rate
of injection.

In most environmental applications,  leakoff generally
controls the size of the gas-driven fractures. For exam-
ple, injecting gas at 25 to 50 cubic meters (883 to 1,766
cubic feet)  per minute into sandstone for approximately
20 seconds typically results in a fracture roughly 10 to
                         Casing
                                      Fracture
Figure 3-2.  Injected fluids leaking out of fracture during
          propagation.
20 meters (32.8 to 65.6 feet) in maximum dimension. A
longer injection period negligibly affects fracture dimen-
sion (3). Thus, the rate of injection  is a critical design
variable  affecting fracture size during air injection (9).
The fast rates of leakoff during air  injection also may
yield the secondary benefit of dilating fractures or pores
adjacent to the main fracture.
Leakoff tends to be more significant when creating frac-
tures by injecting  slurries into sand  or gravel. This is
because in  other situations  (e.g., when creating frac-
tures at shallow depths in silty clay by injecting water or
viscous gel), other effects (e.g., intersecting the ground
surface)  become important before the fractures  become
large enough to be affected by leakoff.

3.1.1.4  Other Fracturing Methods

Although effective, injection of fluid is by no means the
only method of inducing useful fractures  in the vicinity
of wells.  Propagating at high rates can create  multiple
fractures in the vicinity of a bore. This process involves
detonating explosives (10)  or igniting  rapidly  burning
propellants to drive fractures at high rates (11,12). Early
practitioners of this technique lowered glass bottles filled
with nitroglycerin into wells and detonated the explosive
with a sudden shock. This procedure probably helped to
clear the zone adjacent to  the well that  was  plugged
during drilling. A variety of  modern  propellants,  which
rapidly produce gas without an explosive shock, have
been used to fracture wells. In some cases propellants
lowered  into a well perforate the wall of a bore, whereas
in other  cases the propellant rapidly drives a gas-filled
fracture into the adjacent formation (11).
Another  method involves applying an electric field to the
vicinity of a  bore to induce fractures. This method has
been used to a limited extent to affect  the  productivity
of wells  (13). Myriad microfractures  in  low permeabil-
ity  formations accompany  the application of  radio-
frequency heating, presumably resulting from the rapid
boiling of pore water.

3.1.2 Monitoring Fracture Location
The most widely used method of monitoring  fracture
location  is measuring  the displacement of the ground
surface (3, 4, 9). The displacement over a gently dipping
fracture  at shallow depths appears  as  an asymmetric
dome (Figure 3-3).  Net displacements can be  deter-
mined by surveying a field of staffs with finely graduated
scales before  and  after fracturing. Alternatively, tilt-
meters can  measure  extremely  gentle slopes of the
ground surface  in  real time while  the fracture is being
created.  The former method is inexpensive and provides
reliable data on the final displacements, whereas the
latter method supplies information on the  growth  of the
fracture,  which may be necessary when creating frac-
tures in the vicinity of sensitive structures.
                                                    41

-------
       Uplift Measured
                              1 mm
    5 meters
Figure 3-3.  Typical pattern of uplift over a shallow, gently dip-
          ping hydraulic or pneumatic fracture.

The pattern of uplift indicates characteristics of the frac-
ture at depth. A broad, symmetric dome indicates that
the fracture gently dips and is roughly symmetric. Many
fractures propagate in a preferred direction, which is
reflected by  displacements that  are asymmetric  with
respect to the borehole. At shallow depths (where the
ratio of fracture length to depth is roughly 3), the mag-
nitude of uplift appears roughly equivalent to the aper-
ture  of the  fracture. Thus, in these cases, fracture
aperture and extent can be estimated directly from the
uplift. Lesser ratios of length to depth (deeper fractures)
generally require mathematical inversion of appropriate
analyses (14-16) to estimate geometry of the fracture.

Borehole extensiometers fitted to open  bores also can
detect the location of induced fractures. This application
directly detects  the  opening of the  fracture  tip as an
increase in the length of the bore. This provides a direct
determination of the aperture of the fracture that is more
reliable than measurements of uplift.  In one application,
borehole extensiometers placed in the vicinity of a re-
taining wall ensured that induced fractures were termi-
nated before they reached the wall (17).

Monitoring induced fractures through their displacement
field is not without its limitations. Accurately determining
the dip and  aperture of deep fractures is difficult, even
when using numerical inversion methods (14,18). More-
over, details such as fracture bifurcation or intersection
of adjacent fractures generally cannot be  detected.
Heterogeneities  in the subsurface may produce results
that are difficult to interpret.

3.1.3   Equipment

The equipment used to induce fractures consists of an
aboveground system, which must be capable of inject-
ing the desired fluid  at the required pressures and rates,
and a belowground system, which must be capable of
isolating the zone where injection will take place.

3.1.3.1   Aboveground Equipment

The type of fluid to be  injected largely  determines the
aboveground equipment. Pneumatic fracturing, which
entails injecting  air to create fractures,  requires equip-
ment  that rapidly  delivers air to the subsurface.  The
most versatile equipment developed for pneumatic frac-
turing employs  a series of high-pressure gas cylinders
with a pressure regulator to control injection (Figure 3-4)
(9). This equipment injects air at rates of 25 to 50 cubic
meters  (883 to  1,766 cubic feet) per  minute  and at
pressures of 0.5 to 2 MPa (72.5 to 290 psi). The process
can be tailored to site  conditions  and is  particularly
suited to delivering air at high rates. Moreover, this  method
can create fractures with compressed gases other than
air, which may have applications during bioremediation,
in situ oxidization,  or other remedial processes.
                 Pressure Regulator
Figure 3-4.
                                    To the Subsurface
Aboveground equipment used for pneumatic
fracturing (9).
Injecting air directly from  a compressor may induce
fractures under some circumstances. Filters or special-
ized compressors are required to eliminate traces of oil
in the air stream. Typically, compressors are unable to
supply the  pressure or the  rate available from pressur-
ized cylinders; therefore, this approach may be limited
in relatively permeable formations where leakoff limits
the size of fractures that  are  created  by injecting at
modest rates.

Injecting liquid to induce  fractures, hydraulic fracturing,
requires equipment to prepare and inject the liquid. Hy-
draulic fractures created by injecting water alone require
equipment  that  consists primarily  of a  high-pressure
positive displacement pump with associated  pressure
relief devices (19). Hydraulic fractures that are filled with
sand or other granular proppant require a mixer to cre-
ate the slurry. A batch mixer consisting  of one or two
open tanks fitted with agitators can be used to create
                                                    42

-------
the slurry, a labor-intensive approach that can limit field
productivity when desired fracture volumes are larger
than the tanks. An  alternative is  to use  a continuous
mixer, which blends  metered streams of gel, crosslinker,
breaker,  and  sand to form slurry. This type of device
represents a larger capital  investment  than a batch
mixer, although it reduces the time, labor, and thus the
cost required to create fractures. In most cases, positive
displacement pumps  inject slurry to create hydraulic
fractures. Duplex and triplex piston pumps, as well as
progressive cavity  pumps, are  widely  used  for this
purpose.

3.1.3.2   Belowground Equipment

Belowground equipment for inducing fractures consists
of a device for isolating  the zone of injection. Both
pneumatic and hydraulic fracturing use straddle packers
in open  holes (Figure  3-5) (19). A specialized nozzle
assembly (3, 20) fits  between straddle  packers and
improves gas delivery during pneumatic fracturing. This
technique allows fractures to be spaced  approximately
every 0.5 meters  (1.6 feet) along an open borehole. In
some cases,  the  seal that straddle  packers provide in
open holes in saturated silts and clays may be insufficient.

When fractures are created  in  unlithified sediments,
driven casing (4) offers an alternative to straddle pack-
ers. One example of this approach is to drive a casing
Figure 3-5.  Fracture created by isolating a zone with a straddle
          packer.
with an inner-pointed rod to depth  (Figure 3-6). After
removing the rod, a high-pressure pump injects a water
jet to cut a notch in the sediments at the bottom of the
borehole.  The notch reduces the pressure required to
start propagation, much as the notch in plastic packag-
ing reduces the effort required to open the package. The
notch also ensures that the fracture starts in a horizontal
plane at the  bottom of the casing.  A fracture can be
created at the bottom of the casing by injecting either
gas, liquid, or slurry. After creating the fracture, the rod
can be reinserted and driven to greater depth to create
another fracture, or the  casing can  be left in place to
access the fracture during recovery.

Using this approach in unlithified sediments allows cas-
ing to be advanced by either hammering (using a drop-
weight, pneumatic, or hydraulic hammer) or direct push
(using the weight of a drill rig or cone penetrometer).
Packers or related methods are appropriate in rock and
in some unlithified sediments.

3.1.4   Well Completion

The method  used to complete  a well that has been
hydraulically fractured affects the versatility and cost of
the well. Completions that allow access to individual
fractures provide the most versatility, but they can con-
sume more time and money than completions that offer
access to all fractures by one casing. Access to individ-
ual fractures can improve performance during vapor
extraction. For example, this method can allow alternat-
ing between air inlet and  suction on  adjacent fractures,
or can provide dewatering capabilities from lower frac-
tures and vapor recovery from upper fractures. It also
can improve recovery of NAPL by directing aqueous and
nonaqueous phases to separate  pumps. Other applica-
tions, such as water recovery  from a confined aquifer,
may benefit little from access to each fracture.

Completion techniques that access  all fractures simul-
taneously (Figure 3-7a) resemble standard well comple-
tion methods (10).  To  provide  individual access, a
grouted zone along the  bore can isolate each fracture
(Figure 3-7b). Alternatively, casing driven to create frac-
tures in unlithified sediments  can be  left  in place to
access the fracture during recovery (Figure 3-7c).

3.2  Design Considerations

Several factors affect the form and  permeability of in-
duced  fractures. These  factors should be evaluated
when considering using induced  fractures to increase
the performance of wells at a site. Geologic conditions
play a major role in affecting fracture form and dictate
the need  for using a sand  proppant. Site conditions,
particularly structures that may interfere with propaga-
tion, also are important considerations. Table 3-1 sum-
marizes the  important factors and  their favorable or
                                                   43

-------
                                                                                          Rod
                                                                                            Pressure From
                                                                                            Soil Seals
                                                                                            Casing
Figure. 3-6.  Fractures created at the bottom of driven casing.
                                                                  Table 3-1.   Design Considerations for Induced Fractures

                                                                  Factor          Favorable              Unfavorable
                                                                  Fracture form     Gently dipping for
                                                                                   vertical wells
                                                                                  Moderate to low
                                                                                  (k < 10'8 cm2)
Formation
permeability


Fracture
permeability
                                                                  Formation type   Rock or fine-grained
                                                                                   sediment
                                        Vertical (fracture may
                                        reach ground surface)

                                        Unnecessary in high
                                        permeability formations
                                        (clean  sand)

                 \ ,000 times formation k  <100 times formation k
                                        Coarse-grained
                                        sediment
                                                                  Sand proppant   Unlithified, saturated     May be unnecessary
                                                                                                          in rock formations
Figure 3-7.  Methods  of completing wells with  induced  frac-
            tures: a) screen across all fractures b) casing to each
            fracture, and c) driven casing to each  fracture.
                sediments

No proppant     Rock formations


State of stress   Horizontal stress >
                vertical stress
                (overconsolidated)
                                        Soft formations where
                                        fracture may close

                                        Horizontal stress <
                                        vertical stress
                                        (normally consolidated)
                                                                  Well completion  Access to each fracture  Screen to several
                                                                                  most versatile           fractures less versatile
                                                                                                          but less costly than
                                                                                                          individual access
                                                                  Site conditions   Open ground over
                                                                                  fracture
                                        Structures sensitive to
                                        displacement over
                                        fracture

-------
unfavorable results. These factors  specifically target
relatively shallow applications typical of many contami-
nated  sites. The table is  based  on current  findings
and  is subject to change as fracturing techniques are
modified.

3.2.1   Flow to a Gently Dipping Fracture
To better understand the impact of some of the factors
listed in Table 3-1, consider the effects of  creating a
fracture in the vicinity of a well. One perspective ignores
the geometry of the fracture and views the improvement
in well performance as an increase in the effective per-
meability of the material enveloping the well (21). From
another view, the fracture is a discrete layer that affects
the pattern of flow in  the subsurface. The former pro-
vides a simple  method of assessing the improvement
resulting from the fractures, whereas the latter provides
more detailed insight into effects in the subsurface.
For example, consider the effect of a sand-filled fracture
on subsurface air flow during a field test (22). The frac-
ture  in this case was  1.5 meters (4.9 feet)  deep and
3 meters (9.8 feet) in radius and had an average thick-
ness of 6  millimeters  (0.2 inches).  The fracture was
accessed  via a well 5  centimeters (2 inches) in radius
and screened from 1.4 to 1.7 meters (4.6 to 5.6 feet).  An
identical well with no fracture was used for control. Silty
clay, of permeability   10"9  square centimeters (10"12
square feet), underlays the site. The surface of the site
was  exposed to the atmosphere (no cap), and an imper-
meable boundary was assumed at 7 meters (23 feet)
(bedrock was at this depth). Air was pulled from the wells
using a suction  head  of 2.5 meters (8.3 feet) of water
(absolute  pressure head of 7.9  meters  [25.75 feet] of
water). To evaluate subsurface flow, field observations
allowed calibration  of a numerical analysis of these
conditions.
Assuming the conditions cited  above, the discharges
from a  vapor  well were calculated  for fractures with
permeabilities ranging from 10"9 to 10"3 square centime-
ters  (10"12 to 10"6 square feet). The results (Figure 3-8)
indicate that discharge increases only slightly if the frac-
ture  permeability is less than 10"7 square centimeters
(10"1° square feet), or not even 100 times greater than
the formation permeability. The discharge increases sig-
nificantly, however, if the fracture permeability is greater
than 1,000 times that of the formation. Using Figure 3-8,
field  measurements of well discharge, and other analy-
ses,  the permeability of the fracture  used for this test
was  estimated at 5 x 10"6 square centimeters (5 x 10"9
square feet), approximately the permeability  of coarse-
grained sand injected  into the fracture.
The fracture influences each of the essential measures
of subsurface flow—pressure,  flux,  and travel times.
Pressure contours form concentric shells around the
conventional well screen, whereas they elongate around
                                                          12
      -10-9
                   10-7
10-s
10-5   1CH
10-3
                 Fracture Permeability (cm2)
Figure 3-8. Air discharge from a well intersecting a flat-lying
          circular fracture as a function of the permeability of
          the fracture. (Formation permeability is 10~9 cm .)

the fracture. The pattern of flow paths also elongates by
the fracture  (Figure 3-9). The suction felt by the forma-
tion less than 0.5 meters (1.6 feet) from the control well
is equal to the suction in a broad area 3 meters (9.8 feet)
from the well, and more than 1 meter (3.3 feet)  above
and 2 meters (6.6 feet) below the fracture. Remember
that equal suction was applied to both wells; thus, most
of the applied suction was lost within a few tenths of a
meter of the conventional well, whereas significant suc-
tion occurred several meters  from  the well along the
fracture.

The most easily measured formation parameter in the
field is  pressure, or suction. Flux (volumetric flow/unit
area) and travel time, however, are more important than
pressure for environmental applications. The flux pat-
tern resembles  the  pressure  pattern in that it  forms
concentric shells around the conventional well and elon-
gated shapes around the fracture. Fluxes in the vicinity
of the fracture are at least two orders of magnitude more
than fluxes roughly a half meter from the conventional
well. The flux 2 meters (6.6 feet) from the conventional
well is roughly the same as the flux 6 meters (19.7 feet)
from the fractured well.

Travel times, which were determined by tracking  a par-
ticle from its starting location to the  well,  are shorter in
the vicinity of the fracture. Travel  time  is particularly
short in  the region over the fracture, and the travel time
roughly  5 meters (16 feet) from the  well with a fracture
is less  than travel time 2 meters  (6.5  feet)  from the
conventional well. This is significant because some es-
timates  of remediation are  based on  the number of pore
volumes, and the travel time is a measure of the time
required to exchange one  pore volume within that con-
tour. Note that travel times in Figure 3-9 are given in
units of  time/effective porosity, so that the actual time of
travel is obtained by multiplying the  number on the plot
                                                   45

-------
                          Conventional Well
                     6 m
Well Intersecting Sand-Filled Fracture
   6m       4        2         0
                         0.002
                     0.001
                                                        111   \
                                                        Travel Time: Days/Effective Porosity^
                                                        \ ,\  \    \    V   .
Figure 3-9.  Pressure, flux, and travel time to a conventional well and a well intersecting a sand-filled fracture. Shaded areas are for
           comparison (based on the results of a numerical simulation using field data [22]).
by the effective porosity of the site.  The  results are
presented this way because estimates of effective po-
rosity were unavailable from this site.  Estimates of ef-
fective porosity are 0.001 or less (23) at a similar site
underlain by fractured silty clay till, so  that air in the
pores that are available  for flow, shown in the shaded
areas in  Figure 3-9, would be exchanged in 2.4 hours.

The essential conclusions of this study are:

• The fracture  increased the discharge by a factor of
  20 and increased the radial distance where pressure
  is affected by a factor of 10 compared with the control
  well.

• The fracture changed the flow paths and patterns of
  pressure, flux, and travel time in the subsurface. Val-
  ues of those parameters in the vicinity of the fracture
  were typically more than 10 times greater than in the
  vicinity of the control well.

• A theoretical model that treated the fracture as a thin
  layer of coarse-grained sand aided in predicting field
  observations.

• The fracture permeability was critical to well perform-
  ance. Well discharge was nearly unaffected when the
  fracture permeability was  less than 100 times that of
  the formation, whereas it increased abruptly as the
   fracture permeability increased from 100 to  10,000
   times that of the formation.

 3.2.2   Forms of Fractures

 The  effectiveness of an induced fracture depends pri-
 marily  on its form, that is, its shape,  aperture, orienta-
 tion,  length, width, and location with  respect to the
 borehole. In most cases, fractures created by injecting
 fluids consist of one to several fracture surfaces. The
 general forms of fractures range from a steeply dipping,
 elongated feature to a flat-lying  circular disk or bowl-
 shaped feature. The  flat-lying fractures are useful to
 many applications because they can grow to significant
 sizes without  intersecting the ground surface. Con-
 versely, steeply dipping fractures tend to climb upward
 and intersect the ground surface.

 The form of an induced fracture results from both frac-
 turing technique and site conditions. Critical factors re-
 lated to fracturing technique include type of fluid, rate or
 pressure of injection, and configuration of the borehole.
 Critical site conditions  affecting form include loading at
 the ground surface, permeability, formation heterogenei-
 ties,  and subsurface borings. Fractures created by in-
 jecting sand-laden slurry are easy to identify; their form
 is better known than the form of naturally propped frac-
 tures, which  lack a  proppant  and can be difficult to
                                                    46

-------
 identify in split-spoon samples.  Most sand-filled  frac-
 tures used for environmental  applications  have been
 created in glacial sediments or vertisols. Therefore, the
 impression of fracture form is  biased toward fractures
 created under those conditions.

 A database containing characteristics of approximately
 140 fractures was analyzed to  estimate the charac-
 teristics  of a typical fracture  created  using methods
 described in Murdoch et al. (4). According to the analy-
 sis, the "typical" fracture induced by injecting  liquid at
 shallow depths in  overconsolidated silty clay is slightly
 elongated in plan  and  dips gently toward  its parent
 borehole. The major axis of the fracture is approximately
 three times greater than the depth of initiation and 1.2
 times greater than the minor axis (Figure 3-10).
  $ Average Extent of Uplift   , •• ,  5 ,    Maximum '
                                 Uplift  ' "
Plan
             Injection Point
                                            Section
Figure 3-10. Plan and section of a typical hydraulic fracture
           created in overconsolidated silty clay.
Ground overlying the fracture displaces upward to cre-
ate a gentle dome, and the  amplitude of the dome
resembles the  aperture  of the fracture at depth.  The
fracture closes and the dome subsides after injection,
but the  injected sand prevents the fracture walls from
closing completely. The amount of closure depends on
the concentration of sand in the slurry. Here, the ratio of
maximum aperture when the fracture is pressurized to
thickness of sand is similar to the ratio of the total slurry
volume  to the  bulk volume  of  sand in the slurry.  The
major axis of the extent  of uplift ranges from 5 meters
(16.4 feet) to more than  12 meters (39.4 feet), with an
average of 8.5  meters  (27.9 feet).  The maximum uplift
ranges from a  few millimeters to  more than 30 milli-
 meters (1.2 inches), with an average of 19 millimeters
 (0.75 inches).

