United States        Office of Policy,       March 1985
           Environmental Protection     Planning and Evaluation
           Agency          Washington, DC 20460
           PoNcy Planning and Evaluation
vvEPA     Assessment of Incineration
           As A Treatment Method for
           Liquid Organic Hazardous
           Wastes
           Background Report IV:
           Comparison of Risks from
           Land-Based and Ocean-Based
           Incineration

-------
COMPARISON OF RISKS FROM LAND-BASED
AND OCEAN-BASED INCINERATION
March 1985
                      x
A background report for the study by
EPA's Office of Policy, Planning and
Evaluation:  "Assessment of Incineration
As A Treatment Method For Liquid Organic
Hazardous Waste."
Prepared by:

Industrial Economics, Inc.
2067 Massachusetts Avenue
Cambridge, Massachusetts  02140

in association with

Applied Science Associates, Inc.,
Arthur D. Little, Inc., and
Engineering Computer Optecnomics, Inc.
Prepared for:

Office of Policy Analysis
U.S. Environmental Protection Agency
Washington, D.C.  20460

EPA Project Officer:  Dr. Jeff Kolb

-------
                        TABLE OF CONTENTS
INTRODUCTION AND SUMMARY                                CHAPTER 1

     Introduction	1-1
     Incineration Systems Considered	1-2
     Summary of Results and Conclusions	1-5
     Limitations	1-17
     Organization of This Report	1-19


OVERVIEW OP ANALYTIC METHODS                            CHAPTER 2

     Introduction	2-1
     Incineration System Structure	2-1
     Methods to Estimate Quantities Released	2-3
     Human and Environmental Effects Analyzed	2-5
     Uncertainty	2-7
     Summary.	2-10


WASTE QUANTITIES RELEASED PROM LAND TRANSPORTATION      CHAPTER 3

     Data Sources	3-1
     Assumptions	3-2
     Releases from Vehicular Accidents	3-3
     Releases from Container Failures	3-5
     Caveats and Sensitivity Analysis	3-6
     Comparable Hazards	3-8
     Summary of Releases from Land Transportation	3-9


RELEASES FROM WASTE TRANSFER AND STORAGE                CHAPTER 4

     Data Sources	4-1
     Assumptions	4-2
     Transfers To and From Tank Trucks	4-2
     Spills at Transfer and Storage Facilities	4-4
     Fugitive Emissions from Transfer and Storage Facilities....4-6
     Caveats and Sensitivity Analysis	4-6
     Comparable Hazards	4-9
     Summary of Storage and Transfer Releases	4-10

-------
WASTE RELEASES FROM OCEAN TRANSPORTATION                CHAPTER 5

     Data and Assumptions	5-1
     Estimation of Releasing Vessel Casualty (Spill) Rates......5-2
     Waste Quantities Released	5-4
     Caveats and Sensitivity Analysis	5-6
     Comparable Hazards	5-8
     Summary of Releases from Ocean Transportation	5-10


RELEASES FROM INCINERATION                              CHAPTER 6

     Introduction	6-1
     Emissions of Undestroyed Wastes	6-2
     PIC Emission Rates	6-3
     Metals Emissions	6-7
     Hydrochloric Acid Emissions	6-8
     Scrubber Effluent	6-9
     Caveats and Sensitivity Analysis	6-10
     Summary	6-11


EFFECTS OF RELEASES FROM OCEAN TRANSPORTATION           CHAPTER 7

     Introduction	7-1
     Methods and their Limitations	7-4
     Marine Effects of a Release of PCBs in Mobile Bay	7-9
     Marine Effects of a Release of PCBs over the
       Continental Shelf	7-12
     Marine Effects of a Release of PCBs in the Burn Zone	7-15
     Marine Effects of Releases of EDC	7-17
     Human Health Effects of Spills of PCB and EDC Waste	7-17
     Summary	7-18


EFFECTS OF RELEASES FROM INCINERATION                   CHAPTER 8

     Introduction	8-1
     Human Health Effects	8-2
     Environmental Effects	8-9
     Summary	8-13
SELECTED BIBLIOGRAPHY

-------
INTRODUCTION AND SUMMARY                                CHAPTER 1
INTRODUCTION

     The U.S.  Environmental Protection Agency (EPA) currently is
evaluating  ocean-based incineration as a method for disposal  of
hazardous  wastes.  EPA'a  Office of Policy Analysis (OPA)  asked
Industrial Economics,  Incorporated (lEc) to work with OPA  staff
in  developing  a  comparative assessment of the risks  posed  by
land-based  and ocean-based incineration systems.  This  document
presents the results of our work.

     Ocean- and  land-based  incineration systems have  different
physical  characteristics  and  affect  different  locations  and
ecosystems.  As a result,  structuring a consistent comparison is
difficult.  This report integrates existing information and  adds
new  analyses developed using existing methods and data.  No  new
primary research has been completed.  We hope that our method  of
structuring  the  analysis and the results developed will  assist
efforts  to  evaluate  incineration and  other  technologies  for
hazardous  waste  management.   However.   li  i&  important   £fi
understand  that the results of this effort are not sufficient to
determine  tne advisability of any specific |.and- or  ocean-based
incineration proposal.

     The remaining sections of this Chapter summarize:

     o    the  ocean- and  land-based  incineration  systems
          considered,

     o    major results and conclusions,
                              1-1

-------
     o    the major limitations of our work, and

     o    the organization of this report.

All  exhibits referenced in the text are included at the  end  of
the chapter.


INCINERATION SYSTEMS CONSIDERED

Incineration System Components

     We separate  land- and ocean-based incineration systems into
three and four separate physical components,  respectively.  Both
land- and ocean-based systems are defined to include:

     o    Land Transportation:  transport of wastes by truck
          from  the  generator site to  the  incinerator  or
          pierr

     o    Transfer   and  Storage:   transfer  and   storage
          operations at the land-based incinerator, pier, or
          other storage facilities, and

     o    Incineration: incineration of the waste.

In addition to these steps,  ocean-based incineration systems are
defined to include:

     o    Ocean  Transportation:  transport of the wastes by
          ship from the pier facility to the burn zone.

At  each of these stages,  wastes and hazardous by-products  (for
example,   volatilized   fractions  or  products  of   incomplete
combustion)  can be released to the environment.  The  nature  of
these  releases varies from relatively unlikely releases of large
quantities of waste (such as spills from truck or ship accidents)
to  very likely releases of smaller quantities of waste  (such  as
stack emissions,  minor pump leaks,  and so forth). He attempt to
quantify losses from all of these possible release points.

     Our  ocean-based  system has a configuration similar to that
proposed by Chemical Waste Management, Inc. (CWM) to operate  the
I/V Vulcanus II from Mobile,  Alabama to the Gulf of Mexico  burn
zone. However, we assume  that an integrated storage and transfer
facility  is  located at the port.  Our land-based system is  not


                              1-2

-------
based  on any single incinerator but combines characteristics  of
several   facilities.   Exhibit   1-1  displays   the   important
characteristics assumed for each incineration system.

     As  Exhibit 1-1 shows,  land transportation assumptions  for
the  ocean- and land-based systems are similar.  We  assume  that
wastes  are  transported 250 miles from generator  to  land-based
incinerator  or  pier.  Tank trucks are the assumed mode of  land
transport in both cases;  and weather,  road,  and other  driving
conditions are assumed to be "average." While wastes destined for
land- or  ocean-based systems might travel  different  distances,
changes  in  the  250 mile trip length do not alter  our  results
significantly.

     Transfer  and  storage  characteristics for  each  case  are
assumed  to  be  similar  and are  determined  primarily  by  the
configuration of equipment used for handling and storing  wastes.
Ocean-based  incineration  requires  one extra loading  step  	
pumping  wastes  from  an onshore storage  facility or  from  the
current  fixed  piping system at Chickasaw into  the  incinerator
ship.  The type of storage tanks used also is critical, since the
emission characteristics of alternate tanks differ  greatly.  Our
analysis  considers  both accidental spills during  transfer  and
storage and continuous "fugitive" losses from storage tank vents,
pump seals,  and so forth.   We have not considered releases from
major   accidents   involving  fire  or  explosion   at   storage
facilities.  The probability of such events occurring is very low
and,  because  both land- and ocean-based systems require similar
storage facilities,  the potential for events of this type  would
be about the same for each system.

     Ocean  transportation  characteristics  are  unique  to  the
ocean-based system, and  have  been  drawn  directly  from  CWM's
proposed   plan  for  operations  through  Mobile   Harbor   from
Chickasaw,   Alabama.   These  operations  will  require  an  800
kilometer  transit through Mobile Harbor,  across the continental
shelf near the mouth of the Mississippi River, and on to the burn
zone.

     Finally,  Exhibit  1-1 shows the assumed characteristics  of
the  incinerators  themselves.  Both land-based  and  ocean-based
incinerators  are  assumed  to  be liquid  injection  units  with
capacities up to 70,000 metric tons per year.  This capacity  was
selected  based  on  the  characteristics  of  the  Vulcanus  II.
Although  this capacity is greater than any commercial land-based
facility, incinerators of this size are feasible and in operation


                              1-3

-------
 at some on-site facilities.   Consistent with current practice the
 land-based unit is assumed to employ  scrubbers,   while the  ocean-
 based  unit  does   not.    We   assume   that  these  units achieve
 destruction  and  removal  efficiencies (DRE)  of  99.99 to 99.9999
 percent depending  on the waste burned.   In effect,  we assume  that
 either   system   will meet  current  permit requirements  concerning
 DRE and other operating parameters.


 Waste
     The  environmental transport of wastes  and   by-products   and
their   ultimate   fate and effects depend  strongly on  the   precise
composition of the mixture  released. In general,  adequate  data do
not exist to predict the transport,  fate and  effects of mixtures
released  to  the environment.   In view of this,   we   assume   two
"simplified" waste streams with  single hazardous  constituents  for
use in  this analysis.

     1.   A   waste  containing  35  percent  by   weight   of
          polychlorinated biphenyls (PCBs) .  Arochlor  1254  is
          assumed to be the specific PCB,  and the remaining
          65  percent of the waste stream is assumed   to   be
          non-hazardous.  Each   system  is assumed to burn
          56,000  metric tons of  this waste stream each year.

     2.   A   waste  containing  50  percent  by   weight   of
          ethylene  dichloride   (EDO and 50  percent non-
          hazardous  substances.  Each system  is  assumed   to
          burn  68,400 metric tons of this waste  stream each
          year.

In addition, each waste stream is assumed to include  100 ppm each
of arsenic, cadmium, chromium and nickel.  These  metals are among
those  specifically  regulated by the Agency's draft   permit   for
ocean-based  incineration  and each has been designated  a human
carcinogen  by  EPA's Carcinogen Assessment  Group.  Although   all
four have been found in a variety of actual  waste streams,  it is
likely  that our  assumption overstates the average concentration
of carcinogenic metals in liquid incinerable wastes.

      Commercial   incinerators  handle   many  waste   streams   of
varying  composition.   Pour considerations  caused us to  assume
the simplified wastes described above. First,  CWH has  requested a
                              1-4

-------
permit  to  burn  PCB-containing wastes  and  thus  the  possible
release,   transport  and  effects  of  these  compounds  are  of
particular interest to EPA.  Second, EDC is a common component in
many hazardous waste streams currently incinerated and is typical
of  a large volume of wastes generated by the  organic  chemicals
industry.  Third,  the  physical  characteristics  and  resulting
transport  behavior of PCBs and EDC in the marine environment are
quite different, thus allowing us to illuminate how fundamentally
different waste components might behave.    Fourth,  human cancer
potency   factors  are  available  for  both  PCBs  and  EDC  and
information  is  available  on  the  toxic   and  bioaccumulative
effects of the materials in marine organisms.


SUMMARY OF RESULTS AND CONCLUSIONS

     This section summarizes the quantities of waste released and
the possible human and environmental effects from these releases.


Quantities of Waste Released

     Exhibits 1-2 and 1-3 present our estimates of the "expected"
annual average release quantities from ocean-based and land-based
incineration  for  the PCB  and  EDC  wastes,  respectively.  For
releases  due  to accidents,  spills and other infrequent  events
these  estimates  represent the long-term average  release, which
includes  years  with  no  release and years  with  one  or  more
releasing events. As a result, actual releases in any single year
for  these  events  could range from  zero  to  relatively  large
quantities  if,  for example,  a truck is involved in an accident
that results in a spill.  Our calculation of expected  quantities
released  accounts for both the probability of a release and  the
resulting  magnitude  of waste lost, and provides  one  means  of
comparing average long-term releases from infrequent events  with
continuous releases.

     Each   exhibit  shows the expected quantity of  release  for
each component of the land- and ocean-based systems.  All figures
have been rounded to the nearest 100 kilograms  (0.1 metric tons).
For  convenience,  a  subtotal  is  provided  for  releases  from
transport and handling steps and for incinerator stack  releases.
Metals included in scrubber effluent are also reported.
                              1-5

-------
     Overall, comparison of the expected  releases  from  the ocean-
based  versus  the land-based systems for these  two wastes   shows
that   expected   release  quantities  from  the   transportation and
handling  components  range to roughly 15 percent  of the  long-term
average release  expected.  The extra transport and handling  steps
required  by ocean-based systems do  not add significantly to the
long-term expected   release,  but   they  do   create  the  remote
possibility of a major accident and  subsequent release  of waste.

     Incineration itself accounts for the major  release of wastes
and hazardous by-products for both wastes and  systems considered.
The  quantities  released  by  incineration  are  a  function   of
assumptions  about   metals  content  and the  performance  of the
incinerator and  scrubber.  Available estimates of  PIC   generation
are  subject to  very high levels of  uncertainty.   Releases  from
each component are further discussed below.


Land Transportation

     Exhibits  1-2   and 1-3 show that the expected release  from
land transportation  will average 2.1 and  2.7 metric tons  (HT) per
year   for the PCS and EDC  wastes,   respectively.   Again,   these
estimates  represent long-term averages.   Releases in any  year
would  vary from zero to larger quantities if  a  spill  occurred.
The slightly greater release estimated for the EDC waste reflects
the  larger quantity of this waste assumed to  be handled by  each
system.   Because  land transportation has the same configuration
in each system,  there is no difference in the release  quantities
expected  for the land- versus ocean-based system.

     Our  analysis of releases from land transportation  considers
two  types of potential losses — spills  from  vehicular accidents
and  spills  from  enroute  container  failures.    We  base our
estimates of the frequency of such events and  of the size of the
resulting  spills  on data provided by the  U.S.  Department   of
Transportation   (DOT).   These  data pertain to  all  tank  trucks
carrying  hazardous   materials.   Use of  the DOT  data  with our
assumptions  regarding miles travelled results in  an expected .18
and .26 releasing vehicle accidents  per year for the PCS and EDC
wastes, respectively.  The annual number  of container failures  is
estimated at .23 and .32 for the PCS and  EDC wastes.  The average
fraction  of  cargo  released in vehicular accidents is  about   40
percent.   In contrast,  spills from container failures typically
release only about   4 percent of the cargo in  the  container.
                              1-6

-------
     Information  supplied  by  hazardous  waste  services  firms
indicates  that  the  DOT accident rates are  higher  than  those
experienced  by these firms.   This probably is due to management
practices  undertaken by such firms to reduce the probability  of
accidents and to their use of stainless steel tanks that are more
resistant to rupture than are aluminum tanks.   Thus,  we believe
that  our  estimates  of releases from land   transportation  are
conservative.   These estimates represent,  on average, about 0.5
percent  of the number of annual transportation-related  releases
of hazardous substances in EPA Region IV.


Transfer and Storage

     Our  analysis of wastes released from transfer  and  storage
considers three types of releases:   spills when unloading wastes
from  tank  trucks;  spills from equipment at waste transfer  and
storage  facilities;  and  fugitive emissions from  transfer  and
storage.  As shown in Exhibits 1-2 and 1-3, the expected quantity
released  from transfer and storage activities is  slightly  over
one  metric  ton  per  year for both systems  and  waste  streams
considered.   These  release estimates are based  on  information
developed by Arthur D. Little, Inc. and DOT.

     Spills  from  the transfer and storage component are  fairly
infrequent  events.   They  are estimated to occur at a  rate  of
about 0.04 to 0.05 per year for the transfer of wastes from  tank
trucks  and  at  a  rate  of  about 0.03 to  0.04  per  year  for
equipment"and storage tanks.   The ocean incineration system  has
one  additional component — the loading hose to transfer  wastes
to  the ship.   We estimate the rate of  spills from the hose  at
about 0.002 to 0.003 per year,  with the average spill size about
6  to  7  MT.  It is likely that spills of  this  type  would  be
contained  either on the deck of the Vulcanus or by booms  placed
around the vessel during loading.  Spills from truck unloading or
from  equipment or storage tanks are also likely to be  contained
in  the facility.  Fugitive emissions would account for about 0.6
to 0.7 MT per year of this amount.  The number of spills expected
from  this component of  land- or ocean-based incinerators  would
represent,  on  average,  less than 0.1 percent of the number  of
spills of hazardous material likely from fixed facilities in  EPA
Region IV.
                              1-7

-------
Ocean Transportation

     The   ocean   transportation   component   of     ocean-based
incineration  is the only major component of the  system  that   has
no  parallel in a  land-based  operation.   Therefore,   potential
releases  from ocean transportation are of  special  interest  when
comparing   the   relative   risks   of   land- and   ocean-based
incineration systems.

     Incineration  ships have operated o£f  the coast of  Europe in
the  North Sea since 1972.   About 320 voyages have been made  and
about   650,000  metric  tons  of  hazardous  waste   have   been
incinerated.   No  casualties  such  as  collisions,   groundings,
rammings,  or fires have occurred, nor have there been any spills
from  loading these ships in port.   Although a very good  safety
record has been established,  the number of voyages completed   is
too  small  to be used directly in estimating statistically    the
probability of spills.

     In view of this, EPA asked Engineering Computer Optecnomics,
Inc.  (ECO)  to  develop  estimates of spill rates  based on   the
worldwide  historical  record  of tank ships of   a  similar  size
class.    Spill  rates  were  developed  for  three impact  type
accidents  (collisions,  groundings,  and rammings) and  for  non-
impact type accidents (fires,  explosions,  structural   failures,
and capsizings).   Spill rates were  developed for  four  locations
of  interest — pier and harbor.  Mobile Bay,  coastal area,   and
burn  zone.   Estimates were also developed of the  percentage   of
spills likely to involve one,  two,  or three or more  tanks — 80
percent, 15 percent, and 5 percent, respectively.

     The  historical  spill  rates were  adjusted   to  take  into
consideration  the  design of the Vulcanus  (double  hull,  double
bottom construction and the use of a controllable pitch  propeller
and  bow thruster),  operating restrictions to be imposed by   the
Coast  Guard  (escorts  by tug boats and a  Coast   Guard vessel,
imposition  of a 300 foot moving safety zone,  and  limitation   of
transits  to  daylight hours and in conditions of   above average
visibility),  and  the soft bottom conditions in the  Gulf.    The
precise  effect  that these factors may have  in  reducing  spill
rates is difficult to determine.  ECO's adjustments were based on
published studies,   observed differences in spill rates,  kinetic
energy  levels likely for accidents in different  locations,   and
professional judgment.
                              1-8

-------
     The  expected annual releases of 0.6 MT foe PCBs and of  0.8
MT for EDC shown in Exhibits 1-2 and 1-3 are relatively small and
represent  average  releases  expected over  a  very  long  time.
Spills  from  the  vessel would be very  infrequent  events.   We
estimate  that  the frequency of all spills for the  Vulcanus  is
about one per 1,200 operating years.   However,  the frequency of
spills   estimated  for any particular  location   is  less.   For
example,  the  overall spill rates for the pier and harbor  area,
Mobile  Bay,  Coastal area,  and the burn zone are about one  per
3,000,  10,000,  4,000  and 6,000 operating years,  respectively.
These estimated spill rates are for all sizes of spills.   Spills
involving two or three or more tanks would be extremely  unlikely
events.   For  example,  the estimated rates for spills involving
two  and  three  or more tanks in Mobile Bay are   about  one  per
67,000 and 200,000 operating years, respectively.

     The  preceding  estimates of releases are  conservative,  in
that  we  assume that any tank involved in a spill  releases  its
entire  contents and that the entire ship's cargo  is released  in
accidents  involving  three or  more  tanks.   In  addition,  the
estimates  do  not reflect the effects that remedial actions  may
have  in  removing  wastes from the  marine  environment.   As   a
condition  of  their  permit,    hazardous  waste  operators  are
required  to  develop  a contingency plan  for  handling  spills.
Efforts  to  contain  and recover spills are most  likely  to  be
successful  in enclosed areas or in shallow waters,  such as  the
pier  and harbor area and Mobile Bay.   However,   estimating  the
effectiveness  of remedial actions was beyond the  scope  of  this
study.

     The  tonnage carried by the Vulcanus is small in  comparison
to commercial shipments of petroleum and hazardous substances  in
the  Gulf area.   For example,  the cargo carried  by the Vulcanus
would  be  only  about 0.01 percent of  petroleum  and  hazardous
substances  transported  annually in the Gulf  area.   Since  the
Vulcanus has a lower spill rate than other vessels,  the expected
volume of releases from the Vulcanus,  in a statistical sense, is
only  about  0.002  percent of  expected  releases from  ongoing
shipments of petroleum and hazardous substances in the Gulf area.


Incineration

     Incineration  itself  is  the major release   point  in  both
systems. Metals account for the largest releases when burning the
PCB   waste,   with   undestroyed  wastes   providing   a   minor


                              1-9

-------
 contribution.!/  Our   estimates  of  undestroyed wastes  for   both
 systems   assume  99.9999   percent destruction of  the  PCB  waste
 stream.   Our   estimates   of   metals   emissions  result  from  our
 assumptions about metals  concentrations in the waste.  Metals are
 transferred to scrubber effluent  in  the land-based case by use of
 a  scrubber which is  assumed to remove from 50 to 90  percent  of
 the four  metals  considered.  Thus,  stack emissions of organics and
 metals  in the land case  are about 20 percent of those  predicted
 for the ocean-based case.

      Exhibits  1-2  and 1-3  present  the expected  annual  average
 release   quantities   for  the  incineration  component  of   both
 systems.  Total  organics  and metals  released from incineration of
 the  PCB  wastes  is 22.5 metric  tons  per year for both systems  if
 one includes both stack releases  and scrubber  effluent.     Total
 organics  and  metals  released by  the ocean-based incinerator  for
 the  EDC  waste is more than  50  percent greater than that expected
 for  the  land-based  case  (54.8 MT compared  to  34.8  HT).   This
 difference  is  due primarily to  the higher level of PIC  release
 estimated for  the ocean case.

      Products  of incomplete  combustion (PICs)  are not expected in
 significant  quantities   for the  PCB waste based on  EPA's  trial
 burn  data.  However,  the  results  of  these  trial burns are subject
 to  large  uncertainty and  considerable debate  because  of  the
 procedures used  and the limited number of  PIC compounds that were
 considered.  Thus,  our  estimates  of PIC emissions for the  PCB
 waste  (.000000006 metric  ton per  year and .00002 metric ton  per
 year  for  the ocean  and land cases,   respectively)  could be  in
 error by many  orders  of magnitude.    In addition,   we do not know
 of  any complete  explanation  for the  lower  level of PIC generation
 found for ocean-based  PCB  incineration and the higher level  found
 for  ocean  incineration   of  general  organic  waste  relative  to
 incineration   on land.  Alternate  estimates of PIC generation and
 the resulting  human health effects are analyzed in later chapters
 and appendices of this report.
JL/Note  that if incinerated waste streams   contain   significantly
less carcinogenic metals than we have assumed,  the  total quantity
of   stack  emissions  released  from  each  system would   drop
dramatically.  Further,  releases  from  the  transportation  and
transfer/storage  components would become the major contributors
to total release.
                              1-10

-------
     In addition to the compounds reported in these exhibits,  we
also  analyzed  the release of chlorine at incinerators  in  both
systems  and  the disposition of the chlorine to  the  atmosphere
and,  for the land-based system, to scrubber effluent and sludge.
Our   results  show  that  10,505  and  25,034  metric  tons   of
hydrochloric acid (BCD will be released from the incineration of
the PCS and EDO wastes,  respectively.  In the ocean case all HC1
is  released  to  the  atmosphere.  In the  land-based  case  the
scrubber captures 99 percent of the HC1 and neutralizes it. Thus,
most chlorine in the land case is disposed as scrubber sludge  or
effluent.
Effects From Incinerator Releases

     He  estimated and compared possible human health effects due
to  releases  from the incinerator and due to  fugitive  releases
from   transfer/storage  equipment  located  at  the   land-based
incinerator   or   the  pier.   We   also   considered   possible
environmental effects of these releases.
Human Health Effects

     Our analysis of human health risks estimates the incremental
risk  of  developing  cancer for  a  hypothetical  "most  exposed
individual"  (HEI)  who  resides at the location of  the  highest
overall risk due to air concentrations resulting from incinerator
stack and transfer/storage fugitive releases.  For the land-based
system,  the location of the HEI is based on Census data; whereas
for  the ocean-based system the MEI is assumed to reside at  that
point  on  the coast where modelled  concentrations  are  highest
averaged  over a year.   These risk estimates assume 70 years  of
continuous  exposure.  In  the  ocean case  different  areas  are
affected by stack emissions (the coastline downwind from the burn
zone) and transfer/storage fugitive releases  (the area around the
port),  while  for the land-based case the same area is affected.
Our  calculations of risk for the land-based system consider  two
alternative sites for the incinerator, in Texas and Arkansas.

     We chose to estimate risks to the most exposed individual in
order to estimate the largest risks likely to be suffered by  any
person  due  to  the releases considered.  While  we  could  have
considered  the average incremental risk across the entire  human
population  affected  by  each  system,   this  metric   requires


                              1-11

-------
estimation  of  the  total exposed population and all  levels  of
exposure  — a  difficult and controversial task given  the  long
distances and persistence of some compounds considered  here.  In
general,  other  risk analyses have found that average population
risks  range from one to four orders of magnitude lower than  the
risk to the MEI.

     Exhibit  1-4  presents  the incremental risk  of  developing
cancer for the most exposed individual due to releases from land-
and  ocean-based fugitive (transfer/storage) and stack  releases.
As  shown,  the  incremental risks from  land-based  incineration
releases are about three chances in one-hundred thousand for  the
locations and wastes considered. Virtually all of the incremental
risk to the MEI is due to stack releases,  with fugitive releases
resulting  in  increased  risks of less than one  in  a  million.
Incremental risks to the most exposed individual at the coastline
for  the ocean-based system range from one in one million to 6 in
ten million.  As shown,  the risks from fugitives to the MEI near
the port facility are less than two per hundred million.

     The data and methods used to generate these incremental risk
estimates  are highly uncertain and tend to overestimate expected
human health effects. Thus, the absolute risk levels indicated by
these figures must be interpreted with caution. EPA has completed
studies  of  incremental  risks from  other  hazardous  pollutant
releases  using  similar methods with similar  uncertainties  and
biases. For example, a recent study on toxic air pollutants found
that, on average, individuals in the U.S. face incremental cancer
risks  of  about  4 to 6 chances in ten  thousand.2/  Using  this
estimate  as the base for comparison,  incremental risks  to  the
persons  most exposed by the incineration systems considered here
would be one to three orders of magnitude lower.

     Exhibit  1-5  presents more detailed information  concerning
the sources of the incremental cancer risks for the  incineration
systems  considered.  This  exhibit reports the  contribution  of
principal organic hazardous constituents (POHCs),  PIC and metals
emissions to the total incremental risks suffered by the MEI. For
2/EPA,  The Magnitude and Haiiitfi s£. thfi Alt Toxics Problem in thfi
United  States.  Draft Report.   Office of  Air and Radiation  and
Office of Policy, Planning and  Evaluation,  1984.
                              1-12

-------
convenience,  risks  due to fugitive emissions from transfer  and
storage  operations are not included in the figures in Exhibit 1-
5.

     The estimates in Exhibit 1-5 show the relative magnitudes of
incremental  risk caused by each stack component for both  wastes
and  systems considered.  As shown,  POHC and PIC releases  cause
risks  that  are from one to five orders of magnitude  less  than
risks  from metals.  Thus,  metals account for from 90 percent to
virtually  all  of  the incremental risks  calculated  for  stack
emissions.   However,  as  noted earlier,  it is likely that  our
assumptions overstate the average concentrations of  carcinogenic
metals  in liquid incinerable wastes.   Risks from POHC  releases
are  less than 1 per billion for the ocean system and less than 2
per 10 million for the land system.  Risks from  PIC releases are
less  than 4 in one billion for the ocean system and less than  2
in 1 million for the land-based system.   Thus,  risks from  both
PICs and POHCs in each system are low.

     Exhibit 1-6 presents the ratio of the incremental risks from
land-based versus the ocean-based stack releases.  The figures in
this  exhibit  were  calculated by dividing the  land-based  risk
estimates  in  Exhibit  1-5 by those shown  for  the  ocean-based
system.  Thus,  the figures in Exhibit 1-6 indicate the  relative
size of risks estimated for the land-based versus the ocean-based
system considered.  For example,  Exhibit 1-6 indicates that, for
the  PCB waste,  land-based emissions create about 40 times  more
incremental  risk to the HEI than do ocean-based  emissions.  For
the EDC waste, the ratio of land to ocean risk is about 30.

     Exhibits 1-5 and 1-6 show that, given our assumptions, there
is  roughly  30  to  40 times more incremental  risk  from  metal
released from land systems. Changes in the type and concentration
of  metals  in  the waste could reduce these  risk  estimates  by
several  orders  of magnitude but would not change  the  relative
performance of the land and ocean systems.  Different assumptions
about  the  performance of the land-based  scrubber  in  removing
metals,  or  about  the atmospheric transport of metals over  the
ocean  could affect  the relative performance of the two  systems
considered.  While different assumptions could broaden or  narrow
the  differences in metals risk,  it is unlikely that ocean-based
systems  would  generate more incremental  risk  than  land-based
systems.
                              1-13

-------
     These   exhibits  also  show  that  the  land-based   system
generates  about  300   times more  incremental  risk  from  POHC
emissions.  The  transport behavior of PICs is similar to that of
POHCs and thus, for similar quantities and toxicities of release,
PIC risks should show the same ratios.  However,  our analysis of
trial burn data indicates that land- and ocean-based systems  can
generate very different quantities of PICs.

     EPA's  trial  burn  data indicate  wide  variation  in  PICs
generated by different incinerators from different waste streams.
The  results of these burns are subject to great uncertainty  and
considerable debate.  For the PCS waste,  our analysis uses a PIC
generation  rate for the ocean-based system that is 10,000  tiroes
lower  than  that used for land since this rate is  derived  from
trial  burn data.  When combined with the additional advantage of
the  ocean-based system in being further from human  populations,
the  land-based unit generates over 1 million times the  risk  of
the  ocean system for PICs.  Thus,  our relative estimates of PIC
generation  from  the PCB waste would have to be in  error  by  a
factor  of one million for the land and ocean systems to  present
equivalent risks from PICs.  Again,  note that the absolute risks
estimated for PICs from both systems are very low.

     Our assumption regarding the amount of PICs from EDC  wastes
is quite different.  For this waste, trial burn data suggest that
the  ocean-based unit will generate about 30 times more PICs than
the land system.  Despite the ocean system's exposure  advantage,
this  lowers  its ratio of risk compared to the land system to  a
factor  of 8.  While changes in relative PIC generation that  are
greater than an order of magnitude could make PIC risks from land
and ocean systems equivalent or show land systems to be safer, we
believe that such changes are unlikely.