 The typical extent of uplift is  roughly elliptical, with  an
 aspect ratio of 1.2:1. The borehole  used to create the
 fracture and the point of maximum uplift rarely coincide
 with the center of the extent of uplift. Interestingly, both
 the borehole eccentricity (ratio of distance between the
 center of uplift and the borehole to the  major axis) and
 the displacement eccentricity (ratio of distance between
 the center of uplift and the point of maximum uplift to the
 major axis) are 0.14. Those two points, however, typically
 are on opposite sides of the center of uplift (Figure 3-10).

 The average dips of fractures  at seven different sites in
 the Midwest and Gulf Coast  range from 5 to 25 degrees.
 At each site,  however, the dips were fairly consistent,
 with the standard deviations  of  dips approximately 5
 degrees. The dips at some sites were statistically differ-
 ent from dips at other sites, so that dip angle appears to
 depend strongly on site conditions.

 3.2.3   Geologic Conditions

 Geologic  conditions significantly affect the forms of  in-
 duced fractures. Details of all  the  effects are still being
 evaluated, but a discussion  of some of the major geo-
 logic factors follows.

 3.2.3.1   Permeability

 A fracture must be significantly more permeable than the
 enveloping formation to  have a  major  impact on well
 discharge (Figure 3-8). Therefore, the relative  improve-
 ment resulting from induced fractures increases as the
 permeability of the formation decreases. In most cases,
 rock or formations of silt or  clay are  best suited  to
 induced fracturing because  they have the lowest initial
 permeabilities.

 One exception involves using  induced fractures to de-
 liver solid compounds to the subsurface. This applica-
 tion  can address processes independent of formation
 permeability  and may require creating fractures and
 filling them with permeable sand,  gravel, or rock.

 3.2.3.2   State of Stress

 The  state  of stress in the formation affects the orienta-
 tion of an induced fracture once it has propagated away
 from the borehole. Induced fractures are usually flat-
 lying where horizontal formation  stresses are greater
than vertical stresses, whereas they  tend to be steeply
 dipping where vertical stresses are greatest.  In rock
 formations that erosion has  buried or exposed, the lat-
 eral  stresses  are typically  greater  than the  vertical
stresses near the ground surface,  effecting flat-lying
 induced fractures. Most shallow  rock formations, with
the possible exception of recent lava flows, have rela-
tively high lateral stresses.
                                                    47

-------
The  state of stress of soils and unlithified sediments
depends on several factors, including consolidation his-
tory (24) and wetting and drying  history. Soils that were
consolidated under a load greater than the present load
are overconsolidated, and many  such soils contain hori-
zontal stresses that exceed vertical stresses. For exam-
ple, loading by a glacier results in overconsolidation, so
sediments deposited subglacially are good candidates
for high lateral stresses.  Soils containing clay minerals
that undergo a large volume reduction upon drying be-
come overconsolidated with repeated cycles  of wetting
and drying. For instance, vertisols (soils rich in swelling
clays) are particularly susceptible to relatively large lat-
eral stresses.

Glacial sediments are common in the northern Midwest
and Canada, and vertisols are common along the Gulf
Coast of Texas (Figure 3-11). The states of stress found
in  these areas  favor the creation  of gently  dipping
fractures.

Other conditions also result in favorable states of stress.
Local geotechnical engineers can evaluate the state of
stress at a  particular site.

3.2.3.3   Bedding

Induced  fractures  may follow contacts in interbedded
sediments, or they may follow  partings  between rock
beds. The effect of bedding  can be  capricious,  with
fractures following beds in some cases and crosscutting
beds in  others. In some cases, it appears that flat-lying
fractures are created in interbedded sediments with a
state of stress that would  favor vertical fractures.  Gen-
erally, field tests need to establish the effects of bedding
at a particular site.

3.2.3.4   Formation Strength

The strength of the formation plays an important role in
determining whether fractures  can be naturally propped
or if they should be propped with sand. The section on
fracture permeability includes a more  detailed discus-
sion of this issue.

3.2.3.5   Water Content

Water content of a formation appears to have negligible
effect on creating fractures by injecting fluid.

3.2.4   Fracture Permeability

Because the permeability  of the induced fracture is half
the ratio between fracture and formation permeability, it
critically affects the performance of wells. To compare
the permeabilities of real fractures induced around  bore-
holes, it is helpful to express their performance in terms
of a smooth-walled slot with an aperture we that has the
Figure 3-11.  Locations of major areas of soils related to glaciers and vertisols (25, 26).
                                                    48

-------
same effective permeability as the real fracture. Several
methods  relate effective  permeability ke and  effective
aperture (27), but one of the most commonly used (28)
is
                                                (3-1)

The aperture of a real fracture varies greatly along its
length. Thus, the effective aperture is an average that
accounts for how some locations with wide gaps and
other locations that may be closed completely affect flow
through the fracture. Accordingly, only hydraulic ortracer
tests in the field can measure the effective permeability
of natural fractures. The effective apertures of induced
fractures adjacent to boreholes are incompletely known
but have been estimated for several natural fractures. In
one example, several flat-lying fractures in dolomite had
an  effective  aperture of 0.022  to 0.023  centimeters,
(0.0087 to 0.0091 inches), according to constant head
tests (29).

The effectiveness of a fracture propped open with sand
is expressed in terms of the product of proppant perme-
ability kp and aperture w. Using this basis, it follows that
the equivalent aperture of a proppant-filled fracture is
           0.06  1
wp =
             !>1/3
According to field tests and tabulated values (30), the
permeability of medium- to coarse-grained sand used as
a proppant is approximately 10"5 to 10"6 square centime-
ters (1.6 x 10'6 to 1.6 x 10'7 square inches). The actual
aperture of  sand-filled fractures typically ranges 0.5 to
1.0 centimeters (0.2 to 0.4 inches), so that the equiva-
lent aperture is 0.02 to 0.05 centimeters (0.008 to 0.02
inches) (Figure  3-12). This overlaps with  the effective
apertures observed for fractures in rock.

The discussion above suggests that the effective aper-
ture of naturally propped fractures may be similar to that
of sand-filled  fractures in  rock. In soils, however, the
strength of  fracture asperities are less than those  in
rock, so naturally propped fractures may close. The rate
of closure increases with decreasing strength of the soil
or increasing driving stress on the fracture. In general,
the strength of fine-grained soil decreases with increas-
ing water content  or  decreasing  consolidation. Thus,
fractures may stay open in dry soils but may close when
the soil becomes saturated.  The stress  driving closure
is the stress the formation applies (for a horizontal frac-
ture, the unit weight of the formation times the depth)
plus the amount of suction applied to the  fracture. Ac-
cordingly, fractures probably can be  naturally propped
when soil or rock is strong relative to  the closure stress
(9).  If strength decreases, depth increases,  or suction
increases  past a  critical  value,  however,  fractures
should be propped  with granular materials. Figure 3-13
schematically depicts  this concept. Data are currently
           0.00
                       Aperture Proppant-Filled Slot (cm)
        Figure 3-12.  Effective aperture of an open slot that is equiva-
                    lent to that of a fracture filled with proppant.

        unavailable to determine the critical values,  so the axes
        of Figure 3-13  are qualitative.
(3-2)    3.2.5  Fracture Size
        Fracture size is an important design consideration be-
        cause performance generally increases with increasing
        size.  The  rate and  volume of injected fluids are the
        primary variables affecting size. Where significant leak-
        off may occur,  such as  when injecting air to create
        fractures, increasing the rate of injection increases the
        size of the fracture. In other cases, the volume of in-
        jected fluid determines the size of the resulting fracture.
        The major exception  is  when  a fracture  climbs  and

                                         Area Where Naturally
                                         Propped Fractures
                                         Stay Open /
              A
            Soil/Rock
            Strength
          Area Where
          Fractures Should
          Be Filled With
          Sand To Maintain
          Permeability  x
                                       Asperities Collapse
                                       Proppant Needed
                                Depth + Suction •
        Figure 3-13.  Factors that affect how fractures should be propped.
                                                     49

-------
reaches the ground surface, in which case it may be
smaller than anticipated.

The thickness of sand in a fracture can be manipulated
to some extent by controlling the concentration of sand
in the injected  slurry  or  by changing other  process
variables. Sand in the fracture should be thick enough
to provide a large contrast with the permeability  of the
formation. Once the contrast is sufficient,  however, cre-
ating  a thicker sand pack to obtain additional  contrast
provides only minor improvement. Once sand in  the
fracture is several millimeters thick, a decision  must be
made as to whether the cost of additional sand is worth
the incremental benefit achieved by a thicker fracture.

Typical applications (in the depth range of 2 to 5 meters
[6.6 to 16.4 feet]) have created fractures that are 6 to 10
meters (19.7 to 32.8 feet) in maximum dimension and 1
to 2 centimeters (0.4 to 0.8  inches) in maximum thick-
ness. The size of the fracture increases with depth, and
ratios of maximum  length to depth of 3:1  to 4:1  are
typical.

3.2.6   Well Completion

The type of well completion affects the flexibility of sub-
surface control,  versatility in creating additional frac-
tures, and  cost.  Some methods of completion provide
access to each fracture or group of fractures, whereas
others simultaneously access all the fractures in a well.
Individual completions provide versatility by  allowing
each fracture to be used for either injection or recovery
of fluids. This is particularly beneficial during vapor ex-
traction or  the simultaneous recovery of two  separate
phases, particularly NAPL and water. Individual comple-
tions  are also more costly than continuous screening
across all fractures.

In some cases, it may be necessary to  return to a well
and create additional fractures. Additional fractures can
be created in wells that consist of  open bores, and
fracture size can be increased where completions con-
sist of driven casing. Using currently available methods,
it is difficult to create fractures using wells that have
already been completed with a screen and gravel pack.

3.2.7 Site Conditions

Both surface and subsurface structures  (e.g., buildings,
pavement,  buried utilities, wells, piezometers, or back-
filled  excavations)  may affect,  or be affected by,  the
creation of subsurface fractures. Creating fractures be-
neath structures may displace them.  A structural engi-
neer  should  be  involved  to estimate  displacement
tolerances for particular structures. In cases where sur-
face displacements are critical, it is advisable to use real
time  monitoring of the displacements so that the proce-
dure  can  be terminated  before structures reach dis-
placement tolerances.
Surface structures also may affect the propagation of
fractures by loading the ground surface. In one case
where injecting air created a fracture adjacent to a build-
ing, propagation was away from the building apparently
in  response to the surface  loading  by the structure.
Fractures filled with liquid behave in a similar manner.

A fracture also may displace shallow subsurface utilities,
pipe, or related features that lie above the fracture. A
propagating fracture actually may intersect deeper fea-
tures, such as wells, piezometers, or grouted sampling
holes. There is limited evidence regarding the effects of
this type of interaction.

Fracture propagation may terminate or alter markedly if
the fracture  intersects a backfilled excavation. The se-
verity of this effect depends on site details and can only
be evaluated case by case.

3.3   Applications

The typical application for induced fractures is to im-
prove the performance of wells.  Induced fractures can
improve most in situ remedial actions involving fluid flow.
Other applications include placing solid compounds in
the subsurface and enhancing electrical conductivity.
The sections that follow outline the principal applications
that have been either demonstrated or proposed.

3.3.1    Vapor Extraction

Vapor extraction  is one  of the most widespread and
effective methods of remediation, and induced fractures
can improve vapor extraction in a variety of low-perme-
ability formations.  In most cases, applications include
creation of multiple  fractures at various locations along
the length of vertical wellbores.

The  primary purpose of inducing  fractures for  vapor
extraction is to improve the discharge and areas af-
fected by wells. Typically, results increase discharge by
factors ranging from 10 to 100,  and  increase the dis-
tance affected by a well 10 times or more compared to
control wells (1, 2). In addition, coupling individual com-
pletion methods with induced fractures allows air flux in
the subsurface to concentrate in different areas.

Design variables to considerwhen inducing fractures for
vapor extraction remediation include  selection of prop-
pants and details of completion. Some applications use
fractures that  are naturally  propped, whereas  others
prop fractures with sand. Both methods can increase
vapor discharge and contaminant recovery rate  by an
order of magnitude  or more.  In general, the duration of
improvements of vapor discharges by naturally propped
fractures are greatest in competent formations, such as
sandstone  and siltstone. Naturally  propped fractures
stay open  in  competent formations, whereas in soft
sediments they  may  close due to the weight  of the
                                                   50

-------
overburden  and  applied  suction. However, naturally
propped fractures in relatively stiff sediments, such as
glacial drift,  can stay open for many months. Propping
fractures with sand offers an alternative that may increase
the duration  of the discharge in unlithified formations.

Completion  methods range from installing screen over
the entire interval of fractures to installing  individual
casings and screen to service each fracture. Screening
the entire interval containing fractures reduces comple-
tion costs, but also reduces versatility and may limit the
effectiveness of the application. The limitation results
from how suction is applied to a stack of fractures. For
instance, suction is applied  equally when a screen is
placed across all the fractures. As a result, air flux is
particularly great in the region overlying the upper frac-
ture, and it is significant at the ends of the lower frac-
tures  (Figure 3-4). Flux can be  limited, however, in the
region below the upper fracture, even close to the well.
Air from either  a neighboring  inlet well  or from the
ground surface converges toward  the ends of the lower
fractures and avoids the area close to the well (Figure
3-14).  As a  result, screening  a  well across multiple
fractures may result in incomplete remediation in the
vicinity of the well.

An approach to avoiding this problem is to use one or
more  fractures  as air inlet sources. This results in  air
        10ft
     0-0.1    0.1-0.2    0.2-0.4    0.4-0.6    <10,000
Figure 3-14.  Flux of air to three flat-lying fractures (white band)
           shown in cross section: a) suction of 100 inches
           of water on all three fractures, and b) air inlet (at-
           mosheric pressure) at the bottom and suction on
           the other two fractures.
flowing from one fracture to another and flux concentrat-
ing in the area between the fractures (Figure 3-14b). To
use this process, it must be possible to access fractures
individually. Several methods of well completion  can
accomplish individual access. One method includes bor-
ing through the interval containing fractures and placing
sand or gravel at the depth of each fracture and grout in
the intervals between fractures.

Five sites have used fractures filled with coarse-grained
sand for the purpose of vapor extraction, and  naturally
propped fractures have been created at one to several
dozen sites.

3.3.2   LNAPL Recovery

Induced  fractures  can increase the recovery of free-
phase LNAPL by increasing the discharge of recovery
wells. This  application can use either naturally propped
fractures in rock or sand-propped fractures in unlithified
sediments. LNAPL recovery from an aquifer closely re-
sembles oil recovery from a reservoir.

This application requires creating fractures in or slightly
below the contaminated zone. Recovering contaminants
using fractures significantly below the LNAPL should be
avoided because of the potential for trapping NAPL as
a residual phase during drawdown (31).

One strategy for recovering a thick layer of LNAPL is to
create one or several  fractures in  the  contaminated
interval  and to complete the well to access all the frac-
tures  simultaneously. This approach can increase the
recovery of the NAPL phase by an order of  magnitude
compared with a conventional well in a low-permeability
formation.

In many cases, however, the contaminated zone is rela-
tively thin,  and recovery of LNAPL causes the upward
displacement of the underlying  NAPL/water interface.
When the interface reaches the well, the ratio of NAPL
to water diminishes, even though a significant amount
of LNAPL may exist in the vicinity of the  well.

In addressing the problem  of water recovery from a
LNAPL well, one approach is to create and  separately
access  a fracture  below  the LNAPL/water interface.
Water is recovered from  the lower fracture,  thereby
preventing  upward migration of water. This, in turn, pre-
vents the reduction of the ratio of LNAPL to water recov-
ered from the overlying fracture. A test in swelling clays
near Beaumont, Texas, demonstrated this principle. At
this site, several sand-filled  fractures were  installed to
recover gasoline. At one location, the interface between
gasoline and water was at a depth of approximately 3.3
meters (10.8 feet). Fractures were created at depths of
3.0 and 3.6 meters (9.8 and 11.8 feet),  and individual
pumps accessed each fracture. While both pumps were
operating, the ratio of gasoline to water was  between 5
and 6 from  the upper fracture, whereas it was 0.01 from
                                                   51

-------
the lower fracture. The pump in the lower fracture was
turned off while the upper pump continued to operate.
Four hours later, the total rate from the upper fracture
had increased markedly to slightly less than the rate
produced when both pumps were operating. The rate of
recovery of gasoline from the upper fracture, however,
actually diminished, and the gasoline-to-water ratio de-
creased from 5 to 0.3. The case history section provides
more details of this project.

3.3.3   Dual-Phase Recovery

Simultaneous recovery of vapor and liquid is inevitable
when extracting vapor near saturated zones, and it may
occur when accelerating liquid recovery by placing vac-
uum on a well. This process, called dual-phase recov-
ery, uses a well with an inner tube attached to a vacuum
pump (Figure 2-1) (31, 32). The system induces vapor
flow during  normal operation but  also aspirates and
removes liquid should it enter the well (Figure 3-15).

Wells containing induced fractures  in the  vadose zone
of low-permeability formations tend to produce  more
water than conventional wells  (22). Vapor  discharge
diminishes during  water recovery,  so the dual-phase
recovery approach accelerates  dewatering  and en-
hances vapor recovery. Most applications that have ex-
tracted vapor using sand-filled fractures have used the
dual-phase recovery approach.

3.3.4   NAPL Recovery

Dual-phase recovery from wells with induced fractures
is similar to applications from conventional wells. The
                               Air Inlet as Needed
Figure 3-15.  Well configuration to recover liquid  and  vapor
           phases using suction at the ground surface.
design uses a conventional well with a tube that passes
through a seal at the well  head and  extends  to the
bottom of the screened section (31). Suction is applied
either to the inner tube or to the annulus between the
tube and the well casing. Vapor is  recovered  during
normal operation. In addition, water that enters the well
is aspirated in the vapor stream and also recovered.

3.3.5   DNAPL Recovery

Using induced fractures can increase the recovery  of
DNAPL, particularly from low  permeability formations.
However, the benefits of increased recovery must be
balanced against the possible risks. Creating vertical
fractures or intersecting natural vertical fractures may
result in downward migration of DNAPL that cannot be
captured by the well. As a result, induced fractures may
increase the recovery of some DNAPL while making the
remaining liquid more  difficult to recover because it has
migrated to greater depth.

In view of this possibility, only sites where gently dipping
fractures can  be  created should be  considered  for
DNAPL applications. The orientation of an induced frac-
ture depends on a variety of conditions,  including the
state of stress, density of injected fluid, rate of fracture
propagation, depth, and the  nature  of bedding,  fabric,
and preexisting fractures in the soil or rock. These fac-
tors make the orientation of induced fractures difficult to
predict with confidence prior to  site activity. In most
cases, and particularly at DNAPL sites, feasibility testing
is recommended to determine fracture orientation.

A pilot test  (21), described in the "Case Histories" sec-
tion, demonstrated that induced fractures increase the
recovery of free-phase DNAPL.

3.3.6   Bioremediation Applications

Using induced fractures can enhance bioremediation in
a low-permeability formation  by either of two methods:

• Increasing  the  rate of injection of nutrients  and
  oxygen-bearing fluids (33).

• Filling fractures  with solid compounds that provide
  the essential ingredients for  bioremediation (34).

Induced fractures have been used to accelerate biore-
mediation of fine-grained soils by increasing the  rate of
injection of nutrients and oxygen, typically in  the form
of hydrogen peroxide. In one  application  where frac-
tures were  created in  silty clay glacial drift, the rate  of
injection increased by nearly two  orders of magnitude
(33). In this example, one location  contained sand-filled
fractures at depths of 1.2,  1.8, 2.4, and 3 meters (3.9,
5.9, 7.9, and 9.8 feet). The fractures were gently dipping
and reached maximum dimensions of 5 to 7.5 meters
(16.4 to 24.6 feet). The well was screened from 1.2 to 3
meters (3.9 to 9.8 feet) depth, accessing all the fractures
                                                   52

-------
simultaneously. Another well, which lacked induced frac-
tures, was created as a control. A solution  of nutrients
and hydrogen peroxide was injected into both wells at
constant head. The rate of injection into the well contain-
ing induced fractures ranged from 2.5 to 4.6 liters (0.7
to 1.2  gallons) per minute, whereas the rate into the
control well ranged from 0.02 to 0.08 liters (0.005 to 0.02
gallons)  per minute.  This example demonstrated one of
the largest differences of injection rates between wells with
and without induced fractures, with a ratio between 50 and
125.