     Although  the absolute numbers reported in Exhibits 1-4  and
1-5  are uncertain and biased to overestimate incremental  risks,
the  relative  differences shown in Exhibit 1-6 are more  certain
and would be altered only by changes in the relative  performance
of  the  ocean- and  land-based systems.  Overall  these  results
generally  indicate that the human health risks posed  by  either
system  are  relatively low,  with the risk from the  ocean-based
system  about one to two orders of magnitude less than  from  the
land-based system.   This general result was expected,  since the
burn  zone  is  about  200 kilometers  distant  from  the  coast,
allowing residual emissions to disperse and to partly settle  out
before  reaching land,  and since the plume emitted during  trial
burns has never been detected at the shoreline.
                              1-14

-------
     Along  with incremental risks due to inhalation of hazardous
compounds,  we also considered incremental risks due to ingestion
of  foods  contaminated by wastes or hazardous  by-products  from
ocean- and land-based stack emissions.  Although data and methods
in  this  area  are extremely  limited,  we  found  insignificant
incremental risks from this ingestion route of exposure.


Environmental Effects

     In addition to the human health effects summarized above, we
considered  the possible environmental effects that might  result
from  incinerator stack releases.  For the ocean-based  case,  we
asked  Applied  Science Associates,  Inc.  (ASA) to estimate  the
deposition of stack releases to the ocean surface,  the transport
of  these  materials  in  the  water  column  and  sediments  and
resulting  effects  on  the  marine  ecosystem.   ASA's  analyses
indicate  that  no measurable effect on the marine  ecosystem  is
expected  due to stack releases from the EDC waste.  The analysis
of  the  PCB  waste  is complicated by  the  persistence  of  the
compound  and  by  scientific uncertainty about the role  of  the
ocean's surface  (the "microlayer") in capturing and concentrating
atmospheric  pollutants  and  providing these  materials  to  the
marine  ecosystem.  Notwithstanding  these  uncertainties,  ASA's
analyses   indicate that long-term continuous burning of the  PCB
waste at the levels assumed here would not result in a measurable
effect  on the marine ecosystem.    Information developed on  the
background  atmospheric  flux  of  PCBs  into  the  Gulf   waters
indicates  that  it  would  be  about two  to  three   orders  of
magnitude greater than that from incineration of PCBs.

     We were unable to complete a similar analysis of the effects
on  terrestrial ecosystems caused by land-based  stack  releases.
However  we did consider the possible environmental effects  from
the release of scrubber effluent and sludges from the  land-based
system.  Because  discharge  of scrubber effluent and  sludge  is
regulated  by  the Clean Water Act and the Resource  Conservation
and Recovery Act, respectively, disposal of these materials would
have  to  be carried out in a manner  approved  by  environmental
permitting  authorities.  Thus,  we assume  environmental  damage
from these discharges is minor.
                              1-15

-------
Effects From Ocean Transportation Releases

     In  addition  to  the  effects from  incinerator  stack  and
fugitive   releases,  we  also characterized possible  human  and
environmental  effects  resulting  from  spills  in  the   marine
environment.   Although  the probability of a spill is very  low,
the  magnitude of the resulting effects is of interest in  public
deliberations   about   ocean- versus   land-based   incineration
systems.  We  considered  the likely effects of release of  cargo
from  the  vessel  at  sites  within  Mobile  Harbor,   over  the
continental shelf on the path to the burn zone,  and in the  burn
zone itself.


Human Health Effects

     Exhibit  1-7  presents information about the  potential  for
human  health consequences from loss of the entire vessel  cargo.
Volatilization  of such a spill could expose human populations to
high  concentrations of hazardous constituents for short  periods
of  time.   Because of the acute nature of  these  exposures,  we
compared  the  estimated dosage received by human populations  in
the  first  24 hours after a spill to the Threshold  Limit  Value
(TLV) for PCBs and EDC.  The TLV represents the dosage to which a
worker  can  be exposed with no adverse health  effects  such  as
coughing,  dizziness,  and longer-term health damage. We adjusted
the TLVs to  account for continuous exposure rather than exposure
for  only  eight hours per day.   In all calculations  we  assume
that  the  human population is directly downwind from  the  spill
site and that the entire cargo of the vessel is released.

     As shown in Exhibit 1-7,  we estimated the ratio of 24  hour
dosages  to adjusted TLV's for spills in Mobile Harbor at one and
15  kilometers  from  the city of Mobile,  for  spills  over  the
continental  shelf near the mouth of the Mississippi  River,  and
for spills in the burn zone.  The results show that spills of the
entire  cargo  of  either waste one kilometer from  the  city  of
Mobile  could  cause human health problems.  Spills at the  other
locations are not expected to cause acute human health problems.


Environmental Effects

     Exhibit  1-8 presents a summary of the potential effects  to
the  marine  ecosystem from spills of half a tank  in  the  three
locations  described  previously.    The  ecosystem  effects  are


                              1-16

-------
summarized  by changes in biomass levels and in  bioconcentration
levels for PCBs and EDC.   For PQBs we consider both floating and
sinking cases,  since this compound, although heavier than water,
might  float if entrained in lighter-than-water  materials.   For
EDC,  we  consider  only a sinking case,  which results in  rapid
diffusion  since  this  compound is soluble in  water.   We  also
modelled the effects of larger spills of 2 tanks and 8 tanks.  As
noted earlier,  for modelling purposes we assumed no actions were
undertaken to contain and remove the spill.

     Exhibit  1-8 indicates that EDC spills would have relatively
minor  effects on the marine ecosystem.  These small impacts  are
the   result   of  this  compound's  rapid   diffusion   to   low
concentration  levels  and its relatively low toxicity to  marine
species.   In  addition,   bioconcentration  of  EDC  is  not   a
significant  phenomenon.   The same results hold for larger  size
spills.

     In  contrast,  spills  of  PCBs are modelled to  have  major
effects on the marine ecosystem.   These effects range from being
quite severe in the Bay (substantial reduction in benthic species
and  large bioconcentration effects on fish and shrimp)  to  less
severe  in the burn zone area.   Since PCBs are a persistent com-
pound   such   effects  are  expected  to  last  a   long   time.
Bioconcentration  effects in commercial and recreational  species
would be of most concern in the Bay and contaminated shelf areas.
In the event of larger or smaller size spills,  the magnitude  of
bioconcentration  effects is approximately linear with regard  to
quantity released.

     Estimating  the  effects of spills of   persistent compounds
such as PCBs in the marine environment is an imprecise science at
best.   Because  of substantial uncertainties regarding the  long
term  fate of PCBs in the marine environment and  the  biological
mechanisms involved in the food web/ the results of the modelling
effort  should  be  viewed as a general indication  of  potential
effects rather than as a precise measure of those effects.


LIMITATIONS

     All  of  bur analyses are subject to  many  limitations  and
caveats due to uncertainties in the data and methods that we use.
These limitations and caveats are explained fully in the chapters
and  appendices  of this report.  All of our  results  should  b_e_
interpreted  with cautionf  and with a. complete understanding  of


                              1-17

-------
all  ££ .these, limitations. Some of the general limitations of the
analyses are described below.

     1.   By design this analysis is limited to  considering
          only  incineration systems.   It does not consider
          potential  environmental  or  economic  risks   or
          benefits  from  use of other methods of  hazardous
          waste treatment, storage or disposal.

     2.   The analyses reported here are applicable only  to
          the specific land- and ocean-based cases examined.
          Results   for   other   locations,    wastes   and
          technologies  could  vary substantially  from  the
          information reported here.

     3.   We   have  attempted  to  structure   incineration
          systems  and  wastes typical of actual  or  likely
          practice so as to generate an expected rather than
          a  best  or worst  case  analysis.  However,  data
          limitations have required use of many conservative
          assumptions    in   our   estimates   of   release
          quantities, and the methods for estimating effects
          of   releases  generally  err  on  the   side   of
          overestimation.    Thus,   our   overall   results
          overestimate  releases and resulting effects  from
          the incineration systems considered.

     4.   He  have not considered a number of effects  which
          might  result  from releases from  the  ocean- and
          land-based  systems.  In particular,  we have  not
          analyzed  the  possible  effects  of  releases  on
          terrestrial ecosystems.

     5.   Our  analysis of the quantity of and effects  from
          stack  releases  for  both  systems  is  based  on
          assumptions   about   incinerator   and   scrubber
          performance and waste composition,  and on results
          from  EPA-sponsored trial burns.  The data on  PIC
          generation  developed  from  the  trial  burns  is
          extremely uncertain and subject to debate.

     6.   The  estimates  of the effects of spills into  the
          marine environment  assume  that   no   mitigating
          activities are completed.
                              1-18

-------
     In  view  of the above,   the absolute  release  and  effects
estimates  for land- and ocean-based systems are less  meaningful
than the relative differences shown between the two systems.   Our
results  are  particularly  sensitive  to factors  (such  as   PIC
emissions,   scrubber  efficiency,  and so forth)  that  alter   the
relative performance of the two systems considered.
ORGANIZATION OF THIS REPORT

     All of the methods,   data and results for our comparison  of
the  risks of land- and ocean-based incineration are detailed  in
the  remaining  chapters  and appendices of  this  report.   These
materials are organized as follows.
     Chapter 2



     Chapter 3


     Chapter 4



     Chapter  5



     Chapter 6



     Chapter 7



     Chapter 8



     Appendix A
provides an overview of the methods used
in the comparison of ocean-based  versus
land-based incineration systems.

presents    estimates   of   the   waste
quantities   released  from   the   land
transportation component of each system.
presents    estimates   of   the   waste
quantities  released from  the  transfer
and storage component of each system.

presents    estimates   of   the   waste
quantities   released  from  the   ocean
transportation component.

presents  estimates  of the  wastes  and
hazardous   by-products  released   from
incineration itself in both systems.

describes    the   human   health    and
environmental  effects of releases  from
ocean transportation.

describes    the   human   health    and
environmental  effects of releases  from
land-  and ocean-based incinerators.

describes   the  voyage  plan  for   the
Vulcanus  II  and  the  resulting  waste
throughput.
                              1-19

-------
     Appendix B


     Appendix C


     Appendix D



     Appendix E



     Appendix F



     Appendix G
Appendix


Appendix
All  exhibits
referenced.
              H
               presents detailed estimates of  releases
               from the transfer and storage component.

               presents  detailed estimates of releases
               from ocean transportation.

               describes  the human health effects  due
               to  releases  from land-based  incinera-
               tors.
                                       •
               presents estimates of the generation and
               composition  of products  of  incomplete
               combustion (PICs).

               describes  the  generation,  composition
               and  disposition of scrubber effluent at
               land-based incinerators.

               describes  the human health effects  due
               to  releases from ocean-based  incinera-
               tors.
      describes  the human health effects
      to ocean transportation releases.
due
I     describes  the marine ecosystem  effects
      resulting   from  ocean   transportation
      releases.

follow the chapter or appendix where they are first
                              1-20

-------
                             Exhibit 1-1

            SUMMARY OP INCINERATION SYSTEM COMPONENTS
Component
Land Transportation
Ocean-based
System


Tank trucks
(5000 gallons)

250 miles

"Average"
weather, roads/
etc.
Land-based
System


Tank trucks
(5000 gallons)

250 miles

"Average"
weather,roads,
etc.
Transfer and Stroage
Storage at
pier in
two floating-
roof tanks.
Storage at
incinerator in
two floating-
roof tanks.
                              One truck
                              unload

                              One load to
                              vessel
                         One truck
                         unload
Ocean Transportation
800 km  (500 miles)
to burn zone

Specific path from
Mobile to zone

Vessel specifications
and operations plan as
per CWM for Vulcanus II
None

-------
                             Exhibit 1-1
                             (continued)

            SUMMARY OF INCINERATION SYSTEM COMPONENTS
Component


Incineration
Ocean-based
System
Liquid In-
jection,  no
scrubber

DRE-99.99 or
99.9999

Throughput to
70,000 MT/year
Land-based
System
Liquid In-
jection,with
scrubber

DRE«99.99 or
99.9999

Throughput to
70,000 MT/year

-------
                           Exhibit 1-2

       SUMMARY OF EXPECTED QUANTITIES RELEASED PER YEAR *
                            PCB Waste
                     (metric tons per year)
                                   Ocean-based         Land-based
Release Point                      System              System


Land Transportation                2.1                 2.1

Transfer and Storage               1.2                 1.1

Ocean Transportation               0.6


  Subtotal                         3.9                 3.2


Incineration

  Undestroyed Wastes               0.1                 0.1
  PICs                             0.0                 0.0
  Metals                          22.4                 4.5

  Subtotal (Stack)                 22.5                 4.6
  Scrubber Effluent
    Metals                         —                 17.9
Total Organics and Metals         26.4                25.7
     For  releases due to accidents,  spills and other  uncertain
     events  these  estimates  represent  the  long-term  average
     release  which includes years with no release and years with
     one or more releasing events.
Source: lEc Analysis

-------
                           Exhibit 1-3

       SUMMARY OF EXPECTED QUANTITIES RELEASED PER YEAR *
                            EDC Waste
                     (metric tons per year)


                                   Ocean-based         Land-based
Release Point                      System              System


Land Transportation                2.7                 2.7

Transfer and Storage               1.2                 1.1

Ocean Transportation               0.8                 —


  Subtotal                         4.7                 3.8


Incineration

  Undestroyed Wastes               6.8                 6.8
  PICs                            20.6                 0.6
  Metals                          27.4                 5.5

  Subtotal (Stack)                54.8                12.9
  Scrubber Effluent
    Metals                         —                 21.9
Total Organics and Metals         59.5                38.6
     For  releases due to accidents,  spills and other  uncertain
     events  these  estimates  represent  the  long-term  average
     release  which includes years with no release and years with
     one or more releasing events.
Source: IEC Analysis

-------
                           Exhibit 1-4

  SUMMARY OF INCREMENTAL CANCER RISK TO HOST EXPOSED INDIVIDUAL
                    PROM INCINERATOR RELEASES
                                   PCB Waste         EDC Waste

Ocean-based System

  Stack (coastline)                6.37E-7             1.06E-6
  Fugitives (port)                 2.02E-8             4.97E-10



Land-based System (average of two sites)

  Stack                            2.74E-5             3.14E-5
  Fugitives                        7.05E-7             1.69E-8

  Total                            2.81E-5             3.14E-5
Source:  Exhibits 8-4, 8-5

-------
                           Exhibit 1-5

       INCREMENTAL CANCER RISK TO MOST EXPOSED INDIVIDUAL
                    BY TYPE OF STACK RELEASE
Ocean-based System

  POHCs
  PICs
  Metals

  Total Stack
                                   PCB Waste
1.45E-10
1.68E-12
6.37E-7

6.37E-7
                  EDC Waste
5.51E-10
3.36E-9
1.06E-6
1.06E-6
Land-based System (average of two sites)
  PORCs
  PICs
  Metals

  Total Stack
5.13E-8
1.79E-6
2.56E-5

2.74B-5
1.43E-7
2.59E-8
3.12E-5

3.14E-5
Source: Exhibits 8-3, 8-4

-------
                           Exhibit 1-6

                RATIO OF INCREMENTAL CANCER RISK
            FOR LAND- VERSUS OCEAN-BASED INCINERATORS
                    BY TYPE OF STACK RELEASE
                                   PCB Waste         EDC Waste

  POHCs                                354                 260

  PICs                           1,070,000                   8

  Metals                                40                  29


  Total                                 43                  29
Source: Exhibit 1-5, lEc Analysis

-------
                        Exhibit 1-7
            SUMMARY OF HUMAN HEALTH EFFECTS FROM
                  LOSS OF ENTIRE VESSEL *
         (ratio of 24 hour dosage to adjusted TLV)
Release Location

Mobile Harbor
  1 Kilometer
  15 Kilometers
Continental Shelf
Burn Zone
PCB Waste


 1.3
 0.06
 0.019
 0.0019
EDC Waste


 1.9
 0.12
 0.002
 0.0004
     The  probability  of a spill  involving  three  or  more
     tanks  of  the vessel in any  location is about  one  in
     24,000 per year,  and in the  pier, harbor, and bay area
     is about one in 50,000 per year.
Source:  Exhibit 7-18

-------
                                Exhibit 1-8

      SUMMARY OP MARINE ECOSYSTEM EFFECTS FROM SPILLS OF HALF A TANK
Release Location
	PCB Waste	
Effect           Bioconcen-
on               tration
Biomass          Levels
                	EDC Waste	
                Effect       Bioconcen-
                on           tration
                Biomass      Levels
Mobile Bay
  Floating Case
  Sinking Case
Small overall,
severe reduc-
tion for
benthos

Uncertain
3 to 5 orders
of magnitude
Uncertain
Not
Considered
Minor
Not
Considered
Minor
Continental Shelf
  Floating Case


  Sinking Case
Uncertain
Small overall,
substantial
for benthos
Uncertain
2-3 orders
of magni-
tude
Not
Considered

Minor
Not
Considered

Minor
Burn Zone
  Floating Case
  Sinking Case
Uncertain
Minor overallr
substantial
for benthos
Uncertain
1-2 orders
of magni-
tude for
benthos and
demersal fish
Not
Considered

Minor
Not
Considered

Minor
Source:  lEc Analysis

-------
OVERVIEW OF ANALYTIC METHODS                            CHAPTER 2
INTRODUCTION

     This   chapter  introduces  the  analytic  methods  used  to
estimate releases of waste and hazardous by-products from  ocean-
and  land-based systems and to characterize the effects of  these
releases.   In addition, the chapter presents the sources of data
and  the important assumptions used in this study.   The sections
below  review our rationale for structuring the ocean- and  land-
based systems as we have,  the methods and data used to  estimate
releases  from each component of these systems,  and the  metrics
and  methods  selected  to  characterize  the  effects  of  these
releases.   The final section provides some thoughts on the types
and magnitudes of uncertainty in our results.


INCINERATION SYSTEM STRUCTURE

     As  indicated in Chapter 1,  the analysis of the ocean-based
incineration  system  is   based on  Chemical  Waste  Management,
Inc.'s  (CWM) proposal to operate the Vulcanus II from Chickasaw,
Alabama  through   Mobile Bay to the Gulf of  Mexico  burn  zone.
CWM's proposal envisions  delivery of wastes to Chickasaw in tank
trucks;  transfer, storage and loading of wastes at an integrated
port   facility;   and  transport  to  the  Gulf  burn  zone  and
incineration  aboard  the Vulcanus II.  Because CWM has  not  yet
developed waste storage capacity at Chickasaw, we also analyze an
operation  in  which  wastes are  delivered  from  generators  to
intermediate  storage at CWM's existing Emelle,  Alabama  storage


                              2-1

-------
facility—150  miles inland from Chickasaw—and are trucked  from
there  to the port to be loaded directly onto the Vulcanus.  This
operation  increases  the  total  distance that  the  wastes  are
shipped  and  the  number  of  transfer  operations  involved  in
delivering the wastes to port.I/

     The  analysis  of  land-based  incineration is  based  on  a
hypothetical  incineration system,  the characteristics of  which
reflect  (1)  data on several existing incinerators and   (2)  our
assumptions about similarities and differences between a  typical
land-based  system  and CWM's proposed operation.   We  employ  a
hypothetical land-based system for two reasons.   First/  because
numerous  land-based  incinerators  may offer an  alternative  to
incineration  aboard  the Vulcanus,  we did not wish to   tie  our
analysis  to the configuration or location of a single  facility.
Second, the throughput of PCB wastes for the Vulcanus exceeds the
capacity  of each of the three land-based incinerators  currently
permitted to burn PCBs.  Since comparison of the risks from land-
and ocean-based systems requires identical waste throughputs, the
use of a hypothetical land-based unit is necessary.

     Our   comparison   of  risks  from   land- and   ocean-based
facilities is based on annual throughput of 56,000 metric tons of
the  PCB waste or 68,400 metric tons of the EDC waste defined  in
Chapter  1.  These  throughputs  are achievable if CWM  does  not
develop  waste storage capacity at Chickasaw.  If  waste  storage
capacity  is  installed at the port,  our calculations show  that
maximum  throughput for the ocean-based system could increase  by
35  to  55  percent (see Appendix A).  We have  used  the lesser
throughputs as the basis for our analysis for two reasons:
I/We  have not examined potential releases  resulting  from  systems
like those proposed by Seaburn,  Inc.  and  Oceanic  Environmental
Services,  Inc.  Both companies propose to  use Class  1  intermodal
containers  to  transport wastes.   Seaburn plans to  use a  barge
with  two horizontally mounted incinerators as   the   incineration
vessel.   The  barge  would carry approximately  144 5,000  gallon
containers of wastes above deck.   Oceanic  Environmental proposes
to  use a conventional offshore supply vessel outfitted with  one
horizontally  mounted  incinerator.    The  vessel  would    carry
approximately  80,000  gallons of wastes  in 16 containers  stored
above deck.   Each company proposes to use  sea water  scrubbers to
cool incinerator off-gases and direct them  into  the ocean.


                              2-2

-------
     (1)   CWM anticipates that demand for ocean incineration
          services in the Gulf area,   if permitted, would be
          approximately  60,000  metric tons per year  (even
          with an improved port facility); and

     (2)   Use  of the throughput estimate for  the  existing
          facility  allowed  us  to  perform  a  sensitivity
          analysis  examining the risks associated with  use
          of intermediate waste storage at Emelle.

     For    both  the  land- and  ocean-based  systems,   we  have
organized  our  analysis of release quantities by three  or  four
components,  respectively.  As  described  in  Chapter  1,  these
components include:

     o    Land Transportation:  transport of wastes by truck
          from  the  generator site to  the  incinerator  or .
          pier;

     o    Transfer   and  Storage:   transfer  and   storage
          operations at the land-based incinerator,  pier or
          other storage facilities;

     o    Ocean Transportation:  for the ocean-based system,
          transport  of wastes by ship from the pier to  the
          burn zone; and

     o    Incineration: incineration of the waste.
METHODS TO ESTIMATE QUANTITIES RELEASED

     The  releases  from different system  components  vary  from
relatively  unlikely releases of large quantities of waste   (such
as  spills due to accidents involving tank trucks) to very likely
releases  of small quantities of waste (such as stack  emissions,
minor pump leaks, and so forth). In view of the stochastic nature
of  releases from many of these release points,  the quantity  of
hazardous material released to the environment is likely to  vary
from  year  to year.  Estimation of release quantities must  take
these variations into account.
                              2-3

-------
     Ideally,  one would determine the probability of release for
all  possible  release  quantities  at  each  release  point.   A
probability distribution of potential releases could be developed
for  each  release  point and for the system as  a  whole.   This
distribution  would indicate the likelihood of  releases for  all
possible release quantities.  Such distributions can be developed
for  only a few components of the land- and ocean-based  systems.
Thus,  we develop estimates of the "expected value" of release —
the  product  of  the  probability of  release  and  the  average
quantity released for each release point.   In addition, for each
point  we estimate the probability that releasing  events  occur,
the  total quantity at risk (i.e.,  the maximum spill  possible),
and the average fraction of the quantity at risk released when an
event occurs. For convenience, we use the following abbreviations
and definitions throughout this report:

     EN   is  the number of releasing events expected over a
          one year period, and is calculated from underlying
          probabilities of events multiplied by the level of
          activity projected over the year,

     Q    is the quantity at risk,  for example the contents
          of one truck or one ship,

     ERF  is  the expected release fraction if  a  releasing
          event occurs,

     QR   is  the  average quantity released if a  releasing
          event occurs, and is the product of Q and ERF, and

     EV   is the expected value of release over a year,  and
          is the product of EN and QR.

None  of  these  single  metrics  is  adequate  to  describe  the
underlying  probability  distribution  of  releases,   but  taken
together   they   provide   a  more  complete  picture   of   the
probabilities  of  release and quantities of  hazardous  material
involved in the ocean- and land-based systems.

     Releases  from the incineration component of each system are
continuous and certain to occur,  although the magnitude of these
releases  is  uncertain.   We estimate both  stack  and  fugitive
losses of undestroyed wastes,  products of incomplete combustion,
metals  included in the waste stream,  and chlorine (in a variety
of chemical forms).   Our estimates of these quantities are based
on  our  assumptions  about  waste  composition  and  incinerator


                              2-4

-------
performance and on our analysis of trial burn data for  land- and
ocean-based incinerators.

     To  develop information on the  probability and quantity  of
releases,  we  rely on many sources of information including past
EPA and other studies,  trade and academic literature, government
datasets,  and  judgments  from various  experts.   All  specific
sources  are  cited in the following chapters and  appendices  of
this   report.   EPA   contracted   with   Engineering   Computer
Optecnomics, Inc. (ECO) to provide information on vessel casualty
rates and spill quantities; and with Arthur D. Little, Inc. (ADD
to provide information and analyses on releases from transfer and
storage  operations,  and  on the   generation,  composition  and
disposition of scrubber effluent at land-based incinerators.  Dr.
Jeff   Kolb  of  EPA18  Office  of  Policy   Analysis   developed
information  on  comparable  hazards as well as  other  data  and
insights useful in the study.


HUMAN AND ENVIRONMENTAL EFFECTS ANALYZED

     The  release of hazardous wastes and combustion  by-products
into  the  environment can cause a wide variety of  effects.  For
purposes  of discussion,  we group the effects of  major  concern
into four categories:

     1.   acute  human  health effects:  effects  caused  by
          short-term  exposure to  hazardous releases or  by
          explosions,  fires and other accidental events;

     2.   chronic  human health effects:  effects caused  by
          long-term exposure to hazardous releases;

     3.   acute    environmental   effects:    damages    to
          terrestrial  and marine ecosystems due  to  short-
          term exposures to hazardous releases; and

     4.   chronic   environmental   effects:    damages   to
          terrestrial and marine ecosystems due to long-term
          exposures to hazardous releases.


     Hazardous  releases from land- and ocean-based  incineration
can  be  brought  into  contact  with  human  and   environmental
populations  through a number of pathways,  many of which involve
complex  dispersion phenomena.  The estimation of movement  along


                              2-5

-------
these  pathways and eventual human and environmental exposure  is
extremely  difficult,  particularly  at long  distances  and  for
materials  that  are persistent and  that  bioconcentrate.  Given
current  scientific  knowledge  it is not possible to  fully  and
accurately predict the movement of hazardous materials once  they
are  released from incineration systems.  Further,  data to allow
accurate prediction of human health or environmental damage as  a
function  of  exposure to acute and chronic levels  of  hazardous
materials are limited.

     In view of these difficulties, we focus our efforts on those
release   points,    environmental   pathways   and   human   and
environmental  effects  we  judge most important to  the  overall
comparison of land- versus ocean-based incineration.  Exhibit 2-1
illustrates all possible combinations of the four release  points
defined  for incineration systems and the four types of human and
environmental effects outlined above.  As the exhibit shows,  our
major  efforts concern the ocean transport and  land- and  ocean-
based incineration release points.

     As  Exhibit  2-1 shows,  we have not evaluated  the  effects
resulting  from  land  transportation  releases.  Since  we  have
assumed  land transportation release quantities would be the same
for  the  land- and ocean-based systems,  the  resulting  effects
would be equivalent, on average, for the two systems.

     We  evaluate  acute  human  health  effects  for  the  ocean
transport release point,  since inhalation of hazardous materials
volatilized  from spills could affect human health.   Acute human
health effects would not be expected from the other releases  and
resulting   environmental  concentrations,   with  the   possible
exception  of  land  transportation  accidents.   However,  acute
effects  such as injury or death could occur as a result of  land
or  ocean  transportation accidents,  explosions or fires  during
transfer  or  storage operations,  and so  forth.   We  have  not
estimated  the magnitude of these accidental injuries or  deaths;
in  view  of the low probabilities assigned to these  events  the
expected  effects  would be very  small.   In  addition,  similar
accidents  would  be expected (except for those  involving  ocean
transportation) in both systems.

     Our  major  efforts  in  transport  and  effects  estimation
concern  the ocean transportation and land and ocean incineration
release points.  The dispersion of ocean transport spills and the
resulting  effects  on  the marine ecosystem  are  estimated  for
several  spill  locations  and  quantities.    In  addition,  the


                              2-6

-------
bioconcentration of persistent compounds is  estimated to provide
information  on  contamination levels and as a proxy  for  marine
chronic effects.

     The releases from incineration itself occur continuously and
with  certainty  rather than only in the (unlikely) event  of  an
accident.    We analyze the chronic human health effects  expected
due  to inhalation from stack emissions and to ingestion of  food
contaminated  by  fallout  from the  stack  plume.   Further,  we
consider possible effects on the marine ecosystem from deposition
of  organic chemicals,  metals and chlorine from the  plume,  and
effects on the land ecosystem from the final disposal of scrubber
effluent  and  sludge.   We have not been able  to  consider  the
effects   of  land-based  stack  emissions  on  the   terrestrial
ecosystem  due  to limited time and resources.   In  addition  to
these major efforts, we evaluate the chronic human health effects
expected  from inhalation of fugitive releases from transfer  and
storage.

     lEc used several methods to estimate pollutant transport and
effects   from  ocean  transportation  releases  and  incinerator
releases for both land- and ocean-based systems.   First, we used
EPA-developed air dispersion models and human health effects data
to estimate the human health risks due to inhalation of hazardous
materials released from land-based incinerators.  In addition, we
used recent EPA studies to estimate additional human health risks
from ingestion of foodstuffs contaminated from incinerator  stack
emissions.   Further,  lEc  subcontracted  with  Applied  Science
Associates, Inc. (ASA) to provide air dispersion analyses for the
ocean-based incinerator;  lEc used these estimates with EPA human
health   effects  data  to  estimate  human  health  risks   from.
inhalation of ocean stack emissions.

     For   effects   for  ocean  transportation   releases,   lEc
subcontracted  with ASA to provide ocean transport and  ecosystem
modeling  for hypothetical spills at three sites:  Mobile Harbor,
over  the continental shelf on the ship's path to the burn  zone,
and  in  the burn zone itself.   Along with  ASA's  results,  lEc
calculated  the possible human health effects from marine  spills
due to spill volatilization and subsequent human inhalation.


UNCERTAINTY

     Our   estimates  of  release  from   land- and   ocean-based
incineration  systems  and the resulting human and  environmental


                              2-7

-------
effects are highly uncertain.  In view of this uncertainty, it is
useful  to consider how these results should be  interpreted  and
whether  they represent typical or "worst-case" outcomes.   These
issues are discussed in the paragraphs below.

     In  conducting  this  analysis,  we have  structured  system
components and waste streams which are roughly typical of  ocean-
and  land-based incineration.   We have not located or configured
either  system in a manner which would yield worst or  best  case
releases or effects.   In addition, we have not assumed the worst
or best possible waste compositions.   In short,  the systems and
wastes analyzed are typical or average representations.

     In estimating release probabilities and quantities, we often
were   forced  to  use  data  not  directly  applicable  to   the
incineration system as structured.   For example,  in considering
land  transportation releases we assumed a vehicle accident  rate
similar  to  those  experienced by tank  trucks,  and  for  ocean
transport   our  accident  rates  are  estimated   by   adjusting
historical  rates  for tank ships to account for the  design  and
operating characteristics of the Vulcanus, rather than by  relying
on  the operating record of incinerator ships.   Use of such data
was necessary since adequate information for incinerator   systems
does not yet exist.

     When  specific  information for transportation and  handling
was not available,  we selected other data conservatively.  As  a
result,  we err on the side of overestimating release  quantities
and  probabilities.   With  the exception of releases from  ocean
transport, this overestimation error is the same for both  systems
and thus does not effect the evaluation of comparative risk.  The
likely  overestimate of ocean transport releases could  bias  the
analysis against the ocean-based system; however, these estimated
releases  are a  small portion of the expected long-term   release
from the incineration systems.