Pilot-scale field  tests have  been conducted in  which
solid compounds designed  to slowly release oxygen
were injected  into a fracture (33). A bioventing applica-
tion to promote bioremediation of solvents, in  which
sand-filled fractures  will be used to enhance air injec-
tion, is currently in progress.

3.3.7  A ir Injection

The rate  of air injection also can  be increased by induc-
ing fractures in the vicinity of an  injection bore (35, 36).
Injecting  ambient air can stimulate the activity of aerobic
organisms, or heated air can heat the  formation and
increase  the vaporization of VOCs. During one test, air
heated to between 100°C and 130°C (212°F to 266°F)
was injected for 90 hours at 70 standard cubic feet per
minute (scfm) into a fractured well. Temperatures in-
creased from 14°C (57.2°F) to between 21 °C and 25°C
(69.8°F to 77°F), according to measurements 1.5 and
3.0 meters (4.9  to 9.8 feet) from the point of injection
(36).

3.3.8  Steam Injection

Induced  fractures can enable steam  injection to heat
low-permeability formations. This application has been
evaluated during a  pilot-scale test. In addition, a pro-
gram designed  to lead to a field demonstration at  a
contaminated  site is currently underway.

The well design for steam  injection  resembles other
applications except that steel, ratherthan plastic casing,
is recommended because of steam temperature. More-
over, wells should be able both  to inject steam and to
recover condensate. One possible design  involves in-
jecting steam into the middle one of three fractures while
applying  suction to the other two (Figure  3-16). This
approach would induce upward  and downward migra-
tion of steam and a condensate front accelerated by the
applied suction.

The lower fracture plays another important role in this
approach. Active suction on the lower fracture should
intercept DNAPL mobilized in the condensate front,
thereby enhancing the reliability  of the system. Several
fractures could serve this function, depending  on  the
requirements at the site.
  Suction Out
                                                                                        Steam In
          Recovery of Possible Mobilized DNAPL
Figure 3-16.  Schematic application of steam injection into in-
           duced fractures.
A pilot-scale test of a simplified version of this approach
has been conducted. It involved injecting steam into a
well that accessed a sand-filled fracture at a 2.4-meter
(7.9-foot) depth. The test compared the rate of injection
and the temperature field with results from a control well.
An  upper and  lower fracture  were present, but they
received no suction during this preliminary test. The test
site was underlain by silty clay glacial drift with hydraulic
conductivity of 10"6 centimeters per second. Steam was
injected for 17 days at approximately 34 kPa (4.9 psi).
The rate of injection of steam into the fracture was 0.19
to  0.23  kilograms  (0.4  to  0.5  pounds)  per minute,
whereas the  rate of injection into the  control well was
less than 0.006 kilograms (0.013 pounds) per minute.
Thus, the rate  of injection into the fracture  was more
than 30  times greater than  that into the control well.
Moreover, the rate of steam injection was enough to
result in significant heating of the silty clay (Figure 3-17).
A monitoring  point 1.2 meters (3.9 feet) from the point
of injection of the fractured well showed that tempera-
tures reached 80°C (176°F)  in 4 days,  but the tempera-
ture was roughly 25°C (77°F) after 6 days of injection
into the control well. Temperature continued to increase
around the fractured well, and the temperature  profile
formed a 0.6-meter (2-foot) thick zone of 95°C (203°F)
after 17 days of injection. This was interpreted as a zone
of live steam that propagated outward into the silty clay.
(A 5°C temperature loss is associated with the method
of temperature measurement.)

3.3.9  Electrokinetics

Graphite can be used as a proppant to create fractures
that are  electrically conductive. This  application, cur-
rently being evaluated, may enhance migration of water
and contaminants  via electrokinetics. The approach is
to  create two  graphite-filled fractures,  one  over the
other and separated by several meters. To drive vertical
                                                   53

-------
   0


  -2


„ -4
±±_
n
Q. -6
  Fractured-^ 0
  Well at
  Start of  _12
  Test
        -14
                  Control Well
                 / After 6 Days
                                  Fractured Well After.
                                    Days
                Fractured Well
                at 2 Days
                              Fractured Well
                              at 4 Days
              20
                      40      60
                       Temperature (°C)
                                       80
                                               100
Figure 3-17.  Temperature as a function of depth and time.

migration  by  electrokinetics,  a voltage difference  is
maintained between the  two fractures. This process is
analogous to maintaining a pressure difference between
two fractures  to  drive flow, except that migration  by
electrokinetics can be faster than migration by hydraulic
flow in fine-grained soils.

3.3.10  In Situ Treatment Zones

Chemically  or biologically  active compounds injected
into fractures can act as broad sheets in the subsurface.
Examples of these types of compounds include fine-
grained 0-valence iron,  which degrades chlorinated
solvents on contact (37), or encapsulated sodium per-
carbonate,  which releases oxygen to stimulate aerobic
bacteria for several  months (33).  Fractures filled with
these compounds could  be designed to create  in situ
treatment zones.

One approach of this application creates flat-lying treat-
ment zones in the vadose zone through which contami-
nants  migrate as  they flow  downward  by gravity.
Alternatively, electrokinetics can induce contaminants to
migrate through the treatment zones. The development
of in situ treatment zones is in its infancy, but  it can
potentially be a low-maintenance system that offers ma-
jor cost reductions compared to current methods.

3.3.11  Barriers

For years,  induced fractures have been recognized as
a possible consequence of grouting (38,  39).  Tech-
niques that purposefully fill fractures with grout are also
available (40). Most of these grouting techniques call for
creating many cross-cutting fractures to reduce the per-
meability of a zone. In one  application, however,  grout-
filled  fractures  from  neighboring  boreholes  linked
together to form a continuous horizontal sheet (41). The
technique, termed the block displacement method (42),
probably was the first horizontal  barrier successfully
created beneath a site. However, this test involved spac-
ing the boreholes a few meters apart and creating a slot,
which was nearly  equal to the distance between the
boreholes, with a water jet. Accordingly, the fractures
propagated relatively short distances before they  inter-
sected adjacent fractures. With wider spacing between
the wells, which many practical applications would  prob-
ably require, it would be difficult to ensure that the
fractures form a continuous sheet.

The use of grout-filled fractures as barriers to flow has
received limited attention since the initial demonstration.
Bifurcation of a fracture into several lobes  commonly
occurs.  Moreover,  current methods cannot detect this
phenomenon. Accordingly, ensuring that multiple frac-
tures connect to serve ultimately as a flow barrier would
be difficult.

3.3.12   Monitoring

Conventional wells in low-permeability formations are
associated with slow rates of sample recovery. There-
fore, monitoring the chemical composition of fluids in
these formations can be difficult.  Inducing fractures in
the vicinity of a monitoring point would increase the rate
of sample recovery as well as broaden the area repre-
sented by the sample. This application is currently under
investigation, although results were unavailable as of
this writing.


3.4   Case Histories

Induced fractures have  been  used to improve environ-
mental applications since the late 1980s. Most of the
applications to  date have been  pilot-scale tests that
evaluate the technology and provide data for progress-
ing to full-scale. The petroleum industry, however, has
used the general technique for more than 50 years, and
it is a common  means for improving the discharges of
water wells, particularly in granitic or metamorphic ter-
rains. All the applications strive to improve the perform-
ance of wells, and the petroleum and water applications
provide important insights into  possible environmental
applications.

3.4.1   Overview

Records of oil production establish  the benefits of hy-
draulic fracturing;  ratios of production rates before and
after fracturing suitably measure those benefits. Accord-
ing to data collected from several dozen oil wells (43),
recovery ratios (production rate  after  fracture  :  initial
rate) range from 1.4 to 100 or more.  In general, the ratio
ranges from  1.5 to  roughly 10 for wells that were pro-
ducing before fracturing (Table 3-2). The ratios are large,
however,  for wells  that showed  negligible production
prior to fracturing.
                                                    54

-------
Table 3-2. Summary of the Effects of Induced Fractures Used To Improve the Recovery of Wells
Location
Purpose
Injection
Fluid
Fracs/
Well
Formation Type
Performance
Without Fractures3
Performance With
Fractures
Ratio
Environmental Applications
Dayton, OH (33)
Beaumont, TX
Beaumont, TX

Cincinnati, OH
(22)
Chicago, IL (1)
Cincinnati, OH
Frelinghuysen,
NJ (9)
Newark, NJ (9)
Bristol, TN (21)
Bioremediation
Liquid recovery
LNAPL
recovery
Air recovery
Vapor
extraction
Steam injection
Air recovery
Air recovery
DNAPL+
water
Gel + sand
Gel + sand
Gel + sand

Gel + sand
Gel + sand
Gel + sand
Air
Air
Water
4
2
2

1
3
1
2
2
2
Silty clay glacial
Swelling clay
Swelling clay

Silty clay glacial
Silty clay glacial
Silty clay glacial
Silty clay glacial
Sandstone
Sedimentary rock
0.14-0.57 L/min m
0.0027 L/min m
0.0024 L/min m

3.34 L/min mH20
4.4 L/min mH20
<0.0019 kg/min mH20
6.6-11 L/min mH20
27-113 L/min mH20
0.32-0.98 L/min m
18-33 L/min m
0.13 L/min m
0.046 L/min m

67 L/min mH20
56-136 L/min mH20
0.055-0.066 kg/min mH20
87-204 L/min mH20
397-580 L/min mH20
1 .6-2.7 L/min m
50-100
50
19

20
22
30+
13-19
5-14
2.8-6.2
Petroleum Applications
California (43)
California (43)
Alaska
Texas (43)
Texas (43)
Texas (43)
West Virginia (43)
Oil recovery
Oil recovery
Oil recovery
Oil recovery
Oil recovery
Natural gas
Natural gas
Gel
Gel
Gel
Gel
Gel
Gel
Gel
1
1
1
1
1
1
1
Reservoir rock
Reservoir rock
Reservoir rock
Reservoir rock
Reservoir rock
Reservoir rock
Reservoir rock
20 bopd
10 bopd
1,128 bopd
6 bopd
50 bopd
15 mcfd
3.5 mcfd
120 bopd
70 bopd
1 ,584 bopd
65 bopd
130 bopd
1,100 mcfd
66.1 mcfd
6
7
1.4
10.8
2.6
73
19
Water Supply Applications
New Hampshire
(45)
New Hampshire
(45)
Massachusetts
(47)
Australia (48)
Water
Water
Water

Water
water
water
water

water
1
1
1

1
Rock
Rock
Rock

Rock
18 L/min
15.6 L/min
0.46 L/min

1 .8 L/min
109.2 L/min
68.4 L/min
11.4 L/min

13.2 L/min
6
4.4
22-25

6.9
a Performance measured as rate or rate/drawdown
bopd: barrels of oil per day
mcfd: millions of cubic feet of gas per day
1 mH2o head = 1.42 psi
Hydraulic fracturing also increases discharge of water
wells, and the relative increases resemble results from
the petroleum industry. Thirty years ago,  Koenig (44)
examined data  from wells used  for waterflooding  or
waste disposal and  reported that 78 percent of those
wells increased discharge following hydraulic fracturing.
The  recovery ratios ranged up to  100, with a median of
5.0 (44). More recent tests in the United States (19,
45-47) support these  results (Table 3-2).

Recovery rates typically decrease as a  function of time
due to depleting target fluid (oil, gas, water, or NAPL) or
reduced fracture permeability from closure or clogging.
Sediment or mineral  precipitate are known to close or
clog fractures induced for petroleum recovery. In many
cases, however, creating another fracture can restore oil
recovery. Although environmental applications would be
expected to experience  similar  problems, several pro-
jects  demonstrated that shallow fractures  retain their
permeability for more than a year (9, 22, 49).

The recovery ratios for environmental applications resem-
ble some  of the more successful applications from the
petroleum and  water  well industries, with typical values
between 10 and 50. Many of the environmental applica-
tions  involve silty clay or rock of quite low  permeability.
Therefore, the initial specific discharges (ratio of discharge
to head differential at the well) are exceptionally small,
                                                    55

-------
leaving a large margin for improvement. Similar ratios
are observed for both naturally propped fractures in rock
(created by  injecting air or water) and sand-filled frac-
tures  in silty clay.  Interestingly, the ratios  seem to be
relatively independent of the type of fluid recovered and
the remedial method used; applications for vapor extrac-
tion, liquid injection, steam injection, and NAPL recovery
report similar values (Table 3-2).

3.4.2  Selected Examples
Documents published through the Superfund Innovative
Technology  Evaluation (SITE)  program describe case
histories of hydraulic and pneumatic fracturing (36, 49).
These projects chiefly involve  vapor extraction in tight
soils and rock. The following sections present two appli-
cations of NAPL recovery.

3.4.2.1    DNAPL  Recovery From  Bedrock,
          Bristol,  Tennessee
At a site  near  Bristol, naturally propped fractures were
created in July 1991  in  rock at depths of 30.5 to 61
meters (100 to 200  feet) to enhance the  recovery of
free-phase TCE and other DNAPLs. Injecting water into
intervals of the well isolated by straddle packers created
the fractures. Pumping tests and vapor extraction tests
were  conducted to  evaluate the effects of the fractures.
Site Conditions. Sandstone, shale,  and  limestone un-
derlie the vicinity of the site and form a broad fold and
dip approximately  45 degrees  beneath the  site itself.
The site lies in a local recharge area of a bedrock aquifer
characterized by downward vertical hydraulic gradients
of approximately 0.5 units (21). The hydraulic conduc-
tivity of the water-bearing formation is approximately 10"6
centimeters per second, based on constant rate tests.
Contaminants. The site contains a free-phase plume of
TCE,  other solvents, and cutting oil that extends to a
depth of 100 meters (328 feet). Adissolved-phase plume
of primarily TCE extends to greater depths—at least 130
meters (426 feet)—and has migrated at least 300 me-
ters (984 feet) from the suspected source.  The specific
gravity of the free-phase liquid  is 1.3 (21).
Design. Recovery using a  pump  and treat  system
yielded approximately 3.7 liters (1 gallon) per minute of
water  per well and  fewer  than  1.4  kilograms (3.1
pounds) per day of DNAPL. These low rates of recovery
provided  impetus  for using  hydraulic fracturing tech-
niques to stimulate the wells. The intent was to increase
formation permeability that, in turn, would promote liquid
flow and possibly permit sufficient air flow into wells for
recovery though vapor transport.
Three new wells were drilled to 60 meters (197 feet) with
open  hole completion, and the  performance of these
wells was characterized before and after hydraulic frac-
turing. Each well was fractured by setting open hole
packers 15 meters (49.2 feet) apart and injecting 4,500
to 9,000  liters (1,188 to 2,376 gallons) of clean water;
no proppants were injected. Injection pressures ranged
between  0.5 and  5  MPa (73 and 725 psi), and the
injection rate was approximately 280 liters (73.9 gallons)
per minute. This approach resembles methods used to
increase the discharge of water wells (19). Injection was
terminated when water flowed around the upper packer
and began to spill to the surface. In one case, an obser-
vation well 2.5 meters (8.2 feet)  away responded with
discharge of injected water. The  initial discharge was
muddy but cleared with continued injection, suggesting
that fine-grained particles had been removed from frac-
tures in the formation (21).

Results. The specific discharge  of the three wells in-
creased by factors ranging from 2.8 to 6.2. These effects
are typical of naturally  propped fractures created by
hydraulic fracturing of water wells (Table 3-3). Pumping
tests helped determine the effective hydraulic conduc-
tivity after fracturing. In general, the results  indicated
that the  effective hydraulic conductivity increased by
factors of 20 or more. (Actual values depend on the
method of solution  used to analyze the test data.)

Table 3-3.  Specific Capacities Before and After Hydraulic
         Fracturing

TW-1
TW-2
TW-3
Before
L/min m
0.32
0.45
0.98
After
L/min m
1.6
2.8
2.7
Ratio
5.0
6.2
2.8
Inducing fractures appeared to make vapor extraction a
feasible remedial technique. After fracturing, vapor dis-
charges were on the order of 285 to 700 L/min, and
suction was detectable 10 meters (32.8 feet)  from the
recovery well. In contrast, both discharge and suction in
the formation were negligible prior to fracturing.

During a 2-day  test of vapor extraction, DNAPL was
recovered at a rate of approximately 82 kg/day (180
Ibs/day). Concentrations  diminished  during  this test,
probably representing an upper limit of the  recovery
rate. Nevertheless, the combination of hydraulic fractur-
ing to increase conductivity and suction to induce dewa-
tering and  DNAPL  recovery appeared to  be  a viable
method of increasing contaminant recovery at  this site.

Cost. Reportedly, the cost to create the fractures used
during this project was $1,500 per well.

3.4.2.2   LNAPL Recovery From Swelling Clay,
         Beaumont, Texas

In July 1993,  sand-filled hydraulic fractures were created
in swelling  clay to enhance the recovery of free-phase
                                                   56

-------
LNAPL at a site in Beaumont. A pilot test followed in late
February 1994.

Site Conditions. Silty clay of the Beaumont formation
underlies the site to a depth of 6 to 8  meters (19.7 to
26.2 feet), with fine-grained sand below it. In general,
the Beaumont formation consists of kaolinite, illite, cal-
cium smectite, and fine-grained quartz (50). Wetting and
drying  cycles have resulted  in overconsolidation in the
upper 8 to 10 meters (26.2 to 32.8 feet). Drying near the
ground surface largely decreased clay volume there,
commonly resulting in dessicated areas. Moreover, in
the interval containing hydraulic fractures, lateral stress
is two  to three times greater than vertical stress in the
upper few meters (50).

The upper meter of the site is fill, composed ofsiltyclay,
gravel, and shells. From  1  meter (3.3 feet) to approxi-
mately 3.6 meters (11.8 feet), the formation is a firm to
stiff, dark gray, silty clay with reddish to  olive yellow
mottling.  Slickensided partings, which indicate preexist-
ing fractures, are common. A light gray, clayey silt occurs
at  approximately 3.6 meters (11.8 feet).

Fractures were created between 2 and 3.6  meters (6.6
to  11.8 feet) deep so most of them were initiated in the
dark gray, silty clay. The deepest fractures,  however,
were initiated in the light gray, clayey silt. The watertable
was between 1 and 1.5 meters (3.3 to 4.9 feet) deep, so
all the  fractures were created in saturated conditions.

Contaminants. The area of the test contained gasoline
and cyclohexane, which infiltrated from surface spillage.
The contaminant appeared  as  free-phase  NAPL  from
approximately 1.5 to 3 meters (4.9 to 9.8 feet) in depth
in the vicinity of Wells I, C, and PW-1, and it thinned to
the east toward Well G (Figure 3-18).
control well. One of the fractured wells in the test con-
sisted of two casings that accessed fractures at different
depths, one in the LNAPL and the other in the water
bearing zone below. The other well only contained one
fracture nearthe bottom of the NAPL zone (Figure 3-19).
The goal of the two-fracture design was to recover NAPL
from the upper fracture and water from the  lower one.
This approach, if successful, would limit upward coning
of water, which would both increase the rate of recovery
of NAPL and improve the NAPL-to-water ratio (to reduce
costs of phase separation)  compared with  recovering
from one  fracture. Both fractured wells were expected
to produce at greater rates than the conventional well.
Sand-filled fractures were created at six locations at the
site, but only two of them, I and C, were used during the
pilot test. At 1-12 (Figure 3-18), a single sand-filled frac-
ture was created at a depth of 3.6 meters (11.8 feet). At
Well C, four fractures were stacked one above the other,
but the test used only the deepest two, C-10 and C-12.
The fracture at C-10 was initiated at a 3-meter (9.8-foot)
depth, and it curved  upward and cut  through much of
the zone  containing NAPL.  Fracture C-12, which was
initiated at 3.6 meters (11.8  feet), extended  mostly be-
neath the NAPL zone (Figure 3-19). The fractures were
planned to be approximately circular, with diameters of
7 to 8 meters (23 to 26.2 feet) and average thicknesses
of 5 to 6 millimeters (0.20 to 0.24  inches) (Table 3-4). A
conventional well,  PW-1, was screened  from 2 to 4
meters (6.6 to 13.1 feet) in depth and used  as control.
Clusters of multilevel piezometers (depths of 1.2, 2.4,
and 3.6 meters [3.9, 7.9, 11.8 feet]) with short screens
were installed along a line from Well  I through Well C
(Figure 3-18).