     Our  estimates of the releases from incineration itself  are
driven  by our assumption about waste composition and incinerator
performance  (undestroyed waste,  metals,  chlorine) and   by  our
analysis  of  trial  burn  data (PICs).   Our  estimates   of  PIC
emissions are extremely uncertain and could well be biased to the
low side.  It is not possible to judge whether the extent  of this
bias, as well as the overall uncertainty, are similar for  the two
systems  considered,  and thus our PIC estimates  could  strongly
affect the comparative risk assessment.
                              2-8

-------
     Available methods to estimate the dispersion of and  effects
from  releases of hazardous substances generally use conservative
methods.   For  example,  both atmospheric and  ocean  dispersion
models rely on simplified representations of winds,  currents and
other  physical  phenomena.   This results in  overestimation  of
concentrations,   especially   at  longer  distances.    Further,
coefficients  developed to predict human or environmental  damage
as a function of exposure or dose are often calculated at the  95
percent  confidence  interval and assume all of the  compound  is
biologically  active.   As a result,  these dose-response factors
overestimate  the  actual effects likely from a  given  level  of
exposure.

     It is very difficult to consider all dispersion pathways and
effects  for compounds that persist in the environment and  which
bioaccumulate.    For  these  compounds,  available  methods  may
overestimate the effects considered while ignoring other  effects
that  occur  at  longer times and  distances.   If  these  latter
effects  are important,  our analysis may underestimate the total
eventual effects of such compounds.

     In  general,  we have tried to use equally conservative data
and  methods  to  estimate  effects  from  land- and  ocean-based
systems  in  order to generate  comparable  estimates.   However,
precise data on the likely uncertainty in these estimates are not
available.  Further, as explained previously, we were not able to
look  at all possible effects for both systems.    In  particular,
limited  time and resources precluded consideration  of  possible
effects  of  releases  from  either incineration   system  on  the
terrestrial ecosystem.

     As  a  result of these factors,  we believe that  while  the
incineration  systems considered are structured to be  "average",
our  estimates  of release quantities and resulting  effects  are
biased  to the high side.   We have tried to be consistent in our
choice  of methods and assumptions used to analyze  each  system,
and thus the differences shown between systems are less uncertain
and  less  biased.   The  purpose of this  study   is  comparative
assessment of ocean- versus land-based incineration systems.  The
validity  of  this comparison will be most strongly  affected  by
methods  or assumptions (such as PIC rates,  scrubber efficiency)
that change the relative performance of the systems.
                              2-9

-------
SUMMARY

     This chapter has reviewed the measures, analytic methods and
general sources of data used in the study.  In addition, the text
above  provides further information on the assumed  structure  of
the  incineration  systems  and on  the  sources,  directions  and
magnitudes of uncertainty in our results.
                              2-10

-------
                                                           Exhibit 2-1

                                          POSSIBLE EFFECTS EVALUATED IN THE  ANALYSIS
                       —.............. -Human  effects———————    —————Environmental Effects——-—--—-
Release Point          Acute                      Chronic                    Acute                        Chronic
•»««B
-------
WASTE QUANTITIES
RELEASED FROM LAND TRANSPORTATION                       CHAPTER 3
     The  analysis of releases from land transportation considers
two types of potential losses — vehicular accidents and  enroute
container failures from causes such as loose fittings,  corrosion
or internal pressure.   Both of these events are probabilistic  in
nature.   We  base our estimates of potential releases from  such
events  on  existing  studies of accident and failure  rates  and
available  information on the size distribution of the  resulting
spills.   Using these data,  we calculate the expected number   of
annual  releases,  the average quantity of waste released given a
spill,  and  the  expected annual release  quantity.    Below   we
discuss our sources of data/  our assumptions,  and  characterize
the resulting releases.  Final sections of the chapter review the
limitations  of our estimates and compare our results with likely
releases from general commerce.


DATA SOURCES

     The  analysis of releases from land transportation is  based
on several sources of information.  Data concerning the frequency
of  vehicular  accidents and container failures were provided   by
the U.S.  Department of Transportation (DOT).  Information on the
size  of release (the percentage of total truck cargo  discharged
in the event of an accident) was developed using DOT'S  Hazardous
Materials  Incident  File (HAZMAT).   Using this   information   we
estimate annual releases for trucks transporting hazardous wastes
from  waste generators to CWM's, proposed Chickasaw,  Alabama port
facility or to a land-based hazardous waste incinerator.
                              3-1

-------
ASSUMPTIONS

      The  analysis  of  releases  from  transporting wastes on land is
based on  the  following  assumptions:

      o   Wastes are  transported  in  5000  gallon tank  trucks;

      o   In  the ocean-based case,   the trucks  carry   wastes
          an   average  of   250 miles from the   generator   to
          CWM's  proposed Chickasaw  port   facility;   in  the
          land-based  case,  the trucks travel 250  miles from
          the   generator   to    a    commercial   land-based
          incinerator;

      o   All  other conditions  (for  example,  speed,  highway
          composition,   road  surface,    weather,    traffic
          volume,   share   of  day   and   night   travel)   are
          average.

      It   is   important   to  emphasize that we  assume   equal   trip
distances for the land- and ocean-based  systems.   Although   the
average distance from generators  to  the transfer stations serving
land- and ocean-based   incineration may  differ,    analysis  of
typical   trip  distances is  beyond the scope of  the present study.
A  study  of   trip distances would   require information   on   the
location  of   generators,   land-based incinerators,   and future
ocean-based  facilities,  as well as data on the capacity of  each
facility  and the demand for incineration  services.

      To   examine   the  effect of  differing trip   distances,   we
perform   a sensitivity  analysis assuming   waste storage  capacity
does   not exist at the port.    In  this  case,   we assume wastes
would be  trucked 250  miles  to CWM's  existing storage  facility  in
Emelle, Alabama, stored there, and later  transported  150  miles to
Chickasaw for  immediate loading onto the  Vulcanus.  This  scenario
increases the total trip  distance  in the ocean-based case   from
250 to 400 miles.I/
I/CWM  currently  plans to construct storage  tanks  at   Chickasaw.
The  company may be prohibited from doing   so,   however,   if   the
tanks  would  be  located in a floodplain.    If  the Vulcanus   is
permitted before storage tanks are constructed,   CWM plans to  use
its Emelle facility as an intermediate storage site.
                              3-2

-------
RELEASES PROM VEHICULAR ACCIDENTS

     According  to DOT'S Bureau of Motor Carrier  Safety,  trucks
are  involved in enroute vehicular accidents at a rate  of  about
1.2  accidents per million miles traveled.   About 29 percent  of
accidents  involving tank trucks result in a release of  cargo.2/
Therefore,  we estimate the rate of releasing vehicular accidents
for  tank  trucks  to be 0.35 per million  miles  traveled  (0.29
releases per accident x 1.2 accidents per million miles).

     We base our analysis of releases from vehicular accidents on
this accident rate, assuming that trucks that transport hazardous
wastes  at worst experience the same release rate from  vehicular
accidents  as  estimated for tank  trucks.  This  assumption  and
alternative  estimates of releasing vehicular accident  rates  are
discussed later in this chapter.


Frequency of Vehicular Accidents

     Given  a  releasing accident rate of 0.35 per million  miles
traveled,  the  expected  number  of  releasing  accidents  is  a
function of the total number of miles traveled to deliver  wastes
from the generator to Chickasaw or to the land-based incinerator.
In  turn,  this  is a function of route length and the  number  of
shipments  required  to  deliver the  annual  quantity  of  waste
incinerated.

     Exhibit  3-1   summarizes the calculations used to estimate
the number of shipments required to transport 56,000 metric  tons
of  PCB  wastes  or 68,400 metric tons of  EDC  wastes.   As  the
exhibit shows,  151 shipments of PCB wastes would be required for
each  voyage of the Vulcanus.   Assuming 14 voyages per year  for
the PCB waste (see Appendix A),  2,114 truckloads are required to
deliver the 56,000 MT annual throughput.   More shipments  (2,934)
are  required  to deliver the EDC waste each  year,  due  to  the
larger annual throughput and the lower density of the waste.
2/Derived  from a memorandum   from   J.   Nalevanko,   Materials
Transportation    Bureau,    Research   and   Special    Programs
Administration,  Department of Transportation,   to Dr. Jeff  Kolb,
EPA, 30 January 1985.
                              3-3

-------
     Exhibit  3-2  uses  the  above estimates  to  calculate   the
expected  number of vehicular accidents per year for the PCB   and
EDC cases,  assuming each shipment travels 250 miles.   In  the  PCB
case,  the total distance traveled is 528,500 miles annually.  At
a  releasing  accident  rate  of  0.35  per  million  miles,   the
expected  number  of releasing vehicular accidents  is  0.18   per
year.  The expected number of vehicular accidents in the EDC case
is  about  0.26  per year,  higher than the PCB case due   to   the
greater number of shipments and greater total distance  traveled.


Waste Quantities
Released from Vehicular Accidents

     Using  HAZMAT  data on the quantity of truck cargo  lost  in
releasing vehicular accidents, we estimate the  fraction of total
cargo expected to be spilled if a releasing accident occurs.   The
HAZMAT data give the following distribution of release  fractions.

                            Table 3-1
        Quantity Released in a Releasing Truck Accident


       Percent of Capacity
            Released                 Percent of Releases
              0-10%                         31.6%
             10-30%                         18.7
             30-50%                         13.3
             50-70%                          9.8
             70-90%                         11.9
            90-100%                         14.7

       Source:  HAZHAT
These  data indicate that over 30 percent of releasing  vehicular
accidents  involve  less  than 10 percent  of  the   total   cargo.
However,  14.7  percent of releasing vehicular accidents   involve
over  90 percent of the total cargo.   The bimodal nature  of   the
distribution  indicates that accidents tend to be either minor  or
severe enough to cause a substantial loss of cargo.   The  average
(or  expected)  release is approximately 39.5  percent  of total
cargo.
                              3-4

-------
     Exhibit  3-3  shows the calculation of the  expected  annual
release from vehicular accidents for both the PCS and EDC  cases.
These  calculations  are based on the quantity of cargo at  risk,
the expected release fraction,  and the expected number of annual
releasing accidents.  The average release quantity  for PCB wastes
is 10.5 metric tons,  while the average release quantity for  EDC
wastes  is 9.2 metric tons.   The expected annual release is  1.9
metric  tons  and 2.4 metric tons in the PCB case and  EDC  case,
respectively.


RELEASES FROM CONTAINER FAILURES

Frequency of Container Failures

     The   DOT  data  indicate  that  tank  trucks   transporting
hazardous wastes will experience enroute container  failures   from
causes other than vehicular accidents at a rate of  about 0.43 per
million miles traveled.   Again, the expected number of container
failures  per  year experienced by trucks carrying  wastes  is   a
function  of  the total number of miles traveled to deliver  the
waste  from the generator to Chickasaw or to a land-based incine-
rator.    Exhibit  3-4  shows  the  expected  number  of  enroute
container  failures is 0.23 per year in the PCB case and 0.32   in
the  EDC case.   The rate of release for EDC is greater than  for
PCBs due to the greater number of miles driven.


Haste Quantities
Released from Container Failures

     As with vehicular accidents,  we use the  HAZMAT data on the
quantity of truck cargo lost in the event of a container  failure
to  estimate  the  fraction of total  cargo  spilled.   The   data
indicate the following distribution.
                              3-5

-------
                          Table 3-2
            Quantity Released from Container Failure
       Percent of Capacity
            Released                 Percent of Releases
              0-10%                         92.6%
             10-20%                          3.3
             20-50%                          2.0
            50-100%                          2.1
       Source:  HAZMAT
These  data  indicate that over 90 percent of  enroute  container
failures release less than 10 percent of the cargo.   The average
release is approximately 4 percent of total cargo.

     Exhibit  3-5  shows the calculation of the  expected  annual
release  from  enroute  container failures for the  PCB  and  EDC
cases.    The  expected  annual  release  in  the   PCB  case    is
approximately  0.25 metric tons;  and the expected  annual release
in the EDC case is 0.29 metric tons.
CAVEATS AND SENSITIVITY ANALYSIS

Vehicular Accident Rate

     The  accident rate for trucks transporting hazardous  wastes
could be lower than the rates estimated from DOT data for  several
reasons.

     1.   The  DOT data include trucks with aluminum  tanks.
          Trucks carrying wastes to Chickasaw or to a   land-
          based incinerator will have stainless steel tanks,
          making them more resistant to puncture or rupture.

     2.   CWM  plans  to  hire  only  experienced   drivers,
          provide  special safety training,  and take   other
          precautions  to avoid accidents.   These  measures
          should  help reduce the likelihood that CWM trucks
          will be involved in vehicular accidents.
                              3-6

-------
      3.    Data   from  several   commercial   waste  management
           firms     indicate   an    average   accident   rate
           (releasing   and  non-releasing)   for   all   cargo
           vehicles  operated by these  companies of about  0.8
           per  million  miles.    This estimate is  based  on
           about   24   million vehicle  miles   of   travel.
           Additionally,  no releases of hazardous wastes from
           tank trucks  due  to accidents or  container failures
           were   experienced in approximately 9 million miles
           of travel.    These accident and  release rates  are
           significantly lower,  at  the 95  percent confidence
           level,  than  the rates used in  our analysis.   We
           infer  from this  that trucks transporting hazardous
           wastes are   less likely  than  other trucks  to  be
           involved  in  releasing accidents.I/

 In   light  of these  considerations,  we believe our  estimates  of
 vehicular  accident rates  are higher than the   actual  accident
 rate   for  tank trucks  that carry  hazardous wastes.  Any change in
 the  assumed accident rate  will  have a proportional effect on  the
 the  estimated release  quantities.


 Trip  Distance and the  Potential Effect
 of intermediate  Waste  Storage  at  Emelle

      A second factor affecting  the  analysis of releases from land
 transport  is the assumption that  each shipment travels 250 miles.
The  accident  and  container failure  rates  are   expressed on  an
incidents-per-mile   basis.   Therefore,  estimates  of  both  the
expected  number  of   releases  and  the  expected  annual   release
quantity  are linearly  related  to the trip  length assumed.   Given
this   relationshipr   alternative  trip distances  would    pro-
portionally affect  expected annual  release  estimates.

     To  illustrate  this  effect,   we recalculate  the   expected
number of  releasing  accidents and container failures per  year and
the annual expected  release quantity  for the land  transportation
component  of the ocean-based  incineration  system,   assuming that
wastes are stored at CWM's  Emelle storage  facility rather than at
I/Memorandum from Dr.  Jeff Kolb  (EPA) to Michael Huguenin  (lEc),
"Transportation  Data  from  Companies  in  the  Hazardous  Waste
Business," 24 September 1984.


                              3-7

-------
a   newly-built  storage  facility  in   Chickasaw.    As   noted
previously, we assume that this arrangement adds 150 miles to the
average  trip  distance for each shipment of waste,  raising  the
total trip distance from 250 to 400 miles.  The results are shown
in Exhibit 3-6.

     As this exhibit shows, the total distance traveled increases
to 845,600 miles and over one million miles for the PCB case  and
the  EDC case,  respectively.   The expected number of  releasing
accidents  per year is about 0.30 in the PCB case and 0.41 in the
EDC case, with expected annual releases of 3.14 MT of  PCB  waste
and  3.79  MT  of  EDC waste.   The expected  number  of  enroute
container  failures  per year is about 0.36 in the PCB  case  and
0.50 in the EDC case, with expected annual releases of 0.38 NT of
PCB waste and 0.47 MT of EDC waste.   These figures are about  60
percent  greater  than  the expected  releases  in  our  baseline
analysis, reflecting the increase in miles traveled.


Effect of Remedial Action

     As  a  final point,  we note that most commercial  hazardous
waste  management firms maintain spill response teams trained  to
control  and remove wastes released in vehicular accidents or  as
the  result of container failures.   Remedial action will  reduce
the quantity of waste ultimately released to the environment, and
help to control the hazards presented by spills.


COMPARABLE HAZARDS

     To   compare  the  magnitude  of  possible   releases   from
transporting  hazardous wastes to releases from the transport  of
hazardous  materials  in  daily commerce,  we  examined  data  on
releases  of  hazardous substances maintained by the  U.S.  EPA's
National  Response Center (NRC).A/  According to the  NRC,  there
were  67 transportation-related  hazardous substance releases  in
EPA Region IV (an eight-state area of the southeastern U.S.  that
includes   Alabama)   in   the  last  seven   months   of   1982.
A/The  NRC  data  were reported in ICF,   Inc. ,   Release  oJL
Production  Volume  Substances.   April   1983.   These   data   are
adjusted  to remove petroleum and petroleum product   spills,   and
include only CERCLA-designated hazardous  substances.


                              3-8

-------
Extrapolating   these   data  to  a  full   year   suggests   115
transportation-related  hazardous substance releases would occur.
The expected number of vehicular accidents and enroute  container
failures   involving   trucks  serving  a  land- or   ocean-based
incineration  system (0.41 per year in the PCB case and 0.58  per
year  in the EDC case)  would represent about 0.5 percent  of  the
estimated number of transportation-related releases of  hazardous
substances reported to the NRC for Region IV.


SUMMARY OF RELEASES
FROM LAND TRANSPORTATION

     Exhibit  3-7  summarizes the total expected  annual  release
quantities from the land transportation component of a land-based
and   ocean-based  incineration  system.    If  hazardous   waste
generators  are  on average equidistant from the  waste  transfer
stations serving land- and ocean-based incinerators, the expected
releases  from  the  land transportation components  of  the  two
systems are equal.  As shown in this exhibit, the expected annual
releases  are  2.1 MT for the PCB waste and 2.7 MT  for  the  EDC
waste.   The  expected annual release from vehicular accidents  is
about  eight  times the expected annual  release  from  container
failures,  despite  the  projection that slightly more  releasing
container  failures  are likely to occur.   The  larger  expected
release from vehicular accidents reflects the greater probability
that  these events release a significant fraction of total  truck
cargo.

     Releases  from the land transportation component of a  land-
based  or  ocean-based incineration system can  occur  relatively
frequently.  In the PCB case,  an average of about 0.18 releasing
accidents and 0.23 container failures are expected per year.    In
the EDC case, an annual average of about 0.26 releasing accidents
and 0.32 container failures are expected.  However, many of these
releases would involve only a small quantity of waste.  Almost  93
percent  of  enroute container failures are expected  to  release
less than ten percent of the total cargo.  The release quantities
associated with vehicular accidents are larger, but 32 percent  of
such incidents are expected to release less than ten percent of  a
waste shipment.

     Despite  the expectation that the total release from  trucks
transporting  wastes will be small,  there is still a possibility
that a significant release will occur.  For example,  we estimate
that there is about a 0.09 probability that a truck  transporting


                              3-9

-------
EDC wastes will be involved  in a vehicular  accident  that  releases
more  than  50  percent of its 23.4 metric  ton  cargo  during   any
particular  year.   Thus,  it is important  to bear  in   mind   that
releases  during  any single year could  be  significantly   greater
than the expected annual  release quantity.

     If  transportation   distances for the  land- and  ocean-based
systems differ, expected  annual releases  from land transportation
will also differ.   This  effect is illustrated  by the  analysis of
releases should CWM store wastes in Emelle  rather than Chickasaw.
The  60 percent increase  in transport distances  associated   with
this  arrangement would increase expected'annual releases for the
land  transportation  component of the ocean-based   system by  a
corresponding  percentage.   Even  in  this  case,   however,   the
expected  number  of  releasing incidents involving  tank trucks
serving  an incineration  system represents  less than one   percent
of  the transportation-related releases  of   hazardous   substances
currently reported for EPA Region IV.

     Finally,  available  data for the hazardous waste management
industry  indicate  a  relatively  low   frequency  of   releasing
accidents.   These  data  suggest that our calculation  of  expected
releases  is conservative and probably overstates the  quantity of
waste released from land  transportation.
                              3-10

-------
                           Exhibit 3-1

           CALCULATION OF NUMBER OP SHIPMENTS REQUIRED
          FOR TANK TRUCKS DELIVERING PCB OR EDC WASTES
Assumptions:
                                     PCB Waste          EDC Waste
  Truck capacity (gallons)             5,000              5,000
  Specific gravity                     1.4000             1.2351
  Ship capacity (metric tons, MT)      4,000              3,800
  Voyages per year                     14                 18


Calculations:  PCB Case

  Cargo weight (MT)  =  5,000 gallons x 3.785 liters/gallon *
                        1.4 kilograms/liter * 10E-3 MT/kilogram
                     =  26.5

  Truckloads per
    voyage           =  4,000 MT per voyage/26.5 MT per truckload
                     =  151

  Truckloads per
    year             =  151 truckloads per voyage  * 14 voyages
                        per year
                     =  2,114

Calculations:  EDC Case

  Cargo weight (MT)  =  5,000 gallons * 3.785 liters/gallon *
                        1.2351 kilograms/liter  * 10E-3 MT/kilogram
                     =  23.4

  Truckloads per
    voyage           =  3,800 MT per voyage/23.4 MT per truckload
                     =  163

  Truckloads per
    year             =  163 truckloads per voyage  * 18 voyages per
                        year
                     =  2,934
Source:  lEc Analysis

-------
                           Exhibit 3-2

      CALCULATION OF EXPECTED NUMBER (EN) OF RELEASING VEHICULAR
              ACCIDENTS PER YEAR FOR TANK TRUCKS SERVING
              LAND- OR OCEAN-BASED INCINERATION SYSTEMS
Assumptions:
                                  PCB Waste          EDC Waste
  Truckloads per year               2,114              2,934
  Trip length (miles)               250                250
  Incident rate (IR)
  (releasing accidents/mile)        3.5E-7             3.5E-7


Calculations:  PCB Case

  Annual
  Involvement (AI)   -  2,114 truckloads/year * 250 miles/truckload
                     =  528,500 miles/year

                EN   -  AI * IR
                     =  528,500 miles/year * 3.5E-7 releasing
                          accidents/mile
                     =  0.18 releasing accidents/year

Calculations:  EDC Case

                AI   =  2,934 truckloads/year * 250 miles/truckload
                     =  733,500 miles/year

                EN   -  733,500 miles/year * 3.5E-7 releasing
                          accidents/mile
                     =  0.26 releasing accidents/year
Source:  lEc Analysis

-------
                           Exhibit 3-3

        CALCULATION OF EXPECTED ANNUAL RELEASE (EV) FROM
             VEHICULAR ACCIDENTS FOR TRUCKS SERVING
            LAND- OR OCEAN-BASED INCINERATION SYSTEMS
PCB Waste

Assumptions:
     Expected release fraction  (ERF) = 0.395
     Quantity at risk (Q) = 26.5 MT
     Expected number of releases per year  (EN) =  0.18

Calculations:

     Average quantity
     released (QR)       = Q *  ERF          EV = QR * EN

                         = 26.5 MT  * 0.395     = 10.5 MT * 0.18

                         = 10.4675  MT          - 1.89 MT

EDC Waste

     Assumptions:

          ERF = 0.395
          Q   = 23.4 MT
          EN  =0.26

     Calculations:

                   QR = Q * ERF         EV = QR * EN

                      = 23.4 *  0.395        = 9.2 MT * 0.26

                      = 9.243 MT            = 2.39 MT
Source:  lEc Analysis

-------
                           Exhibit 3-4

    CALCULATION OP EXPECTED NUMBER (EN)  OF ENROUTE CONTAINER
            FAILURES PER YEAR FOR TANK TRUCKS SERVING
            LAND- OR OCEAN-BASED INCINERATION SYSTEMS
PCS Waste

     Assumptions:

          Annual involvement (AI)  - 528,500 miles/year
          Incident rate (IR) = 4.3E-7 container failures/mile

     Calculation:

          EN =  AI * IR

             »  528,500 miles/year * 4.3E-7 container failures/mile

             »  0.23 container failures/year


EDC Waste

     Assumptions:

          AI =  733,500 miles/year
          IR *  4.3E-7 container failures/mile

     Calculation:

          EN -  AI * IR

             «  733,500 miles/year * 4.3E-7 container failures/mile

             =  0.32 container failures per year
Source:  lEc Analysis

-------
                          Exhibit 3-5

        CALCULATION OF EXPECTED ANNUAL RELEASE (EV)  FROM
          ENROUTE CONTAINER FAILURES FOR TRUCKS SERVING
            LAND- OR OCEAN-BASED INCINERATION SYSTEMS
PCB Waste

     Assumptions:

          Expected release fraction (ERF)  = 0.04
          Quantity at risk (Q) = 26.5 MT
          Expected number of releases per year (EN)  = 0.23

     Calculations:

          Average quantity released (QR) =  Q * ERF

                                         =  26.5 MT * 0.04

                                       ,  =  1.06 MT

          EV =  QR * EN

             =  1.1 MT * 0.23

             =  0.25 MT


EDC Waste

     Assumptions:

          ERF =  0.04
          Q   =  23.4
          EN  =  0.32

     Calculations:

          QR  =  Q * ERF

              =  23.4 MT * 0.04

              =  0.936 MT

          EV  =  QR  * EN

              =  0  .9 MT * 0.32

              =  0.29 MT



Source:   lEc Analysis

-------
                             Exhibit 3-6

                 EFFECT OF STORAGE AT EMELLE, ALABAMA
                    ON EXPECTED RELEASE QUANTITIES
                IN THE OCEAN-BASED INCINERATION SYSTEM
Assumptions
                                        PCB Waste           EDC Waste
     Truckloads per year                    2,114               2,934

     Trip length (miles)                      400                 400

     Quantity at risk (MT)                   26.5                23.4

     Releasing accident rate               3.5E-7              3.5E-7

     Container failure rate                4.3E-7              4.3E-7

     Fraction released
       in event of accident                 0.395               0.395

     Fraction released
       in event of container failure         0.04                0.04



Calculations

     Total miles traveled                 845,600           1,173,600

     Expected number
       of releasing accidents per year       0.30                0.41

     Expected number
       of container failures per year        0.36                0.50

     Annual expected release
       from accident (MT)                    3.14                3.79

     Annual expected release
       from container failure (MT)           0.38                0.47
Source:  lEc Analysis

-------
                             Exhibit  3-7

        SUMMARY OF  EXPECTED RELEASES  FROM LAND TRANSPORTATION
Expected Number  of  Releasing Incidents

     Vehicular Accidents
     Container Failures
                                              PCB Waste
0.18
0.23
         EDC Haste
0.26
0.32
Annual Quantity Released (MT):

     Vehicular  Accidents
     Container  Failures

          Total Release
1.89
0.25

2.14
2.39
0.29

2.68
 Source:   lEc Analysis

-------
RELEASES FROM
WASTE TRANSFER AND STORAGE                              CHAPTER 4


     This  chapter  describes  our analysis  of   releases  during
transfer  and  storage of wastes for both  ocean- and  land-based
incineration  systems.   The analysis considers  three sources  of
releases  —spills  when  transferring wastes to and  from  tank
trucks,  spills  from  equipment at waste  transfer  and   storage
facilities,  and  fugitive  emissions from transfer  and  storage.
Fugitive  emissions are slow leaks from pump  fittings,  sampling
connections,  flanges,  storage tanks and other   equipment.  Some
of these releases are probabilistic,  while others are relatively
certain.

     The  following  discussion  describes our data  sources  and
assumptions   and  the  resulting  estimates  of waste    release
quantities for these three sources.  This chapter also includes a
sensitivity analysis examining releases from transfer and  storage
should CWM not build storage tanks at its port facility.


DATA SOURCES

     The  analysis of storage and transfer releases  is  based  on
several sources.   Data on release rates during  truck loading and
unloading operations are based on data supplied  by DOT,  as cited
in  Chapter 3.  Information on the quantity of waste likely to be
released in the event of such spills was obtained from the HAZMAT
file, also described in Chapter 3.  Arthur D. Little, Inc.  (ADD
analyzed  spills and fugitive emissions at storage   and  transfer
                              4-1

-------
 sites.     Appendix   B  summarizes  the  AOL  analysis   of   these
 releases.I/
ASSUMPTIONS

      The   analysis  of  waste releases  during transfer  and  storage
is   based  on  operation of  the facility that CWM proposes to build
at   its port  in Chickasaw.    We assume that similar transfer  and
storage  equipment   is  located at  the  land-based   incinerator.
Appendix B gives a  full description of the probable configuration
of  this facility.

      We also  consider  storage of waste at Emelle,  rather than at
Chickasaw.    This  analysis  considers the following   sources   of
releases:

      o     Spills when unloading and  loading tank  trucks  at
           Emelle and when  unloading at Chickasaw;

      o     Spills from  transfer  and storage  equipment  at
           Emelle and   for  transfer equipment at  Chickasaw;
           and

      o     Fugitive   emissions  from all  equipment at  both
           Emelle and Chickasaw.

As  this list  suggests,   intermediate  waste storage at Emelle adds
several  potential  release  points to  the  ocean-based  incineration
system.


TRANSFERS  TO  AND FROM  TANK  TRUCKS

Frequency  of  Loading/Unloading Spills

      The DOT  data provide  information on  the rate  at  which.spills
at   transportation   terminals  occur.   The  data  indicate that
releases involving  the loading and  unloading of tank  trucks occur
I/We used the DOT and HAZMAT  data  rather  than  the ADL  analysis  to
describe  releases  from  loading  and  unloading  tank  trucks  because
the  ADL analysis is limited  to  releases   from  loading/unloading
arm  failures,  and does not consider  other potential causes   of
loading/unloading spills.


                               4-2

-------
at  a  rate of 1.7 per hundred  thousand  handlings.   Thus,  the
number  of handlings involved in delivering waste to Chickasaw  or
to  a land-based incinerator will determine the probability of  a
transfer  release in a given year.

     The   assumptions for land transport given in Exhibit 3-1  of
Chapter  3  yield  the  following  estimates  of  the  number  of
shipments  required  to deliver annual throughput  quantities  to
Chickasaw or to a land-based incinerator.

     o    PCB wastes:  2,114 shipments of approximately 26.5
          NT each;

     o    EDC wastes:  2,934 shipments of approximately 23.4
          NT each.

Each  shipment must be unloaded at the termination of  the  trip.
(The  original  loading  of  the tank truck is  excluded  in  our
analysis  since this step occurs at the generator site.)

     Based  on these estimates.  Exhibit 4-1 shows calculation of
the expected number of transfer spills in the PCB  and EDC  cases
for  both ocean- and land-based incinerators.   In the PCB  case,
the expected number of spills  is 0.04 per year.  In  the EDC case,
the  expected  number of transfer  spills is 0.05  per  year.  The
slightly  higher chance of a spill in the EDC case  is a result of
the greater number of handlings required.

Waste Quantities Released
from Truck Dnloading Spills

     HAZMAT data on the fraction of  cargo lost  in the event of   a
transfer  release  suggest the following distribution of  release
fractions.
                          Table 4-1
            Quantity Released  from Tranfer Operations

       Percent of  Capacity
            Released                 Percent of Releases
              0-10%                          94.4%
             10-20%                           2.9
             20-50%                           2.0
             50-100%                           0.7

   Source:  HAZMAT


                               4-3

-------
These  data  indicate  that over 94 percent  of   transfer   spills
release less than 10 percent of the cargo.  Less  than one percent
involve  50  to 100 percent of the total  load.    The HAZHAT data
show  that  the  average  fraction  released  is   approximately   3
percent.

     Exhibit  4-2  illustrates the calculation  of  the  expected
quantity  of  waste  released while loading  and  unloading tank
trucks.   The  expected release quantity  in the event PCB   wastes
are  spilled is 0.8 metric tons.   In the EDC case,  the expected
release quantity is 0.7 metric tons.  Over  a year  the  expected
release is approximately  0.03 metric tons in both the PCB and EDC
cases.
SPILLS AT TRANSFER AND STORAGE FACILITIES

     Appendix  B  describes  CWM's   "improved"  Chickasaw   waste
storage  and  transfer   facility  and  the   storage   and  transfer
facility assumed for  the land-based  incinerator.  The  analysis  of
the potential for spills at  each  site  based on  available   failure
rate data for valves,  pipes,  loading hoses,   and  liquid  storage
tanks  also  is  provided in Appendix  B.    The  results  of   this
analysis are summarized  below.


Frequency of Spills at
Transfer and Storage Facilities

     Exhibit  4-3 summarizes the  expected number of  releases  per
year  for the storage and  transfer facilities at ocean-based  and
land-based  systems (exclusive of spills from unloading  trucks).
As  the exhibit indicates,   the total  expected  number  of releases
from   all  storage and  transfer  components of the  ocean-based
system  is 0.03 per year  in  the PCB  case and 0.04 per  year in the
EDC case.   The total expected number  of releases from the  land-
based  storage  and transfer facility  is approximately 0.04  for
both  the  PCB  and  EDC  cases.   The  frequency of  failures  is
slightly  higher  for the  land-based system since we  assume  the
pumps and pipelines at the land-based  facility  constantly contain
wastes, whereas approximately half of  the pumps and  pipelines for
the  sea-based  system   (those required for  vessel   loading)  are
assumed  to carry wastes  only during the 36-hour period  required
to load the incinerator  ship.
                              4-4

-------
     As  shown,   the  ocean-based  system  has  one   additional
component  — the  loading  hose used to transfer wastes  to  the
ship.   We  estimate the probability of a spill from the hose  to
be roughly two to three in one-thousand per year.

     Pump failures  are the single most likely cause of a release
at both Chickasaw and the land-based incinerator,  with the total
expected  number of releases from such failures ranging from  .02
to  .03  per year.   In contrast,  the- least likely sources of  a
release  are the two 800 thousand gallon waste storage  tanks  at
each site.   For each system,  the expected number of spills from
these tanks is 2.0E-4 per year,  a long-term average of one spill
every 5,000 years.


Waste Quantities Released from
Spills at Transfer and Storage Facilities

     Exhibits  4-4 and 4-5 present the average  spill  quantities
for  each component of the storage and transfer networks for  PCB
and  EDC  wastes,  respectively.   The  exhibits  also  show  the
expected  annual  release quantity  for each component using  the
frequency of spills from Exhibit 4-3.  The exhibits indicate that
the  expected annual release at storage and  transfer  facilities
for   either   ocean- or  land-based  incineration   systems   is
approximately 0.5 metric tons in both the PCB and EDC cases.   In
each  case,  releases  from storage tanks account for roughly  80
percent  of the expected annual release.   While a  storage  tank
failure  is  expected  to occur only once  in  5,000  years,  the
expected  release in the event of a failure is so large  (1,870 to
2,115 metric tons) that this low probability,  high release event
dominates  the total expected annual release for both  systems.2/
Again,  the expected annual release from the land-based system is
slightly  larger than the expected annual release for the  ocean-
based  system because of the greater likelihood of  releases  from
valves and pipelines at the land-based operation.
2/According to Appendix B,  ADL's estimates of  the  probability of
a storage tank failure and the quantity  of waste  released  in  the
event  of  a  failure apply to   "catastrophic"  releases.    Minor
releases are not significant and are  ignored.
                              4-5

-------
FUGITIVE EMISSIONS FROM
TRANSFER AND STORAGE FACILITIES

     In addition to hazardous waste  spills,   storage and transfer
facilities  are  sources  of  airborne   fugitive   emissions.   As
described  in  Appendix  B,  intermittent  fugitive emissions  at
Chickasaw  or  the  land-based incinerator  are   associated  with
occasional small leaks from pumps and valves  in the waste  storage
and transfer network.   Relatively continuous sources  of fugitive
emissions  include ordinary breathing losses  from storage  tanks,
working  losses  that occur in filling  and  emptying   tanks,  and
activities such as manual gaging of  tank truck cargo levels.

     Exhibit  4-6  summarizes expected  annual fugitive emissions
for  the  improved  Chickasaw facility   and   for   the   land-based
incineration  site.   As  the exhibit   shows,  expected fugitive
emissions for the ocean-based system are approximately 0.7 metric
tons  per  year in both the PCB and  EDC cases.    In   comparison,
expected  fugitive  emissions for the land-based  system are  0.6
metric  tons  per year for both wastes.    Fugitive releases  are
slightly  higher for the ocean-based system due to the emissions
associated  with  loading  the incinerator  ship's cargo  tanks.
Otherwise,  fugitive  emissions  from   the transfer  and   storage
facilities are identical.   Emissions from storage tanks dominate
total   fugitive  emissions  in  both    cases,   accounting   for
approximately  two-thirds  of total  releases  for  the   ocean-based
system  and  80  percent of total  releases   for   the   land-based
operation.
CAVEATS AND SENSITIVITY ANALYSIS

Releases from an Ocean-Based System
Assuming Intermediate Storage at Emelle

     As  described  in Chapter 3,  CWM would  store wastes at   its
Emelle,  Alabama storage facility  if development  of waste storage
capacity at Chickasaw is not feasible.  This  arrangement requires
I/Note that the ADL analysis contained  in Appendix B  assumes   the
same  vapor  pressure for the  PCB and EDC wastes,  and   therefore
does not consider the effect that different volatilization  rates
could have on emissions from storage tanks.
                               4-6

-------
tank trucks to unload wastes at Emelle,  reload to deliver wastes
to  the  Vulcanus,  and unload again at the  port.  Tripling  the
number  of truck  loading/unloading steps in this manner  triples
the  expected  number and annual  quantity  of  loading/unloading
spills,  as  shown  in  Exhibit 4-7.    The  expected  number  of
loading/unloading  spills annually is 0.11 and 0.15 for  the  PCB
and EDC wastes,  respectively.   The expected release quantity is
0.09 HT of PCB waste and 0.11 NT of EDC waste.

     Intermediate  waste  storage  at Emelle  also  would  affect
expected  releases  from storage and transfer  equipment.   Under
this arrangement,  the existing storage and transfer equipment at
both  Emelle  and Chickasaw would be potential  release  sources.
The  expected  number of releases from the various components  of
the  two facilities is summarized in Exhibit  4-3.   The  exhibit
indicates that the total  number of releases expected for the two
facilities is 0.07 per year in the PCB case and 0.08 per year  in
the  EDC case,  slightly more than double the expected number  of
releases in the baseline analysis of the ocean-based system.  The
higher number of releases is primarily due to a greater number of
pump   failures,   since  the  combined  Emelle/Chickasaw  system
requires  more  than twice as many pumps as  an  integrated  port
facility.

     Exhibits  4-9 and 4-10 summarize expected release quantities
for the Emelle/Chickasaw transfer and storage system for the  PCB
and  EDC wastes,  respectively.   The expected quantity released,
excluding storage tanks,  is about 0.16 and 0.18 metric tons  for
the  PCB and EDC wastes,  respectively.   These figures are about
double  those  estimated  for the  improved  Chickasaw  facility.
However,  as the exhibits show, the expected total annual release
in both cases is approximately 0.4 metric tons,  0.1 metric  tons
less  than  the  release  expected  for  the  improved  Chickasaw
facility.    The  lower  total  release  for  the  non-integrated
facility  is due to the assumption that wastes would be stored at
Emelle in four 200 thousand gallon storage tanks rather than  the
two   800  thousand  gallon  tanks  assumed  for  the  integrated
Chickasaw operation.   This assumption reduces by half the  total
quantity of waste "at risk" in storage tanks at any point in time
— 800  thousand gallons at Emelle versus 1.6 million gallons  at
                              4-7

-------
 an  integrated  Chickasaw  facility  —  and therefore reduces  by  half
 the expected annual  release  from  this  source.!/

      Exhibit 4-11  illustrates  expected annual  fugitive   emissions
 for the ocean-based  incineration  system,  assuming that  wastes are
 stored at Emelle.    The  exhibit indicates that fugitive emissions
 would total   4.9  metric  tons per year in the PCB case and   5.5
 metric tons per year in  the  EDC case.     These estimates are  more
 than seven times higher  than the  fugitive emission estimates  made
 for  an  integrated  Chickasaw  facility.    Fugitive emissions   are
 considerably   higher for storage  at  Emelle since ADL  assumed   the
 storage tanks  at Emelle  would  have fixed roofs,  as opposed to the
 tanks with floating  internal covers  assumed for  the new Chickasaw
 operation.   As  explained  in Appendix B,  the  use  of floating
 covers minimizes vapor spaces  and significantly  reduces fugitive
 emissions.

      Total  expected annual  releases for  the   Emelle/Chickasaw
 storage and transfer system  are summarized below.

                             Table 4-2
          Summary  of Storage and  Transfer Releases for
                 Emelle/Chickasaw Configuration

                             PCB  Haste        EDC Waste
                            (MT per year)     (MT  per year)
     Truck loading/
      unloading spills            0.1               0.1

     Spills at transfer/
      storage facilities          0.4               0.4

     Fugitive emissions           4.9               5.5
     Total releases              5.4               6.0


     Source:  lEc Analysis
4/This  assumption by ADL  introduces a minor  inconsistency  in  our
analysis.  Assuming that the  improved port  facility  has  twice  the
storage capacity  implies that the wastes are  held  in storage   for
a longer period of time or that the tanks are only filled to half
capacity compared to the Emelle storage facility.


                              4-8

-------
The  net  effect of intermediate waste storage at  Emelle  rather
than  development  of an integrated storage and port facility  at
Chickasaw  is an increase in the expected annual releases for the
storage  and  transfer component of an  ocean-based  incineration
system  from 1.2 metric tons to 5.4 metric tons in the  PCB  case
and to 6.0 metric tons in the EDC case.  Both cases present about
a  fivefold  increase  in  releases  from  storage  and  transfer
operations,  and  yield  total releases about five times  greater
than that estimated for the land-based incineration system.


Effect of Remedial Action

     It  is important to note that hazardous waste  transfer  and
storage  facilities typically are equipped to contain most  waste
spills.   For  example,  the existing storage tanks at Emelle are
located  in  spill  containment areas,  and  the  existing  waste
transfer  station  in  Chickasaw has a  spill  containment  area.
Spill  containment equipment is also likely to be present at most
land-based incineration facilities.   These precautions limit the
health and environmental effects from hazardous waste releases.
COMPARABLE HAZARDS

     During  the  last  seven months of  1982,  132  releases  of
hazardous  substances occurred at  "fixed facilities"   (i.e.,  not
vehicles)  in EPA Region IV and were reported to the   U.S.  EPA's
National Response Center.i/  Extrapolating over a full year,  we
estimate  that 226 hazardous materials spills of this  type  would
occur in Region IV.   In contrast,  the expected number of spills
from  the  waste storage and transfer systems at Chickasaw  or   a
land-based  incinerator  (including truck loading/unloading spills
and spills from transfer and storage equipment) is  approximately
0.1 in both the PCB and EDC cases.   Thus, the expected number of
spills  from  ocean- or land-based incinerator waste storage  and
transfer  operations  represents   less than 0.1  percent  of  the
region's reported fixed facility hazardous materials spills.
5/ICF, Inc., Releases £f flicjh. Production Volume Substances. April
1983.   As  explained  previously,  these  are  CERCLA-designated
hazardous substances.
                              4-9

-------
     Data  to  compare the total quantity of  fugitive  emissions
from   the   ocean- and  land-based  incineration    systems   are
unavailable.   However,  thirty waterfront facilities in the Port
of  Mobile  currently  are  authorized  to  handle   and/or  store
federally-designated  hazardous materials.fi/  The number of these
facilities  near Mobile suggests that other sources  of  hazardous
fugitive emissions exist in the area.


SUMMARY OF STORAGE
AND TRANSFER RELEASES

     The  estimated  releases  associated with   the  storage  and
transfer component of ocean- and land-based incineration  systems
are summarized in Exhibit 4-12.  The summary does not distinguish
between  the  PCB  and  EDC wastes  since  the   expected  release
quantities  (rounded to the nearest 100 kilograms) are  identical.
As this table indicates, expected annual releases from  the ocean-
and  land-based systems differ only slightly,  due to the  higher
level of fugitive emissions for the ocean-based  alternative.  The
table   also  shows  that  spills  and  fugitive emissions   are
approximately  equal  sources of storage and  transfer  releases,
each accounting for roughly half the expected annual storage  and
transfer losses in both the ocean- and land-based case.

     As   shown  in  our  sensitivity  analysis,   these  release
estimates  are  very  sensitive  to  the  configuration  of   the
storage/transfer  facilities.   If  CWM stored wastes   at  Emelle
rather  than Chickasaw,  the release quantities  would be  greatly
increased, primarily due to fugitive emissions resulting from the
use  of  fixed  rather than floating roof storage  tanks  at  the
Emelle facility.  Also, the expected number and  quantity of truck
loading/unloading  spills would triple,  as a result of  tripling
the number of handlings required.

     Again,  it  is  important to bear in mind that  waste  spills
during  any single year could be significantly greater  than  the
expected annual release quantity.   For example, a single storage
tank  failure  could release approximately 2,000 metric  tons  of
hazardous  waste.   However,  the probability of such an event is
extremely low — about one in 5,000 per year.
^./Personal  communication with Captain  W.J.  Ecker,   U.S.   Coast
Guard, Captain of the Port, Mobile, Alabama.


                              4-10

-------
                           Exhibit 4-1

    CALCULATION OF EXPECTED NUMBER (EN)  OP LOADING/UNLOADING
             SPILLS PER YEAR FOR TANK TRUCKS SERVING
           LAND- AND OCEAN-BASED INCINERATION SYSTEMS
PCS Waste

     Assumptions:

          Annual involvement (AI)  =•  2,114 transfers/year
          Incident rate (IR)  =  1.7E-5 spills/transfer

     Calculation:

          EN  =  AI * IR

              =  2,114 transfers/year * 1.7E-5 spills/transfer

              =  0.04 spills/year
                                          /

EDC Waste

     Assumptions:

          AI  =  2,934 transfers/year

          IR  =  l*7E-5 spills/transfer

     Calculation:

          EN  =  AI * IR

              =  2,934 transfers/year * 1.7E-5 spills/transfer

              =  0.05 spills/year
Source:   lEc Analysis

-------
                           Exhibit 4-2

        CALCULATION OF EXPECTED ANNUAL RELEASE  (EV) FROM
               UNLOADING SPILLS FOR TRUCKS SERVING
            LAND- OR OCEAN-BASED INCINERATION SYSTEMS
PCB Waste

     Assumptions:

        Expected release fraction (ERF)                -  0.03
        Quantity at risk (Q)                           -  26.5 MT
        Expected number of releases per year  (EN)      =  0.04

    Calculation:

        Average quantity released (QR)  =  Q  * ERF
                                        =  26.5 MT * 0.03
                                        =  0.795 MT

        EV  - QR * EN

            =  0.8 MT * 0.04

            -  0.032 MT
EDC Waste

     Assumptions:

          ERF  =0.03
          Q    «  23.4 MT
          EN   =  0.05

     Calculation:

          QR   =  Q * ERF              EV  =  QR * EN

               =  23.4 MT * 0.03           =  0.7 MT * 0.05

               «= 0.702 MT                  »  0.035 MT



Source:  lEc Analysis

-------
                           Exhibit 4-3

             EXPECTED NUMBER OF SPILLS (EN) PER YEAR
            FOR WASTE TRANSFER AND STORAGE FACILITIES
       SERVING LAND- AND OCEAN-BASED INCINERATION SYSTEMS
Ocean-Based System-Improved Chickasaw Facility
     Component
     Valves
     Pipelines
     Pumps
     Loading Hose (ship)
     Storage Tanks

          Total
Frequency for
  PCB Waste

   4.8E-3
   4.2E-3
   2.1E-2
   2.0E-3
   2.0E-4
Frequency for
  EDC Waste

   4.8E-3
   4.3E-3
   2.7E-2
   2.6E-3
   2.0E-4
   3.22E-2
   3.98E-2
Land-Based System:
     Component
     Valves
     Pipelines
     Pumps
     Storage Tanks

         Total
Frequency for
  PCB Waste

   8.9E-3
   6.7E-3
   2.1E-2
   2.0E-4
 Frequency for
  EDC Waste

   8.9E-3
   6.7E-3
   2.7E-2
   2.0E-4
   3.7E-2
   4.3E-2
Source:  ADL Analysis, Appendix B.

-------
                           Exhibit 4-4

    EXPECTED RELEASE QUANTITIES FOR SPILLS AT WASTE TRANSFER
                     AND STORAGE FACILITIES:
                            PCB WASTE
Ocean-Based System — Improved Chickasaw Facility:
     Component
     Valves
     Pipelines
     Pumps
     Loading Hose (ship)
     Storage Tanks

          Total
 Spill Size
    (MT)

     3.4
     6.9
     1.1
     6.9
 2,114.6
Expected Annual
    Release
     (MT)

    1.63E-2
    2.90E-2
    2.31E-2
    1.38E-2
    4.23E-1
                    5.05E-1
Land-Based System:
     Component
     Valves
     Pipelines
     Pumps
     Storage Tanks

         Total
Spill Size
   (MT)

    3.4
    6.9
    1.1
2,114.6
Expected Annual
    Release
      (MT)

    3.03E-2
    4.62E-2
    2.31E-2
    4.23E-1
                    5.23E-1
Source:  ADL Analysis, Appendix B

-------
                           Exhibit 4-5

    EXPECTED RELEASE QUANTITIES FOR SPILLS AT WASTE TRANSFER
                     AND STORAGE FACILITIES:
                            EDC WASTE
Ocean-Based System — Improved Chickasaw Facility:
     Component
     Valves
     Pipelines
     Pumps
     Loading Hose  (ship)
     Storage Tanks

          Total
 Spill Size
    (MT)

     3.0
     6.1
     1.0
     6.1
 1,869.8
Expected Annual
    Release
     (MT)

    1.44E-2
    2.62E-2
    2.70E-2
    1.59E-2
    3.74E-1
                    4.58E-1
Land-Based System:
     Component
     Valves
     Pipelines
     Pumps
     Storage Tanks

         Total
Spill Size
   (MT)

    3.0
    6.1
    1.0
1,869.8
Expected Annual
    Release
      (MT)

    2.67E-2
    4.09E-2
    2.70E-2
    3.74E-1
                    4.69E-1
Source:  ADL Analysis, Appendix B.

-------
                                 Exhibit  4-6

              EXPECTED  ANNUAL  FUGITIVE  EMISSIONS  FROM TRANSFER
                          AND STORAGE  FACILITIES
Ocean-Based System —  Improved Chickasaw  Facility:
Source
Truck gaging
Pumps
Valves
Storage tanks
Ship cargo tanks

     Total
	PCB Waste	   	EDC Waste	
Expected      Percent of   Expected     Percent of
Emissions       Total      Emissions       Total
  (MT/yr)       Emissions     (MT/yr)      Emissions
  3.2E-2
  3.0E-2
  4.2E-2
  4.6E-1
  1.1E-1

  6.7E-1
4.8%
4.5
6.3
68.7
16.4
4.4E-2
3.0E-2
4.3E-2
4.7E-1
1.4E-1
100.0%
7.2E-1
  6.1%
  4.2
  6.0
 65.3
 19.4

100.0%
Land-Based System:
Source
Truck gaging
Pumps
Valves
Storage tanks

     Total
	PCB Waste	
Expected      Percent of
Emissions       Total
 (MT/yr)       Emissions
 3.2E-2
 3.0E-2
 4.2E-2
 4.6E-1

 5.6E-1
  5.7%
  5.3
  7.4
 81.6

100.0%
            	EDC Waste	
            Expected     Percent of
            Emissions       Total
             (MT/yr)      Emissions
4.4E-2
3.0E-2
4.3E-2
4.7E-1

5.8E-1
   7.5%
   5.1
   7.4
  80.0

 100.0%
Note:  Totals may not sum due to rounding.

Source:  ADL Analysis, Appendix B.

-------
                           Exhibit 4-7

              EFFECT OF STORAGE FACILITY AT EMELLE
               ON TRUCK LOADING/UNLOADING RELEASES
                IN OCEAN-BASED INCINERATOR SYSTEM
Annual Involvement
  (transfers/year)

Incident Rate

Expected Number of
  Releases

Quantity at Risk  (MT)

Expected Release
  Fraction

Expected Quantity
  Released (MT)
PCB Waste


  6,342

  1.7E-5


  0.11

 26.5


 0.03


 0.09
EDC Waste


  8,802

  1.7E-5


  0.15

 23.4


  0.03


  0.11
Source:  lEc Analysis

-------
                          Exhibit 4-8

               EXPECTED NUMBER OF SPILLS PER YEAR
      FROM UNIMPROVED WASTE TRANSFER AND STORAGE FACILITIES
                     AT EMELLE AND CHICKASAW
               Component
                 Frequency for
                    PCB Waste
               Frequency for
                 EDC Waste
Emelle:
Valves
Pumps
Storage Tanks
7.9E-3
3.0E-2
4.0E-4
7.9E-3
3.9E-2
4.0E-4
               Subtotal
                    3.83E-2
                   4.73E-2
Chickasaw;
Valves
Pumps
Pipelines
Loading Hose
1.4E-3
1.5E-2
1.2E-3
l.OE-2
1.7E-3
1.9E-2
1.5E-3
1.3E-2
               Subtotal
                    2.76E-2
                   3.52E-2
Total:
Valves
Pumps
Pipelines
Loading Hose
Storage Tanks
9.3E-3
4.5E-2
1.2E-3
l.OE-2
4.0E-4
9.6E-3
5.8E-2
1.5E-3
1.3E-2
4.0E-4
               TOTAL
                    6.59E-2
                   8.25E-2
Source:  ADL Analysis, Appendix B.

-------
                          Exhibit  4-9

             EXPECTED RELEASE QUANTITIES FOR SPILLS
     FROM UNIMPROVED WASTE TRANSFER AND STORAGE FACILITIES
                    AT EMELLE AND  CHICKASAW:
                            PCB  WASTE
Emelle:
Component

Valves
Pumps
Storage Tanks

  Subtotal
Spill Size
   (MT)

   3.4
   1.1
   529.9
Expected Annual
    Release
      (MT)

   2.69E-2
   3.30E-2
   2.12E-1
                                                  2.72E-1
Chickasaw:
Valves
Pumps
Pipelines
Loading Hose

  Subtotal
   3.4
   1.1
   6.9
   6.9
   4.76E-3
   1.65E-2
   8.28E-3
   6.90E-2

   9.85E-2
Total:
Valves
Pumps
Pipelines
Loading Hose

  Subtotal

Storage Tanks

TOTAL
   3.4
   1.1
   6.9
   6.9
                                   529.9
   3.16E-2
   4.95E-2
   8.28E-3
   6.90E-2

   1.59E-1

   2.12E-1

   3.70E-1
Source:  ADL Analysis, Appendix B.

-------
                          Exhibit 4-10
             EXPECTED RELEASE QUANTITIES FOR SPILLS
      FROM UNIMPROVED WASTE TRANSFER AND STORAGE FACILITIES
                     AT EMELLE AND CHICKASAW:
                            EDC WASTE
Emelle:
               Component
Valves
Pumps
Storage Tanks

  Subtotal
                 Spill Size
                    (MT)
3.0
1.0
467.4
Expected Annual
    Release
      (MT)

   2.37E-2
   3.90E-2
   1.87E-1
                                                  2.50E-1
Chickasaw:
Valves
Pumps
Pipelines
Loading Hose

  Subtotal
3.0
1.0
6.1
6.1
   5.10E-3
   1.90E-2
   9.15E-3
   7.93E-2

   1.13E-1
Total:
Valves
Pumps
Pipelines
Loading Hose

  Subtotal

Storage Tanks

TOTAL
3.0
1.0
6.1
6.1
                                   467.4
   2.88E-2
   5.80E-2
   9.15E-3
   7.93E-2

   1.75E-1

   1.87E-1

   3.62E-1
Source:  ADL Analysis,  Appendix B,

-------
                                 Exhibit 4-11

      EXPECTED ANNUAL FUGITIVE EMISSIONS FROM UNIMPROVED WASTE TRANSFER
                AND STORAGE FACILITIES AT EMELLE AND CHICKASAW
Emelle:
            Source
Chickasaw;
Truck gaging
Pumps
Valves
Storage tanks
Truck loading

Subtotal
Truck gaging
Pumps
Valves
Ship cargo tanks

Subtotal
                     	PCB Waste	
                     Expected     Percent
                     Emissions    of Total
                       (MT/yr)     Emissions
TOTAL
3.2E-2
1.5E-2
8.0E-2
4.0E+0
5.7E-1

4.7E+0
3.2E-2
1.5E-2
1.1E-2
1.1E-1

1.7E-1

4.9E+0
                                                96.6
  0.7
  0.3
  0.2
  2.2

  IT?

100.0%
	 6UV, *
Expected
Emissions
(MT/yr)
4.4E-2
1.5E-2
8.0E-2
4.4E+0
7.3E-1
5.3E+0
4.4E-2
1.5E-2
1.5E-2
1.4E-1
2.1E-1
5.5E+0
Percent
of Total
Emissions
0.8%
0.3
1.5
80.0
13.3
96.2
0.8
0.3
0.3
2.5
3.8
100.0%
Note:   Totals may not sum due to rounding.

Source:  ADL analysis,  Appendix B.

-------
                          Exhibit 4-12

  EXPECTED RELEASES (EV) FROM STORAGE AND TRANSFER COMPONENT OF
         LAND-BASED AND OCEAN-BASED INCINERATOR SYSTEMS
Source                            Ocean-Based     Land-Based
*    Less than 0.05 MT


Source:  lEc Analysis
Truck Unloading/Loading
Spills                                ~ *           — *

Spills at Transfer/Storage
Facilities                            0.5            0.5

Fugitive Emissions                    0.7            0.6
  Total Quantity Released (MT)        1.2            1.1

-------
WASTE RELEASES
PROM OCEAN TRANSPORTATION                               CHAPTER 5
     The    ocean   transportation   component   of   ocean-based
incineration  is the only major component of the system that  has
no  parallel  in a  land-based  operation.  Therefore,  potential
releases  from ocean transportation are of special interest  when
comparing   the   relative   risks   of   land- and   ocean-based
incineration systems.   These potential releases are discussed in
this chapter.


DATA AND ASSUMPTIONS

     The  analysis  of spills from the Vulcanus as a result of  a
vessel casualty is based on an analysis of vessel casualty  rates
and  expected release quantities prepared by Engineering Computer
Optecnomics, Inc. (ECO).  This analysis, contained in Appendix C,
groups   vessel   casualties   into   three   impact   categories
(collisions, groundings and rammings) and one non-impact category
(fires,  explosions,  structural failures,  and capsizings).  The
analysis  provides  estimates of vessel casualties and  resulting
spills at four different locations:

     o    Mobile Harbor and the Vulcanus1 pier in Chickasaw;

     o    Mobile Bay;

     o    The Gulf coastal area; and

     o    The burn zone.
                              5-1

-------
All  estimates are based on a proprietary,  worldwide  tank  ship
accident and spill database (entitled ECOTANK) that is maintained
by  ECO.   This  database  presently contains  information  about
10,000  worldwide  tank ship accidents and over 1/000  tank  ship
spill events that occurred  from 1969 to 1982.   As described  in
Appendix  C  and later in this chapter,  ECO has  adjusted  these
spill  rate estimates to account for the design and operation  of
the Vulcanus.

     Incinerator  ships  have operated off the European coast  in
the North Sea since 1972.   They have made about 320 voyages  and
have incinerated about 650,000 metric tons of hazardous wastes to
date.  No casualties such as collisions, groundings, rammings, or
fires have occurred,  nor have there been any spills from loading
incinerator ships in port.   Although incinerator ship operations
have  established  a good safety record,  the number  of  voyages
completed   is  too small to be used in estimating  spill  rates.
Thus, we rely on accident data for tankers to estimate spills for
the Vulcanus.
ESTIMATION OF RELEASING VESSEL
CASUALTY  (SPILL) RATES

     ECO  selected  from its database accidents and spill  events
for tank  ships of a size similar to the Vulcanus — from 2,000 to
10,000  deadweight long tons.   These data were used to  estimate
releasing  vessel  casualty  or "spill" rates for four  types  of
accidents  and  for  areas  corresponding  to  the  locations  of
interest  for the proposed operation of the Vulcanus.  These spill
rates were then adjusted for the design,  location, and operating
features  for  the  Vulcanus  that are  expected  to  reduce  the
frequency  of  spills below that experienced  worldwide  by  tank
ships of  similar size.

     The  design  features adjusted for include the double  hull,
double  bottom  construction  of  the  Vulcanus  and  the  ship's
controllable pitch (CP) propeller with bow thruster.   The double
hull,  double  bottom  construction  is expected  to  reduce  the
probability  that cargo is released in the event of an  accident,
and the CP propeller with bow thruster is expected to reduce  the
probability  of impact-type accidents,  especially at low speeds,
by improving the maneuverability of the vessel.   With regard  to
location,  the  generally soft-bottom conditions in the Gulf  are
expected  to reduce the rate of spills due to groundings below the


                              5-2

-------
 rate experienced worldwide.   Finally,  operating  restrictions  that
 the Coast Guard plans  to  impose  also are expected  to  reduce spill
 rates.

     As an example  of  the operating restrictions noted above,  the
 Coast  Guard  will  require all movements of  the  Vulcanus   in   the
 Port  of  Mobile,   both inbound  and outbound,    to be under   the
 direction  of  the  Captain  of the  Port,   and  a  24 hour   advance
 notice will be required prior to the ship entering Mobile  Bay.I/
 This  notice  will  include   a report on  the  condition   of   the
 vessel's propulsion, steering, navigation and  cargo systems.   The
 Coast  Guard  will  allow  the Vulcanus to transit the  harbor   only
 during  daylight hours,   with a  minimum of two miles  visibility.
 In addition, the Captain  of  the  Port will issue  a  local Notice To
 Mariners  establishing a moving safety zone of  300 feet   in   all
 directions  around  the ship.    Vessels will be  prohibited   from
 entering this zone  without specific approval.    The effect of the
 safety  zone  is  to prohibit oncoming vessels  from  the  Mobile
 Harbor ship channel while the Vulcanus is in the harbor,   and to
 keep  vessels traveling in the same direction  as the  Vulcanus at
 least 300 feet away.   The Coast  Guard will escort  the Vulcanus to
 enforce  the safety zone.    The  Coast Guard  also will require the
 ship to be escorted by two tugs  while it is  in the harbor.  While
 the ship is in transit in the Gulf  or in the Gulf  burn zone,   the
 USCG  will broadcast a Notice To Mariners notifying them   of   the
 ship's position and advising them to steer clear of the vessel.2/
I/The  discussion  of USCG provisions  governing  the  Vulcanus   is
based on the statement of Rear Admiral  Clyde  T.  Lusk,  Jr.,  Chief,
Office  of  Merchant Marine Safety,  United States   Coast   Guard,
before   the  House  Subcommittee  on   Fisheries  and  Wildlife
Conservation  and the Environment,   Subcommittee on  Oceanography,
December 7, 1983.