Table 3-4.  Specifications of Fractures  Used During the Pilot
         Test
design. The pilot test was designed to compare the Average
erformance of two designs of fractured wells with a _ .. M™™"m T1?*nd Average Sand
a Depth Uplift Thickness Diameter Volume
m mm mm m m3 (ft3)

r Storage
Tank
C-10 G_85 G-10
C-8.5 I m ^^^
• "Pc-12 " " " OP'12
•
.'
f Storage
I .10 i 	 - i -p i i
1 0 Meters J
N H
1 A • Well Accessing Fracture 1|
I PW-1 tH
^ • Multilevel Piezometer ||
A Control Well %
C-10 3 16 6 7 0.23(8)
C-12 3.6 24 5.5 8 0.28(10)
1-12 3.6 22 5.5 8 0.28(10)
A constant head was maintained approximately 1 0 cen-
timeters (4 inches) above each fracture, producing a
drawdown of 1 .5 to 2 meters (4.9 to 6.6 feet). Fluid was
pumped from the wells to storage drums and periodically
diverted to a graduated cylinder to determine discharge
rate. The proportion of NAPL and water in the beaker
was measured to estimate the discharge of each phase.
Figure 3-18.  Area of LNAPL recovery using fractured and con-
ventional wells.
Results.  Both wells  containing  fractures produced
LNAPL at rates an order of magnitude greater than the
conventional well (Table 3-5). The C location was par-
ticularly noteworthy,  producing both the greatest NAPL
                                                   57

-------
                        Two-Fracture Design
                        (fractures C-10 and C-12)
     One-Fracture Design
     (fracture 1-12)
                     Water
                                         LNAPL
                                                       Water + LNAPL
Figure 3-19.  Methods of completing a well containing hydraulic fractures to recover LNAPL.
Table 3-5. Average Discharges and Ratios of Discharge

C-10
C-12
C (combined)
1-12
PW-1
NAPL
L/hr
4.33
0.07
4.40
1.85
0.23
Water
L/hr
0.34
8.24
8.58
5.61
0.03
Total
L/hr
4.7
8.3
13.0
7.5
0.26
NAPL/
Water
13
0.008
0.5
0.3
7
NAPL/
PW-1
19
0.3
19
8

Total/P
W-1
18
32
50
29

rate  and  the  greatest  NAPL-to-water  ratio. Fracture
C-10 produced a high concentration of LNAPL at a rate
that was  19 times  greater than the control, whereas
C-12 produced almost completely water. The combined
rate of liquid recovery from the C location was 50 times
greater than from the control.
As an  additional  test, the rates were evaluated  as a
function of time, then the pump in the C-12 fracture was
turned  off. These results show that Well C initially rapidly
recovered water,  presumably  as it drained out water
used to created the  fractures. Then C-12 recovered
water at a constant rate while C-10 primarily recovered
NAPL  (Figure  3-20). After the pump was  turned off at
116 hours, the discharge from C-10 changed abruptly;
the total discharge from C-10 increased, but the recov-
ery of NAPL actually decreased. Apparently, turning off
the pump  in C-12 caused water to flow upwards, effec-
tively reducing the area at C-10 available to recover NAPL.
The distribution of head was consistent with the rela-
tively large NAPL recovery by wells intersecting sand-
filled fractures. Bowl-shaped zones  of  relatively  large
drawdown occurred in the vicinity of the fractures (Fig-
ure 3-20). In  addition, significant drawdown occurred
throughout the area between Wells I and C, and draw-
down of 7  centimeters (2.8 inches) occurred over a band
25 meters (82 feet) long in the vicinity of the two wells.
                                                         15
                                                         10
                                                         5
1 '
1
, 0
• 6

. i > i i i
i i
1 1 1 ,
Water
LNAPL
ll
' 1 '

?
O
C
1
"^ O CM T O CM V O CM O CM OCM TO CM O CM T
~> 6 6 £ 66o_6666 66o_66 66c

•j
)
~0
LO '
, '
             20
                    40     60

                        Hours
                                 80     100
                                             120
Figure 3-20.  Discharges from C-10, C-12, and PW-1 as func-
           tions of time.
Drawdown in the vicinity of PW-1 was unavailable, but
similar tests in the area have shown that drawdown  is
negligible within 1 to 2 meters (3.3 to 6.6 feet) of con-
ventional wells.
It is noteworthy that the fractures caused large vertical
head  gradients  (Figure 3-21).  Multilevel  piezometers
with short screens (25 centimeters [9.8 inches]) were
required to characterize the  head distribution. Conven-
tional piezometers screened over a large interval would
have  missed the vertical gradients, resulting in a mis-
leading estimate of the effects of the fractures in the
subsurface.
Cost.  The fractures cost approximately $800 to $1,000
each to create and complete  as wells. This cost includes
mobilization, materials, labor, and equipment. The pilot
test itself cost approximately $40,000.
                                                    58

-------
                  1-12
                                              C-12
                                                                   10m
                                                   C-10
                                                              Piezometric Monitor
                                                          	  Sand-filled Fracture
                                                         Contour-Head Change (cm)
                                                                                  3m
Figure 3-21.  Cross section along line of piezometers showing change in head after 5 days of pumping.
3.5   References

When an  NTIS number is cited in  a reference, that
document is available from:
  National Technical Information Service
  5285 Port Royal Road
  Springfield, VA 22161
  703-487-4650

 1.  U.S.  EPA. 1993. Hydraulic fracturing technology:
    Applications analysis and technology evaluation re-
    port.  EPA/540/R-93/505.

 2.  U.S. EPA. 1993. Accutech pneumatic fracturing and
    hot gas injection, Phase I. Technology demonstra-
    tion summary. EPA/540/SR-93/509.

 3.  Schuring, J.R.,  V.  Jurka, and  PC.  Chan.  1991.
    Pneumatic fracturing to remove VOCs.  Remedia-
    tion Winter:51-67.

 4.  Murdoch, L.C., G.  Losonsky, P.  Cluxton, B. Patter-
    son, I. Klich,  and B. Braswell. 1991. The feasibility
    of hydraulic fracturing  of soil to  improve remedial
    actions.  Final   report.  EPA/600/2-91/012  (NTIS
    PB91-181818).

 5.  Frappa, R.H., and  T.H. Forbes. 1994. Innovative
    ground water remediation solutions for the Mercury
    aircraft site. Presented  at NYS WEA Spring Confer-
    ence, Corning, NY (June).

 6.  Begor,  K.F.,  M.A.  Miller,  and  R.W. Sutch. 1989.
    Creation of an artificially produced fractured zone
    to prevent contaminated ground water  migration.
    Ground Water 27(1):57-65.

 7.  Zinck, E. 1984. Method of and device for loosening
    agriculturally used  soil. U.S. Patent 4,429,647.

 8.  Waltz, J.P, and T.L Decker.  1981. Hydro-fracturing
    offers many benefits. Johnson Drillers J. 53:4-9.
 9. Schuring, J.R., and  PC. Chan. 1992. Removal of
    contaminants from the vadose zone by pneumatic
    fracturing.  NTIS  PB92-161207. Reston, VA:  U.S.
    Geological  Survey,   Water  Resources  Division
    (January).

10. Driscoll, F.G.  1986.  Ground water and wells, 2nd
    ed. St. Paul, MN: Johnson Division Publisher.

11. Passamaneck, R.S.  1993.  Controlled propellant
    fracturing for vertical wells. SPE Paper 26271.

12. Chu, T.Y.,  N.  Warpinski, R.D. Jacobson.  1988. In
    situ experiments of geothermal well stimulation us-
    ing  gas  fracturing  technology.  Sandia  Report
    SAND87-2241. Albuquerque, NM: Sandia National
    Laboratories.

13. Sarapu, E. 1985. Electrofrac rock-breaking  sys-
    tems.  Skillings Mining Rev. January:2-8.

14. Du, Y, A.  Aydinm, and L. Murdoch. 1993. Incre-
    mental growth of shallow hydraulic  fracture  at a
    waste remediation site. 34th  Symposium  on Rock
    Mechanics, Madison, Wl (June).

15. Davis,  P.M. 1983. Surface deformation associated
    with a dipping hydrofracture. J.  Geoph. Resour.
    88:5,826-5,834.

16. Sun, R.J.  1969.  Theoretical size of hydraulically
    induced fractures and corresponding surface uplift
    in  an  idealized  medium. J. Geophys.  Resour.
    74:5,995-6,011.

17. Warpinski, N.R.  1985. Measurement of width and
    pressure in a propagating hydraulic fracture. Soc.
    Petrol. Eng. J. February:46-54.

18. Holzhausen, G.R., C.S. Haase, S.H.  Stow, and G.
    Gazonas. 1985. Hydraulic-fracture growth in dip-
    ping anisotropic strata as viewed  through the sur-
    face deformation field. Proceedings of the 26th U.S.
    Symposium on Rock Mechanics, Rapid City, SD.
                                                  59

-------
19. Smith, S.A. 1989. Manual of hydraulic fracturing for
    well stimulation and geologic studies.  Dublin, OH:
    National Water Well Association.

20. Schuring, J.R., P.C. Chan, J.W Liskowitz,  P. Pa-
    panicolaou, and C.T. Bruening. 1991.  Method and
    apparatus for eliminating non-naturally occurring
    subsurface, liquid toxic contaminants from soil. U.S.
    Patent 5,032,042 (July 16).

21. Lundy, D.A., C.J. Carleo, M.M. Westerhiem. 1994.
    Hydrofracturing bedrock to enhance DNAPL recov-
    ery. Proceedings of 8th Annual NGWA National Out-
    door Action Conference (May).

22. Wolf, A., and L. Murdoch. 1993. Afield test of the
    effect of sand-filled hydraulic fractures on air flow
    in silty clay till. Proceedings of the Seventh National
    Outdoor Action Conference, Las Vegas, NV (May).

23. McKay, L.D., J.A. Cherry, and R.W Gilham. 1993.
    Field experiments in a fractured clay till, 1. Hydrau-
    lic conductivity and fracture aperture. Water Resour.
    Res. 29(4):1, 149-1, 162.

24. Brooker, E.W, and  H.O. Ireland. 1965. Earth pres-
    sures at rest related to stress history. Can. Geotech.
    J.
25. U.S. Geological Survey. 1967. Distribution of prin-
    cipal kinds of soils: Orders, suborders, and great
    groups. Sheet 86.

26. Flint, R.F.  1971. Glacial and quaternary  geology.
    New York,  NY: John Wiley and Sons.

27. Tsang, YW 1992. Usage of "equivalent apertures"
    for rock fractures as derived  from hydraulic  and
    tracer tests. Water Resour. Res. 28(5):1, 451 -1,455.

28. Muskat, M. 1949. Physical principles of oil produc-
    tion.  Boston, MA: McGraw-Hill.

29. Novakowski, K.S. 1988. Comparison of fracture ap-
    erture widths determined from hydraulic measure-
    ments and tracer experiments. In: Hitchon, B.,  and
    S.  Bachu,  eds. Proceedings of the Fourth Cana-
    dian/American  Conference on Hydrogeology. Dub-
    lin, OH: National Water Well Association, pp. 68-80.

30. Bear, J. 1979.  Hydraulics  of ground  water. New
    York, NY: McGraw-Hill.

31. American Petroleum Institute. 1989. A guide to the
    assessment and remediation of underground petro-
    leum releases. API Publication 1628. Washington,
    DC.

32. Blake., S.B,  and M.N. Gates. 1986. Vacuum en-
    hanced hydrocarbon recovery. Proceedings of the
    Second Hazazardous Materials Management Con-
    ference, Long  Beach, California (December), pp.
    17-25.
33. Vesper,  S. J., L.C. Murdoch, S. Hayes, and WJ.
    Davis-Hoover. 1994. Solid oxygen source for biore-
    mediation in subsurface soils. J. Haz. Mat. 36:265-
    274.

34. Davis-Hoover, W.J., L.C. Murdoch,  S.J. Vesper,
    H.R.  Pahren, O.L. Sprockel, C. L. Chang, A. Hus-
    sain,  and W A. Ritschel. 1991. Hydraulic fracturing
    to improve nutrient and oxygen delivery for in situ
    bioreclamation. In situ bioreclamation. Boston, MA:
    Butterworth-Heinemann. pp. 67-82.

35. Mack, J.P., and H.N. Apsan. 1993. Using pneumatic
    fracturing extraction to  achieve regulatory compli-
    ance and enhance VOC removal from low-perme-
    ability formations. Remediation Summer:309-326.

36. U.S.  EPA. 1993. Accutech  pneumatic fracturing and
    hot gas  injection,  Phase  I.  Applications analysis.
    EPA/540/AR-93/509.

37. Gillham, R.W, and D.R. Burris. 1992. In situ treat-
    ment wells: Chemical  dehalogenation,  denitrifica-
    tion,  and  bioaugmentation.  Proceedings of the
    Subsurface Restoration Conference,  Dallas, TX
    (June), pp. 66-68.

38. Morgenstern, N.R., and PR. Vaughan. 1963. Some
    observations on allowable grouting pressures. Pro-
    ceedings of the Conference on Grouts and Drilling
    Muds. London, England: Institute of Civil Engineer-
    ing, pp. 36-42.

39. Wong, H.Y, and I.W Farmer. 1973. Hydrofracture
    mechanisms in rock during pressure grouting. Rock
    Mechanics 5:21-41.

40. Zuomei,  Z., and H. Pinshou. 1982. Grouting of the
    karstic caves with  clay fillings. In: Baker, W.H., ed.
    Proceedings  of the Conference on Grouting  in
    Geotechnical Engineering, New Orleans, LA (Feb-
    ruary), pp. 92-104.

41. Brunsing, T.P, and R.B. Henderson. 1984. A labo-
    ratory technique for assessing the in-situ construc-
    tability of  a  bottom  barrier for waste isolation.
    Proceedings of the National Conference on Manag-
    ing Uncontrolled Hazardous Waste Sites, Washing-
    ton, DC. pp. 135-140.

42. Brunsing,  T.P.  1987.   The  block  displacement
    method field demonstration and specifications. U.S.
    Environmental Protection Agency Project Summary
    600-S2-87-023. pp. 1-6.

43. Howard, G.C., and C.R. Fast. 1970. Hydraulic frac-
    turing. New York,  NY:  Society of Petroleum Engi-
    neers, AIME.

44. Koenig, L.  1960. Survey and analysis of well stimu-
    lation performance.  J.  Am. Water Works Assoc.
    52:333-350.
                                                  60

-------
45. Stewart, G.W. 1974. Hydraulic fracturing of crystal-
    line rock stimulates yield of two test wells drilled in
    New Hampshire. Ground Water 12:46-47.

46. Stewart, G.W. 1978. Hydraulic fracturing of drilled
    water wells in crystalline rocks of New Hampshire.
    New Hampshire Department of Resources  and
    Economic Development and Water Resources Re-
    search  Center.

47. Mony, P.M. 1989.  Hydro-frac: One man's opinion.
    Water Well J. Feb:48-50.
48. Williamson, W.H., and D.R. Woolley. 1980. Hydrau-
    lic fracturing to improve the  yield of bores in frac-
    tured  rock. Australian Water  Resources Council
    Technical Paper 55:1-76.

49. U.S. EPA.  1993. Hydraulic  fracturing technology:
    Applications analysis and technology evaluation.
    EPA/540/SR-93/505.

50. Mahar, L.J., and M.W O'Neill. 1983. Geotechnical
    characterization of desiccated clay. J.  Geotech.
    Engin. 109(1):56-71.
                                                  61

-------
                                              Chapter 4
                                       Interceptor Trenches
Other than vertical wells, trenches are the most widely
used method of controlling subsurface fluids and recov-
ering contaminants; trenches are remarkably effective,
and widely  available  construction  equipment can  be
used to install them. The construction and agricultural
industries have long used trenches or drains for dewa-
tering, and many environmental practices are based on
dewatering methods. However, new methods of install-
ing and completing trenches for environmental purposes
have recently been developed. For instance, techniques
for analyzing a trench's  ability to control ground-water
flow directions and travel times differ from the dewater-
ing techniques.

Atrench provides a long zone with which to collect fluids.
Perhaps  the most common applications of this  linear
feature are to increase the rate of recovery from low-per-
meability formations or to intercept the migration of a
plume. The former uses the  large surface area available
for drainage, whereas the latter places the trench normal
to the  regional gradient to enhance  plume  capture.
Trenches also can have a  significant vertical compo-
nent, which  cuts across and can allow access to the
permeable layers  in interbedded sediments. Alterna-
tively, the vertical component can facilitate LNAPL  re-
covery   from   areas  with  considerable   seasonal
fluctuations in the watertable. Although trenches primar-
ily recover liquids, vapor extraction also benefits from
the large area  exposed by a trench.

Although trenches have various  applications, most of
this chapter discusses the trenches  used to recover and
deliver fluids:  interceptor trenches. In their  simplest
form,  interceptor trenches  are slots excavated in the
earth and filled with highly permeable  material. More
sophisticated designs involve placing perforated pipe or
casing  at the bottom of the excavation (Figure 4-1); the
casing commonly slopes  downward  from the ends of the
trench  towards a sump, which  collects  and removes
water or  contaminant through a vertical access casing
(1). Commonly, low-permeability material caps trenches
to prevent surface water or air  (for vapor extraction
applications) from diluting the fluids recovered from the
formation.
                                                                        Recovered Fluids
                                                                                          Access Casing
                                             Pump
Figure 4-1.  Cross sections of a basic interceptor trench con-
          figuration  showing permeable  backfill and per-
          forated casing.

This chapter presents methods of constructing trenches
and discusses some of the factors that affect their de-
sign and the decision to use trenches  at a site. It also
presents an overview  of some of the applications of
trenches and  several  case histories  of sites where
trenches have been used.

4.1   Trench Construction

Most interceptor trenches are created with construction
methods adapted to environmental applications. Some
specialized techniques are available  for particularly
deep installations, streamlining the installation and com-
pletion process, or for forming trenches in rock.

4.1.1   Conventional Methods

Creating an interceptor trench typically involves:

• Excavating material

• Supporting the trench walls

• Backfilling with permeable material

• Installing casing and  pumps

• Sealing  the surface over the trench

Standard construction  equipment, either a backhoe or
a trenching  tool, commonly is used  to perform  the
                                                  63

-------
excavation. Backhoes capable of digging a trench up to
5.5 meters (18 feet) deep are readily available in most
locations. Extended-reach hydraulic excavators (2) may
reach depths approaching  15.2 meters (50 feet), with
widths of 0.9 to  1.5 meters (3 to 5 feet).  Trenches of
intermediate depth (6.1  to  9.1 meters  [20 to  30 feet])
may require an excavation  wide enough to accommo-
date the backhoe or trencher if the equipment is to reach
the desired depth.

Trenching tools  are available to construct trenches of
various  depths  and  widths. Small   vibratory  plow/
trencher units create slots a few inches in width and up
to 0.6 meters (2 feet)  deep. Many  of these small units
were designed for utility installation (gas line, electric,
telephone) and can install the pipe or cable while creat-
ing the slot.

Medium-sized,  tractor-mounted  trenching  units  may
reach depths of 2.4 to 3 meters (8 to 10 feet), with widths
of up to 0.6 meters (2 feet).  Attachments for these units
permit trench excavation through concrete and asphalt
pavement.  Large trenching systems are discussed in
detail in the next section.

Vertical walls on the side of a trench may stay open if
the trench is relatively shallow and cuts through  cohe-
sive soil. In many cases,  however, the walls of the trench
collapse if they are unsupported. Traditional methods of
support include  a sliding trench  box, sheet pilings, or
timber beam bracing  (shoring).  Bentonite slurry also
sometimes supports trenches, although the bentonite
may significantly reduce the permeability of the forma-
tion  adjacent to the trench; therefore, this technique
should be avoided for interceptor trenches.

Recently, a slurry formed from guar gum, a cellulose-
based  polymer, has been used to support deep trench
excavations in an effort to  avoid the problems associ-
ated with bentonite (2,  3). Guar-based slurries have high
gel strength and viscosity (greater than 40 centipoises)
and low water loss (filtrate less than 25 milliliters); these
properties permit efficient transfer  of hydrostatic head
from the slurry to the trench walls. Slurry head of 0.9
meters (3 feet) or more over ground-water head should
stabilize  most soil types. Guar gum used to support  a
trench  is treated so  that it degrades when  construction
is completed; the guar decomposes to water and biode-
gradable sugars  and is pumped from the trench. Usually
three trench pore volumes circulate to remove residual
slurry,  in a  process  similar  to well  development.  Small
amounts of degraded slurry may remain as excess fluid
in the  trench and must evaporate, solidify, or be  dis-
posed  of through a wastewater treatment facility (2-4).

Permeable backfill material  (coarse sand or gravel)  and
injection/extraction structures (perforated casing, sumps,
access pipes, and  wells)  are placed in  the  trench
through the slurry. The fill material  typically consists of
clean washed gravel (e.g., crushed stone, pea gravel).
When anticipating fine sediment clogging, engineered
gradation  or a geotextile filter fabric may be  included.
Engineered gradation involves placing the fill material
into the trench by sliding it down the slope of previously
deposited fill or by tremie tube. (Sand and finer material
must be pre-wetted; gravels may be tremied dry.) Tremie
tube emplacement is preferable around wells and per-
forated casing. Using woven geotextile fabrics can in-
hibit  siltation.  Geotextile  filters  generally   must be
weighted to sink through the slurry. Overlapping sheets
of geotextile is usually sufficient to ensure filter continuity.

The  backfill should extend near the ground surface and
always above ground-water level.  Excavated soil can
usually fill the remainder of the trench, although regula-
tions may prevent backfilling with contaminated material
at some sites. Backfill or clay usually seals the  top of the
trench to prevent infiltration of surface runoff or air.

Vertical access pipes installed in the fill material are the
most common injection/extraction structures. Perforated
pipe, which is commonly laid along the bottom of the
trench, connects to the vertical access pipes. This tech-
nique is  particularly  prevalent  in  relatively shallow
trenches created in  cohesive soil  with  free-standing
walls. Slurry in the trench may buoy the perforated pipe
and make installation cumbersome in deep excavations.
Practitioners disagree on the need for perforated pipe in
trenches;  some  argue that permeable backfill is suffi-
cient. A variety  of casing types exist, including  PVC,
polyethylene, and galvanized and stainless steel.  Sub-
mersible ejector and  progressive  cavity  pumps have
been successfully implemented.

4.1.2  Specialized Methods

Most interceptor trenches are installed using  methods
similar to the ones outlined above. Some cases, how-
ever, require  specialized  methods to increase the effi-
ciency of the installation procedure orto create trenches
that are particularly deep or in rock.

4.1.2.1   Continuous Excavation and Completion

A recently developed technology combines trench exca-
vation, well  installation,  and barrier installation  proc-
esses in a single step (5). The equipment digs a trench
up to 6.1 meters (20 feet) deep, lays in a flexible hori-
zontal well screen, and  backfills the trench  with the
original soil. Typically, a vertical excavation 35.6  centi-
meters (14 inches) wide is completed to the desired
depth. Then,  a vertical casing is installed through the
trenching  head. The well is coupled to the flexible well
screen.  The trenching machine then moves along the
trench line (Figure 4-2),  installing the horizontal well
screen at an  average rate of 0.9 to 1.5 meters (3 to 5
feet) per minute  (7). Sections of screen can be coupled
to create  continuous well  screens in excess of 610
                                                   64

-------
     Well Screen

        1°^^
Figure 4-2. Schematic of continuous excavation and comple-
          tion technique (6).
meters (2,000 feet). A permeable polyester and glass
filter system encases the flexible well screen to reduce
or eliminate clogging by fine sediments. Moreover, the
well screen/filter system may be installed within a sand
or gravel filter pack in the trench. (The sand or gravel is
placed at the same time as the well screen.)

Reportedly, continuous excavation and completion sys-
tems are about half the cost of bored (drilled) horizontal
well systems, and as much as 70 percent less expensive
than vertical well systems (5). At a  site in North Carolina,
initial remediation plans called  for 100 vertical wells to
recover a hydrocarbon plume at an estimated cost of $1
million. Instead, a continuous  excavation and comple-
tion system was installed at a cost of under $350,000. A
second benefit under some settings is faster remedia-
tion. A fuel storage facility in Florida opted for a continu-
ous excavation and completion system with nine wells
over a conventional vertical well system requiring  41
wells. Actual time for successful remediation was just 8
weeks, compared with an  estimated 30 months for the
vertical well system (5). Finally, this technology reduces
worker exposure to contaminants.

This technique has some limitations:

• The trenching head cannot penetrate bedrock.

• The equipment cannot access  beneath buildings or
  landfills.

• The equipment is currently limited to a depth of less
  than 6 meters (19.7 feet).

4.1.2.2   Deep Applications

Traditional excavation of deep trenches requires the use
of chisels or grabs suspended  on cables or guided  by
kelly bars. Recent developments have elicited new tech-
nologies  for deep excavation for trench and slurry wall
construction.  An example is  "trench  cutting."  Large-
toothed cutting wheels loosen the soil material and mix
it with a  bentonite slurry (Figure  4-3). The equipment
then pumps the soil-slurry suspension to the surface,
where oil particles  are  removed  and the slurry is  re-
turned to the trench for reuse. This technique may be
used to construct trenches 0.5 to 1.5 meters (1.6 to 4.9
feet) wide and up to 120 meters (394 feet) deep. Exca-
vation rates in unconsolidated sediments and soft rock
of 25.2 to 39.8 cubic meters (33 to 52 cubic yards) per
hour have  been reported (8); excavation rates  in rock
with high compressive strength 145 to 179 MPa (21,000
to 26,000 psi) are reported to be 1.1 to 1.9 cubic meters
(1.5 to 2.5 cubic yards) per hour. This technique  is
effective in all types of unconsolidated sediments and in
medium  to  hard  bedrock formations (e.g.,  limestone,
sandstone). To create cutoff walls, the trench may be
filled with low-  permeability materials (e.g., bentonite,
concrete). Alternatively,  to create an interceptor trench,
fill may consist of high permeability materials. Additionally,
a crane can be used to install precast/preconstructed pan-
els. Benefits of the trench cutting technique include:

•  Higher rates of productivity than traditional deep ex-
   cavation methods.

•  Greater control on verticality of the trench because
   of the  rigid excavation platform.

•  Reduced shock and vibration (commonly associated
  with grab and chisel excavation), resulting  in  de-
   creased risk of damage to adjacent structures and of
  trench wall collapse.

4.1.2.3   Trenches in Rock

Economic reasons generally preclude excavating
trenches in rock. Trenchlike structures can be created in
                           To Setting Pond
                                  Steel Frame
                               Hydraulic Motor


                               Centrifugal Pump


                             Spade With Pump Inlet

                             Cutting Wheels
Figure 4-3.  Schematic of deep trench excavation equipment
          (redrawn  from  material  provided  by  Coastal
          Caisson Corp., Clearwater, FL).
                                                   65

-------
rock,  however,  by detonating  explosives in  closely
spaced boreholes. One case in New York (9) created an
elongated fracture zone resembling a trench by placing
approximately 900 kilograms (1,984 pounds) of explo-
sive in 60 boreholes on a line 100 meters (328 feet) long
and perpendicular to ground-water flow. The bores were
7 meters (23 feet) deep  and penetrated a sandstone
aquifer with a hydraulic conductivity of 2.8 x 10"4 centi-
meters per second. After detonation, fractures extended
between boreholes, from 7 meters (23 feet) deep to the
ground surface, and 3 to 4 meters (9.8 to 13.1 feet) into
adjacent  rock. A conventional well placed in the zone
recovered water. Other instances of blasting to create a
permeable wall  in rock exist, although  published ac-
counts are scarce.


4.2   Design Considerations

The design  of an interceptor trench system requires
information related to geology, hydrology, trench hydrau-
lics, and nature of contaminants. These factors influence
the length, depth,  and location of a trench, as well as
the method of operation and type of completion.

4.2.1  Pattern of Flow to a Trench

The advantage a trench offers over other physical meth-
ods of recovery mostly results from the pattern of flow
that the trench induces in  the contaminated formation.
Unlike a  vertical well,  which causes  flow along radial
paths, the pattern of flow to a trench changes with time.
Under ideal conditions, the flow pattern to a trench
progresses through three basic periods: linear flow, tran-
sition, and radial flow. These periods are best recog-
nized  when the trench  penetrates a laterally extensive,
confined  aquifer  with  minimal recharge  and regional
gradient.  In the presence  of boundaries,  recharge, or a
regional gradient, the flow pattern differs from the ideal-
ized case; however, the pattern still changes with time,
and this change still affects the trench performance.
Figure 4-4.  The pattern of flow to a trench with no regional
          gradient: a) linear flow, b and c) transition, and
          d) radial flow (11).
The period of linear flow occurs early on, shortly after
the onset of pumping, when ground water flow is nearly
perpendicular to the  plane of the trench (Figure 4-4a).
During the linear flow period, flow vectors are essentially
parallel, and the region affected is small when compared
with the entire length of the trench.

In contrast,  the  period of radial flow occurs after the
trench has been operating for a relatively long time.
During this period, flow converges from great distances
toward the trench (Figure 4-4d), much as flow converges
toward a well. Except in the vicinity of the trench, flow
vectors are essentially radial during this period, causing
the trench to behave like a well of large diameter.

Details of the flow pattern during the transition period,
as the flow  changes from linear to radial,  depend on
details of the trench geometry. For example, if the trench
partially penetrates the aquifer, the linear period is fol-
lowed by a period when the flow is linear and adjacent
to the trench, radial and converging upward beneath the
bottom of the trench, or radial and converging horizon-
tally toward the ends of the trench (Figure 4-4b). With
increasing time,  the  lower radial zone  expands down-
ward and affects the bottom of the aquifer, and the area
influenced by the trench increases (Figure 4-4c). Even-
tually, the area of horizontal radial flow grows until it is
dominant and the period of radial flow begins. The tran-
sition  period is somewhat simpler when the trench fully
penetrates the aquifer because it lacks the early period
of upward flow toward the bottom of the trench.

The pattern of flow underlies much of the hydrodynamic
performance of  an  interceptor trench. For example,
analyses that consider an interceptor trench in section
view only include flow normal to the trench and omit flow
from areas beyond  the ends of the trench  (published
analyses of trenches are summarized in Beljin and Mur-
doch  [10]). The consequences of this assumption are
relatively minor early on, but amplify as time increases.
Neglecting the effects  of the ends  of the trench may
significantly underestimate the discharge (11).

One approach to estimating the discharge of a trench at
late times represents it as a well of large diameter. It is
possible to represent a trench of infinite hydraulic con-
ductivity (1,000 or more  times the conductivity of the
aquifer) by considering a well whose radius is one-quar-
ter the total length of the trench (11).

The actual times when the flow pattern changes depend
on the length of the trench, the hydraulic conductivity of
the aquifer K, aquifer thickness h, aquifer  storage S,
trench length 2\, and perhaps other quantities. Rather
than  examining all the different possibilities of trench
length and aquifer properties, it is convenient to combine
those quantities with time t to give dimensionless time:
td =
(4-1)
                                                   66

-------
The periods of flow to a fully penetrating trench follow
(11):

Linear flow (early time): fd < 0.25
Transition: 0.25 < td < 25
Radial flow (late time): td > 25

This framework shows that the duration of the different
flow periods in real time t can vary widely, depending in
particular on the hydraulic conductivity of the aquifer and
the length of the trench. Consider, for example, a trench
200 meters (656 feet) long (x, = 100 meters [328 feet]),
in a layer of silty sand that is 5 meters (16.4 feet) thick,
with K=  10"4 centimeters per second and S= 0.1. The
linear flow period would last approximately 0.4 year, and
the radial flow period would not begin  before 40 years.
However, if the trench were half that long (x, =  50 meters
[164 feet]) and installed in a clean sand (K=  10"2 centi-
meters per second), the linear flow period would be only
8.7 hours, and the radial flow period would begin after
36 days. Note that when the trench is very long, the
linear flow period lasts an  extremely long time and the
cross-sectional analyses are appropriate.

4.2.2  Site  Geology

Site geology  affects the design process in two  areas:
performance  efficiency and trench construction. Several
geologic  factors affect the performance of the  trench
system, including heterogeneity of the aquifer, formation
composition,  and degree of saturation  (12).

4.2.2.1   Heterogeneities

Formation heterogeneity is one of the most notorious
obstacles to contaminant recovery, and trenches prob-
ably present the best physical method of addressing this
obstacle. This is particularly true when the heterogeneity
involves horizontal  stratification because the trench can
cut across the strata and access all the permeable units
simultaneously.

Vertical  heterogeneities,  such  as fractures, may be
drained effectively if the trench is roughly perpendicular
to the fracture strike. Pumping  tests or geological map-
ping may help delineate the  orientation of vertical frac-
tures when locating the trench.

Vertical fractures or other heterogeneities may cause
preferential flow of fluids in the subsurface. Therefore,
even though  a trench can access many fractures and
produce  fluids at an impressive rate, vertical fractures
may cause the flow to channel and avoid unfractured
areas. In this case, diffusion from the unfractured areas
can control  recovery  of  contaminants.  Accordingly,
trenches are an excellent method of controlling fluid flow
in fractured or heterogeneous formations, although even
the advantage of accessing many of the fractures (com-
pared with a vertical well) may be insufficient  to recover
contaminants from unfractured areas.
4.2.2.2   Formation Composition

The composition of the formation  primarily affects how
the trench is constructed. Most excavation methods are
capable of removing formations from unlithified sedi-
ments to weathered rock. However, they are unable to
effectively penetrate well-cemented sediments or meta-
morphic or igneous rocks.  In the latter case, it  is still
possible to use a trenchlike approach for recovery, al-
though the approach requires specialized trench-cutting
methods (8) or blasting to create the permeable zone (9).

To prevent collapse of the trench  walls during excava-
tion,  many sediments must be supported, typically by
filling the trench with a dense fluid. The need for support
of the trench  walls increases as the depth and  the
degree of saturation increase. Moreover, granular sedi-
ments, such as sand, are more prone to collapse than
cohesive sediments, such  as  clay. Should a fluid be
necessary to hold  open the trench, the type of fluid
selected should minimize potential reductions in perme-
ability adjacent to  the trench. Accordingly, bentonite
slurry, which supports trenches in standard construction
practice,  should  be avoided.  Proper support  of the
trench is  critical  when  installing  perforated pipe and
access casing for pumps, which require that the trench
remain open for several days.
4.2.2.3    Degree of Saturation

The degree of saturation affects both the construction
and application of trenches. The stability of trench walls
increases as the soils become drier, so it may be possi-
ble to create and complete trenches in relatively dry soils
without holding them open using fluids. When fluids are
used to hold open trenches in unsaturated soil, some of
the fluids may leak out of the trench during construction.
In many cases,  the fluids  are designed  to  form  filter
cakes, or they have sufficient gel strength to minimize
penetration into even relatively dry soils.

The  degree of saturation also affects the  method of
remediation, although the extent to which this is appli-
cable is similar to other methods of recovery.

Vibratory plow trenching units are used where the per-
centage of large rocks near the ground surface are low.
Earth saws are commonly used when the trench exca-
vation must cross pavement, such as a sidewalk, road-
way, or  parking  lot.  Earth saws  have also allowed
trenching through shallow, "soft" bedrock, such as lime-
stone. Using chain-type trench cutters is possible with
most unconsolidated sediments; the use of tungsten-
carbide tipped bits allows chain-type cutters to be used
in hard,  rocky soils,  and even to trench through asphalt
paving.
                                                   67

-------
4.2.3  Hydrologic Conditions

Hydrologic conditions affect the  location, depth, and
sizing of a trench. Of particular importance are the ef-
fects of hydraulic conductivity, regional flow, and aquifer
extent.
4.2.3.1   Hydraulic Conductivity

A trench is an effective method of enhancing the recov-
ery of subsurface fluids if it is more permeable than the
enveloping formation.  Based on field  experience, a
trench  requires a contrast of at least 100 between the
permeability of trench filling and that of the  aquifer,
although extremely long trenches may require  either a
greater contrast or several pumps. Accordingly, it may
be difficult to obtain sufficient permeability contrast when
the trench is in a very permeable sand or gravel. Con-
sequently, a trench oriented  perpendicular to regional
gradient may fail to capture the regional flow, with water
flowing in one side and out the other. One option in this
situation  is to  install  a low  permeability liner on the
downgradient  side  (or the  side containing the least
amount of contamination) of the trench.

The permeability and thickness of the  aquifer also play
central roles in determining the operating parameters of
the recovery system. For example, trenches installed in
relatively permeable material typically operate at con-
stant discharge, and the drawdown decreases with time.
The drawdown decreases rapidly in trenches installed
in low-permeability formations, so a float switch is com-
monly  used to  hold the drawdown  constant above the
intake of the pump. In the latter case,  the drawdown is
held constant, and the discharge decreases with time.
Similar behavior occurs with  wells.  When a well oper-
ates at constant discharge, the Theis solution  is com-
monly used to estimate the drawdown as a function of
time.

A solution similar to the Theis solution for a well can help
predict the drawdown at a trench operating at constant
discharge (11). Following the assumption of Theis, the
trench  is analyzed as a sink over which the inflow is
uniform. This is an idealization,  but it results  in head
gradients along the trench that approximate field condi-
tions where the permeability of material in the trench is
approximately  100 times  greater than the aquifer. A
similar analytical solution can predict drawdowns  if ma-
terial of infinite permeability fills  the trench. (A perme-
ability contrast between the trench filling and the aquifer
that is greater than 1,000  is  essentially an infinite con-
trast.)

As  in a well, the drawdown increases with time (Figure
4-5). The  major difference, however, is that drawdown
increases  at a much slower  rate  than it does for a well
of reasonable size. Table 4-1  lists the variables used to
analyze trenches, including the ones represented in
                                 Uniform
                                 Flux
                Pd = 47tKhAP/Q
     10-2
1CH
10°
102
Figure 4-5.  Drawdown at the midpoint of a trench where dis-
          charge is constant (11).
Table 4-1.  Variables Used in Trench Analysis
Q

S

K

R

n

h

a
 Pumping Rate

 Storage Coefficient

 Hydraulic Conductivity

 Retardation Factor

 Effective Porosity

 Aquifer Thickness

 Regional Gradient
Figure 4-5. The results in Figure 4-5 allow forecasting of
the drawdown so that the pump can be sized and lo-
cated properly.  Forecasting drawdown can also help in
estimating aquifer parameters from a pumping test, just
as aquifer parameters are estimated from pumping tests
using wells. The uniform flux case in Figure 4-5 resem-
bles a trench with a permeability contrast of roughly 100.
Keep in mind, however, that the results assume that the
trench does not interact with lateral boundaries, nor is it
affected by recharge. A more  comprehensive analysis
should consider both conditions  because they  affect
drawdown.

The performance of a trench held at constant drawdown
is  markedly different from one operating  at constant
discharge. The discharge from  a trench held at constant
drawdown is greatest at  the  onset of  pumping. Dis-
charge then decreases with time  (Figure 4-6) as head
gradients in the vicinity  of the  trench flatten. Note that
this effect may  decrease the discharge  by an order of
magnitude or more. The  real time  of this  decrease
                                                   68

-------
   10.0
    1.0
    0.1
                         Qd = Q/(APw8Kh)
      1C-2   -10-1
10°
101    102   103    1Q4    105
Figure 4-6.  Discharge from a trench held at constant  draw-
          down (11).

depends on the aquifer properties, just as the real time
separating the  different flow periods depends on the
properties described earlier. As a result, an appreciable
decrease  in  discharge may take place over several
minutes or  several years, depending on the  aquifer
properties. Over  long  periods, fluctuations  in  natural
hydrologic conditions may mask this effect.

Nevertheless, the discharge from a trench held at con-
stant drawdown may decrease with time as an inevitable
consequence of the dynamics it creates in the aquifer.
Therefore, this effect should  be considered  along with
others when trying to diagnose diminishing  discharge.
Other factors that may contribute to diminishing  dis-
charge  include biofouling  or siltation  of the  material
filling the trench.
The vertical extent of the aquifer is an important consid-
eration when designing a trench system to establish
control of the site. If the vertical extent of the aquifer is
shallow, then the trench should fully penetrate the aqui-
fer thickness. In circumstances where it is impractical to
penetrate the full extent of the aquifer,  however, con-
taminated fluids escaping beneath the trench may be a
concern (13). The severity of this concern depends on
the location and nature of the contaminants, which must
be determined prior to designing the recovery system.

The path of the dividing streamline  (Figure 4-7), which
separates the area where the trench captures water
from the region where water escapes capture, depends
at least on the depth of the trench, pumping rate, hy-
draulic conductivity,  regional gradient,  and  recharge.
Methods of  estimating the  position of  the dividing
streamline  are available in Zheng and others (13)  and
have been validated with field tracer studies by Cham-
bers and Bahr (14).
                                     Figure 4-7.  Cross section of a trench partially penetrating an
                                               aquifer in a regional flow, showing the vertical ex-
                                               tent of aquifer captured by the trench.