2/In  addition  to these operating precautions,   USCG   inspection
requirements  for incineration vessels  include annual   inspection
of navigational and fire-fighting equipment,  biennial inspection
of the exterior of the hull and  interior  of the  cargo  tanks,   and
ultrasonic  testing of hull thickness  every   five years.   These
steps  should ensure the  maintenance  of  the  ship's  navigational
and structural integrity as long as  it  is permitted  to incinerate
hazardous  wastes,  and  reduce  the  probability of   non-impact
related casualties.
                              5-3

-------
     The  precise effect that these factors may have in  reducing
the  rate of spills is difficult to determine.   The  adjustments
made  are  based on published studies,  observed  differences  in
spill  rates,  kinetic  energy  levels likely  for  accidents  in
different  locations,  and  professional  judgment.  No  downward
adjustments  were  made  to the spill rate  for  non-impact  type
accidents because of insufficient information.   The calculations
of  spill  rates  and  the adjustments made  to  them  are  fully
discussed in Appendix C.

     The results of ECO's calculations are summarized  in Exhibit
5-1.   These  data  give  spill rates (any size spill)  for  each
location and type of casualty on a per voyage basis.  The overall
spill rate is quite low — about 6 per 100,000 voyages.  The data
in Exhibit 5-2 show the relative frequency of spills by  location
and  type  of  casualty,  i.e.,  which  locations  and  types  of
casualties  have  greater  or lesser  probability  of  occurence.
These  data  indicate  that the pier and coastal areas  have  the
highest expected spill rates — about 70 percent of spills  would
be expected to occur in those areas.I/


WASTE QUANTITIES RELEASED

     ECO  also estimated the probability that accidents resulting
in  a spill involve one,  two,  and three or more  tanks.   Their
estimates for impact type casualties  (collisions,  groundings, or
rammings) are as follows:

          o    one tank - 80 percent

          o    two tanks - 15 percent

          o    three or more tanks -  5 percent

There  is little information on which to base such estimates  for
non-impact type casualties.   Accordingly,  the same distribution
is  assumed,  but there obviously is some amount  of  uncertainty
regarding this assumption.
I/Most of the risk of spill in the  "pier and  harbor"  area  is  from
non-impact  type  accidents during  loading  at  the  pier.   These
rates were not adjusted downward,   as explained in  the   text,   so
they are conservative upper-bound estimates.


                              5-4

-------
     The  data in Exhibit 5-1,  along with the data above on  the
percentage of spills involving one,  two,  or more tanks,  may be
used  to  calculate  the spill rates  by location and  number  of
tanks involved.  Exhibit 5-3 expresses these spill rates on a per
voyage basis,  while Exhibit 5-4 gives estimated spill rates  (for
the PCB case) in terms of operating years.   These data  indicate
that  spills  in  any location are very  unlikely.   The  overall
average spill rate is 6 per 100,000 voyages, or about 1 spill per
1,200  operating years.   The rate of spill events where  two  or
three  or more tanks are involved is even lower — 9 and 3 spills
per million voyages,  respectively.  These rates equal an average
of one two-tank spill per 8,000 operating years, and one spill of
three or more tanks per 24,000 operating years.   Spill rates for
particular locations are,  of course, less than the overall spill
rate.  For example, the spill rates for Mobile Bay when one, two,
or three or more tanks are involved are 1 per 13,000, 67,000, and
200,000 operating years, respectively, with an overall spill  rate
of 1 per 10,000 operating years.

     The  amount  of  cargo likely to be  released  from  a   tank
affected  by  an accident depends on many factors,  such  as  the
nature of the accident (the location and extent of damage to  the
tank), the specific gravity of the waste, sea conditions, and the
length  of  time  before  some action is taken  to  lessen  cargo
outflow.   For example,  in many collisions the impact is  severe
enough  to put the cargo in free communication with the  sea.  In
such accidents virtually the entire cargo of a tank could be  lost
in a relatively short period of time.   Groundings,  on the other
hand,  usually  crack  or puncture the hull far below  the  water
line.  The pressure from cargo in the tank that resides above the
water  line usually will force a portion of the cargo out of  the
tank.   The  amount  of additional cargo loss and the  time   over
which  it will occur depends on the extent of the damage and  the
density  of  the  waste  relative  to  sea  water.   Wastes   with
densities  greater  than sea water will tend to flow out  as  sea
water  flows in and floats  to the top of the tank.   Outflow  of
waste  of  lesser  density probably would not  be  as  extensive.
Finally,  rammings tend to create punctures at or near the  water
line.   Cargo  above the water line would be lost rather quickly.
Cargo  below the water line could also be lost if the density  of
the waste were less than sea water and the sea water was able  to
circulate into the tank, allowing the lighter waste to float  out.
                              5-5

-------
     We  have not developed spill size distributions or  expected
spill  sizes  for accidents involving various numbers  of   tanks,
since data are not available to make such estimates.   Rather, we
have made the conservative assumption that  the  entire contents of
tanks affected by accidents are released.

     The  "expected" amount of waste released on an annual   basis
is  calculated  for  PCS  and EDC  wastes   in   Exhibit  5-5.   We
emphasize  that  the expected values represent  an annual  average
used  only  as  an index to compare  releases   from  the  various
components of incineration systems and do not suggest that  spills
would occur each year.  Expected annual  releases are estimated to
be  about  0.63  MT for PCB wastes and 0.77 MT  for  EDC  wastes.
These  calculations are conservative,  in that  we assume that any
tank involved in a spill releases its entire contents,  and that
the  entire ship's cargo is released in  accidents involving three
or more tanks.  The expected values reflect the large quantity of
waste  that  could be released in a spill and the  extremely low
probability  that  a spill would  occur.   As   noted  above,  the
estimated  spill rate for the Vulcanus is 6 per 100,000  voyages,
or  (in the PCB case) 1 per 1,200 operating years.   Given   these
estimates,  it  is  unlikely that a spill will  occur  within the
ship's operating life.

CAVEATS AMD SENSITIVITY ANALYSIS

Vessel Casualty Rates

     The  vessel  casualty rates employed  to   analyze  potential
releases  from  the Vulcanus are based on historic ship  casualty
data,  with consideration of the specific conditions under   which
the  Vulcanus would operate.   The casualty estimates account for
such factors as U.S. Coast Guard (USCG)  requirements, the double-
hulled,  double-bottomed design,  the use of a  CP propeller with
bow thruster,  the bottom conditions in  the Gulf, and the size of
the  vessel.   As  a result,  the estimated spill rates  for the
Vulcanus are lower than for a typical tanker.

     For purposes of comparison, Exhibit 5-6 shows the spill rate
estimated  for the Vulcanus as a percent of the historical   spill
rates  for  typical tank ships operating in similar areas as the
Vulcanus.  The  data indicate that the likely spill rate for the
Vulcanus  is substantially less than that experienced by  tankers
world  wide.   This  is  especially  the  case  for  impact type
accidents in the bay and pier/harbor areas.   No adjustments were
made for non-impact type casualties, so  the data reflect only the


                              5-6

-------
fact that small size vessels have lower non-impact casualty rates
than the average for the world fleet of tank ships.  Overall, the
Vulcanus  is expected to have a spill rate of about 10 percent of
that for typical tank ships.


Fugitive Emissions from the Vulcanus

     While our analysis of marine transportation releases focuses
on  vessel  casualties,  we also  considered  potential  fugitive
emissions   from  the  Vulcanus  while  the  ship  is   underway.
According  to  an ADL analysis,  the ship  may  release  fugitive
emissions  if  an accidental discharge of waste liquid  into  the
pumproom  bilge occurs and a portion of the release  volatilizes,
or  if a build-up of pressure in the cargo tanks necessitates use
of pressure relief valves.   Fugitive emissions of this type  are
difficult  to estimate accurately.   A rough estimate made by ADL
suggests that these sources are likely to release less than about
4  kilograms of fugitive emissions per voyage.   Annual  fugitive
emissions  would be about 50 kilograms and 70 kilograms of  waste
in the PCB and EDO cases,  respectively»   Since these  emissions
are small,  would consist mainly of the more volatile portions of
the waste,  and would be released and dispersed,  on average, all
along  the ship's path of transit,  their environmental and human
health effects would be neglibible.A/  Further, any material that
did  not  volatilize  when accidentally discharged to  the  bilge
would  be  retained  in tanks on board and  eventually  would  be
pumped   to  an  onshore  facility  for  storage  or   subsequent
incineration.

Remedial Action

     The preceding analysis does not consider remedial action  to
remove waste materials released to the environment as a result of
spills,  as  this was beyond the scope of this study.   Hazardous
waste  operators are required to develop a contingency  plan  for
handling spills as a condition of the permit.   The effectiveness
of  remedial  actions would depend on many factors,  such as  the
location   and  type  of  accident,   sea   conditions,   weather
conditions,  currents,  properties of the material,  and response
time.   We  think that efforts to contain and recover spills  are


1/IEc copy of memorandum from J.  Hagopian  (ADL) to J.  Ehrenfeld
(ADL),  "Fugitive Emissions from Incinerator Ships," 19 September
1984.


                              5-7

-------
most  likely   to  be  successful  in   enclosed areas  or   in   shallow
waters,  such  as  the pier and harbor  area and Mobile  Bay.   To  the
extent that such  mitigating activities  are effective,   the amount
of  waste remaining  in the marine  environment after a spill will
be less than the  releases we estimate.

COMPARABLE HAZARDS

     To  characterize the relative effect that operation   of   the
Vulcanus  would have on expected releases of hazardous substances
in the Gulf of Mexico or Mobile Harbor,   we examined  data  on   the
shipments  of  petroleum and hazardous materials in Mobile   Harbor
and the Gulf.  These data show  that there is a substantial volume
of shipments of petroleum and hazardous substances in both Mobile
Harbor and the Gulf  of Mexico.

     The table below provides data on the average  volume of crude
petroleum,  petroleum  derivatives,   and chemicals shipped in  and
out of Mobile Bay over the period  1977  to 1981.5_/

                            Table  5-1

         AVERAGE  ANNUAL PETROLEUM  AND CHEMICAL SHIPMENTS
                          IN MOBILE BAY

                                Imports,  Exports       Internal
                                &  Coastwise            & Local
                                (Tankers)               (Barges)
Commodity
	   	 (000 MT) 	

Crude Petroleum                    1,445                2,203

Gasoline, Jet Fuel,  Kerosene,
  Fuel Oil & Solvents               386                3,632

Benzene, Toluene, &
  Basic Chemicals                    42                   113
Total                             1,873                  5,948
Ji/Derived from Water borne Commerce of  the  United States,   Part 2_t
Waterways and Harbors — Gulf Coast. Mississippi River  System .and
Antilles, Calendar Years 1977-1981, Department  of the Army,  Corps
of Engineers.


                              5-8

-------
The  expected  volume  of wastes to be carried  by  the  Vulcanus
(56,000  to  68,400 HT) is less than 4 percent of the volumes  of
the  above  commodities  carried by tankers and is  less  than  1
percent of the volumes carried by both tankers and barges.  Since
the  relative frequency of spills from the Vulcanus is less  than
from  tank ships generally (and assuming spill rates  for  barges
are  similar),  the expected volume of releases from the Vulcanus
in  the pier and harbor areas and Mobile Bay,  in  a  statistical
sense,  is  only  about  0.1 percent of  expected  releases  from
ongoing shipments of these commodities.

     Data  on shipments of petroleum and hazardous  substances in
the Gulf area in FY 1983 are shown in the table below.£/

                            Table 5-2

         SHIPMENTS OF PETROLEUM AMD HAZARDOUS SUBSTANCES
                 IN THE GULF OF MEXICO, FY 1983

                                                      Hazardous
                                 Petroleum            Substances
Volume of shipments (MM MT)          270                   274

Number of shipments
  Tankers                          8,290                1,947
  Barges                          36,627                13,031


The number of voyages by the Vulcanus per year would be about  0.2
percent  of total tanker shipments  (about 0.03 percent  of   tanker
and  barge  shipments)  and the volume of waste   carried   by   the
Vulcanus  would  be about 0.01 percent of the volume of petroleum
and  hazardous substances shipped by both tankers and   barges   in
the  Gulf  area.   Given  the lower expected spill rate for   the
Vulcanus,  the  expected volume of releases from  the Vulcanus   is
less  than  0.002  percent  of  expected  releases from   ongoing
shipments of petroleum and hazardous substances in the  Gulf area.
I/Information  from  U.S.  Coast Guard,  Port   and   Environmental
Safety Division.
                              5-9

-------
SUMMARY OF RELEASES
FROM OCEAN TRANSPORTATION

     Our  analysis indicates that the overall spill rate for  the
Vulcanus  (all  locations combined) is very low — 6 per  100,000
voyages or,  in the PCB case,  about 1 per 1,200 operating years.
Spill rates estimated for particular locations are even lower, as
are  spill  rates for accidents involving two or  three  or  more
tanks.   For  example,  the overall spill rate for Mobile Bay  is
about  1  per  10,000 operating years,  with the spill  rate  for
accidents involving one,  two, or three or more tanks being about
1 per 13,000,  67,000, and 200,000 operating years, respectively.
The overall spill rate for the coastal area is about 1 per  4,000
operating years, with the spill rate for accidents involving one,
two,  or three or more tanks being about 1 per 5,000, 25,000, and
75,000 operating years, respectively.

     The  overall spill rate estimated for the Vulcanus is  about
one  order  of magnitude less than the historical spill rate  for
tank  ships worldwide.   This is due to the adjustments  made  to
historical  spill  rates  to reflect the likely  effects  of  the
design  and operating characteristics of the Vulcanus.   However,
no downward adjustments,  other than for the size of the  vessel,
were  made to spill rates from non-impact casualties,  because of
insufficient information.

     The amount of expected annual releases of PCB and EDC wastes
are also low,  about 0.63 MT and 0.77  MT,  respectively.   These
numbers  represent long-term averages,  rather than the  releases
likely  to  occur  in  any  year.   They  are  also  conservative
estimates,  as  we  assumed complete loss of the cargo  of  tanks
affected by an accident.

     The Vulcanus would operate in a region of the country having
substantial  commerce in petroleum and hazardous materials.   The
volume  of material transported by the Vulcanus would be a  small
fraction  of shipments of such commodities in the  Gulf  — about
0.01 percent.   Further,  because the spill rate for the Vulcanus
is  less than for other vessels,  the potential releases from the
Vulcanus  would be less than 0.002 percent of those from  ongoing
shipments of petroleum and hazardous materials in the Gulf.
                              5-10

-------
Location
Pier & Harbor
Mobile Bay
Coastal
Burn Zone
All Locations
                                     Exhibit 5-1
                  ESTIMATED SPILL RATES PER VOYAGE FOR THE VULCANUS
                          BY TYPE OP CASUALTY AND LOCATION
	 Type of Casualty

Collisions   Groundings   Hammings
               All
Non-impact  Casualties
4.312-6
1.9E-6
6.4E-6
4.5E-6
1.7E-S
2.2E-7
3.4E-6
6.7E-6
-
l.OE-5
1.6E-7
1.4E-6
2.4E-6
-
4.0E-6
1.8E-5
3.6E-7
3.6E-6
7.1E-6
2.9E-5
2.3E-5
7.1E-6
1.9E-5
1.2E-5
6.0E-5
Source:   Appendix C, Table 7

-------
                                     Exhibit 5-2

                       DISTRIBUTION OP EXPECTED SPILLS BY TYPE
                              OF CASUALTY AND LOCATION
Location
Pier & Harbor

Mobile Bay

Coastal

Burn Zone

All Locations
Collisions
0.071
0.032
0.106
0.074
0.284
	 -ryi
Groundings
0.004
0.056
0.111
-
0.171
pe oz ^asuaxi
Hammings
0.003
0.024
0.040
-
0.066
-y 	
No n- impact
0.297
0.006
0.059
0.118
0.479
All
Casualties
0.375
0.118
0.315
0.192
1.000
Note:     This  table indicates the relative frequency of spills by casualty and
          location.   The  data are calculated from Exhibit 5-1 by dividing  the
          estimated  spill  rate for a given casualty type and location  by  the
          overall spill rate.   For example, relative fequency for collisions in
          pier and harbor area of 0.071 is calculated by dividing the spill rate
          of 4.3E-6 by the overall spill rate of 6.0E-5.
Source:
Exhibit 5-1

-------
                                     Exhibit 5-3
            ESTIMATED  SPILL RATES  PER VOYAGE FOR  THE VULCANUS BY LOCATION
                            AND NUMBER OF TANKS AFFECTED
Location

Pier & harbor
Mobile Bay
Coastal
Burn Zone
 One
1.8E-5
5.7E-6
1.5E-5
9.3E-6
 Number of Tanks Affected
 Two        Three or More
3.4E-6
1.1E-6
2.9E-6
1.7E-6
1.1E-6
3.6E-7
9.5E-7
5.8E-7
Any Number

 2.3E-5
 7.1E-6
 1.9E-5
 1.2E-5
     Total
4.8E-5
9.0E-6
3.0E-6
 6.0E-5
Source:   Appendix C,  Table 7 and lEc Analysis

-------
                                  Exhibit 5-4

                            SPILL RATES IN TERMS OF
                           NUMBER OF OPERATING YEARS *
Location


Pier & harboc

Mobile Bay

Coastal

Burn Zone
One
-  Number  of  Tanks  Affected  	
Two      Three or  More     Any Number
4,000
13,000
5,000
8,000
21,000
67,000
25,000
41,000
63,000
200,000
75,000
120,000
3,000
10,000
4,000
6,000
     Total
1,500
 8,000
24,000
1,200
*    The  spill  rate is one per the number of operating years shown  above
     and is calculated assuming 14 voyages per operating year.

Source:   Derived from Appendix C, Table 7

-------
                           Exhibit 5-5
          CALCULATION OF EXPECTED ANNUAL RELEASES (EV)
                     FROM VESSEL CASUALTIES
          PCS CASE
          Expected release fraction (ERF) = 0.1875
          Quantity at risk (Q) = 4,000 MT
          Average quantity released (QR) = ERF * Q = 750 MT
          Expected number of releases per year (EN) = 8.4E-4
          Expected annual release  (EV) = QR * EN = 0.63 MT
          EDC CASE
          Expected release fraction (ERF) = 0.1875
          Quantity at risk (Q) = 3,800 MT
          Average quantity released (QR) = ERF * Q = 712 MT
          Expected number of releases per year (EN) = 1.1E-3
          Expected annual release  (EV) = QR * EN = 0.77 MT
Source:  Appendix C, lEc Analysis

-------
                             Exhibit 5-6

          ESTIMATED SPILL RATE OF THE VULCANUS AS A PERCENT
             OF THE HISTORICAL SPILL RATES FOR TANK SHIPS
                   BY TYPE OF CASUALTY AND LOCATION *
Location


Pier and Harbor

Mobile Bay

Coastal

Burn Zone

All Locations
                  Impact


                    3.6%

                    4.5

                   10.3

                   68.8

                    7.2
Type of Casualty 	
Non-Impact **    All Casualties
   75.5%

   13.7

   13.5

    5.8

   16.3
14.6%

 4.7

10.8

 8.9

 9.8
**
We  compute  these  percentages by comparing  the  estimated
spill  rates  for the Vulcanus to the spill  rates  obtained
from  ECOTANK  for  tank  ships  between  2,000  and  10,000
deadweight  tons.   ECOTANK has five   location  categories:
pier;  harbor;  entranceway to a harbor;  coastal (within 50
nautical miles of land);  and at sea.   For purposes of this
analysis  we  compare  ECOTANK pier and harbor data  to  the
estimated  spill rates for the Vulcanus in Mobile  pier  and
harbor,  ECOTANK  entranceway  data to our  estimated  spill
rates for Mobile Bay,  ECOTANK coastal data to our estimated
spill  rates for coastal waters,  and ECOTANK at sea data to
estimated spill rates for the Vulcanus in the burn zone.

The only adjustment affecting the spill rate of the Vulcanus
from non-impact casualties is for vessel size.
Source:  Derived from Tables 7 and 8, Appendix C

-------
RELEASES FROM INCINERATION                              CHAPTER 6
INTRODUCTION

     Unlike  some  releases  from other components  of  ocean- or
land-based incineration systems,  stack emissions are expected as
part of normal operations.  This chapter presents our analysis of
expected stack emissions, addressing four classes of pollutants:

     o    Undestroyed constituents of the waste stream;

     o    Products of Incomplete Combustion (PICs);

     o    Heavy metals; and

     o    Hydrochloric acid (HC1).

We use existing or proposed EPA permit requirements and available
test  burn  data  to estimate annual releases of  each  of  these
materials.

     In  addition to stack emissions,  the analysis  of  releases
from  land-based incinerators includes estimates of the  quantity
and content of scrubber effluent generated annually.   Land-based
incinerators  typically  are equipped with scrubbers  to  control
emissions  of HC1,  metals,  and particulates.   The analysis  of
incineration  aboard the Vulcanus does  not consider releases  of
scrubber water because the Vulcanus has no scrubber.
                              6-1

-------
      The   analysis  ignores  releases  of  ash  from both  the Vulcanus
and land-based  incinerators.    CWM does not expect  the  Vulcanus1
liquid  injection incinerators  to  generate  ash  residues.    If any
ash is  generated,   EPA's  draft  permit for the Vulcanus  prohibits
the   ship  from  dumping  it at  sea and requires CWM to  dispose  the
ash   on land   in accordance  with  applicable   EPA  regulations.
Similarly,  we   assume  that  any  ash generated by  a land-based
incinerator  would  be  properly   disposed   in   an  EPA-permitted
hazardous  waste landfill.
 EMISSIONS  OF  UNDESTROYED WASTES

      Exhibit  6-1   illustrates our  calculation  of  emission   rates
 for   undestroyed  wastes for  both the  PCB  and   EDC wastes.   The
 calculation   assumes  that  the  Vulcanus  meets   EPA's  proposed
 performance standards for ocean-based  incineration,  and that the
 land-based incinerator meets  existing  EPA  standards for hazardous
 waste incinerators.   The two standards  are  identical,  requiring
 both  ocean-  and land-based incinerators to  achieve a destruction
 and   removal  efficiency   (DRE)  of 99.99  percent.I/  The  only
 exception  to this  requirement is for  incinerators burning   PCBs,
 dioxins,   and dibenzofurans,  which EPA  requires to achieve  a DRE
 of  99.9999 percent.   Based  on  these  requirements and the   total
 annual  throughput  for the PCB and  EDC wastes,  we calculated the
 following  emissions of undestroyed wastes for  the Vulcanus and
 the land-based  incinerator:

      o     Emissions for incineration of  the  PCB waste -
           .056  metric tons per year;

      o     Emissions for incineration of  the  EDC waste -
           6.84  metric tons per year.

Annual  emissions of undestroyed wastes  are  significantly  higher
 in  the  EDC  case since the DRE  required for EDC   wastes  is two
orders of  magnitude lower than that for  PCB  wastes.

      As  described  previously,  we assume  that   the  PCB   waste
contains   35  percent by weight of PCBs and EDC  waste contains  50
I/The standard for ocean incineration is in terms of  destruction
efficiency,  rather  than  destruction  and   removal  efficiency,
because  scrubbers  are not used.   We use the  term DRE for  both
systems to simplify the discussion.


                              6-2

-------
 percent by weight of ethylene dichloride.   The  remainder of each
 waste stream is assumed to be non-hazardous.  Thus,  each of these
 wastes  contains a single "primary organic hazardous constituent"
 (POHC)  as  defined in EPA's current  regulations   for  land-based
 incinerators.  In the calculations above, we assume  both the POHC
 and   the  non-hazardous  constituents  are  destroyed   equally.
 As  shown on Exhibit 6-1,  the resulting annual  release of  POHCs
 only  is 0.02 metric tons for the PCB waste and  3.42 metric  tons
 for the EDC waste.   These POHC releases are used  in our analysis
 of  human  health and environmental effects from   stack  releases
 presented in Chapters 7 and 8.


 PIC EMISSION RATES

     The formation of products of incomplete combustion (PICs) by
 hazardous  waste incinerators is a poorly understood  phenomenon,
 and  the   topic of considerable debate.   By strict  definition,
 PICs  are organic compounds formed by chemical   reactions  during
 the incineration process.  Trial burn data, however, may identify
 as  PICs  all organic compounds other than POHCs present  in  the
 stack  gases.   These compounds may include not  only those formed
 during combustion,  but also compounds from contaminated scrubber
water or non-combusted trace components of the waste feed itself.
We  developed  our estimates of PIC emissions based on  the  best
available  trial burn data for ocean- and land-based  facilities.
These  data,   however,  are  too limited to present a  definitive
portrayal of  PIC emissions.   Specifically,  the data are plagued
by the following  uncertainties:

     o    The  PIC  emission  data are  based  on  differing
          definitions  of what constitutes a PIC.    In  some
          trial  burns  PICs  are defined as  compounds  not
          found  in the waste itself,  while in others  PICs
          are  compounds  not  detected  in  the   waste   in
          concentrations greater than 100 ppm.

     o    The  trial  burn data are not based on   consistent
          sampling procedures.   Further, the trial burns did
          not sample for a consistent set of  compounds,  so
          the  results  depend  upon  whether  investigators
          tested for the presence of a particular  compound.
                              6-3

-------
     o    The  trial burn data are not based on burns  of  a
          consistent  set of wastes,  and none of the wastes
          analyzed  are identical to the PCS or  EDC  wastes
          of concern in this study.

     o    The  emission  rates measured in trial  burns  are
          based   on  several  types  of  incinerators  with
          different operating parameters.  PIC generation is
          believed  to  be a function of  incinerator  type,
          configuration, and operating conditions.

     o    Finally,  there is disagreement over the  validity
          of the results of some test burns due to  possible
          contamination of samples.

With these uncertainties in mind,  our estimates of PIC  emission
rates should be viewed only as rough approximations.  However,  in
the   absence   of  better  data  or  an   improved   theoretical
understanding  of PIC generation,  we believe these estimates are
the best available.2/


PIC Emissions from the Vulcanus

     We base our estimates of PIC emission rates for the Vulcanus
II  on available trial burn data for the ship and for its  sister
ship,  the  Vulcanus I.   As explained in Appendix E,  these data
yield the following estimates:

     o    For  PCS  wastes burned at  99.9999  percent  DRE,
          hazardous  PICs  are emitted at a rate no  greater
          than 1.13E-11 percent of total waste feed;

     o    For  non-PCB  wastes burned at 99.99 percent  DRE,
          PICs  are emitted at a rate of 3.01E-2 percent  of
          total waste feed.

     As these estimates show, ocean-based incineration is assumed
to  generate  significantly less PICs when burning the PCB  waste
than when burning EDC waste.  This large difference is partly the
result of two factors:
2/Appendix  E  reviews the test burn data and fully explains  the
methodology employed to derive the PIC emission estimates.


                              6-4

-------
     o    Past  trial burns  of  PCB-containing  wastes  on   the
          Vulcanus  I  and II tested  for  a   single  compound
          onlyr  while  burns of  other  organic  wastes  were
          tested for many compounds,  and  I/

     o    Some  experts believe that  the  higher  ORE required
          for  PCB wastes should  result in lower generation
          of PICs.

However,  the  large  discrepancy between the  two   wastes  raises
concern for the accuracy of  the estimates.   As shown  below  and  in
Appendix E, additional cause for  concern  is  a  further discrepancy
in  relative  PIC  generation   rates   for land- and  ocean-based
incinerators.   The  test burn  data for PCB  wastes  indicate  that
land-based  incinerators  generate more  PICs  than  ocean-based
incinerators,  but  the  data for non-PCB wastes show that  land-
based incinerators generate  considerably  fewer PICs.  At  present,
we   know   of  no  complete explanation for  these    apparent
inconsistencies.

     We  used  these PIC emission rates and  the  throughput  data
described  above to calculate total annual releases of  PICs  for
both the PCB and EDC wastes.    These  calculations,  summarized  in
Exhibit 6-2, produce the following estimates:

     o    PIC  emissions from PCB burns:  6.3E-9 metric tons
          per year;

     o    PIC emissions from EDC  burns: 20.6 metric tons  per
          year.
I/As described in Appendix E, ten  PCB   burns  onboard  the Vulcanus
revealed  no  detectable  levels   of   tetrachlorodibenzo-p-dioxin
(TCDD),  the PIC of interest  in this analysis.   The estimated  PIC
emission  rate  given above is based on the maximum   quantity   of
TCDD that could have gone undetected.     The  trial burns did   not
analyze other PICs.
                              6-5

-------
     In  light of the anomalies in the test burn data,  and given
the  highly carcinogenic nature of PICs that may be generated  by
PCB incinerators,  we include in Chapter 8 a sensitivity analysis
for  at-sea  incineration of PCB wastes which employs the  higher
PIC generation rate given for land-based incinerators.   As shown
in  Exhibit  6-2,   use  of this higher rate increases the annual
quantity  of  PIC emissions in the PCB case by  more  than  three
orders  of  magnitude  (to 1.9E-5 metric tons per year)  for  the
ocean case.


PIC Emission Rates for Land-Based Incinerators

     Like  the estimates of PIC emissions from the Vulcanus,  our
estimates of PIC emissions from land-based incineration are based
on  the analysis of test burn data described in Appendix  E,  and
are  subject to the same uncertainties.   With these  caveats  in
mind,  we  calculated the following PIC emission rates for  land-
based incineration:

     o    For  PCB  wastes burned at  99.9999  percent  DRE,
          hazardous  PICs  are emitted at a rate of  3.40E-8
          percent of total waste feed; £/

     o    For  non-PCB wastes burned at 99.99  percent  DRE,
          PICs  are emitted at a rate of 8.94E-4 percent  of
          total waste feed.

     Based on these emission rates,  we calculated the  following
estimates  of  total  annual PIC emissions for the  PCB  and  EDC
cases:
4/The  PIC  emission rate reported above is for  TCDD  only.   As
described in Appendix E,  the test burn data indicate that furans
and  other  dioxins  also  are generated as  PICs  at  land-based
incinerators  burning PCB wastes.   We based our estimate of  PIC
emission  rates solely on TCDD emissions because  EPA  recommends
that  risk  analyses for emissions containing mixtures of  furans
and  dioxins be based on TCDD emissions and the unit risk  factor
for TCDD.   Since the PIC emission rate for PCB burns aboard  the
Vulcanus also reflects only TCDD emissions, the estimates for the
two systems are directly comparable.
                              6-6

-------
     o    PIC emissions from PCB burns:  1.9E-5 metric tons;
          and

     o    PIC emissions from EDC burns: 0.61 metric tons.

Exhibit  6-3 contains the calculations employed to develop  these
estimates.
METALS EMISSIONS

Ocean-Based Incineration

     Because  the  Vulcanus  is not equipped with  scrubbers,  its
stack  emissions include any metals contained in the waste  feed.
To   control  metals  emissions,   EPA  plans  to  issue   metals
concentration limits for wastes burned at sea.  Current proposals
include individual limits on a number of metals,  with a  maximum
limit  for total metals on each manifested shipment of wastes  of
500 ppm.   For purposes of analysis, we assumed that both the PCB
and EDC wastes contain arsenic,  cadmium, chromium, and nickel —
the four metals for which EPA's Carcinogen Assessment Group (CAG)
has  developed  cancer unit  risk factors — at concentrations  of
100   ppm  each.5/    The  concentration   assumptions   probably
overstate  the  average concentrations of carcinogenic metals  in
liquid incinerable waste streams.

     Based on this assumption,  Exhibit 6-4 shows expected annual
metals  emissions  for the Vulcanus.   As the exhibit  indicates,
estimated  metals emissions  in the PCB case are 5.6  metric  tons
per  year  for  each  metal,  or a total  of  22.4  metric  tons.
Estimated  annual emissions  in the EDC case are 6.84 metric  tons
of each metal, for a total of 27.36 metric tons per year.


Land-Based Incineration

     Our   analysis   of   metals   emissions   from   land-based
incinerators  assumes  the   same concentration of metals  in  the
waste feed as the ocean-based analysis,  but considers the effect
of a scrubber in removing  metals.   As explained in Appendix  D,
I/Appendix D gives a fuller explanation of the rationale employed
in selecting these metals.