                                     4.2.3.3   Regional Flow
4.2.3.2   Aquifer Extent

The lateral extent of the aquifer affects the performance
of atrench, causing it to differ from the theoretical results
based  on the infinite aquifer given above.  In general,
aquifer  boundaries  that  are  within  several  trench
lengths, or perhaps longer, probably have a major effect.
Constant head boundaries, provided by a lake or river,
reduce the drawdown or increase  the recovery rate in
Figure 4-6. No-flow boundaries, such as where the aqui-
fer terminates,  have the opposite  effect,  increasing
drawdown or decreasing rate.

In addition to the operational considerations, lateral
boundaries of an aquifer should affect the  decision of
where  to locate  a trench. For example,  the boundary
between a contaminated aquifer and a surface water
body may be a critical point of exposure that a trench
could protect when installed  parallel to the boundary.
                                     Many applications of trenches involve installations nor-
                                     mal to the regional gradient in an effort to cut off migra-
                                     tion of a plume (12, 15) or to flush fluids through a zone
                                     to enhance recovery (16). In these cases, the areas the
                                     trench influences and the time required to  reach  the
                                     trench are  of interest. It  is possible to estimate those
                                     areas and  times for  various  regional  gradients  and
                                     trench orientations using relatively simple methods (11).
                                     The  most useful  presentation of the results involves
                                     combining  the  various  parameters  in  dimensionless
                                     groups. Because it is independent of actual aquifer pa-
                                     rameters, this presentation allows analysis of many field
                                     scenarios using just a  few graphs. The following exam-
                                     ple illustrates how it works.

                                     Assume that the trench cuts completely through a con-
                                     fined  aquifer of thickness h and infinite lateral  extent.
                                     A regional  gradient is present  and flows  either north
                                     to  south,  northeast  to southwest,  or  east to west
                                                   69

-------
(Figure  4-8). The dimensionless  strength  of  the  re-
gional gradient is
  = x,aS/4hRn.
(4-2)
The variables of the equation are presented in Table 4-1.
A discussion earlier in this chapter pointed out that the
pattern  of flow to a trench changes with  time, so this
problem requires a transient  analysis.  Therefore, the
analysis must specify the strength of the trench relative
to  aquifer properties, whereas a steady-state analysis
incorporates strength in the travel time. The dimension-
less strength is
  = QSrtQKh2nRn.
(4-3)
Figure 4-8 presents a series of maps where the strength
of the trench is fixed at 0.01 and both the orientation and
magnitude  of the regional gradient vary. Regional flow
sweeps away some particles. The areas where particles
are captured form U-shaped zones, with the principal
axis in the  direction  of regional flow. The shape of the
capture zone changes with increasing p by shortening
in the downgradient dimension and lengthening in the
upgradient dimension.  Shortening of the capture zone
also occurs normal to the regional flow, but to a lesser
extent than shortening in the  downgradient.  The con-
tours of equal arrival time can change shape but not
area as a result of the changing direction of  a uniform
and constant regional flow. This is because contours of
equal arrival time must circumscribe the same volume
                          ₯=0.01
                     Areas Where Regional Flow
                     Sweeps Away Particles
                                  = -0.003
                    Flow Is
                    East to
                    West
                  Flow Is
                  North to
                  South
Figure 4-8. Dimensionless arrival times as a function of location for various strengths and directions of regional flow (11).
                                                    70

-------
of water in  a  system operated at constant discharge
(17). It follows that the arrival time contours for a well
would be the same area as those for a trench, except
the zones for a well would be longer and narrower than
those fora trench.

To apply the maps of arrival times, assume a hypotheti-
cal situation in which a trench is installed in an aquifer
with the following parameters:

K = 10"4 meters per second
S=0.1
n = 0.1
h = 5 meters
x, = 50 meters

For ₯ = 0.01, it follows that Q = OAQnnKtf/S = 0.07539
cubic meters  (20 gallons) per minute  . Assuming the
regional gradient is 0.004, then p = -0.01, and the map
at the bottom of the middle column in Figure 4-8 repre-
sents the site  conditions. The arrival times for specific
points can be estimated by obtaining fda from Figure 4-8.
For example,  a  point x = 50 meters, y =  50 meters,
where y is up the regional gradient (Xd = 1.0, yd = 1.0),
yields tda = 38. According to the equation td=4tKh/S%2,
this corresponds to 55 days in real time. The effect of
the regional gradient is  illustrated by taking a point of
equal distance down the regional gradient  (xj = 1.0, yd
= -1.0), where fda = 164, or 237 days. Alternatively, the
area affected  after 6  months is circumscribed by the
contour of tda = 126.

The maps in Figure 4-8  can be applied to most aquifer
settings with valid limitations set by assuming confined
conditions and an  absence of nearby boundaries. The
effect of regional gradient is negligible and probably can
be ignored for Ipl <0.001. At large values of regional
gradient (Ipl >0.05),  a trench operating at a strength of
*P = 0.01 is unable to generate a sink strong enough to
effectively capture particles. The maps in Figure 4-8 are
limited to *P = 0.01, so other strengths require calculating
other maps.

The versatility  and value that interceptor trenches may
offer is  noteworthy.  Interceptor trenches  oriented  at
acute angles to the regional  flow may be  valuable for
contaminant recovery. An  interceptor trench oriented
parallel to the  regional gradient is an effective sink for
areas upgradient  of the  trench.  Moreover,  in  areas
where access or other factors prevent installing a trench
normal to the regional flow, it may be possible to achieve
the desired performance by installing the trench oblique
to the flow. Where the regional gradient is inclined 45
degrees to the axis of the trench, the arrival times  in the
region  upgradient of  the  trench  are  similar  to  those
where p = 90 degrees (Figure 4-8).
4.2.4   Trench Hydraulics

The hydraulics of the trench itself are relatively unimpor-
tant for dewatering applications but become increasingly
important when the trench must maintain control over
long periods. Because the objective of dewatering ap-
plications is to remove water as quickly as possible, the
trench is operated at maximum drawdown. Alternatively,
the objective during long-term control is to affect hydrau-
lic gradients at the site while removing as little water as
possible and  still maintaining some factor of safety. As
a result, the traditional methods of designing trenches
for dewatering applications typically ignore trench hy-
draulics. Therefore, much of the detail of this process
has yet to be analyzed.

The hydraulics of a trench involve coupling flow  in the
trench with flow in the aquifer. There is a point of lowest
potential, either head or air pressure, at the pump, and
then potentials increase along  the trench until reaching
maximum values at the end. The flow per unit area, or
flux, into the trench  is relatively large adjacent to the
pump, and then decreases towards the end. If the trench
terminates in  an aquifer  boundary the  influx will  be
lowest  at the end of the trench  (Figure 4-9). In  the
absence of a boundary, however, the  influx may  in-
crease as it approaches the end of the trench. This is
because flow converges from distant points toward the
end of the trench. In this respect, the flow paths resem-
ble Figure 2-17.

In general, flowrate  along  the trench increases as flow
approaches the pump from the end  of the trench. This
occurs unless fluid flows out of the trench at some point.
For example, isolated perched water that may flow into
a trench locally may be lost from the trench as it passes
an  unsaturated interval. Another possibility is to lose
DNAPL out the bottom of a trench.

The magnitude of the variation of influx depends on the
permeability of the  trench filling (including perforated
casing  if it is used),  permeability of the aquifer, amount
of drawdown, trench length, and location  of pump. The
variation increases as:
                           Flux Into Trench
  Pump'
              V   V   V   V   V   V   V
                              Flow Along Trench
Figure 4-9.  Map of the distribution of influx and flow along an
          idealized trench.
                                                   71

-------
• The permeability of the trench filling is reduced com-
  pared with that of the aquifer.

• The drawdown increases.

• The trench  becomes longer.

Heterogeneities in the aquifer result in local fluctuations
of influx that differ from the effects described above. In
highly  heterogeneous aquifers,  the  heterogeneities,
rather than trench  hydraulics, may dominate the vari-
ation of influx.

The distribution of influx becomes a concern when trying
to operate the trench to maintain control while recover-
ing a minimal amount of water. These efforts minimize
power as well as, possibly, treatment or disposal costs.
Currently,  analytical methods of designing the hydrau-
lics of the trench are unavailable. Consequently, moni-
toring the heads within and adjacent to the trench is
recommended to ensure maintenance of the head gra-
dients toward  the trench.

4.2.5  Contaminants

The type of contaminant mostly affects  the method of
completing a trench. Trenches completed with methods
developed for dewatering can recover aqueous phase
contaminants  (18),  whereas specialized completions
can most effectively recover NAPLs.

NAPL completions may involve one pump and resemble
dewatering applications. This approach is commonly
used to recover LNAPL. Another possibility for comple-
tion involves the use of two pumps: one to recover water
and the other to recover the nonaqueous phase.  Two
perforated pipes can also be effective, with each  pipe
installed at the approximate depth of water and NAPL.
Recovery of water and  NAPL simultaneously inhibits the
upward or downward migration of the NAPL/water inter-
face. This effect tends to pinch off the NAPL layer and
reduce the effectiveness of recovery.

Sealing the upper portion of the trench is a key consid-
eration when recovering vapors. Details of the vapor-re-
covery design may also include components to recover
water, depending on site conditions.

The depth of  the trench  is another factor that partially
depends on contaminant type. Trenches designed to
recover LNAPL must extend at least as deep as  the
lowest seasonal depth of the water table; it may be
unnecessary for this trench to be much deeper. A design
for DNAPL recovery, however,  must key trenches into
the formation on which  the DNAPL has pooled.  The
trench  must not be cut too deeply and penetrate  the
lower formation, because this could lead to loss of the
DNAPL from the trench. In cases where this could occur,
it may be advisable to place a liner along the bottom of
the trench.
Flammability  of the contaminants  is a consideration
when designing  trench  components,  particularly  the
pump and controller. Flammable vapors can accumulate
in the trench or access casing, then be ignited by a spark
from the  pump. Design of the casing and pipe, as well
as the pump itself, must ensure they can withstand
contaminants that are corrosive compounds or vigorous
solvents.


4.2.6  Site Conditions

Site  conditions need consideration  before the start of
trench construction. Access for excavation equipment
along the path of the trench is important. Buildings can
pose a major obstacle, although trenches can be placed
inside some buildings. Prior to excavation, underground
structures (e.g., utility lines, sewers, or tanks) must be
located and marked. Space must be  sufficient to accom-
modate storage of excavated material during construc-
tion,  or provisions must be made to haul the material as
it is excavated.

In  some  locations,  regulations prevent using  contami-
nated material removed during excavation as backfill for
the trench. Regulations  may require disposing of the
excavated material as contaminated waste, which can
significantly increase the cost of installation.

Some sites are covered with concrete, pavement, or a
building,  which may act as a cover to confine air flow.
The presence of such structures and how they affect the
sources of air should be considered when trenches are
used for vapor extraction. Many older paved  surfaces
are cracked and  underlain by gravel, so they may pro-
vide  a poor seal.

4.2.7  Trenches and Horizontal Wells

The  geometry of a  trench can resemble that of a hori-
zontal well. In fact, both  trenches and  horizontal wells
can  be used for  similar applications. In many applica-
tions, the decision between using  a trench or horizontal
well  hinges on economic rather than technical issues.
Depth can increase the cost of trench installation signifi-
cantly, and there is a maximum depth below which
trench installation is impossible. In contrast, depth has
less  effect  on the  cost of a horizontal well (although
exceeding the 8-meter [26.2-feet] depth limit for elec-
tronic beacon locators may increase  the cost). Rock can
present problems to both technologies, although blast-
ing can create trenchlike structures at shallow depths in
rock, and it is certainly possible to bore horizontal wells
in  rock. In both cases, however, rock can increase in-
stallation cost substantially. Access  problems, too, can
increase the cost of a trench or make installation impos-
sible, whereas access requirements  for a horizontal well
are restricted  to entry and possible exit points.
                                                  72

-------
Trenches present significantly more vertical exposure of
the contaminated  area than horizontal wells. This is
particularly important in stratified formations, where a
trench can cut across many thin permeable layers while
a horizontal well  directly intersects only a few layers.
Moreover, where contaminants are present over a large
vertical extent, trenches may be preferable, although the
increase in trench cost with depth should be considered.
For example, the vertical extent of trenches would make
them perform better than a horizontal well when  recov-
ering LNAPL where large fluctuations in the water table
occur seasonally.  Interception of a contaminated  plume
migrating with regional flow is an application for which a
trench has no equal, particularly when excavation equip-
ment can reach the bottom of the  aquifer.

Methods of creating both horizontal wells and trenches
for environmental applications are evolving rapidly, with
both improvements in capabilities  and lower costs.

4.3   Applications

Interceptor trenches have  a wide variety of environ-
mental applications, including hydrodynamic control and
dewatering,  plume capture, NAPL recovery, and vapor
extraction. Trenches also have been central to an inno-
vative remedial application that decontaminates ground
water in situ as it flows through reactive material.

4.3.1   Hydrodynamic Control

Interceptor trenches have been used extensively to con-
trol local, site-scale, hydraulic gradients (3,19, 20). One
site-scale application involves reducing the flow through
a contaminated area by pumping water from the up-
gradient side of the contaminated area and injecting the
water on the downgradient side (Figure 4-10). This es-
sentially produces a stagnant area in the region between
the trenches, which inhibits movement of contaminants.

Interceptor trenches can also be used in conjunction
with  barriers around a site to control offsite migration
(19, 20). In this case, barriers placed around the site and
the interceptor trenches reduce piezometric levels within
the barriers. Vertical wells can supplement this applica-
tion, which is similar to traditional dewatering applica-
tions for trenches. The barriers should be keyed  into
underlying an aquitard, if possible.

In some cases, using barriers can restrict flow into  one
side of the trench (21),  or can divert ground water mov-
ing under a regional gradient away from a contaminant
source (3). This configuration (Figure 4-11) is used when
clean surface water (e.g.,  lake or river) provides signifi-
cant ground-water recharge within the area influenced
by the trench. This application is most effective when the
barrier is keyed into a low permeability formation so that
ground water does not  travel underneath the barrier.
             Interceptor Trench
                                   Barrier
Figure 4-11.  Interceptor trench coupled with a hydraulic barrier
           to  minimize  recharge from an adjacent surface
           waterbody.

4.3.2  Plume Interception


The  classic, common application of a trench  involves
placing a trench perpendicular to the  regional  gradient
on the downgradient side of a plume  to intercept con-
taminated water (3, 6,12,15). Figure 4-12 illustrates the
concept. The application serves the same function as a
line of wells but is more reliable, because water may
escape between the wells in the presence of permeable
zones unaccounted for during design  or as a  result of
other factors. This application, therefore, is useful when
high reliability is required,  such as  when  protecting a
                                                        Well

                                                         O
                                                                         Property Boundary
                                                                   Trench
Figure 4-10.  Controlling the flow through a contaminated area
           by pumping from an upgradient trench and inject-
           ing into a downgradient trench.
Figure 4-12.  Trench used to capture plume and prevent migra-
           tion to an offsite well.
                                                   73

-------
downgradient well. Installing the trench, however, by no
means guarantees that it will totally intercept the plume.
Monitoring enough locations  of hydraulic heads in the
vicinity of the trench  should ensure that  the  flow is
inward toward the trench. A barrier on the downgradient
side of the trench helps to ensure the reliability of plume
capture.

The idealized trench is a single straight feature, such as
in Figure 4-12, but in practice many applications involve
creating networks of trenches oriented to best achieve
remedial objectives and  avoid obstacles to  excavation.
Networks may have several segments oriented normal
to regional flow to arrest migration,  or an irregular pat-
tern of contamination may dictate their location. In cases
where the ideal path  for  the trench is obstructed, a
trench network is used instead.

4.3.3  NAPL Recovery

The recovery of NAPL is one of the most widely used
applications  of interceptor trenches, because trenches
offer an effective method of ensuring capture and arrest-
ing  possible  offsite migration of NAPL. In cases where
NAPL is relatively shallow (LNAPL  in particular),  com-
monly available excavation equipment can be used to
install the trench.  Moreover, trenches can cut across a
significant depth  range,  making them  well suited to
NAPL recovery at sites where seasonal  hydrological
factors cause wide variations in piezometric levels.

The simplest trench design to recover DNAPL calls for
constructing  a trench on top of an underlying imperme-
able unit. The product collects in drain lines within the
trench, then flows laterally to sumps, where it is pumped
to the surface (Figure 4-13a).  Using two recovery (drain)
lines may enhance DNAPL recovery (22): one drain in
the DNAPL itself and the second drain in ground water
above the DNAPL drain (20). Pumping ground water
from the upper drain produces drawdown  of the water
and upwelling  or mounding of the product  (Figure
4-13b). Pumping both drains results  in drawdown in both
ground water and DNAPL (Figure 4-13c).

To enhance  DNAPL recovery, water removed from the
upper drain of the two-drain system can be reintroduced
to the system through  a nearby trench or drain  (Figure
4-14).  The reintroduction  of the water to  the  aquifer
system enhances the hydraulic gradient, further driving
the mobile DNAPL (20).

Caution is  required when pumping DNAPL recovery
systems; excessively high pumping  rates can cause the
drawdown of the  overlying ground  water to "pinch-off'
the DNAPL  plume (Figure  4-15).  This results in  a
marked decrease in the ratio of NAPL to water in the
recovery fluid, even though large volumes of NAPL re-
main  relatively close to the trench.
                 a.
                 Ground Surface
   b.
  Ground Water
   Ground-Water Drawdown,
      DNAPL Mounding
                  Ground-Water Drawdown,
                    DNAPL Drawdown
Figure 4-13.  Trench systems for DNAPL recovery: a) one-drain
           system, b)  two-drain system with upper drain
           pumping, and  c) two-drain  system  with both
           drains pumping (22).

Trenches are also widely used to recover LNAPLs (2, 3).
This application may make use of a single pump, or it
can  use a two pump system to reduce the upward
migration of the LNAPL/water interface. In this sense,
the LNAPL  application resembles the approach  de-
scribed  above for DNAPLs.

4.3.4   Vapor Extraction

Interceptor trenches have been used successfully to
recover VOC, including petroleum hydrocarbons  and
halogenated compounds (e.g., tetrachloroethane, 1,1,1-
trichloroethane, TCE). Vapor extraction with an intercep-
tor trench  requires an  impermeable trench  cap. In
practice, low-permeability soils from trench excavation
have capped the trench.  This requires compaction to
maximize the integrity of the seal. The combination of
           Ground Wate
             -~~»~—
             DNAPL

               I^ISP*
               Permeability U
                              Ground Water
Figure 4-14.
Enhanced recovery  of DNAPL  by reintroducing
water to increase the hydraulic gradient (20).
                                                  74

-------
                Ground Water Drawdown,
                   DNAPL "Pinch Off
Figure 4-15.  Excessive pumping rates from the ground-water
           and DNAPL drains causing "pinch off" of the
           DNAPL plume.

using an extremely low-permeability liner material (e.g.,
HOPE) along with the natural soils can result in a cap
with a sufficiently low permeability. Multiple  soil-liner
layers also  may  be sufficient. Another technique in-
volves using a bentonite slurry in the upper few feet of
the trench. Liner material, such as HOPE, may  also
supplement the bentonite slurry method.

To initiate vapor extraction, suction applied to perforated
casing within the permeable trench material induces air
flow from the adjacent soil  into the trench. This results
in vaporization of the VOCs. Air and contaminant vapors
are collected and delivered to secondary treatment sys-
tems, just as in applications using conventional wells.

The path of air flow through the soil is critical to this
application. To estimate the path, it is best to evaluate
the location of a source of air as well as possible het-
erogeneities in the site. If the air flows in from the ground
surface,  then much of the flow would follow a steeply
dipping path close to the trench. This effect is particu-
larly acute if the formation is homogeneous or contains
vertical fractures. Alternatively, if a  naturally occurring
low-permeability unit overlies  the contaminated area,
the effect is less serious. In the absence of such a  unit,
it may be possible to  add a cap to induce air through
more of the subsurface. A similar situation  exists  with
flow patterns of horizontal wells (e.g., Figure 2-19).