                              6-7

-------
there is significant uncertainty associated with scrubber removal
efficiencies.  Based on available trial burn data and discussions
with  EPA experts,  we assumed that scrubbers achieve 90  percent
removal of cadmium,  chromium and nickel,  and 50 percent removal
of  arsenic.   Removal of arsenic is lower because this metal  is
more volatile.   These assumptions yield the following  estimates
of annual metals emissions for a land-based incinerator:

     o    PCS  Case:  0.56  metric  tons  each  of  cadmium,
          chromium  and  nickel,  and  2.8  metric  tons  of
          arsenic;

     o    EDC  Case:  0.684  metric  tons each  of  cadmium,
          chromium  and  nickel,  and 3.42  metric  tons  of
          arsenic.

In  the  PCS case,  metals emissions total 4.48 metric  tons  per
year.  In the EDC case, annual metals emissions total 5.47 metric
tons.   Exhibit  6-5 summarizes the calculations used to  develop
these estimates.
HYDROCHLORIC ACID EMISSIONS

Ocean-Based Incineration

     The lack of scrubbers on the Vulcanus implies that  chlorine
contained  in the waste feed will be emitted from the incinerator
stacks.   Host  chlorine is  emitted as hydrochloric acid   (HC1),
while  the rest may be emitted as chloride salts  or  chlorinated
organic matter.   We did not analyze the  forms in which chlorine
may  be  emitted,  but  simply calculated the total  quantity  of
chlorine  released and assumed that it would be released as  HC1.
Based on a mass balance calculation, we estimated that in the PCB
case,  10,505 metric tons of chlorine per year are released.   In
the  EDC  case,  25,034  metric  tons of chlorine  per  year  are
released.   We  base  these  estimates  on  the  assumption  that
Arochlor  1254  (53.6  percent chlorine)  or  EDC  (73.2  percent
chlorine)  is the only source of chlorine in the PCB or EDC waste
(that is,  we assume that the balance of the waste is composed of
non-chlorinated organics).
                              6-8

-------
Land-pased Incineration

     The presence of scrubbers on land-based incinerators greatly
reduces  emissions  of  HC1  and  other  chlorinated   compounds.
According to an ADL analysis (see Appendix F),  scrubbers achieve
99 percent removal of hydrochloric acid emissions.  Based on this
removal efficiency, ADL estimated annual emissions of 105 and 250
metric tons of chlorine,  in the form of HC1, for the PCB and EDC
cases, respectively.


SCRUBBER EFFLUENT

     Appendix  F  describes  ADL's analysis  of  the  generation,
composition, and disposition of scrubber effluent from land-based
incinerators.   The  analysis  assumes the scrubber  achieves  99
percent removal of hydrochloric acid emissions,  as EPA requires.
Based on these and other assumptions explained in Appendix F, ADL
estimated  that an incinerator burning  56,000 metric tons of PCB
wastes  annually generates approximately 1.5 million metric  tons
of scrubber blowdown.   A unit incinerating 68,400 metric tons of
EDC  wastes  annually produces approximately 3.6  million  metric
tons of scrubber blowdown.  Waste water accounts for more than 99
percent of both these quantities.

     As discussed in Appendix F,  chlorine is the principal  non-
aqueous  constituent  of  scrubber  blowdown,   representing  0.7
percent of the total effluent.   In the PCB case,  ADL  estimated
that  10,400  metric tons of chlorine per year are  contained  in
scrubber  effluent.   In the EDC case,  ADL estimated that 24,784
metric tons of chlorine are removed by scrubbers and contained in
scrubber blowdown.

     EPA  regulates  disposal of scrubber sludge or  brine  under
RCRA and controls the discharge of scrubber waste water under the
Clean Water Act, limiting the probability that hazardous scrubber
wastes will be released to the environment.   In fact,  three  of
four   incinerator  operators  contacted  by  ADL  sell  scrubber
effluent to be used,  reused,  recovered or recycled, eliminating
the release of any wastes to the environment.   If scrubber waste
water  is  discharged,   the operators  typically  neutralize  it,
raising the pH to 7.0 or 7.5.   As indicated above,  its chlorine
content  is  normally  maintained below one  percent,  with  most
chlorine contained in low concentrations of dissolved salts.   At
                              6-9

-------
 a   pH  of  7.0 or above there is virtually no hydrochloric  acid   in
 the effluent.   Scrubber  effluent  also  contains   virtually   no
 organic material.

     Treated  scrubber effluent also can contain metals   removed
 from stack  emissions.   Assuming the scrubber removal efficiencies
 described  above,   the  maximum quantity of metals   contained   in
 waste   scrubber  water  will   equal  40  percent   of  the  total
 throughput  of cadmium,  chromium,   and nickel,   and one-half  the
 total  throughput of arsenic.    Therefore,  the maximum quantity of
 metals contained in scrubber  effluent from PCB  incineration  would
 be  17.92 metric tons per year (5.04 metric tons each of  cadmium,
 chromium,   and  nickel, and 2.8 metric tons of arsenic); and  21.89
 metric tons per year (6.156 metric tons of cadmium,  chromium,  and
 nickel,   and  3.42  metric tons of arsenic)  in the  EDC  case.
 Exhibit 6-6 illustrates the calculation of these estimates.
CAVEATS AND SENSITIVITY  ANALYSIS

      In  addition  to  the   caveats  regarding   estimates   of   PIC
emissions,  this  analysis assumes incinerators  operate  to meet  EPA
permit  conditions.    With  regard to this assumption,   there   has
been  concern  that  periodic incinerator  malfunctions or "upsets"
could lead to significant  releases  of unburned wastes.    Federal
regulations require   incinerators to be equipped with automatic
waste  feed  cutoff   systems  that   prevent   operation of    the
incinerator in  the event that  critical  operating conditions   are
not met.    In  addition,   the thermal inertia of a hazardous waste
incinerator is  such that any wastes introduced into   the  system
immediately before and during an upset would be  substantially
combusted.   If   automatic   waste feed cutoff  normally occurs   in
less  than one  second and destruction efficiency in the event   of
an  upset   is  at least  99.9 percent,   incinerator  malfunctions
should  have an  insignificant effect on  annual stack   emissions.
In  our  worst case analysis—shutdown of operations   within   ten
seconds,  maintaining  only 90  percent DRE—emissions  during   the
upset period  would be equivalent to normal emissions  over three
hours of   operation   at 99.99 percent   DRE,   or  280  hours   of
operation   at  the  more  stringent PCB level  of 99.9999   percent
DRE.£/  In  this  case,  about ten such upsets over the  burn period
It/Memorandum from J.R. Ehrenfeld  (ADD to  Incineration Risk  Study
File,  "Potential  Emissions Due  to  Incinerator Malfunction,"   15
August 1984.


                              6-10

-------
for  PCBs  of 296 hours would have the effect of reducing DRE  by
one  order of magnitude.   There would have to be about 600  such
upsets over the 169 hour burn period to reduce the DRE for EDC by
about an order of magnitude.    Unfortunately, we have no data on
the number of upsets likely to occur.
SUMMARY

Ocean-Based Incineration

     Our  estimates of annual stack emissions from an ocean-based
incineration system are:

                            Table 6-1

             Summary of Ocean-Based Stack Emissions
     Undestroyed Wastes

     PICs

     Metals

     Chlorine  (as HC1)

     Source:  lEc Analysis
  PCB Waste
(MT per year)

   5.60E-2

   6.33E-9

   2.24E+1

   1.05E+4
  EDC Waste
(MT per year)

   6.84E+0

   2.06E+1

   2.74E+1

   2.50E+4
The  quantity  of  chlorine  emitted  as  HC1  clearly  dominates
releases of the three other categories of emissions.  Undestroyed
wastes,  PICs,  and  metals emissions,  however,  are of  greater
concern  in characterizing the magnitude of potentially hazardous
emissions.    As  the  table shows,  our assumptions result  in  a
substantial quantity of metals  emitted from the Vulcanus in both
the PCB and EDC cases.   PIC emissions are also large in the  EDC
case, but represent a far smaller share of total emissions in the
PCB case.  Emissions are higher in the EDC case because the waste
is   burned  at  a  lower  DRE,   the  PIC  generation  rate   is
substantially  higher,  and the waste contains more chlorine than
the PCB waste.
                              6-11

-------
Land-Based  Incineration

     Our estimates of likely annual stack emissions from a  land-
based hazardous waste incinerator are:

                            Table 6-2

              Summary of Land-Based Stack Emissions
     Undestroyed Wastes

     PICs

     Metals

     Chlorine  (as HC1)
  PCB  Waste
(MT per  year)

   5.60E-2

   1.90E-5

   4.48E+0

   1.05E+2
  EDC Waste
(MT per  year)

   6.84E+Q

   6.11E-1

   5.47E+0

   2.50E+2
     Source:  lEc Analysis
The  stack  emissions for both wastes contain more  chlorine   (as
HC1)  than any other category of emissions.   The quantity of HC1
released,  however,  is  likely  to be of less concern  than  the
quantity of undestroyed wastes,  metals and PICs. The table shows
that a substantial quantity of metals is released in both the PCB
and EDC cases,  with metals emissions exceeding emissions of both
PICs  and  undestroyed wastes in the PCB case.   The  table  also
shows  that emissions of undestroyed wastes exceed PIC  emissions
by  two  orders of magnitude in the PCB case,  and one  order  of
magnitude in the EDC case.  Total annual PIC emissions in the EDC
case are approximately three orders of magnitude higher than  the
PCB case.

     Releases  from  land-based  incinerators  can  also  include
scrubber wastes containing substantial quantities of chlorine and
metals.  Because the disposal of scrubber sludge, brine and waste
water  are  regulated  by EPA,  uncontrolled  releases  of  these
materials  in  hazardous  form are  unlikely.   The  presence  of
scrubbers reduces stack emissions  of chlorine as HC1 two  orders
of   magnitude   below  similar  emissions  for   the   Vulcanus.
                              6-12

-------
Similarly, scrubbers reduce expected annual metals emissions from
a  land-based unit to 20 percent of the expected metals emissions
from the sea-based system.

     As  noted above/  emissions of undestroyed wastes from land-
based  incinerators  are  assumed  to be the  same  as  from  the
Vulcanus.   PIC emissions,  however, differ markedly.  In the EDC
case,  annual PIC emissions are two orders of magnitude lower for
land-based incinerators.   In the PCB case,  the opposite appears
to hold;  PIC emissions for land-based incinerators are estimated
to be three orders of magnitude higher than PIC emissions for the
Vulcanus.  As described previously, this apparent anomaly and the
uncertainties associated with the PIC analysis raise considerable
questions  regarding the accuracy of the PIC estimates  for  both
the land-based and ocean-based systems.
                              6-13

-------
                           Exhibit 6-1
        CALCULATION OF ANNUAL EMISSIONS OF UNDESTROYED WASTES
PCB, Was f:p
          Assumptions:
               ORE  =  99.9999%
               Annual throughput  -  56,000 MT
          Calculation:
               Annual emissions
               of undestroyed wastes  =  56,000 HT * U-.999999)
                                      =  .056 MT
               Annual emissions
               of POHC (PCB)           =  .020 MT
EpC Waste
          Assumptions:
               ORE  =  99.99%
               Annual throughput  =  68,400 MT
          Calculation:
               Annual emissions
               of undestroyed wastes  -  68,400 MT *  (1-.9999)
                                      -  6.84 MT
               Annual emmisions
               of POHC  (EDO          -  3.42 MT
Source:  lEc Analysis

-------
                           Exhibit 6-2

          CALCULATION OP ANNUAL PIC EMISSIONS FROM THE
                        INCINERATOR SHIP
PC
     Assumptions:

          PIC emission rate  =
          Annual throughput  =

     Calculation:

          Annual PIC emissions
1.13E-11% of waste feed
56,000 MT
=  56,000 MT * 1.13E-13
=  6.33E-9 MT
EDC Waste

     Assumptions:

          PIC emission rate  =
          Annual throughput  -

     Calculation:

          Annual PIC emissions
3.01E-2% of waste feed
68,400 MT
   68,400 MT * 3.01E-4
   20.59 MT
                Aa.sis
     Assumptions:

          PIC emissions rate  =
          Annual throughput   =

     Calculation:

          Annual PIC emissions
 3.40E-8% of waste feed
 56,000 MT
   56,000 MT * 3.40E-10
   1.90E-5 MT
Source:  IEC Analysis

-------
                           Exhibit 6-3

               CALCULATION OF ANNUAL PIC EMISSIONS
                  FROM A LAND-BASED INCINERATOR
PCB, ftaste

     Assumptions:
          PIC emission rate  =  3.40E-8 percent of waste feed
          Annual throughput  =  56,000 MT
     Calculation:
          Annual PIC emissions  =  56,000 MT * 3.40E-10
                                =  1.90E-5 MT
EDC Waste

     Assumptions:
          PIC emission rate  =  8.44E-4 percent of waste feed
          Annual throughput  =  68,400 MT
     Calculation:
          Annual PIC emissions  =  68,400 MT * 8.94E-6
                                =  6.11E-1 MT
Source:  lEc Analysis

-------
                           Exhibit 6-4
      ANNUAL METALS EMISSIONS FROM OCEAN-BASED INCINERATION
Metal
Arsenic
Cadmium
Chromium
Nickel

TOTAL
Concentration
  In Waste
    (ppm)
     100
     100
     100
     100
                                           Annual Emissions
PCB Waste
   (MT)
   5.6
   5.6
   5.6
   5.6

  22.4
EDC Waste
  (MT)
   6.84
   6.84
   6.84
   6.84
                                     27.36
Source:   IBc Analysis

-------
                           Exhibit 6-5
           CALCULATION OP ANNUAL METALS EMISSIONS FOR
                     LAND-BASED INCINERATION
Assumptions
                   —Annual Throughput—     Scrubber
                           (MT)              Removal
     Metal         PCS Waste   EDC Waste     Efficiency
Arsenic
Cadmium
Chromium
Nickel
5.6
5.6
5.6
5.6
6.84
6.84
6.84
6.84
50 %
90 %
90 %
90 %
       Total         22.4        27.36

CfrJ.culaEJ.ons
     Annual Emissions  =  Annual Throughput  *  (1  - Removal Efficiency)

PCB Waste - Cadmium, Chromium and Nickel:
     Annual Emissions  =  5.6 MT *  (1 -  .9)
                       «  0.56 MT

PCB Waste - Arsenic:
     Annual Emissions  «  5.6 MT *  (1 -  .5)
                       =  2.8 MT

EDC Waste - Cadmium, Chromium and Nickel:
     Annual Emissions  =  6.84 MT  * (1 - .9)
                       =  0.684 MT

EDC Waste - Arsenic:
     Annual Emissions  =  6.84 MT  * (1 - .5)
                       =  3.42 MT
 Source:   lEc Analysis

-------
                           Exhibit 6-6
          CALCULATION OP THE MAXIMUM QUANTITY OF METALS
             CONTAINED IN TREATED SCRUBBER EFFLUENT
Assumptions
                   —Annual Throughput—     Scrubber
                           (MT)               Removal
     Metal         PCB Waste   EDC Waste     Efficiency
Arsenic
Cadmium
Chromium
Nickel
5.6
5.6
5.6
5.6
6.84
6.84
6.84
6.84
50 %
90 %
90 %
90 %
       Total         22.4        27.36
     Quantity in Effluent  «  Annual Throughput * Removal Efficiency
PCB Waste - Cadmium, Chromium and Nickel:
     Quantity in Effluent  «  5.6 MT * .9
                           -  5.04 MT

PCB Waste - Arsenic:
     Quantity in Effluent  =  5.6 MT * .5
                           =  2.8 MT
EDC Waste - Cadmium, Chromium and Nickel:
     Quantity in Effluent  =  6.84 MT *  .9
                           -  6.156 MT

EDC Waste - Arsenic:
     Quantity in Effluent  =  6.84 MT *  .5
                           =  3.42 MT
Source:  lEc Analysis

-------
EFFECTS OF RELEASES FROM
OCEAN TRANSPORTATION                                    CHAPTER 7
INTRODUCTION

     The  probability  that  hazardous wastes might  be  released
accidentally during the transit of the Vulcanus from the pier  at
Chickasaw  to   the  Gulf burn zone was described in  Chapter  5.
Although spills are very unlikely,  they could involve relatively
large quantities of waste.  In addition,  this type of release is
unique  to  ocean-based incineration systems and  has  no  direct
parallel  in land-based systems.  Thus,  we have included in  our
analysis  an  evaluation of the possible environmental and  human
health  effects from spills of hazardous wastes into  the  marine
environment.

     lEc subcontracted with Applied Science Associates, Inc.(ASA)
to  analyze the pollutant transport and marine ecosystem  effects
that  could  result from a loss of cargo from the ship  at  three
points along the vessel's transit path:

     o    in  the Mobile Bay ship channel,

     o    over  the continental shelf near  the  Mississippi
          River  delta  (29 degrees 45 minutes North and  88
          degrees 30 minutes West), and

     o    in  the  center of the designated  burn  zone  (26
          degrees 40 minutes North and 93 degrees 40 minutes
          West).
                              7-1

-------
Since  the  characteristics  of  the marine  environment  differ along
the vessel's  proposed path,  analysis  of  the  effects  of  spills in
these  three   areas  was  necessary  to determine   the   range  of
potential effects  of accidental  releases  during  the voyage to the
burn area.I/

     We asked ASA  to evaluate  three  release quantities — a  loss
of  half  a tank,  a loss of two tanks and a  loss of  the  entire
cargo.   As  discussed  in Chapter 5,   the estimated overall spill
rate  for the Vulcanus  is very low —  about 1 per 17,000  voyages
or 1 per 1,200 operating years.  The estimated spill  rate for any
particular  location is  even lower.   Further, our analysis of the
distribution  of  spill sizes suggests that about  80  percent of the
time only one cargo tank will  be affected.    Casualties  resulting
in  damage  to  two or  to three or more cargo tanks  would  occur
about 15 percent and 5  percent of the  time,   respectively.  Thus,
the probability  for a release  of cargo from one  tank,  two tanks,
or  three or  more  tanks in  Mobile Bay  is  estimated  to be  on  the
order  of one per  11,000,   67,000,   and 200,000  operating  years,
respectively.  Similar  estimates for other locations  are provided
in Chapter  5.

     The analysis  in Chapter 5 indicates  that the probability  of
a  casualty  affecting  one or  two  tanks is  very  low  and  the
probability  of  a  casualty  resulting in a complete  loss  of  cargo
is  extremely low — even  less  than the  1 in 200,000   operating
years  shown   above  for  a loss  involving  3  or  more  tanks.
Nonetheless,   we  have  estimated the  potential  effects  from  the
loss  of an entire cargo to bound the  possible damage that  could
result from an ocean transportation  release.

     ASA's  calculations  were carried out for each  of  the  two
waste streams selected  for  this study:
I/Due  to time and budget constraints,  we  have  not  evaluated  the
effects of releases in the pier area, Chickasaw  Creek,  or  harbor.
Information   in  Appendix  C  indicates  that about  80 percent   of
spills in the pier and harbor area would  be expected at the  pier.
We  believe that containment  or remedial  actions undertaken  as  a
response to spills at the pier are likely to substantially reduce
the amount of contaminants remaining in the environment.
                              7-2

-------
     1.   a  waste  containing  35  percent  by  weight   of
          polychlorinated  biphenyls  (Arochlor 1254) and  65
          percent  non-hazardous  material (it  is  expected
          that  PCB  concentrations in waste streams in  the
          future will be in the 10 percent range), and

     2.   a  waste  containing  50  percent  by  weight   of
          ethylene  dichloride and 50 percent  non-hazardous
          material.

The   dispersion   of  a  pollutant   through   the   environment,
particularly  in  the ocean,  can be  strongly influenced  by  the
mixture  of substances in which the pollutant is included and the
resulting  physical  and  chemical state of  the  pollutant  when
released.  We  asked  ASA to ignore this  "mixture"  problem  and
simply  calculate  dispersion and effects for the two  individual
chemicals of interest. Although Arochlor 1254 is heavier than sea
water  and  would be expected to sink,  we have  considered  both
"floating"  and "sinking" cases to reflect the fact that the PCBs
might  be entrained in a lighter-than-water fraction  that  would
float  much  like an oil slick.2/  Since ethylene  dichloride   is
soluble in water, ASA calculated EDC  concentrations assuming that
the  compound  quickly dissolves and  diffuses once released  from
the ship.

     Along  with  the  specific composition  of  the  waste,  the
dispersion  and  ultimate effects of  material released  from  the
ship  during an accidental spill would depend upon other factors,
including the nature of the accident,  weather and sea conditions
at  the time,  and the effect of any  mitigating  actions  carried
out. We asked ASA to assume in their  analysis that:

     o    the  waste is released to the surface of the ocean
          at a single point, instantaneously;

     o    no mitigating activities are carried out; and
2/JRB Associates,  Inc. tested seven  samples  of  PCB  liquid wastes
to  determine   behavior  in  water.    Four of   the   samples   had
specific  gravities  greater  than  seawater  and  sank   readily,
whereas  three  of the samples had  specific gravities less   than
seawater and floated (See JRB Associates,  Inc.,  Expanded  Modeling
£f  Incineration ai S_ea Impacts; Glllf QL   Mexico,  Final  Report,
June 1, 1984, p. 2 - 32).


                              7-3

-------
     o    weather  and  sea  conditions reflect  the  annual
          average.


     These  assumptions simplify the analysis considerably.   For
example,  the discussion in Chapter 5 indicates that the quantity
of  waste released and the speed of release depends on a  variety
of factors, such as the nature of the accident and the density of
the  material  relative to  seawater.   Further,  the  amount  of
contaminants  that remain in the marine environment after a spill
depends  on  the  extent to which spills  can  be  contained  and
recovered.

     Methods   for  containing  and  recovering  floating  spills
include  the  use  of floating booms to  contain  slicks  and  of
skimmers  to remove the slick from the surface.   Techniques  for
recovering  materials  that  sink include the use  of  hand  held
dredges,  suction dredges, and hydraulic dredges. Such mitigating
activities  are  likely  to be most successful in  the  pier  and
Mobile  Bay  areas and least successful on the continental  shelf
and   in  the  burn  zone.    However,   evaluating  the   likely
effectiveness  of mitigating activities in removing  contaminants
from  the marine environment was beyond the scope of this  study.
Our  analysis of several release amounts provides information  to
bracket the likely effects of spills and to estimate the  effects
of various amounts of wastes remaining in the marine environment.

     In  addition to the analysis provided by ASA,  lEc  analyzed
the  potential  for short term human health effects from a  spill
through inhalation of pollutants volatilized from the spill site.
ZEc's calculation of risks from spill volatilization is described
in  detail in Appendix H.

     The remaining sections of this chapter describe the  methods
used by ASA and lEc and the limitations of these methods; and the
dispersion,  marine effects and human health effects from  spills
at each of the three sites.   A more detailed discussion of ASA's
data, methods, assumptions and results is provided in Appendix I.


METHODS AND THEIR LIMITATIONS

     ASA  used a computerized ocean transport model  to  estimate
the movement of pollutants released over the continental shelf or
in  the burn zone.   This model represents the Gulf of Mexico  in
square grids of size .25 degree,  and uses data on average surface


                              7-4

-------
and  subsurface  current  fields to move  and  disperse  released
material across the area. Mixing in the vertical dimension occurs
through dispersion,  settling, and (for shallow areas such as the
continental    shelf)   resuspension   due   to   wave    action.
Concentrations in the sediments for pollutants that settle assume
perfect mixing to a depth of 5 centimeters.

     In  the case of a floating PCB spill,  the waste material is
assumed  to be perfectly mixed in the top one meter of water  and
to be moved and dispersed only by the seasonally averaged surface
currents  and  wind.  In the case of a  sinking  PCB  spill,  all
hazardous  material  is assumed to fall through the water  column
and reach the ocean floor.  Some of the PCBs will sink rapidly to
the  sea floor,  whereas others would settle at  different  rates
depending  on the settling velocities of particulate matter  onto
which  they  absorb.    Accordingly,  ASA  used  three  different
settling velocities  (0.1,  0.01 and 0.001 m/s) and assumed  equal
amounts  of PCBs would settle out at each velocity.   In the  EDC
case,  the material is assumed to dissolve  and disperse rapidly.
It  is important to note that none of these calculations  include
the  effects of weathering,  volatilization or other physical  or
chemical reactions which might reduce concentrations in the water
column or sediments.2/

     Estimating  the  dispersion  of releases in Mobile  Bay  for
persistent  materials,  such as PCBs,  is difficult.   Among  the
factors  that would influence the dispersion of materials in  the
short-term  are the density of the material  (sinking or  floating
spill),  physical or chemical reactions of the material while  in
the water column, settling velocities, the location of the spill,
wind  speed  and  direction,  and the direction and  velocity  of
surface  and  subsurface currents.   ASA developed  a  number  of
scenarios  for the short-term dispersion of floating spills based
on most probable wind speeds and directions in Mobile  Bay.   The
long-term  transport  of  PCBs  in  sediments  was  qualitatively
assessed  based  on  information on average  hydrodynamic  energy
I/The  JRB analysis cited previously  indicates  that  only  a   small
percentage  of PCBs in a surface  slick would volatilize and  most
of  these would be the less  chlorinated  PCBs.    Thus,  we do  not
expect  such  processes to materially reduce the amount   of  PCBs
remaining  in the marine environment  after  a spill.    However,   a
significant amount of EDC could volatilize  if there  were  a spill,
particularly in the shallow  waters  of Mobile Bay.


                              7-5

-------
levels  in the Bay.   Estimates of the dispersion of spills  that
sink to the bottom of the Mobile ship channel are taken from  the
JRB analysis.

     Once  transport and dispersion of the hazardous  constituent
was  calculated using the methods outlined above/  ASA  estimated
the   effects   of  the  resulting  water  column  and   sediment
concentrations on ten trophic levels of marine organisms found in
Mobile Bay and the Gulf of Mexico.  The trophic levels considered
include:

     o    phytoplankton,

     o    benthic flora,

     o    benthic deposit feeders (e.g. mollusks),

     o    demersal detritivores (e.g. shrimp, crabs),

     o    herbivorous zooplankton,

     o    carnivorous zooplankton,

     o    small pelagic fish (e.g. menhaden, herring),

     o    large pelagic fish (e.g. tunas, sharks),

     o    benthic invertebrate predators  (e.g. corals), and

     o    demersal fish (e.g. flatfish, deep sea species).


     ASA  used a computerized marine ecosystem model to  estimate
effects   on   these  trophic  levels.    The   model    includes
formulations  of  physiological and metabolic processes for  each
trophic  level.   Seasonal cycles are important in the  areas  of
interest  and  are reflected in the ASA model.    The  ecosystems
model  estimates   whether pollutant concentrations  in the  water
column  and in sediments are acutely toxic to marine species   (in
the  event  of sufficiently high  concentrations)  and  estimates
bioconcentration    through   the   food    web.    Dose-response
relationships  for  acute effects  and  bioconcentration  factors
were  estimated  by  ASA from available data in  the  literature.
Chronic  marine  effects (for example,  changes  in  reproductive
rates)   due to long-term exposure to sublethal concentrations  or
to bioconcentration are not included due  to lack of  quantitative
                              7-6

-------
information  relating  cause and effect.   A diagram showing  the
interrelationships among trophic levels in the model is  provided
as Exhibit 7-1.

     ASA's analyses describe the effects of marine spills through
the following measures:

     1.   physical  descriptions  of the dispersion  of  the
          hazardous constituent;

     2.   changes  in  the  level of  biomass  in  the  area
          affected  by  the  spill,   reflecting  the  acute
          toxicity  of  the hazardous constituent to  marine
          organisms as well as effects through the food web;
          and

     3.   changes   in  the  tissue  concentration  of   the
          hazardous  constituent in each of the ten  trophic
          levels  to reflect bioconcentration and  potential
          chronic effects.

The transport model delineates the area affected by the spill and
estimates the contaminant concentrations in the sediments in this
area.   For  the continental shelf and burn zone cases,  sediment
concentrations  exceeding 1 ppb after settling of  the  pollutant
are  used  to define the area over which the ecosystems model   is
run.    Changes   in   the  level  of  biomass  and   in   tissue
concentrations  are  calculated assuming that  the  pollutant   is
uniformly  concentrated  in  the sediments over  this  area  (the
average  concentration  is calculated by dividing the  amount   of
pollutant  within  the  area  by the volume of the top  5  cm   of
sediments  within the area).   Marine species are assumed neither
to  migrate into or out of the contaminated area.

     For  the  Mobile  Bay case,  the ecosystems  model  was  run
assuming  uniform concentration across the entire bay (calculated
as   a  weighted  average  of  contaminated  and   uncontaminated
sediments).   This simplification was necessary because there   is
insufficient  information  to  predict accurately  the  long-term
transport  of  PCBs in sediments and to assess  how  the  feeding
habits   of  marine  species  would  affect  bioconcentration   if
portions  of the Bay were contaminated at different  levels.    In
all  analyses,  biomass  is estimated in units of  milligrams   of
carbon  per  square  meter of area.  This  can  be  converted   to
milligrams  wet  weight  of  living matter per  square  meter   by
                              7-7

-------
multiplying   by   a  factor of 20.  As  stated   above,   all   of   the
methods   and   data   used  and the results  obtained   by  ASA   are
explained in  more detail in Appendix  I.

      In   the  event  of  a spill some hazardous   constituents might
volatilize and   subsequently could be  inhaled by  humans   causing
possible   short  term health effects.  We estimated the  potential
for   such effects  by  estimating the  rate  of   volatilization   for
each  constituent  for the first 24 hours  following a spill   and
modeling   downwind   ambient concentrations.   For the  harbor   and
continental   shelf  spills,  we  used  a  simplified  virtual point
source air dispersion  equation to calculate ground level downwind
concentrations  at  the plume centerline.   For the  burn zone spill
we adapted ASA's air dispersion modeling results (see Appendix G)
to generate similar  outputs.  In all  three cases,  concentrations
for the first 24 hours following a spill were calculated for   the
nearest point at which humans might be  exposed, and the resulting
estimates were   compared  with allowable  limits   for short-term
exposure   to  the constituents of interest.  Appendix  H  describes
lEc's volatilization analyses in detail.

     The   models and  methods used by  ASA and lEc   to  estimate
dispersion and  effects from ocean transportation  releases are by
nature  incomplete  and uncertain.!/ The major strengths of these
tools are that they:

     o     unify   existing  scientific  knowledge  into    an
           internally consistent framework,

     o     force  explicit recognition  of assumptions and data
           sources,  and

     o     allow   investigation of the effect  of  alternative
           data,   methods   or   policies  on  the   results
           predicted.
A/The  ASA model is the most comprehensive  currently   available.
However,  because it has only recently been developed,  it has not
yet undergone a comprehensive peer review.
                              7-8

-------
Because of the extremely complex processes being analyzed and the
significant  scientific uncertainties that remain in many  areas,
the  results  obtained should be viewed as a general  indication,
rather  than  a precise measure,  of pollutant transport  and  of
potential effects.  Ideally, these models and methods should only
be  used  when  time and resources  allow  extensive  sensitivity
analyses  to be carried out,  coupled with  verification  studies
utilizing   field  measurements.   Such  studies  have  not  been
performed for the applications reported here.  The marine models,
however, have been run using data on background concentrations of
PCBs in sediments and the water column and have yielded estimates
of  concentrations  of  PCBs in fish that  generally  agree  with
measured concentrations.

     All of the methods and results are described in more  detail
in the appendices cited previously. These appendices also contain
discussions  of  the specific limitations which pertain  to  each
step  of  the analyses.  All results should be  interpreted  with
caution, and with a full understanding of all of the limitations.


MARINE EFFECTS OF A RELEASE
OF PCBS IN MOBILE BAY

     To analyze the effects of a release in Mobile Bay, we assume
that  the  spill occurs in the ship channel at the point shown in
Exhibit  7-2.   Three release quantities are analyzed  — half   a
tank  (88  metric tons of PCBs),  two tanks  (350 metric  tons  of
PCBs),  and  the entire cargo  (1,400 metric tons of  PCBs).   The
paragraphs  below describe the likely dispersion of the hazardous
constituents and the potential ecological effects.


Pollutant Dispersion

     A spill of PCBs would float or sink depending on its density
relative  to seawater.   For the floating case,  ASA assumed  the
spill  would  behave  much like an oil spill  and  estimates  its
likely  path  using most probable wind directions and speeds  for
the Bay.   Exhibit 7-2 shows the likely paths of spills occurring
at low tide (the likely paths at high tide are shown in  Appendix
I).   The  path  of the spill depends both on wind direction  and
water circulation, which is counter clockwise in the Bay.  Case  A
reflects winds coming out of the north and Case B reflects  winds
coming  out  of the southwest.   Since Case A represents a  worst
case  in  terms of the initial area affected by a spill  (in  the


                              7-9

-------
absence  of  efforts to contain the slick) and in  terms  of  the
Bay's  retention of PCBs,  we restricted further analysis to that
case.

     The  amount  of  PCBs likely to settle  out  into  sediments
before  the slick reaches the shoreline is difficult to  estimate
precisely.   ASA estimates that from 30 to 70 percent of the PCBs
could  settle  before  reaching shore.   We  assume  for  further
analysis  that  50  percent  of  the  PCBs  become  entrained  in
sediments  on the floor of the Bay and the remaining  50  percent
become stranded on the shoreline.

     Exhibit  7-3 shows the area of the Bay's sediments initially
affected by a spill of 8 tanks.   As shown,  about 15 percent  of
the floor of the Bay is affected.   A spill of half a tank  would
affect  about 5 percent and a spill of 2 tanks would affect about
9  percent  of  the floor of  the  Bay.   Under  the  assumptions
outlined above, the initial concentration of PCBs in the top 5 cm
of  sediments averaged over the contaminated zone would be  about
13  ppm,  23 ppm,  and 61 ppm for the half,  2 tank,  and 8  tank
spills, respectively.

     The  long-term dispersion of PCBs in the Bay is difficult to
predict  accurately.   Data on net water circulation in  the  Bay
suggest  that resuspended PCBs would move generally northward and
eastward  on  flood tides,  though some PCBs could move  out  the
mouth of the Bay on ebb tides.   Marine organisms may also play a
role  in  redistributing  PCBs in the  Bay.   PCBs  reaching  the
shoreline  probably  would  gradually  penetrate  into  the  fine
sediments and remain there for a long time.

     PCB  wastes  with a density greater than  seawater  probably
would sink to the bottom of the ship channel, although some could
settle  in shallower waters outside the channel depending on wind
conditions.   The  distance the wastes would be transported  away
from  the spill site would depend on the specific gravity of  the
material,  the extent of adsorption onto particulate matter,  and
current direction and velocity at the time of the  spill.   Under
maximum current conditions, JRB estimated that some PCBs could be
transported  as  much as 7 to 18 kilometers from the spill  site.
However,  most of the waste probably would settle out much closer
to the spill site.

     We  have not attempted to evaluate the long-term  dispersion
of  "sinking" spills nor effects on the marine  environment.   We
expect  that remedial actions would remove a substantial  portion


                              7-10

-------
of the wastes and  routine dredging of  the  ship  channel each  year
by  the Army Corps  of  Engineers would  remove  further amounts from
the  Bay.   However,   determining the  likely  effects of  dredging
activities  on   removal  and  resuspension of PCBs   is  beyond  the
scope of this study.


Marine Effects

     Estimating  the   potential  effects of spills in  a  complex
estuarine  ecosystem   such as  Mobile Bay is   difficult  at  best.
Nonetheless,  to  develop  a  general  indication   of  the  likely
effects  of  PCB  spills we  asked  ASA   to  make a  number  of
simplyifying  assumptions and  then to  apply their  ecosystem model
to  the  Bay.    The modeling assumptions are  fully described  in
Appendix  I,  but   two  of particular  interest  are  (1)  PCBs  in
sediments  are   assumed  to be  fully "available" to benthos   (in
terms  of  mortality   response)  and   (2)  marine   organisms,  on
average,  are  assumed equally likely  to feed in contaminated  or
clean areas.   Since some PCBs are made biologically  unavailable
through  binding with organic carbon  in sediments,  the  modeled
results tend to  overstate biomass and  bioconcentration effects.£/
Assuming  "average  feeding" allows the model to be run using  an
average PCB concentration in sediments across the  entire Bay  and
tends to account for the potential spreading  of PCBs over time.

     Exhibits  7-4  and  7-5  present the model's estimates of  the
effects  of a spill on the biomass of  various trophic levels  and
on concentrations of PCBs in tissues.  The results in Exhibit 7-4
indicate  that   benthic  feeders and predators  would  be  largely
eliminated in the Bay.   While  this exhibit shows results for an  8
tank  spill,  results  are similar for  spills  of one half and  two
tanks.  Other trophic  levels would not be  significantly affected.
However,  the  model   predicts that PCB concentrations in  marine
organisms  would increase by  about 3  to 5 orders of  magnitude
i/Studies  of the effects of PCBs in sediments on marine life   in
the  Puget  Sound and New York Bight areas  indicate less  adverse
effects  than  normally would  be  expected.   These  preliminary
findings  suggest  that  the toxicity of PCBs can be  reduced   by
organic carbon binding,  but existing data  are not sufficient   to
estimate precisely the extent of the reduction.
                              7-11

-------
depending  on  the  size of the spill.   Exhibit  7-5  shows  PCB
concentrations  increasing  in small pelagic fish and  shrimp  by
about  4  to  5  orders of magnitude   (to  tissue  concentrations
slightly in excess of 2 ppm) for a half tank spill.   For  larger
spills,  such  as  an  8 tank spill,   tissue  concentrations  are
estimated  at about 15 ppm.   These effects are expected to  last
for an extended period of time.

     The  above  results  represent an "average"  across  all  of
Mobile Bay.   It is likely that portions of the Bay would be more
or  less  contaminated  than the  average.   Thus,  some  benthic
species  probably  would  survive  in  less  contaminated  areas.
Avoidance  of  contaminated areas by marine  species  also  could
reduce  average  contamination  levels.   However,  we  have  not
attempted  to determine potential adverse effects on spawning and
nursery  areas  or  on species in the  affected  low-lying  marsh
areas.
MARINE EFFECTS OF A RELEASE OF PCBS
OVER THE CONTINENTAL SHELF

     The   continental   shelf  stretches  roughly  100  to   150
kilometers  offshore from the Gulf of Mexico  coastline,  and  is
defined  by water depths less than 100 meters.  These  relatively
shallow waters are rich in marine life and are important for both
ecological  and commercial reasons.  To evaluate the consequences
of  a marine accident and spill in these waters,  we modeled  the
dispersion  and effects of a spill at a point near the  mouth  of
the Mississippi River (29 degrees 45 minutes North and 88 degrees
30  minutes West),  The paragraphs below summarize the results of
the analysis.


Pollutant Dispersion

     The  dispersion  of PCBs was calculated for both a  floating
and  a sinking case.   In the floating case,  we assume that  the
PCBs  would be confined to the top one meter of the water  column
and  would be dispersed by surface currents and winds.  The slick
would spread to cover about two square kilometers within  several
hours  of  the spill,  and then would break up into  patches  and
disperse  over a wider area.   Assuming average wind and current,
some portions of the slick  would come ashore on the  Mississippi
delta  while  the  remainder would move in a  generally  eastward
direction.   Exhibits 7-6 and 7-7 show the position of the  slick


                              7-12

-------
at  10  and  50  days after  the  spill.    (These  exhibits  show
concentrations  calculated  for the release of the entire  cargo;
concentrations  for lesser size spills would  be  proportionately
lower  than  those shown and the area of dispersion of the  slick
would  be slightly smaller.)   After some time much of  the  PCBs
probably  would  sink to the bottom of the continental shelf  due
to weathering,  volatilization of the lighter  hydrocarbons,  and
adsorption onto particulate matter.   The actual behavior of  the
slick . would  be governed by the specific wind and sea states  at
the time of the spill.fi/

     In the sinking case,  we assume that the PCBs would sink  to
the  bottom and lodge in the top 5 centimeters of the  sediments.
Because  of  the relatively shallow depths at the assumed   spill
location,  all  the pollutant would reach the ocean floor  within
two days after the spill event. Thus, concentrations in the water
column  would  be elevated for only a  short  time.  Surface  and
subsurface currents would move and disperse the PCBs as they fall
through the water column to the bottom.  In addition,  because of
the  relatively shallow depth of the continental shelf,  the PCBs
in the sediments would be  continually resuspended by wave action
and  storm  events.  Thus,  PCBs would  continue  to  spread  and
disperse on the bottom for many years.

      Maps provided as Exhibits 7-8 and 7-9 display the estimated
extent  of PCB-contaminated sediments  at 1 and 10 years for an  8
tank  spill.   The area of contamination would be about the  same
for  lesser  size  spills,   but  the  concentrations  would   be
proportionately  less.   Over  time,  sediments  contaminated  to
levels  greater than 1 ppb would include an area on the shelf  of
about 17,500 square kilometers.
fi/An analysis performed by JRB  estimates  the  percentage of spills
occurring on different segments of  the  ship track  that would come
ashore  (anywhere)  in 3,  10 and 30  days.   The analysis indicates
that  a very high  percentage of spills  occurring close to  Mobile
Bay  would come ashore (over 90 percent within  10 days).   This
percentage becomes lower as the spill location moves  further from
Mobile Bay — closer to the Mississippi delta area.   For example,
only  about 15 to  30 percent of spills  off the Mississippi  delta
are estimated to come ashore within 10  days.   However,  about 70
percent  of  these are estimated to come  ashore within  30  days.
(See the JRB study cited previously, pp.  2-7  to 2-10.)


                              7-13

-------
Marine Effects

     We  have not quantified the effects on the marine  ecosystem
that  a floating spill might have.   A temporary effect of a  PCB
slick  would be to interfere with the "microlayer" on the surface
of the Gulf.  Predicted concentrations of PCBs would be lethal to
the  organisms in the microlayer,  but at this time data  do  not
exist  to  quantify  the role of the microlayer  in  the  overall
marine  ecosystem.  As   PCBs leave the slick and settle  on  the
ocean floor,  the ecosystem effects expected would approach those
predicted  for the sinking case.  Indeed,  the area of the  slick
shown  in  Exhibit 7-7 corresponds fairly closely to the area  of
sediments  contaminated by a sinking  spill as shown in  Exhibits
7-8  and  7-9.    If  the  slick comes  into  contact  with   the
shoreline,  and PCBs still remain in it,  significant,  long-term
coastal impacts are likely but have not been estimated.

     Exhibits  7-10  and  7-11  summarize  the  marine  ecosystem
effects  over the continental shelf predicted by the model for  a
one  half tank spill of PCBs that sinks.   As reported above,  an
area of sediment of roughly 17,500 square kilometers is likely to
be contaminated.   The results indicate that there probably would
be a small decrease in the total biomass levels in the area, with
benthic  organisms suffering a substantial  reduction.   However,
the  major  impact  would be an increased level of  PCBs  in  the
tissues of most trophic levels — as shown in Exhibit 7-11,  most
trophic levels experience an increase of from two to three orders
of magnitude.   Large pelagic and demersal fish would  experience
the greater increase, followed by small pelagic fish and demersal
detritivores  (e.g.  shrimp  and  crabs).   These levels  of  PCB
contamination might cause chronic effects in the marine  species.
Spills   of   greater   amounts  are  estimated  to   result   in
concentrations  in tissues  proportionately  higher,  i.e.,  four
times  greater  for a 2 tank spill and 16 times greater for an  8
tank spill.   The effects on the biomass of benthic organisms are
also predicted to be more severe.   Specific modeling results for
these cases are shown in Appendix I.

     There  is  substantial uncertainty concerning the length  of
time  and  concentrations  at  which PCBs  would  remain  in  the
sediments.   The increased PCB tissue concentrations projected in
Exhibit  7-11  to last for only several years are  based  on  the
simplifying assumption made for modeling purposes that PCBs would
be  released from  sediments linearly at a fixed rate over  time.
A   decaying  exponential  release  probably  would  be  a   more


                              7-14

-------
appropriate  functional  form.    Under  such  an  approach,  the
duration of PCBs in sediments and of the bioaccummulation effects
would be much longer than shown in the exhibits.


MARINE EFFECTS OF A RELEASE OF PCBS
IN THE BURN ZONE

     The  center  of the Gulf of Mexico burn zone is  located  in
1,400  meters  of  water at 26 degrees 40 minutes  North  and  93
degrees  40  minutes  West.   At  this  position,   the  zone  is
approximately 315 kilometers south southeast of Galveston, Texas.
To  evaluate  the consequences of a spill in the  burn  zone,  we
modeled  the dispersion and effects of a spill at the  center  of
the  zone.   The  paragraphs below summarize the results  of  the
analysis.


Pollutant Dispersion

     As in the prior cases, the dispersion  of PCBs was calculated
for both a floating and a sinking case.  In the floating case, we
assume  that the PCBs would be confined to  the top one  meter  of
the  water column and would be dispersed by surface currents  and
winds.  The spread of the slick would occur much as described for
the continental shelf spill.  Assuming average wind and currents,
the  slick from a burn zone spill would move generally  westward.
Some  portion  of the slick could  come ashore within 50 days  of
release.  However, most of the PCBs would sink to the ocean floor
by  this  time due to weathering,  volatilization of the  lighter
hydrocarbons,  and adsorption onto particulate matter.   Exhibits
7-12  and 7-13 show the position of the slick at 10 and  50  days
after  the spill.   These exhibits show concentrations calculated
for release of the entire cargo; concentrations for lesser spills
would be proportionately lower and the area of dispersion of  the
slick  would be slightly less.   The actual behavior of the slick
would be governed by the specific wind and  sea states at the time
of the spill.2/
I/The  JRB  analysis  indicates that about  22  percent   of   slicks
originating  in the burn zone would come ashore within   30  days.
(See the JRB study cited previously, p. 2-10.)
                              7-15

-------
     In the sinking case for PCBsf  we again assume that the PCBs
would  sink to the bottom and lodge in the top 5  centimeters  of
the sediments. Because we assume a distribution of sinking times,
the  PCBs would arrive at the bottom from 1 to 25 days after  the
spill  event.  Thus,  concentrations  in  the  water  column  are
elevated  for  a longer time than in the harbor and shelf  cases.
Surface  and  subsurface currents move and disperse the  PCBs  as
they  fall through the water column to the bottom.  Once  on  the
bottom,  resuspension  is negligible due to the greater depths in
the area.

      Maps  provided in Exhibits 7-14 and 7-15 display the extent
of  PCB-containinated sediments at 1 and 25 days after an  8  tank
spill.  At 25 days all PCBs have reached the bottom and sediments
contaminated  to levels greater than 1 ppb would include an  area
of about 57,000 square kilometers. For lesser spills, the area of
contamination  would  be about the same,  but the  concentrations
would be proportionately less.


Marine Effects

     He have not quantified the effects that a floating spill  in
the  burn zone might have on the marine ecosystem.   A  temporary
effect    of  a  PCB  slick  would  be  to  interfere  with   the
"microlayer" on the surface of the Gulf.  The possible effects of
this  interference  would be similar to those discussed  for  the
continental  shelf spill.   As PCBs leave the slick and settle on
the  ocean  floor,  the ecosystem effects  would  approach  those
predicted for the sinking case.

     Exhibits  7-16  and  7-17  summarize  the  marine  ecosystem
effects modeled for a one half tank spill in the burn zone.    As
reported  above  an area of sediments of  roughly  57,000  square
kilometers would be contaminated. The results indicate that there
probably  would  be  a  small decrease in the   biomass  of  some
benthic  communities.   The  effects on benthic  communities  are
estimated  to be more severe for larger spills,  as is  shown  in
Appendix  I.   As shown in Exhibit 7-17,  PCB concentrations  are
projected to increase slightly in only benthic species and bottom
feeders.   Larger  spills  are projected to increase  PCB  tissue
concentrations  in benthic species and bottom feeders by three to
four orders of magnitude,  but there is little effect on  pelagic
fish.    As  mentioned earlier,  there is considerable uncertainty
                              7-16

-------
over the duration of PCBs in sediments and it is likely that  the
effects  modeled for benthic communities would persist for fairly
long time periods.


MARINE EFFECTS OF RELEASES OF EDC

     ASA  also  analyzed the marine effects of  releases  of  the
entire cargo of EDC wastes in Mobile Bay,  the continental shelf,
and the burn zone.  The behavior of EDC in the marine environment
is  very different from that of PCBs — it is highly soluble,  it
is  volatile,   it  is  not  highly  toxic,   and  it  does   not
bioaccumulate.   The  net effect of these characteristics is that
spills  of  EDC  have very little adverse effect  on  the  marine
environment.

     For  a spill of EDC in Mobile Bay,  we assume that  all  the
material  would  go  into  solution and  would  rapidly  disperse
throughout the bay.  The initial concentration would be about 700
ug/1.   However,  flushing  of the bay and  volatilization  would
rapidly  reduce  EDC concentrations,  so that within 50 days  the
concentration  of EDC would be virtually  zero.   The  relatively
brief  elevated water concentrations would cause a brief increase
in  marine  tissue concentrations of EDC.   These  tissues  would
return  quickly (roughly 60 days) to background  levels.   As  no
acute  toxic effects are expected,  the marine effects of the EDC
spill are likely to be transient and relatively minor.

     The effects of spills of EDC over the continental shelf  and
in  the burn zone are similar to the effects in the  bay,  though
the   initial   concentrations  and  their  duration  are   less.
Consequently,  marine effects of an EDC spill in either of  these
locations are likely to be transient and minor.


HUMAN HEALTH EFFECTS OF SPILLS
OF PCB AND EDC WASTE

     To  evaluate short-term human health risks resulting from   a
spill,   we   have  compared  the  possible   concentrations   of
volatilized  PCBs and EDC to the short-term exposure limit  (STEL)
and  the Threshold Limit Value (TLV) for  these  chemicals.   The
STEL represents the maximum concentration to which workers can be
exposed  for short periods without suffering  irritation,  tissue
damage  or  narcosis.   The TLV represents the  concentration  to
which  workers  may be repeatedly exposed without  adverse  acute


                              7-17

-------
effects.   We  have adjusted the TLV values used for EDC and PCBs
downward to reflect continuous exposure rather than exposure  for
a 40-hour work week.

     Atmospheric concentrations of PCBs and EDC are calculated in
Appendix  H.   We  estimated the rates of volatilization of  each
material  in  the  three spill  locations  and  modeled  downwind
ambient   concentrations   using   average   and   "worst   case"
meteorological assumptions.

     Exhibit  7-18  displays the ratios of these  calculated  air
concentrations  to  the STEL and adjusted TLV  values.   As  this
exhibit  shows,  concentrations  of  both  PCBs  and  EDC  exceed
adjusted  TLV when the spill occurs at 1 km  from  Mobile.   Only
under    worst   case   meteorological   conditions   would   the
concentrations  of EDC exceed the STEL.   For the spill at 15  km
from  the  city,   only  EDC  under  worst  case   meteorological
conditions would exceed the TLV.  The ratios of concentrations of
PCBs  or  EDC due to spills over the continental shelf or in  the
burn  zone  are  far below one.   In view of the  fact  that  the
population  would be exposed to volatilization from a  spill  for
period  of  a  week  or  more,  the  adjusted  TLV  is  the  most
appropriate threshold to consider.   Thus, it is likely that only
for  spills  closer  than 15 km would air  concentrations  be  of
concern from a human health standpoint.


SUMMARY

     This  chapter  presents our estimates of the  dispersion  of
pollutants  and resulting marine and human effects from  releases
during ocean transportation. As stated in the Introduction, it is
important to understand that these releases are unlikely to occur
and  that  attempts to characterize the possible effects of  such
releases have many limitations.  Overall,  the results show  that
PCB spills in Mobile Bay would have extremely severe consequences
for the marine ecosystem.   PCB spills over the continental shelf
could  significantly  harm the ecosystem,  whereas spills in  the
burn zone would have lesser effects,  limited to benthic communi-
ties  and to bottom feeders.   In contrast,  EDC spills do  rela-
tively  little damage to the marine ecosystem.  Short-term  human
health  effects  from inhalation of volatilized  wastes  are  un-
likely,  except in the case of an EDC spill very close to Mobile.
All of these results are explained further in Appendices H and I.
                              7-18

-------
                            Exhibit 7-1
     SUMMARY  OF ENERGY  (CAR3O!!)  FLOW IH  THE ECOLOGICAL
           MODEL AS APPLIED TO THE  GULF  OF MEXICO
                               LIGHT
        Herbivorous
        ZoopUnkton
        Carnivorous
        ZoopUnkton
    * Only Mobile Bay
   " Only Mobile Bay and Shelf
Only Shelf and Offshore
Source: Appendix I,  Figure  3-1

-------
                           Exhibit  7-2
               PROJECTED PATH OF SPILLS  OCCURRING
                   IN  MOBILE BAY AT LOW  TIDE
                                                      SCALED
                                                     Kilometers
                                                     r~=3" >—:  r-H
                                                    0246

                                                 DEPTH CONTOURS'
                                                     	2m
                                                     	3m
                                                     	 4m
                                                       • 5 5m

                                                 (Reference Plone-
                                                   Mean Low Water)
                                                                so-
                                                                so'
                      GULF     |    OF    MEXICO
Source:  Appendix I, Fiqure 6-3a

-------
                          Exhibit 7-3
          AREA OF BAY  SEDIMENTS INITIALLY AFFECTED BY
                  A  FLOATING SPILL  (CASE A)
                                                    SCALE:
                                                  Kilometers
                                                  '  -— ' >-< r-l
                                                  0246

                                               DEPTH CONTOURS'
                                                  --- 2m
                                                  ---- 3m
                 MOBILE
                 • 5 5m
            (Reference Plone-
              Mean Low Water)
                                                             30*
                                                             30''
                     GUL F
                              88°
OF   MEXICO
Source:  Appendix I,  Fiqure 6-3d

-------
                                              Exhibit 7-4
                            PROJECTED BIOMASS LEVELS IN MOBILE  BAY AFTER
                                 A  SPILL OF AN ENTIRE CARGO OF  PCBS
  r!.3'  9IOU.OUT U*
           10'
o
o

t/)
       1
       CD
           10' -
           10"

                                                                                             I..i i, n,.
                                                                                                       ~1
                                                                                               XI.I.I. Mix..,
                                                                                             *»•!
                                                                                             I'M
                                                                                                       _ 5

                                                                                                10

              0.
               5.
10.
                                         IS.
                                          20.       25.       30.
                                             TIME  (yrs)
                                                                            35.
                                                     40.
                                                                                              45.
                                                                                                       50.
Source: Appendix I, Figure  6-4a

-------
                                           Exhibit 7-5

                        ESTIMATED PCB CONCENTRATIONS IN TISSUE  AFTER A
                            HALF  TANK SPILL OF PCBS IN MOBILE BAY
[PLOT: BIOU.COHC.U6
1.
1.
1-



lit
Ikl
• * r
1 1
l«

rut*
*«p*'' t
| »«I»MI
•l«|l* fl
•l«flr fl

1
• •*«r »
• • • •

'••

10 ->
105-,
JD :
Q. ,1
Q. 10 I
_ )
~Z. 10 i
o i
ce :
V— . l
^ 10 -i
*-
LJ
i '0'
O
0
10
GO
u
Q. ,0-i
	 	 . 	 	 	 	 	 	 10
i
i
* 9
\\ 	 I
'• i • , ^ „ . , - . _^. 	 , J
i !• i
ji 	 " " ""
1

1> 	 t
. .1,, f 	
[\!
— i
	 _l ^ 	 	 __ ^ — •• ^ ^™ — ^— — — — — 1
1 . _, - — -— , —»^— — ^— —» — • • '• •• — . •
1 	 ' 	 ' ' '. ' !, in 9?;. 30. 35. 40. 45- 50
                      5.
10.
                                        15.
                                                   TIME  (yrs)
           Source:  Appendix I, Figure 6-4d

-------
                           Exhibit  7-6
           CULT OP UCXICO  —  SPILL  ALONG  SHIP  TRACK
                  id OAT fCB CONC(NI»AT IONS  ») UAIC* su»f*cf
                     ASSUUCS MO SCMUNC Of POLLUTANT
ON SHUT
                                 Concentrations  in ppb
Hource:  Appondix  I,  Ficjurc 6-9

-------
                               Exhibit 7-7


             GULF  OF  UEXICO  —  SPILL ALONG SHIP 7R/VCK.  ON SHELf
                   SO 0»V PCB COMCCNTMT10NS AT HATCH
                      ASSUWCS MO SETHINC of POLLUTANT
                           Concentrations  in ppb
Source: Appendix  I,  Figure 6-10

-------
                             Exhibit  7-8


          GULF OF MEXICO  —  SPILL  ALONG  SHIP  TRACK. ON SHELF
                  1 YCA* PCB COMCCWTIUT JONS ON BOTIOW
                  ASSUMING VARUIIC SIHllNC VtLOCIIHS
                              Concentrations  in ppb
Source: Appendix  I, Figure  6-5a

-------
                            Exhibit  7-9
          CULF OF UEXICO  —  SPILL ALONG  SHIP  TRACK. ON SHELF
                  10 YEA* fCt CONCCWT HAT IONS OK BOTTOM
                  ASSUMING VAUIABLC SCTUIMC VtLOCITICS
                         Concentrations  in ppb
Source:  Appendix  I,  Figure 6-6a

-------
FPL or
                                            Exhibit  7-10

                         PROJECTED BIOMASS LEVELS AFTER A HALF TANK SPILL
                                OF PCBS  OVER THE CONTINENTAL SHELF
      siou.our.SB
                                                                                            "1
      CN
      O

      o
      y,
      O
      •— <
      CQ
          10' -,
          10-)
10'
                                                                                              »•!«f It fI ik*
          10"
             0.
                     10.
~\—
 15.
~T	
 20.
                                                          25.       30
                                                    TIME  (yrs)
                                                                  35.
40,
~T	
 45
                                                                                             50
      Source: Appendix I, Figure  6-16d,

-------
                                     Exhibit 7-11


                   ESTIMATED  PCB CONCENTRATIONS IN TISSUE AFTER A
                 HALF TANK  SPILL OF PCBS  OVER THE CONTINENTAL SHELF
PLOT•  B:OU.CONC.sa
         10'
^  10' ^
JD
a

£  ios H
     2  10'H
     h—
     <
         10' -
w    o
O   i 0
z
o
O    -i

CD
O
^   io-J-
                 r\
                 •  \   V
           V« l>-  '*
           •/;/   \'..
           /:.' ?•
                .
                /
                ;t If, \
                'V ''  i
                .f 'J:
                 -._,.
               J	L.
                                                                                         i g.»..t.t '.
                              10
                                  15
20       25

   TIME  (yrs)
                                                                 3C
35
40
—i—

 45
      Source: Appendix I, Figure  6-15c.

-------
                          Exhibit 7-12
           CULT OF  UEXICO  --  INCINERATION SITE SPILL
               10 OH* *CB COMCtWT*»UOWS AT *» U •
                        MO sciniNc or P
                       Concentrations  in ppb
Source:  Appendix  I,  Figure  6-20

-------
                              Exhibit  7-13
               CULF  OF  UEXICO   —  INCINERATION SITE  SPILL
                   SO OAV CCB COWCCWTIUnoNS »1 WMC« SURfACt
                       A54UWCS MO  5CTTL1MC Of POUU1»WT
                        Concentrations  in ppb
Source: Appendix  I, Figure  6-21

-------
                          Exhibit 7-14
            GULF or MEXICO   —   INCINERM ION SITE  SPILL
                 i  DAY PCB coNctwmuoNs ON BOTTOM
                       VAR1ABIC SCTH1MC VC I OC 1 T 1C S
                         Concentrations  in ppb
Source:  Appendix I,  Figure  6-17a

-------
                           Exhibit  7-15
            CULT  OF  MEXICO      INC1NERA 1 I ON SITE  SPILL
                 ?5 O^Y PCB COUC CWlfUIlOMS ON BOMOU
                       VARUBU itMLlMC VfLOClllCS
                           Concentrations in ppb
Source:  Anpondix I,  Fif|ui'c G-19a

-------
                                  Exhibit 7-16
                                                       TAHK  SPIU,
9iou.owr.o8
 CM
 o
 o
10' -j
 o
 S
    )0
                                                                                          .'0
                     10.       IS.
                                          20.       25.       30.
                                             TIME  (yrs)
35.       40.      45.      50
 Source: Appendix I,  Figure 6-26d.

-------
IT.
PLOT•  9IOU.CONC.08
                                         Exhibit 7-17

                       ESTIMATED PCB CONCENTRATIONS IN TISSUE  AFTER A
                          HALF TANK SPILL OF  PCBS IN THE BURN  ZONE
                                                                                           i*»i<»>i••
                                                    ~1
      2
      O
      o
      2
      O
      O

      CD
      O
          101-,
         10' -
          10
           -'
                                                                                  -)—
                                                                                   40.
H	

 45.
                               10.
                                       15.
20.       25.

   TIME  (yrs)
30.
                                                                           35.
50.
       Source: Appendix I, Figure  6-25c.

-------
                                    Exhibit 7-18

               RATIOS OF AIR CONCENTRATIONS TO SHORT-TERM LIMITS
Waste and Atmospheric Conditions
 - Mobile Bay -
 1 Km     15 Km
          Continental    Burn
              Shelf      Zone
STEL

 PCB Waste
     Average
     Worst

 EDC Waste
     Average
     Worst
 0.2
 0.3
 0.4
 3.8
0.01
0.03
0.03
0.33
0.003
0.021
0.0003
0.004
0.0008
0.00009
TLV

 PCB Waste
     Average
     Worst

 EDC Waste
     Average
     Worst
 1.3
 1.9
 1.9
18.3
0.06
0.19
0.12
1.6
0.019
0.13
0.002
0.017
0.005
0.0004
Source: Appendix H, Exhibits H-ll, H-12

-------
EFFECTS OF RELEASES FROM
INCINERATION                                            CHAPTER 8
INTRODUCTION

     The quantity and composition of hazardous materials released
from  ocean- and land-based incineration systems are described in
Chapters  3  through  6.   As  shown  in  those  chapters,  large
quantities of waste are released from incineration operations for
both ocean- and land-based incineration systems. For this reason,
we  investigate and compare possible effects  from these releases.
This chapter describes the results of this work.

     We estimate both human and environmental effects that  could
result  from  incinerator  releases.  For  human  populations  we
estimate  the  incremental  risk  of  developing  cancer  through
inhalation  of atmospheric concentrations of  hazardous  materials
from incinerator releases.  In addition,  for stack releases from
land-based incineration we consider possible  human health effects
resulting  from  ingestion of food contaminated by deposition  of
pollutants on crops and subsequent bioconcentration in  livestock
and  poultry.  In the ocean incinerator case we perform a similar
analysis for ingestion of pollutant-contaminated fish.