Using wells or another trench to provide a source of air
below the subsurface may augment flow through the
contaminated region. Perhaps the most effective imple-
mentation involves two  parallel trenches, with suction
applied to one trench and atmospheric or positive pres-
sure applied to the other.  Keep in mind, however, that
even with two trenches and a cap, heterogeneities be-
tween the trenches (such  as fractures or sand lenses)
may cause an uneven distribution of flow.

One effect of vapor extraction may be the production of
water or other liquid phases, either by draining perched
zones or lifting the water table. The presence of water
reduces pneumatic conductivity, and thus performance,
during vapor extraction. Therefore, if the site may pro-
duce water, the trench design should aim to recover the
water as  well as the vapor.  Using a sump  pump to
recover water while applying  suction with  a  blower
should accomplish this goal. Alternatively, a dual-phase
extraction system (23) can recover both water and other
liquid phases using suction from the blower.

Adding  air sparging can  enhance a vapor extraction
system   (24).  Injecting  air  into  VOC-contaminated
ground water enhances remediation by:

• Stripping the dissolved-phase VOCs.

• Increasing  dissolved oxygen levels to promote bio-
  degradation.

Design   guidelines  for   installation  of  a  coupled
sparge/vapor extraction system are similar to those for
standard vapor extraction  and ground-water pump and
treat systems. The lateral and vertical extent of contami-
nation as well as the nature of the soil affect the sparging
radius of influence, which controls the overall scale of
the system.  Individual design  parameters that these
factors affect  include depth and length  of the screen,
well construction, and air pressure. All these variables
are necessary to  predict  rates of VOC removal and
project duration. For a coupled sparge/vapor extraction
system, a  recommended  vacuum-to-injection ratio is
10:1 (24). This should meet the objective  of adequate
airflow into the contaminated  ground water without in-
creasing the  potential for  offsite migration of the con-
taminant.

Operation of a combined sparge/vapor extraction sys-
tem should include monitoring of the following parame-
ters (24):

• Injection: Air temperature, pressure, and air flow.

• Extraction: Temperature, vacuum, air flow, and con-
  taminant concentration.

4.3.5   Other Applications

Trenches have a variety of other environmental applica-
tions that are  beyond the  scope of this  document  be-
cause they are not directly  related to  delivery and
recovery. In many cases, trenches filled with low-perme-
ability materials form vertical barriers to flow (3, 19, 20).
Such barriers can enhance the control of fluids within a
site, in many cases in conjunction with recovery systems.
                                                   75

-------
Recently, trenches have been filled with reactive com-
pounds capable of degrading or adsorbing contami-
nants in situ (25, 26). This application involves placing
a trench filled with permeable material perpendicular to
a regional gradient so that natural flow sweeps contami-
nants through the trench.  Ideally, compounds  in the
trench remove the contaminants from the flow, and the
ground water is "clean" after it  passes through the
trench. Several variations of this concept have been
proposed or are currently under development (16, 27-30).

4.4   Case Histories
This section  provides an  overview of several  dozen
applications and describes five case histories  in which
trenches have been used for various environmental pur-
poses.

4.4.1  Overview
Tables 4-2 through 4-4 summarize information on repre-
sentative applications of interceptor trenches from pub-
lished  descriptions  and  case  histories  provided  by
consultants. Table 4-2 lists the general site character,
including contaminant,  site geology,  and  hydrologic
characteristics. Table 4-3 details the installed trenches,
including dimensions and fill material. Table 4-4 high-
lights operational parameters of the trenches and prod-
uct recovery.

4.4.2  Selected Examples
Following is a more detailed discussion of site  imple-
mentations of interceptor trench technologies.

4.4.2.1   LNAPL and Dissolved-Phase Recovery,
         Texas

At a former petroleum refinery located in southeastern
Texas (12), the predominant soil type was a massive,
laterally extensive  marine clay with very thin  silt and
sand  interbeds. Slug tests from the site indicated hy-
draulic conductivity from 9.3 x 10"6 to 2.5 x 10"4 centime-
ters  per second (0.0266 to  0.720 feet per day). The
saturated clay zone varied in thickness from 15.2 to 30.5
meters (50  to 100 feet). Depth to ground water typically
was  no more than 3 meters (10 feet). Average annual
precipitation was 88.9 centimeters (35 inches), and the
regional ground-water gradient was 1.7 x 10"3. Contami-
nation consisted of petroleum hydrocarbons, which ex-
isted both in  the dissolved  phase and as free-product
phase on ground water. The total free-product  plume
was estimated to contain more than 1,514 cubic meters
(400,000 gallons)  and  to cover approximately 8,094
square meters (87,120 square feet).
The trench  was oriented  normal to the regional ground-
water gradient and located near the downgradient end
of the free-product plume. It was constructed using a
backhoe and was backfilled with pea gravel. The trench
was 0.9 to 1.2 meters (3 to 4 feet) wide, 3.7 to 4.3 meters
(12 to 14 feet) deep, and 4.7 meters (16 feet)  long. A
30.5-centimeter (12-inch) diameter recovery  well was
located  in the center of the trench, which was deepened
to a depth of 5.5  meters (18 feet) in the vicinity of the
well. Adding a low-permeability cover minimized infiltra-
tion of surface waters.

After 2 years of operation, approximately 9.8 cubic me-
ters (2,600 gallons) of free-product hydrocarbons were
recovered. With a product-to-water ratio of about 1:91,
the trench  recovered approximately 53.9 liters (14.24
gallons) of petroleum hydrocarbons per day.

4.4.2.2   LNAPL and Dissolved-Phase Recovery,
         Westminster, Colorado

At a municipal service center in Westminster (15), the
site geology consisted of 3 to 3.7 meters (10 to 12 feet)
of sandy clay underlain  by interbedded sandstones and
claystones. Ground water occurred at depths between
3.7 to 6.1 meters (12 to 20 feet) during the operation. In
situ slug tests of the upper 3 meters  (10  feet) of the
saturated zone indicated an average hydraulic conduc-
tivity of 7  x 10"5  centimeters per  second.  Regional
ground-water gradient was approximately  0.01.  Con-
tamination at the site was free-product gasoline  floating
on ground water. The gasoline plume covered an area
91.4 to  122  meters (300 to 400 feet) in width and 122
meters (400 feet) in length that was estimated to contain
189 to 379 cubic meters (50,000 to 100,000 gallons) of
product (Figure 4-16). A gasoline vapor plume extended
15.2 meters (50 feet) beyond the limits of the liquid
plume.

The trench was located nearly normal to the contami-
nant plume and was constructed using a backhoe with
a sliding trench box. The trench was 0.9 to 1.2 meters
(3 to 4 feet) wide, 5.5 to 6.1 meters (18 to 20 feet) deep,
and 183 meters (600 feet) long. The base of the trench
contained 10-centimeter (4-inch) perforated pipe, which
sloped at a grade of 0.01 toward a corrugated metal pipe
sump in the  middle of the trench. The trench was back-
filled with 1.9-centimeter (3/4-inch) gravel to within 2.4
meters (8 feet) of the ground surface (688 cubic meters
[900 cubic yards] required); the remainder of the trench
was filled with materials excavated from the trench. In
addition to the  trench,  two 12.2-meter (40-foot)  deep
recovery wells were installed elsewhere in the plume.

The thickness of the original plume varied  from 0.6 to
1.8 meters (2 to 6 feet). After 1 year of trench operation,
most monitoring wells showed less than 15  centimeters
(6 inches) of product. Total flow of product  and  ground
water from  both  the trench  and  recovery wells was
approximately 7.6 cubic meters (2,000 gallons) per day,
and after 1 year total product recovery was estimated to
be 114  cubic meters (30,000 gallons). It will take  an
                                                   76

-------
Table 4-2. Summary of Applications of
Reference Site Name
Hydrodynamic Control
Piontek and Simpkin,
1992
Day and Ryan, 1992

Laramie Tie
Ohio
Trenches for Delivery or Recovery: Site Characteristics
Hydraulic
Contaminant Soil Type Conductivity

Creosote/pentachlorophenol
mixture (DNAPL)
Mixed-waste landfill
leachate

Alluvial sands
Glacial till/bedrock

0.25-1 0 cm/sec

Gradient Depth
<3 m


Ground-Water Remediation
Mast, 1991
Mast, 1991
Mast, 1991
Ganser and Tocher,
1988
Rawl, 1994
Rawl, 1994
Day and Ryan, 1992
Vapor Extraction
Barrera, 1993
Barrera, 1993
Barrera, 1993
Barrera, 1993
Site A
Site B
Site C
Westminster
Rinker Concrete
Langley AFB
Northern
California

Case #1
Case #2
Case #3
Case #4
Free product and dissolved
petroleum hydrocarbons
Dissolved petroleum
hydrocarbon
Free product and dissolved
petroleum hydrocarbons
Free product and dissolved
gasoline
Dissolved-phase PAHs
Free product and dissolved
jet fuel (JP-4)
"Spilled processing
chemicals"

VOCs (PCE, TCE, TCA)
BTEX, petroleum
hydrocarbons
Toluene
VOCs (acetone, MEK,
toluene, xylene, e-benzene)
Marine clay
Clay/sand/clay
Deltaic clay/sand
Sandy
clay/sandstone
Limestone/sandstone
Sands and silts
Clays, silts

Clay/sand and gravel
Clay/silt and fine
sand
Sand and silt
Silt and clay/sand
0.0001-0.00001
cm/sec
<0.0001 cm/sec
0.003-0.04
cm2/sec (T)
0.00007 cm/sec
>0.1 cm/sec




0.00001-0.0001
cm/sec

0.0001 cm/sec
0.0017 <3m
0.0625 1 m
0.0645 1-3 m
0.01 4-6.5 m
1 m



2.5 m
0.04 5-6 m
5-7 m
1-2 m
Combined Ground-Water Remediation and Vapor Extraction
Rawl, 1994
LNAPL Recovery
Hanford and Day,
1988
Day and Ryan, 1992
DNAPL Recovery
Meiri et al., 1990
Rawl, 1994
Sale and Applegate,
1994
Flow Barrier
Rawl, 1994
Law Center

San Jose
South Texas

Rock Creek
Gray Iron Works
Laramie Tie

Star Lake
Free product and dissolved
BTX

Free product diesel fuel
Free product waste oil

TCE
TCE
Creosote/pentachlorophenol
mixture

Free product diesel fuel
Sand and
clay /granite

Sands, silts, clays


Glacial till/bedrock
Silty clay glacial till
Alluvial sands

Sands, silts, clay
0.000007-0.03
cm/sec




0.00002-0.00008
cm/sec
<0.0003 cm/sec
0.25-1 0 cm/saec




9-11 m
Shallow

0.02 1-2 m

<3 m

4.5 m
77

-------
Table 4-3.  Summary of Applications of Trenches for Delivery or Recovery: Trench Parameters
Reference              Site Name         Shape           Length       Width            Depth
Fill Material
Hydrodynamic Control
Piontek and
Simpkin, 1992
Day and Ryan,
1992

Laramie Tie
Ohio

Closed loop 5,100 m
Linear (2) 1 m


14 m

Drainline with
bentonite barrier wall
Sand, geotextile,
bentonite barrier wall
Ground-Water Remediation
Mast, 1991
Mast, 1991

Mast, 1991
Ganser and Tocher,
1988
Rawl, 1994

Rawl, 1994
Day and Ryan,
1992
Vapor Extraction
Barrera, 1993
Barrera, 1993
Barrera, 1993
Barrera, 1993
Combined Ground-Water
Rawl, 1994
LNAPL Recovery
Hanford and Day,
1988
Day and Ryan,
1992
DNAPL Recovery
Meiri et al., 1990
Rawl, 1994
Sale and
Applegate, 1994
Flow Barrier
Rawl, 1994
Site A
Site B

Site C
Westminster
Rinker
Concrete
Langley AFB
Northern
California

Case #1
Case #2
Case #3
Case #4
Remediation
Law Center

San Jose
South Texas

Rock Creek
Gray Iron
Works
Laramie Tie

Star Lake
Linear 5m 1m
Linear:
N-S 48 m 1m
E-W 45 m 1m
Linear 9m 1m
Linear 180 m 1m
Linear 15m 0.3 m

Linear (19) 1,000 m 0.3 m
Linear 0.75 m

Linear (5) 150 m
Linear (8)
Linear (9)
Linear (8)
and Vapor Extraction
Linear (7) 360 m 0.3 m

Linear (2) 180 m 1-1.5 m
Linear 400 m

Model only
Linear (4) 250 m 0.3 m
Linear (12) 670 m

Curvilinear 400 0.3 m
3-4 m

3 m
3 m
3.2 m
5-6 m
3 m

6 m
9 m

2.5 m
3 m
5 m
1 m

5-7 m

1 3-1 4 m
6 m


7 m


4.5 m
Pea gravel

Pea gravel
Pea gravel
Pea gravel
3/4-in. gravel
Sand

Sand
Gravel with
geotextile, slurry wall

Pea gravel
Pea gravel
Perforated sheet
piling
Pea gravel



Gravel over sandy
gravel
Geomembrane panels


Sand
Perforated drain only

40-mil HOPE
                                                          78

-------
Table 4-4.  Summary of Applications of Trenches for Delivery or Recovery: Product Recovery
Reference            Site Name         Pump/Well Type       Pump/Well Location    Pump Rate
Product Recovery
Hydrodynamic Control
Piontek and
Simpkin, 1992
Day and Ryan,
1992
Laramie Tie

Ohio


Recovery wells


In trench


5 gal/min per
trench
Ground-Water Remediation
Mast, 1991
Mast, 1991
Mast, 1991
Ganser and
Tocher, 1988
Rawl, 1994
Rawl, 1994
Day and Ryan,
1992
Vapor Extraction
Barrera, 1993

Barrera, 1993
Barrera, 1993
Barrera, 1993

Site A
Site B
Site C
Westminster
Rinker Concrete
Langley AFB
Northern
California

Case #1

Case #2
Case #3
Case #4

Combined Ground-Water Remediation and
Rawl, 1994
LNAPL Recovery
Hanford and Day,
1988
Day and Ryan,
1992
DNAPL Recovery
Meiri et al., 1990
Rawl, 1994
Sale and
Applegate, 1994
Flow Barrier
Rawl, 1994
Law Center

San Jose
South Texas

Rock Creek
Gray Iron Works
Laramie Tie

Star Lake
12-in. recovery well
12-in. recovery well
24-in. recovery wells
Metal sump
8-in. recovery well
8-in. recovery well
12-in. recovery well

Vacuum

Vacuum
Vacuum
Vacuum

Vapor Extraction
8-in. recovery well

12-in. recovery well
Recovery wells
with skimmers

N/A
8-in. recovery well


N/A
Middle of trench
Bend in trench

Middle of trench
End of trench
End of trench
328-ft intervals in
trench

Out of trench

Out of trench
Out of trench
Out of trench




100-ft intervals in
trench
Wells in trench

N/A
End of trench


N/A
9 gal/min 14 gal/day
0.5 gal/min
0.033 gal/min
1 .4 gal/min 30,000 gal, first
12 months
0.05-0.16 gal/min None, below MCL
Not yet on line


1 5 HP/1 ,000 cfm 1 00 Ibs, first 30
days
15 HP/500 cfm 270 Ibs/day

10 HP/250 cfm 25 Ibs, first 120
days

30 gal/min




N/A N/A
5 gal/ min
280,000 gal
(1992-1993) (50%
after 30 days;
90% after 90 days)

N/A N/A
                                                       79

-------
                                      Underground Gasoline Storage Tanks
                                                  Extent of Free Product
                                                                       Monitoring Wells
                   Interceptor Trench
                                              100ft
Figure 4-16.  Plan of the Westminster gasoline recovery site (15).
estimated 5 years to recover 50 to 70 percent  of the
product (15).

4.4.2.3   DNAPL Recovery,  Laramie, Wyoming

At a former wood treating facility (31) in Laramie, the site
geology consisted of approximately 2.4 meters (8 feet)
of alluvial sediments underlain  by fine-grained  sand-
stone.  Surficial alluvial sediments of sands, silts, and
clays graded downward to coarse sands and fine grav-
els at the base. The lower several feet of the alluvial
aquifer were contaminated with wood-treating oil (Figure
4-17),  a DNAPL  that sank below the surface  of the
ground  water.  The  initial residual  oil concentration
ranged from 5,000 to 100,000 ppm in soil boring sam-
ples  (32).
An initial field demonstration was conducted in 1988 to
test the effectiveness of product recovery.  The demon-
stration used  horizontal drains for waterflooding fol-
lowed  by chemical flooding.  Two 4.6-meter (15-foot),
parallel horizontal drains were installed 4.6 meters (15
feet) apart, and a  sheet piling wall separated them from
the surrounding alluvial aquifer. During this demonstra-
tion,  approximately 6.1 cubic meters (1,600 gallons) of
oil were recovered. This was estimated to be 94 percent
of the oil in the test cell.
In 1989, a second field demonstration  was conducted.
A 39.6-meter (130-foot) square sheet pile  wall isolated
this test cell from the  surrounding aquifer. Three 10-cen-
timeter (4-inch) horizontal drains were installed; these
drains  were oriented parallel, spaced 18.3 meters (60
feet) apart, and located at the aquifer-bedrock contact.
The initial volume of wood treating oil was estimated to
be approximately  259 cubic meters (68,500 gallons). An
initial waterflood removed the mobile  phase of the oil;
recovery from this step was 151 cubic meters (40,000
                                  100 meters
  Sheet Pile
Figure 4-17.  Plan of the Laramie DNAPL recovery site (20).

gallons.) Chemical flooding removed the residual phase
oil; this step removed an  additional 90.8 cubic meters
(24,000 gallons) of oil. In total, the horizontal drain sys-
tem coupled with water and chemical flooding removed
242 cubic meters (64,000 gallons) or 93.5 percent of the
initially estimated product  reservoir.
                                                    80

-------
Full field implementation and system operation began in
1991 with installation of 12 oil recovery units. Drainlines
in  the system were 54.9 meters  (180 feet) long, and
spaced 27.4 meters (90 feet) apart. During 1991, three
units were operated that recovered 227 cubic meters
(60,000 gallons) of oil. From the summer of 1992 to the
summer of 1993, all 12 units were operated, recovering
1,136 cubic meters (300,000 gallons) of oil.  Approxi-
mately 3.79 x 10s cubic meters  (100 million gallons) of
water circulated through the alluvial aquifer.

Production records from the 1992-1993 pumping period
indicate that 50 percent of the mobile oil was recovered
after approximately 30 days of system operation, and 90
percent recovery  was  achieved in 90 days. Pumping
beyond 130 days produced only minor increases in oil
recovery.

Total recovery from all pilot studies and the first 2 years
of system  operations was  approximately 3,407  cubic
meters (900,000 gallons) of oil.

4.4.2.4   Vapor Recovery, Illinois

In  northeastern Illinois  (24), the site geology consisted
of clayey surface sediments underlain by a plastic clay
that contained  trace coarse-grained materials. A 0.6-
meter (2-foot) thick sand and gravel unit occurred near
the water table  at a depth of about 2.4 meters (8 feet).
Contamination at the site included VOCs and semivola-
tile organic compounds (SVOCs) in the soil and ground
water (Figure 4-18).

Initial investigation indicated that dissolved phase VOCs
could be  removed  from the aquifer by air sparging;
followup  investigations concluded that because of lat-
eral geologic  heterogeneities, sparging was not a viable
option unless the extraction wells  could be placed very
near the seasonal high water level. The sand and gravel
layer located  at the water table would serve as a path-
way for VOC  removal by vapor extraction.

The implemented system consisted of nine air sparging
wells and five vapor extraction trenches with a cumula-
tive length of 168 meters (550 feet). The trenches, which
were excavated by backhoe, were 0.9 to 1.2 meters (3
to  4  feet) wide, and 2.1  meters (7 feet) deep (Figure
4-18). Installation  of the sparging wells and extraction
trenches  required about 3 weeks. On the basis of air-
flow analysis, a 10 horsepower, 5.7 cubic meters (200
cubic feet)  per  minute  blower was  selected for
sparging, and  a  15 horsepower, 28.3  cubic meters
(1,000 cubic  feet) per minute blower was selected for
vacuum extraction. Cost for the system as described
was less than $190,000.

The sparging/vapor extraction system removed  more
than 45.4 kilograms (100 pounds) of VOCs in the first
30 days of operation and reduced VOC concentrations
by 75 percent in the first 4 months of operation
     VOC Plume Limit
                                      Regional Flow
                                        30ft
                                  Building
                               Monitoring Well
Figure 4-18.
              Access Pipe

                  Compacted Natural Soil
                 4-in. HOPE Slotted Casing
                 With Filter Sock

                  Naturally Washed, Rounded
                  Pea Gravel
Map of the northeastern Illinois vapor extraction
site and detail of trench completion (24).
4.4.2.5   TCE Recovery, Modeling, and Field
         Results, Ohio

Modeling of ground-water flow at a site can be a valu-
able tool during the design of a delivery or recovery
system. Therefore, this case study presents site charac-
terization, predictive numerical modeling to support the
design of an interceptor trench system, field data from
the installed system, and a comparison of the predicted
and observed results. The material presented here has
been taken from the following references: Krishnan and
Siebers, (33); Meiri and others, (34); Weston, (35-37);
and Woodward-Clyde Consultants (33-38).

Geology. The site, located  in northeastern  Ohio, in-
cluded two adjacent properties totalling  approximately
13 acres (33). The site geology (Figure 4-19) consisted
of an upper fractured till unit (upper aquifer  0  to 2.1
meters  [0 to  7 feet] below ground surface), a lower till
aquitard (2.1 to 3.7 meters  [7 to 12 feet] below ground
surface), weathered shale bedrock (lower aquifer, 3.7 to
8.5 meters [12 to 28 feet] below ground surface), and
competent shale bedrock (34). Depth to ground water in
                                                   81

-------
                                      Shallow Aquifer
                                      Aquitard
                                       Lower Aquifer
   28ft
       I Competent Shale =
Figure 4-19. Typical subsurface profile (34).

the upper unconfined aquifer ranged from 0.6 to 1.5
meters (2 to 5 feet).

Regarding the site's hydraulic profile, slug tests of wells
screened in the upper aquifer indicated  hydraulic con-
ductivities of 2 x 10"5 to 8 x 10"4 centimeters per second
(0.065 to 2.4 feet per day); porosity was estimated to be
0.30.  Regional ground-water gradient in the upper aqui-
fer, determined  from water levels  in  screened  wells,
averaged about 0.02, with flow to the west. Depth to the
piezometric surface of the lower confined aquifer was
slightly more than  0.6 meters (2 feet). Pump tests on a
well screened in the lower aquifer indicated a transmis-
sivity  of about 4.1 square  meters (13.5 square feet) per
day, or a hydraulic conductivity of about 0.3 meters (0.85
feet)  per day (assuming  an aquifer thickness of 4.9
meters  [16 feet]). Lower aquifer  storativity was esti-
mated to be 3 x 10"4. Regional gradient in the lower
aquifer,  based on data from three deep wells, was cal-
culated  to  be 0.01, also to the west. Pump tests using
adjacent wells screened in the upper and lower aquifers
indicated that the two aquifers were separated hydrau-
lically; the vertical  hydraulic conductivity of the aquitard
was calculated to be 3 x 10"8 centimeters per second (8
x10'5feet per day) (34).

Distribution of Contaminants. Contamination at the site
included TCE and heavy metals  in  soil.  The ground
water contained TCE, tetrachloroethene, trans-dichlo-
roethene,  1,1-dichloroethene,  vinyl  chloride,   1,1,1-
trichloroethane,  ethylbenzene, and  xylene (33).  TCE
was the  principal ground-water  contaminant. The con-
taminants apparently  leaked from concrete silos (Figure
4-20).

As of 1984, TCE in the upper till aquifer had  migrated
45.7 to 76.2 meters (150 to 250  feet) from the probable
source  area (Figure  4-21). It was assumed that the
migration had occurred over an 8- to 10-year  period,
suggesting an average plume front velocity of about 4.6
to 9.1 meters (15 to 30 feet) per year. Average ground-
water flux (based on  an average hydraulic conductivity
of 0.37 meters [1.2 feet] per day, a  regional gradient of
0.02,  and a porosity  of 0.30) was estimated to be 9.1
                            Regional Gradient
                            Shallow Aquifer
           Monitoring Well

           Storage Silos

           Probable Source Area
Figure 4-20.  Site map showing locations of monitoring wells,
           storage silos, and probable contaminant source
           area (34).
                                           Regional
                                            Flow
            400ft
                                Probable Source

                                1984 Plume Limit

                                1986 Plume Limit
            Prescribed Head Model Boundary

            No Flow Model Boundary

            Monitoring Wells
Figure 4-21.  Extent of TCE plume in 1984 and 1986; also shown
           is the  ground-water  model  domain, with  pre-
           scribed (fixed) head and no-flow boundaries (34).
                                                    82

-------
meters (30 feet) per year. This indicated a retardation
factor between ground-water velocity and TCE plume
front velocity of 1 to 2. Comparison of plume front posi-
tion  between  1986 and  1984 indicated  a retardation
factor of 1.2 to 2. Based on the 1986 data,  the TCE
plume covered an area of approximately 12,077 square
meters  (130,000  square feet).  Assuming a  uniform
plume thickness of 1.5 meters (5 feet) and a porosity of
0.30, the volume of the plume was approximately 5,522
cubic meters (195,000 cubic feet)  (34). Data from the
deep weathered-shale aquifer were insufficient to define
the extent of the TCE plume.

System Design. Vertical  extraction wells  and  trenches
were considered as remedial alternatives. The vertical
extraction well system was deemed impractical because
modeling indicated that discharge rates would be on the
order of 0.4  liters (0.1 gallons) per minute per well. This
low rate would limit the radius of influence to a few feet and
would require several thousand wells for remediation.

Drains installed in shallow trenches also were consid-
ered  as  a   remedial   alternative.  Two-dimensional
ground-water  flow and contaminant  transport models
were used to predict water table  drawdown, trench dis-
charge, and TCE concentration in the extracted water.
The  two-dimensional coupled flow  transport models
used both advective and dispersive contaminant migra-
tion  and  retardation. Model  boundary conditions were
based  on surface topography and  configuration of the
water table  surface, with  boundaries selected beyond
the expected region of influence of the trench system
(Figure 4-22). The base of the uppertill was selected as
the bottom of the  aquifer. A finite  element mesh was
established, with 212 node points and 204 elements. All
existing shallow monitoring wells were located at nodes.

The flow model was calibrated using empirical data with
an assumption of steady-state conditions. Calibration
was  completed by varying precipitation recharge, initial
boundary head values, and hydraulic conductivity. The
predicted heads in the final calibrated model differed from
the observed field values by 0.3 meters  (1 foot) or less.

Proposed trench  locations were near the plume front
and  along the plume center at a depth of 2.1 meters
(7 feet) below ground surface (Figure  4-23a). Operation
of the trench was simulated as a transient (non-steady-
state) system for 420 "model" days. Beyond 420 days,
water level and flow rate  changes with time were insig-
nificant. Predicted trench  discharge rates dropped from
an initial  value of approximately 37.9  liters (10 gallons)
per minute  to a steady-state  rate of about 3.8 liters
(1 gallon) per minute.

Then, a preliminary transport analysis used the flow
model to estimate  changes  in TCE concentration
resulting from operation of the trench system. Assumed
TCE-related  transport  model  parameters  were:
        Model Domain
                      t^__A/  Site Boundary
Figure 4-22.  Model domain showing location of the interceptor
           trench, the 1986 TCE plume limit, and extent of the
           capture zone (34).
    Observed -1986
                             Calculated -130 months
                                     0  400 ft
          Calculated - 286 months
  TCE>10ppb
                    Interceptor
                    Trench
Capture
Zone
Figure 4-23.  TCE  concentration distribution  through  time
           (Note: TCE  10 ppb outside of capture zone.)
retardation factor = 1.5, longitudinal dispersivity = 9.1
meters (30 feet), and transverse dispersivity = 3 meters
(10 feet). Distribution of TCE in 1986 was used  as the
initial concentration for the model. The model simulated
                                                   83

-------
TCE movement for 8,580 days (23.5 years). Results of
the preliminary simulation indicated an overall reduction
in  TCE concentration and predicted locations of con-
tamination during the recovery system operation (Figure
4-23). The results also indicated, however, that  some
marginal areas of the TCE plume could avoid capture
as advective flow and dispersion moved it beyond the
trench capture zone.

The transport modeling  indicated that the  operation
would require approximately 24 years to decrease aver-
age discharge TCE concentrations to about 8 ppb (from
an initial value of about 1,550 ppb). This would require
the removal of approximately 9 pore  volumes of con-
taminated ground water. Because the analysis assumed
rapid TCE desorption from soil,  no TCE transformation
or degradation, and  other simplifying  factors, the esti-
mates  presented are rough and  to be used only for
planning purposes.

Implementation. Plans for constructing and implement-
ing the interceptor trench  (Figure 4-24) were developed
in  1987 (38). Designs called for all trench segments to
be approximately 2.1 meters (7 feet) deep to fully  pene-
trate the shallow aquifer.  The initial trench system was
approximately 565 feet in total length  and consisted of
two  trench segments 2.1 meters  (7  feet) deep, with
separate sumps (designated Martin and Henfield sumps
on Figure 4-24).

In  addition to a depth of 2.1 meters (7 feet), each trench
segment was  0.9 meters (3 feet)  wide. A filter  fabric
(geotextile) lined the excavation. Then, 15 centimeters
(6  inches) of 1.9-centimeter (3/4-inch) gravel was placed
at  the bottom of the trench followed by a 15-centimeter
(6-inch) diameter, continuously perforated PVC  drain
pipe (Figure 4-25).  The  trench  was backfilled with
1.9-centimeter (3/4-inch) gravel to within about 0.3 me-
ters (1  foot) of the ground surface.  The remaining 0.3
meters  (1  foot) of backfill consisted of natural soils to
minimize infiltration of surface waters.

Results. The ground-water extraction system began op-
eration  in  January 1989.  Figure 4-26  summarizes the
ground-water  extraction   from  the  two interceptor
trenches in the shallow, upper till aquifer from Septem-
ber 1989 through September 1993.  In general, several
observations could be noted  for this period:

• Data  from monitoring piezometers indicated that the
  regional flow in the shallow aquifer  was north-north-
  west, rather than to the west as assumed during the
  modeling phase of the  project.

• Temporal variations were observed in the concentra-
  tion of total VOCs  (monthly averages typically varied
  from  1,000 to 6,000 ppb),  but no consistent trend of
  decreasing concentration through time seemed  to oc-
  cur. On the basis  of an initial "best fit" first-order rate
  constant,  it was estimated that lowering VOCs to a
              Martin Sump
                                Regional Gradient
                                Shallow Aquifer
           Monitoring Wells

           Storage Silos

           Probable Source Area
Figure 4-24.  Site plan showing locations of initial interceptor
           trench segments (38).
Figure 4-25.  Schematic cross section of trench construction
           (38).
  residual concentration of 20 ppb would take approxi-
  mately 15 years (an extraction  of approximately 1.2
  x 10s cubic meters [32 million gallons]).


• The  property containing the storage silos appeared
  to be the principal region of  contamination  of the
  shallow (upper till) aquifer, but insufficient data ex-
  isted to generate a valid spatial model.
                                                   84

-------
   80,000
         Sept89  Mar90  SeptQO  Mar91  Sept91  Mar92  Sept92 Mar93  Sept93
          Key:  a Martin Sump
              * Henfield Sump
The actual trench system is recovering 155 percent of
the predicted value. These differences probably result
from  parameter estimates  that the  modeling  used,
including:

• Rate of recharge by precipitation (model  value  ap-
  pears to underestimate the actual value).

• Hydraulic conductivity  of the shallow aquifer (model
  value appears to underestimate the actual value).
• Regional gradient (a different direction as indicted by
  the observed data).
Additional examination of the data could possibly yield
additional causes for this difference.
Figure 4-26.  Monthly extraction from interceptor trench sumps.    4.5  References
• The initial interceptor trench system was insufficient
  to capture the VOC plume. Consequently, in Decem-
  ber  1991  additional trench segments were consid-
  ered, followed by a recommendation in December
  1992 to construct these segments on  an  adjacent
  property. In October 1993, the design of the additional
  trench segments was completed, and the  trenches
  were installed during the winter of 1994.

• As of September 1993, approximately 40.8 kilograms
  (90 pounds) of VOCs were treated at the site.

For the first  year of operation of the interceptor trench
system, the predicted extraction (from modeling)  ex-
ceeded the  actual extraction. After the first year,  the
actual extraction exceeded the predicted volume. Figure
4-27  compares  the actual and  predicted  volumes. A
linear regression through the observed data and fixed at
the origin yielded an average flow rate of about 67,000
gallons per month. In contrast,  a linear model through
the steady-state portion of the predicted data (after the
first 8 months) yielded an average flow rate of approxi-
mately 164 cubic meters (43,200 gallons) per month.
  3,500,000


  3,000,000


  2,500,000

(A
_g 2,000,000
"ro
CD
  1,500,000


  1,000,000


   500,000
                Observed Discharge
         »Tl I I I I
        Sept89  Mar90 Sept90  Mar91  Sept91  Mar92  Sept92  Mar93  Sept93


Figure 4-27.  Observed and predicted  discharges from  trench
           system.
 1. Gilbert,  S.G.,  and J.J.  Gress. 1987.  Interceptor
    trenches for positive ground-water control. Ground
    Water Monitor. Rev. 7:55-59.

 2. Hanford, R.W., and S.R.  Day.  1988. Installation of
    a deep  drainage trench  by the biopolymer slurry
    drain technique. Proceedings of the National Water
    Well Association Outdoor Action Conference, Las
    Vegas, NV.

 3. Day, S.R., and C.R. Ryan. 1992. State of the art in
    biopolymer drain construction.  In:  Paul, D.B., R.R.
    Davidson, and N.J. Cavalli, eds. Slurry walls: De-
    sign, construction, and quality  control. ASTM STP
    1129.  Philadelphia, PA: American  Society for Test-
    ing  and  Materials.

 4. LaGrega, M.D., PL. Buckingham,  and J.C. Evans.
    1994. Hazardous waste  management. New York,
    NY: McGraw-Hill.

 5. Farrell, M. 1993. Breaking new ground in ground-
    water remediation. Hazmat World  (January).

 6. Rawl, G. 1994. Personal  communication. Horizon-
    tal Technologies, Inc.,  West Palm  Beach, FL.

 7. Soils.  1993. Heres how it works:  Horizontal wells
    linear contaminant recovery system. Soils (Novem-
    ber), p. 48.

 8. Bauer Spezialtiefbau.  1994. Diaphragm and cutoff
    walls.  Personal communication with Badwi Khouri,
    Coastal   Caisson  Corporation,   Clearwater,  FL
    (March).

 9. Begor,  K.F., M.A.  Miller,  and  R.W Sutch. 1989.
    Creation  of an artificially  produced fractured zone
    to  prevent  contaminated  ground-water migration.
    Ground Water 27(1):57-65.

10. Beljin, M., and L.C. Murdoch. 1992. Analytical mod-
    els for interceptor trenches.  Final report. EPA Con-
    tract No. 68-C9-0031-WA1.
                                                   85

-------
11.  Murdoch, L.C. 1994. Transient analysis of an inter-
    ceptor trench. Water Resour. Res. In press.

12.  Mast,  M.K. 1991. The prediction of the  effective-
    ness of interceptor  trenches in the remediation of
    ground-water contamination by petroleum hydro-
    carbons. M.S. thesis. Texas A&M University.

13.  Zheng, C., K.R. Radbury, and M.P.Anderson. 1988.
    Role of interceptor ditches in limiting the spread of
    contaminant  in  ground   water.  Ground  Water
    26:734-742.

14.  Chambers, L.W, and  J.M. Bahr. 1992. Tracer test
    evaluation of a drainage ditch capture zone. Ground
    Water 30:667-675.

15.  Ganser, D.R., and R.J. Tocher.  1988. Design and
    construction of a subsurface gasoline recovery sys-
    tem. Presented  at FOCUS Conference on South-
    western Ground-Water Issues, Albuquerque, NM.
    pp. 479-486.

16.  Gilham, R.W, and D.R. Burris. 1992. Recent devel-
    opments  in permeable in  situ treatment walls for
    remediation  of contaminated ground water.  Pro-
    ceedings of the Subsurface  Restoration Confer-
    ence,    Third   International   Conference   on
    Ground-Water Quality Research, Dallas, TX.

17.  Keely, J.F. 1984. Optimizing pumping strategies for
    contaminant studies and remedial actions. Ground
    Water Monitor. Rev. 4(3):36-73.

18.  Repa, E., and C. Kufs. 1985. Leachate plume man-
    agement. EPA/540/2-85/004.

19.  Piontek, K.R., and T.S. Simpkin. 1992. Factors af-
    fecting  the practicability of in situ bioremediation at
    a wood preserving  site. Proceedings of the 84th
    Meeting of the Air and Waste Management Asso-
    ciation, Kansas City, MO.

20.  Sale, T.C., and D. Applegate. 1994. Oil recovery at
    a former wood-treating facility. Proceedings of the
    Innovative Solutions for Contaminated Site Man-
    agement  Conference  of the Water Environment
    Federation, Miami, FL.

21.  U.S. EPA.  1985. Leachate plume  management.
    EPA/540/2-85/004.

22.  Huling,  S.G., and  J.W  Weaver.  1991.  Dense
    nonaqeous phase liquids. EPA Ground Water Issue.
    EPA/540/4-91/002.

23.  American Petroleum Institute. 1989. A guide to the
    assessment and remediation of underground petro-
    leum releases, 2nd  ed. API 1628. Washington, DC
    (August).
24. Barrera, J.A. 1993. Air sparging and vapor extrac-
    tion as a means of removing chlorinated and BTEX
    compounds in  complex ground-water conditions.
    Proceedings of the Petroleum Hydrocarbons and
    Organic Chemicals in Groundwater Conference,
    Houston, TX.

25. Starr, R.C., and J.C. Cherry. 1993. Funnel-and-gate
    system directs plumes to in situ treatment. Ground-
    water currents. EPA/542/N-93/006.

26. Starr, R.C., and J.C. Cherry. 1994. In situ remedia-
    tion of contaminated ground water: The funnel-and-
    gate system. Ground Water. In press.

27. Pankow, J.F, R.L. Johnson, and J.A. Cherry. 1993.
    Air  sparging in gate  wells  in cutoff walls  and
    trenches for control of plumes and volatile organic
    compounds (VOCs). Ground Water 31:654-663.

28. Devlin,  J.F, and J.F. Barker. 1994.  A semipassive
    nutrient injection scheme for enhanced in situ biore-
    mediation. Ground Water. In press.

29. Blowes, D.W, and  C.J. Ptacek. 1992. Geochemical
    remediation of ground water by permeable reactive
    walls: Removal of  chromate by reaction with  iron-
    bearing solids. Proceedings of the Subsurface Res-
    toration Conference, Third International Conference
    on Ground-Water Quality Research, Dallas, TX.

30. Gilham, R.W,  S.F. OHannesin, and  WS. Orth.
    1993. Metal-enhanced abiotic degradation of halo-
    genated aliphatics: Laboratory tests and field trials.
    Proceedings of the HazMat  Central  Conference,
    Chicago, IL.

31. Pitts, M.J., K. Wyatt, T.C. Sale, and K.R. Piontek.
    1994. Utilization of chemical-enhanced oil recovery
    technology to remove hazardous oily waste from
    alluvium.  Proceedings of the  Innovative Solutions
    for Contaminated Site Management Conference of
    the Water Environment Federation,  Miami, FL.

32. Union  Pacific Railroad. 1990. Laramie tie plant: In
    situ treatment process development program. Mile-
    stone IV report (appendices).

33. Krishnan, P., and D. Siebers. 1989. Ground-water
    remediation at a Superfund site. HMCRI.

34. Meiri, D., M. Ghiasi, R. Patterson, N. Ramanujam,
    and M. Tyson. 1990.  Extraction of TCE-contami-
    nated  ground water by subsurface drains and a
    pumping well. Ground Water 28:17-24.

35. Weston. 1991. Annual evaluation report for Old Mill
    site, Rock Creek, OH. Submitted to U.S. EPA, Re-
    gion V,  Chicago, IL (December).

36. Weston. 1992. Second annual evaluation report for
    Old Mill site, Rock Creek, OH.  Submitted to  U.S.
    EPA, Region V, Chicago, IL (December).
                                                 86

-------
37.  Weston. 1993. Third annual evaluation report for   38. Woodward-Clyde  Consultants.  1987.  Old  Mill
    Old Mill site, Rock Creek, OH.  Submitted to U.S.       ground-water extraction system: Henfield, revision
    EPA, Region V, Chicago, IL (October).                  2. Submitted to U.S. EPA, Region V, Chicago, IL.
                                                 87

-------