     We  also  consider the  possible  environmental  effects  of
incineration  releases,  albeit  at a lower level of  effort  and
resulting  accuracy.  For  ocean-based  stack  releases,  Applied
Science Associates,  Inc.  (ASA) characterized quantitatively the
possible  effects resulting from deposition to the ocean  surface
of  POHCs,  metals and chlorine.  For land-based stack  releases,


                              8-1

-------
Arthur D.   Little,  Inc.  (ADD characterized  the  capture  of metals
and  chlorine   in scrubber waters and  the  possible   environmental
effects  of final disposal of scrubber effluent  and  sludge.   No
work  has   been completed to date on  the  effects  of  land-based
POHCs, PICs or  metals on terrestrial or marine ecosystems.

     All of the transport and effects  estimates  presented in this
chapter  are calculated wherever possible  using  methods  and  data
developed or approved by EPA.   Our use of these methods attempts
to:

     o    unify  existing  scientific knowledge  into   an
          internally consistent framework,

     o    force explicit recognition of assumptions  and  data
          sources,  and

     o    allow investigation of the  effect of  alternative
          data,   methods   or   policies  on  the   results
          predicted.

Ideally, such methods should only be used  when time  and  resources
allow  sensitivity  analyses  of  results  to  be  completed  and
compared  to field verification studies.  Such  studies  have  not
been completed  for  the applications reported here.   Thus,  all  of
the  results obtained are uncertain,   and should  be  viewed   as
general  characterizations of the transport  and  effects  possible
from  incineration  releases.  The sections  below and  supporting
appendices   provide a discussion of the limitations  that pertain
to each step of the analysis.   All results  should be  interpreted
with  caution and with a complete understanding  of all  of  these
limitations.

     The  remaining  sections of this  chapter describe the  human
health   effects   expected  from  ocean-based   and  land-based
incinerator  releases,  and then review the environmental effects
from  each  of these release points.  The final section provides a
brief summary and comparison of the results  obtained for the  two
incineration systems.


HUMAN HEALTH EFFECTS

     Our  estimation of human health effects via inhalation  for
ocean- and  land-based systems includes three steps:
                              8-2

-------
      1.   estimation   of  atmospheric concentrations  due   to
          stack  releases,

      2.   calculation  of  resulting doses received by  humans,
          and

      3.   calculation   of   the  increase in cancer risk   if
          human   populations  received these  dosage   levels
          over a 70 year  lifetime.

Each  of these steps and  associated interim results are presented
in  detail   in   Appendix  D (land-based systems)   and   Appendix  G
(ocean-based  systems).    Although  we  calculate  a   variety   of
concentration  and  health risk  metrics in  the   appendices,  the
discussion   in the sections below is limited to  the  increase   in
cancer risk  for  the most  exposed individual (MEI). Risks  for the
MEI   are the highest estimated for any person and thus provide  an
upper  bound  on the   individual  risk  increases likely  from
inhalation of incinerator  releases.

     All  human   health effects  are estimated assuming that  the
population   is exposed over a 70 year lifetime.   Thus,  in these
calculations  the annual  incinerator release quantities presented
in  Chapters  3  through 6  are assumed to continue for 70  years.
Although  it  is unlikely that  incinerator  operations  and  the
resulting  releases  will  remain constant over such a long  time
horizon,  data and methods to predict human effects from   shorter
exposures   to   chronic  levels   of  these  pollutants are   not
available.
Ocean—Based Incineration

     The atmospheric transport  of  POHCs,  PICs and  metals  released
from  the  stack of the Vulcanus operating in the  center   of   the
Gulf  of  Mexico  burn  zone was   estimated  by  Applied   Science
Associates, Inc. (ASA) using a  numerical  model of  monthly surface
winds in the Gulf. The model incorporates two atmospheric removal
processes — washout due to rainfall  and  dry  deposition.   Washout
is  modeled  as a stochastic process  based on long-term   weather
observations  in the Gulf.   Dry deposition is a function of   the
settling  velocity  of the pollutant.   All pollutants  that   come
into contact with the ocean surface are assumed to be absorbed by
it.   Dispersion modeling results  are based on an  average mixing
height in the western Gulf between 1000  meters  (winter)  and  2000
meters (summer).  Ambient air concentrations  of POHCs,  PICs,  and
                              8-3

-------
 metals  at coastal  locations  are estimated for each month of   the
 year.

     ASA's transport  results  show that  the position of  the plume
 changes  monthly  due to  changes  in the prevailing winds.   Thus,  no
 single  area of  the coast  is  exposed continuously  throughout   the
 entire  year.    Further,   differences in  the  variability of   wind
 direction,  within   the  months alter the concentration levels   ex-
 pected  and  the duration  of  exposure.  To estimate worst   case
 exposure  levels,   ASA  identified the location on  the Gulf coast
 that receives  the highest  modelled concentrations  averaged over a
 year   from  ship releases in  the burn  zone.   This   occurs  at
 approximately  24.5  north latitude and 97.8 west longitude, on the
 coast  near Brownsville,  Texas.   lEc used  ASA's  air   modeling
 results  to determine  health risks to a  hypothetical "most exposed
 individual" at this maximum dose location.

       We  use    the cancer unit risk factors  published   by EPA's
 Carcinogen Assessment Group for the POHCs,  PICs and metals found
 in  the  plume  to calculate the  increase in lifetime cancer   risk
 experienced at  the  MEI annual dosage  level.   These incremental
 lifetime  risks  assume  continuous exposure for 70  years   and   are
 calculated for a 70 kilogram  adult. The results of these calcula-
 tions  are  summarized  in Exhibit 8-1.

     As  shown  in  Exhibit 8-1,  the hypothetical  most  exposed
 individual   incurs  total increased risk of cancer  of 6.4  in   ten
 million  for   the   PCB  waste  and 1.1. in one million for   the   EDC
 waste.   For both wastes,   over  99 percent of  this  increased   risk
 results    from  exposure   to  metals  emitted  from the  stack.
 Incremental risks  for persons living  at other locations on   the
 coast  will  be  even  lower.

     Although  the risks from  metals are dominant in our  analysis,
 note   that  these results are  sensitive  to our   assumptions about
 metals  concentrations.    MEI  risks  would be  lower   for waste
 streams  having significantly  lower concentrations  of carcinogenic
 metals.    In addition,  the results are sensitive  to the settling
 velocity  assumed   for metals.    For  modeling  purposes,  metals
 emitted  from the stack are assumed to rapidly  take on the charac-
 teristics of background metal particles in the  atmosphere.  These
 background  particles are  small,   approximately 1.5  micrometers,
 and  have  slow settling velocities.   If  plume  sampling  indicates
metals emitted from the Vulcanus  have higher  settling  velocities
 than   the   background particles,   MEI risks from metals  could  be
lower.
                              8-4

-------
     In addition to these calculations of  human health risks from
stack emissions, we analyze  the  risks possible through inhalation
of releases from transfer and  storage of wastes at Chickasaw.  We
use  EPA's Atmospheric Transport Model  (ATM)  with  meteorological
data for the Chickasaw area  to estimate air  concentrations out to
a  50 kilometer radius,  assuming that the fugitive emissions are
released from a single point source.   (Information about the  ATM
model  and  our  use  of  it is   included  in Appendix  D.)  The
concentrations  predicted  and the aggregate population  exposure
are presented in Exhibit 8-2.    For comparison, this exhibit also
includes  the  exposure estimates for fugitive emissions  at  the
land-based incinerators considered in this study.

     Given  the concentration  for the maximum exposed  individual
at  Chickasaw  as   shown  in  Exhibit  8-2,   we  calculate  the
incremental  risk  of  cancer  for the MEI  assuming  70  years  of
exposure. These estimates are  shown on the bottom line of Exhibit
8-3.   Exhibit 8-3 also includes the results presented in Exhibit
8-1 for purposes of comparison.   As this exhibit shows,  releases
from Chickasaw result in incremental risks from 2 in  100,000,000
to 5 in 10,000,000,000 for both  waste streams, i.e., at least one
order  of magnitude lower than the risks from stack emissions  of
metals.    Note  that the location of the  most exposed individual
differs  for  ocean-based stack  emissions  and for  releases  from
Chickasaw. No person at a  single  location  would experience both
of these risk levels.

     In  addition  to  direct  inhalation   of stack,  storage  or
transfer  releases,  it is possible that human populations  could
ingest foods contaminated by these releases.   ASA  characterizes
the  effect that deposition  of incinerator stack emissions  could
have  on  pollutant concentrations in the  water column  or  ocean
sediments,  and  the subsequent  bioaccumulation of pollutants  in
fish  and  other  compartments  of  the  marine  ecosystem.   The
analysis focuses on PCBs,  the pollutant of  interest that is most
likely  to bioaccumulate in  marine life,   and examines the effect
of incineration of PCS wastes  at sea over  a  50-year period.   The
results of the analysis suggest  that at the  expected PCS emission
rate there will be no observable change in the concentrations  of
                              8-5

-------
PCBs  in fish,  shrimp or other Gulf marine life.I/  As a result,
seafood consumers would not face an increased cancer risk due  to
deposition of stack emissions.


Land-Based Incineration

     The  atmospheric  transport  of  POHCs,   PICs,  and  metals
released  from land-based incinerator stacks was estimated by lEc
for incinerators located in Arkansas and Texas.  In addition,  we
model  the transport of transfer and storage  emissions  released
from  these sites.   Texas and Arkansas are selected as "typical"
locations  because  incinerators  located  at  these  sites   are
currently  permitted  to  burn  PCBs.   Also,  these  sites  show
variations  in meteorology,  stack parameters and the size of the
surrounding population.

     The  analysis of risks from land-based incineration  closely
parallels the incinerator ship analysis.   The primary difference
between  the ocean- and land-based analyses relates to the volume
of metals emissions.   We assume that the land-based incinerators
are equipped with scrubbers which are capable of removing some of
the  metals from the stack gases.   The removal effectiveness  of
scrubbers  is  a function of the size of metal particles  in  the
stack  gases.   For very volatile metals,  such as  arsenic,  the
removal  efficiency  of  scrubbers is  low.   For  less  volatile
I/To  facilitate  scaling  the results  to  different  levels  of
incinerator performance, the ASA analysis presented in Appendix I
is  based  on a PCB emission rate of one kilogram per  hour.   At
this rate, PCB concentrations in some trophic levels — including
demersal  fish  and  shrimp,   staples  of  the  Gulf  commercial
fisheries  — are predicted to increase by more than an order  of
magnitude  above background concentrations,  posing an  increased
cancer risk to seafood consumers.   However, the one kilogram per
hour  emission rate is more than two orders of magnitude  greater
than  the  expected emission rate — 5.25E-3 kilograms  per  hour
given  a 99.9999 percent DRE.   According to ASA,  the  increased
concentration of PCBs in marine species is approximately linearly
related  to  the  emission  rate.   Therefore,  at  the  expected
emission   rate,   ASA  expects  no  observable  change  in   the
concentration of PCBs in fish, shrimp and other Gulf marine life.
                              8-6

-------
compounds,   removal  efficiency  is  higher.   Analysis  of  trial burn
data  suggests  that  metal  emissions  can  range  from less than  one
percent to as much as 50 percent of  the  metals in the waste feed.
For  the purposes of this  analysis,   we  assume 50 percent of  the
arsenic  (a  relatively  volatile  metal)  in the  waste feed  is
emitted  in  the  stack  gases  and 10 percent   of  the   chromium,
cadmium and  nickel.  The possible  environmental effects resulting
from  the ultimate disposal  of  the scrubber effluent and  sludge
are discussed in a later section of  this chapter.

     Our  analysis   of  pollutant  dispersion from the   Texas  and
Arkansas   sites  uses  the   ATM   atmospheric   dispersion   model
developed for EPA's  Office of Toxic  Substances by researchers  at
Oak   Ridge  National  Laboratory.     ATM  is   a  Gaussian  plume
climatological   model  linked  to  1982  census  data   and   to
meteorology  data from 394 weather stations in the United States.
We  use   the model  to determine atmospheric  concentrations  and
resulting    human    exposures  within 50   kilometers    of   each
incinerator  site.    As in  the analysis of  ocean-based systems, we
assume  both wet and dry deposition.   Wet deposition is based on
average annual  rainfall at each of the sites,   and dry  deposition
is  a function  of the settling  velocity  for gases and   particles.
Our use of the  ATM model is  explained in Appendix 0.

     As  in  the ocean-based case,    we  use the  cancer potency
factors  published   by EPA's Carcinogen  Assessment Group for  the
POHCs, PICs, and metals found in the stack gases to  calculate the
increase in  lifetime cancer  risk experienced by the  most exposed
individual.   These  incremental risks assume continuous exposure
for  70  years  and are calculated  for a  70 kilogram  adult.   The
results  of  these calculations  are summarized  in Exhibit 8-4  and
explained in further detail  in  Appendix  D.

     As shown in Exhibit 8-4,   the most  exposed individual incurs
total  increased  risks of cancer  ranging  from 2 to  3 in 100,000
for  the PCB waste and from  3 to 4 in 100,000  for the EDC  waste.
Risk levels  do  not differ  greatly  for the  two  locations modeled.
At  both  sites,  metals account for  at  least  90 percent of  the
total risk.   Risks  from POHCs,  PICs, and fugitive  emissions are
approximately an order of  magnitude  lower  than risks from metals.
Because  the overall estimate of risk is determined  by  the volume
of  metal  emissions  from  the incinerator,   the   results  are
sensitive to our assumptions about the quantity of metals emitted
in the stack gases.  If the  removal  efficiencies of  scrubbers are
higher than  those assumed  here,  risk from metals is   decreased.
Analysis  of the trial burn  data,  however,  suggests that it  is


                              8-7

-------
unlikely  that removal efficiencies would be higher by more  than
an  order of magnitude.   This,  in turn,  would result in  metal
risks  approximately  ten times lower than those in  Exhibit  8-4
reducing the overall incremental risks of land-based incineration
to roughly one in a million.

     In  addition  to  the  human health  risks  from  inhalation
presented  above,  we  consider  possible human  effects  due  to
ingestion  of foods contaminated by stack releases of  POHCs.  No
PICs,  metals   or  fugitive  releases are  considered  in  these
calculations.  The ingestion risk estimates are adapted from  two
studies  recently  completed  for  EPA  which  consider  possible
ingestion   risks   around  incinerators  located   in   Southern
California,  Wisconsin  and  Missouri.   Our procedures to  adapt
these results for this case study are described in Appendix D.

     The   results  of  our  ingestion  calculations  show   that
ingestion  risks  are generally at least two orders of  magnitude
smaller  than  risks  expected  through  inhalation.  EDC  wastes
generate  approximately 5 times more risk through  the  ingestion
route than the PCS wastes.
Summary of Human Health Effects

     The  sections  above summarize the human health  risks  from
releases   of   hazardous  materials  at  ocean- and   land-based
incinerators.  These estimates are highly uncertain.  The results
of air dispersion models are generally assumed to be accurate  to
plus  or minus 100 percent.  Further,  the GAG cancer  unit  risk
factors  are  generally considered certain only to plus or  minus
one order of magnitude.  As a result,  the health risk  estimates
have  more  meaning  as relative indicators of risks  from  land-
versus  ocean-based  systems than as absolute estimators  of  the
levels of those risks.

     Exhibit  8-5 summarizes the human health risk estimates  for
releases from land- and ocean-based systems.  The figures in this
exhibit  are for inhalation risks to the most exposed  individual
only.  As the exhibit shows,  the incremental risks generated  by
the  inhalation  of  pollutants  are lower  for  the  ocean-based
incinerator  than  they  are for either  our  Arkansas  or  Texas
incinerators.   For  both the ocean and land case the total risks
are dominated by the risk from metals (as mentioned earlier,  our
analysis probably overstates risks from metals).   Although  more
metals are emitted from the stack in the ocean case,  they travel
                              8-8

-------
several  hundred  kilometers  from the burn site  to  Brownsville
before  they are inhaled by the most  exposed  individual.  Thus,
despite the higher emission rates,  the risks to the roost exposed
individual  from  an ocean-based incinerator are below those  for
the land-based case.

     A  recent  study prepared by EPA estimated  risks  to  human
populations  across  the  United States from  inhalation  of  air
toxics. (The jggope and Magnitude of the Air Toxics Problem in the
United States,  E.  Haemisegger,  A.  Jones,  B.  Steigerwald, V.
Thomson,  September 1984) .  In this study the total risk from air
toxics  at various locations throughout the U.S.  is estimated to
range from 4 to 6 in 10,000 for the average individual.  The most
exposed  individual  risks estimated  for  land- and  ocean-based
incinerator  releases are lower than risks to the average  person
from air toxics estimated in this broader study.


ENVIRONMENTAL EFFECTS

Ocean-Based Incineration

     Environmental  effects could occur from deposition of  ocean
incineration  stack  emissions onto the ocean's surface  or  from
other releases which enter the ocean during normal operations. As
explained in Chapters 5 and 6, we assume that stack emissions are
the  only release expected during normal incineration activities.
(The effects of accidental releases due to a vessel casualty  are
discussed in Chapter 7.)  This section of Chapter 8 considers the
possible effects  on the marine environment of POHCs,  metals and
HC1 contained in the stack emissions.  ASA evaluated the possible
effects  of these materials on the marine  environment.     Their
analysis  in  this area is  included as Section 5 of Appendix  I,
and is summarized below.

     The  entry  of  atmospheric pollutants  into  the  ocean  is
mediated   by  the  surface  layer,   and  particularly  by   the
"microlayer."  The  microlayer,  generally defined as the top 100
micrometers at the air-sea interface, has been shown to have very
high  levels of biological activity and pollutant concentrations.
Physical and biological processes in the microlayer are not  well
known, and time and resources have not permitted the inclusion of
the  microlayer  in the marine ecosystems model developed by  ASA
and  described  in  Chapter  7  and  Appendix  I.    Thus,   only
qualitative  and relatively uncertain conclusions can be drawn at
                               8-9

-------
 this time concerning the effects of POHC and metals  deposition on
 the  ocean.  The  effects  of  chlorine   deposition are   better
 understood.

     In  analyzing  possible  marine  effects  from   POHCs,   ASA
 considered  only PCBs.   (Because EDC is highly soluble  in  water
 and  volatilizes rapidly,  the effects of deposition of  EDC  into
 the  ocean is not of concern.)   Analysis using ASA's atmospheric
 and marine transport models indicates that concentrations of  PCBs
 in  the  water  column during incineration would  be about  four
 orders  of  magnitude  below the  concentrations  resulting  from
 background  deposition of PCBs.2/    Once the burn is completed,
 these  concentrations  decline  to an even  smaller   fraction of
 background.   Thus, concentrations of PCBs in the water  column as
 a   result  of  incinerator  plume  deposition  should   have   no
 observable effect on the marine ecosystem.

     A second potential ecosystem effect of PCS deposition  is the
 accumulation  of PCBs in the ocean flow sediments,   resulting in
 increase in PCB concentrations in the benthos,  demersal fish and
 pelagic  fish.   At a PCB emission rate of 5.25 E-3  kilograms per
 hour,  ASA's analysis indicates that PCB deposition  will have no
 significant impact on the marine ecosystem.!/
2/These  results  were  derived  by  scaling  the  concentrations
presented  in  Appendix  I   (based on an  emission   rate  of  one
kilogram  per hour of PCBs) to reflect the assumed   PCB   emission
rate of 5.25E-3 kilograms per hour.

I/Once  again,  these conclusions are based on ASA's analyses   in
Appendix I,  which employ a PCB emission rate of one kilogram per
hour.   At  the one  kilogram per hour emission rate,  deposition
has  a  significant impact on the  ecosystem.   However,  at  the
assumed  emission  rate,  the impact  is  negligible.   (Personal
communication, Dr. Mark Reed, November 1984).
                              8-10

-------
     This  conclusion   is  supported by  analysis  showing that  the
atmospheric flux of  PCBs to  the  Gulf waters  is much  greater  than
that  expected  from  the  Vulcanus.    For   example,   the  annual
deposition  of  PCBs  from stack emissions to  the  area affected  by
the  plume  (about   100,000  square kilometers) would be  about  2
orders of magnitude  less than the  annual atmospheric flux to that
area.   Similarly,   the deposition of  PCBs  from stack emissions
would  be about 3 orders of  magnitude less than   the  atmospheric
flux  of  PCBs  to the  entire Gulf.    (In the mid-seventies,  PCB
concentrations  in the atmosphere  above the  Gulf were higher  by
about an order  of magnitude,   suggesting that background flux, at
one time, was 3 to 4 orders  of magnitude greater than the deposi-
tion of PCBs expected  for  the Vulcanus.)

     In assessing the  possible effects  of deposition of metals on
the ocean, ASA  assumes that  arsenic,  cadmium, chromium and nickel
each are present in  the waste feed at concentrations of 100 parts
per million and are  emitted  from the stack as free metal ions. In
general,  these metals will  be complexed as  oxides or adsorbed to
particulates  as they  fall out from the stack gas plume.  Once on
the  ocean,  remaining free  metals will largely  be adsorbed  onto
organics in the raicrolayer.   Thus,   it  is likely that much of the
metal fallout will be  rendered non-toxic.

     For modeling purposes,   ASA employs a "generic" metal having
the average specific gravity,  particle diameter,  and molecular
weight   of  the  four metals  of   interest.     Based on  these
characteristics,  ASA's atmospheric transport model  predicts that
60 percent of the metals released  from  the Vulcanus1  stack during
a single burn will be  deposited  in  a 7,500 square kilometer  area
of the Gulf before the plume  reaches shore.   ASA then calculates
the  concentration  of  each   metal if  mixed to depths  of  100
micrometers (i.e.,  all metals are  retained  in the microlayer), 1
meter,   and   60  meters.     Cadmium  would exceed  background
concentrations at all  three mixing  depths,   albeit by less than a
factor  of two in the  60-meter case.    None  of the other  metals,
however,  would  exceed background  levels at either  the 60 meters
or 1 meter mixing depths.    If mixed only in the microlayer,  all
metals  would  greatly  exceed background levels,   as would  be
expected.    The  impact  of   these   high concentrations  on  the
microlayer  is  unknown at  this  time,   though   ASA  notes  that
measured  concentrations  of  metals in  the microlayer range  from
one to fifty times surface-water background  levels.    Again, most
metals would not be in a reactive  state when entering the  ocean,
and  most  reactive  metals   that  enter would  be  complexed  and
rendered non-toxic in  passing through the microlayer.


                               8-11

-------
     Because  their  calculations  assume  that  all  metals  are
retained  within the mixing layer for the duration of  an  11-day
burn,  ASA  believes  that the 100 micrometer and 1 meter  mixing
depths are unrealistic, and that the results of the 60-meter case
provide   a   reasonable   worst-case   prediction   of    metals
concentrations.   The  toxic  effects of the slight increases  in
metals  concentrations  shown in this case are dependent  on  the
degree of coroplexation and adsorption the metals undergo,  but in
any case are likely to be small.

     As shown previously,  large amounts of chlorine in the  form
of hydrochloric acid (HC1) will be emitted during incineration of
highly  chlorinated wastes such as PCBs and EDC.  Seawater has  a
large  buffering capacity and can absorb up to about 1 M mole  of
HC1  per liter before the pH changes  significantly.  Given  this
figure  and  assuming daily renewal of the Gulf's surface  layer,
ASA estimates that all of the HC1 from a 6-day burn of EDC wastes
deposited within 14 kilometers of the ship could be mixed in  the
top  one  centimeter  of ocean with no effect on the  water's  pH
level.   Given the normal renewal of the sea surface  layer,  the
turbulence  of  the  upper 2-3 meters of the Gulf  and  the  high
probability  that  the HC1 will be deposited over a  much  larger
area,  mixing would be sufficient to absorb all HC1 released with
no noticeable effect on the marine ecosystem.


Land-Based Incineration

     Hazardous  materials  released from land-based  incinerators
could  damage  both terrestrial and marine  ecosystems.  However,
land-based  units  are  generally equipped  with  scrubbers  that
capture  metals and chlorine before these materials are  released
from  the stack.   While we did not consider the possible effects
of   POHCs,   PICs  or  metals  from  land-based  units  on   the
environment,  we  asked Arthur D.  Little,  Inc.  (ADD to  judge
qualitatively  the  possible environmental  effects  of  scrubber
effluent and sludge.  ADL's analyses in this area are included in
Appendix F. The paragraphs below summarize ADL's findings.

     Scrubber  effluent  or "blowdown" contains small amounts  of
chloride  and is disposed in several ways at existing  commercial
incinerators. At one location the effluent is sold as a feedstock
to a nearby chemical plant; while at other locations the effluent
is discharged via deep wells,  evaporation ponds,  or waste water
treatment  plants.  Direct  or indirect discharges  to  receiving


                              8-12

-------
water  bodies  must  meet  permit  requirements   and   thus  adverse
environmental    effects    are   unlikely  from    these   releases.
Similarly, disposal  to deep  wells is  regulated  and  is  unlikely to
cause adverse environmental  effects.

     Scrubber  sludges   are  generated through the   neutralization
and  precipitation   of scrubber effluent.   The   resulting  solids
include  particulate matter,   metal precipitates and  crystaline
calcium or sodium compounds  (depending on the alkali base used in
the  system).  In  general,  these sludges must be  disposed in  a
surface impoundment  or landfill meeting RCRA permit requirements.
If  these permit requirements  are met,  no adverse   environmental
effects are  expected.


-Summary of Environmental Effects

     The  sections   above  consider   some  of   the   environmental
effects possible from releases of hazardous materials  from  land-
based and ocean-based incinerators. The analysis of these effects
is  qualitative.  For ocean-based systems,  environmental effects
will be determined largely by  the effect of incinerator  releases
on  the microlayer.  The  magnitude of these effects is uncertain
but is likely to be  small.  For land-based systems, the impact of
scrubber  operations on  the  environment is also   judged  to  be
small.   Other   possible  effects  on  the  environment  are  not
considered.
SUMMARY

     This  chapter presents  estimates of the effects of   releases
from ocean- and land-based incinerator operations.   Human,  health
effects are expected  to  be relatively small  for  land- and  ocean-
based systems.  Environmental  effects from ocean-based operations
are   uncertain at present due to lack of scientific data on   the
importance and role of the microlayer but are likely to  be  small.
Risks  from  POHCs  emitted  from  the  ocean-based  system will
typically  be smaller than POHC risks at  land-based  facilities.
Greater  uncertainty  is associated with the relative risks from
PICs.   However,  it  is  unlikely that the risk from PICs emitted
from  an  ocean-based system would exceed the risk  from   PICs   at
land-based facilities.    Risks from metals are somewhat  lower  for
the  ocean-based  incinerator although these results are  sensitive
to  changes  in   the  efficiency of scrubbers  at  the land-based
                               8-13

-------
facilities.   Environmental  effects  from land systems  are  not
considered completely, but the effects of scrubber operations are
expected to be small.
                              8-14

-------
                           Exhibit 8-1

           INCREMENTAL RISK TO MOST EXPOSED INDIVIDUAL
                PROM OCEAN-BASED STACK EMISSIONS
      Constituent                   PCB Waste      EDC Waste
      Stack POHCs                   1.45E-10       5.51E-10
      Stack PICs                    1.68E-12       3.36E-9
      Metals                        6.37E-7        1.06E-6
      TOTAL                         6.37E-7        1.06E-6
Source: Appendix G, Exhibit G-30

-------
                           Exhibit 8-2

      EXPOSURES DUE TO FUGITIVE EMISSIONS AT CHICKASAW AND
              AT THE TEXAS AND ARKANSAS FACILITIES
              Aggregate
              Exposure
              (person-ug/m3)
             Maximum
             Concentration
               (ug/m3)
            MEI
            Concentration
              (ug/m3)
Chickasaw

  PCB Haste
  EDC Waste
3.70E+3
5.90E+3
8.47E-4
1.35E-3
2.20E-5
3.50E-5
Texas Incinerator

  PCB Haste         6.68EO
  EDC Haste         9.48EO
                   7.96E-3
                   5.87E-3
                   1.51E-4
                   9.32E-5
Arkansas Incinerator

  PCB Haste         4.59EO
  EDC HAste         6.60EO
                   1.69E-3
                   1.19E-3
                   1.38E-3
                   9.72E-4
Source: lEc Analysis

-------
                           Exhibit 8-3

           INCREMENTAL RISK TO MOST EXPOSED INDIVIDUAL
                 FROM ALL OCEAN-BASED EMISSIONS
      Constituent                   PCB Waste      EDC Waste
      Stack POHCs                   1.45E-10       5.51E-10
      Stack PICs                    1.68E-12       3.36E-9
      Metals                        6.37E-7        1.06E-6
      TOTAL                         6.37E-7        1.06E-6
      Transfer/Storage

        Chickasaw                   2.02E-8        4.97E-10
Source: Appendix G, Exhibit G-30 and lEc Analysis

-------
                           Exhibit 8-4

           INCREMENTAL RISK TO MOST EXPOSED INDIVIDUAL
                    FROM LAND-BASED EMISSIONS
Constituent
Stack POHCs
Stack PICs
Metals
Fugitives
     Total
	 Texas 	
PCB Waste    EDC Waste
6.10E-8
2.13E-6
2.14E-5
1..39E-7
2.37E-5
           	 Arkansas  	
           PCB  Waste    EDC Waste
1.70E-7
3.07E-8
2.61E-5
2.96E-9
4.16E-8
1.45E-6
2.98E-5
1.27E-6
1.16E-7
2.10E-8
3.63E-5
3.09E-8
2.63E-5
3.25E-5
3.65E-5
 Source:  Appendix  D,  Exhibit D-8

-------
                      Exhibit 8-5

       SUMMARY OF INCREMENTAL HUMAN HEALTH RISKS
              TO MOST EXPOSED INDIVIDUAL
Location                      PCB Waste      EDC Waste
Land-based System

  Texas                         2.37E-5        2.63E-5
  Arkansas                      3.25E-5        3.65E-5
Ocean-based System

  Coastline                     6.37E-7        1.06E-6
  Chickasaw                     2.02E-8        4.97E-10
Source: Exhibits 8-3, 8-4

-------
                      SELECTED BIBLIOGRAPHY
Air Resources Board,  State of California, An Alt Resources Board
     Pol icy  Regarding  Incineration as an Acceptable  Technology
     for PCS Disposalr 1981.

American   Conference  of  Governmental  Industrial   Hygienists,
     "Threshold Limit Values for Chemical Substances in the  Work
     Environment Adopted by ACGIH for 1983-84", revised 1983.

Arthur D. Little, Inc., "Volatilization of EDB for EDB-Containing
     Gasoline Following Small Spills on Human Skin", 1984.

Battelle Pacific Northwest Laboratory,  An Assessment of the Risk
     of Transporting Gasoline by Truck, 1978.

Battelle Pacific Northwest Laboratory,  An Assessment of. £h£ RJLsJs.
     of Transporting Propane by Truck, 1980.
                                                t
Bond, Desmond  H.,  "At-Sea Incineration of Hazardous Wastes," pp.
     148A-152A,  Environmental Science and Technology,  Vol.  18,
     No. 5, 1984.

Charles  T.  Main,  Inc.,  Environmental Risk Assessment for  the
     Incineration oJL Polychlorinated Biphenylsf 1983.

Chemical  Waste Management,  Inc.  and Ocean Combustion  Service,
     Contingency  Plan  for the Ocean Incineration Operations  of
     the Vulcanus 1 and Vulcanus XI, 1983.

Department of Transportation, Logistical Systems to Support Ocean
     Incineration  of Liquid Hazardous Wastesr  Office of  Marine
     Technology. 1982.

Ecker,  W.J.,  Captain of the Port,  Mobile,  Alabama,  Letter to
     Brian   Morrison  regarding  "Hazardous   Material   Storage
     Facilities in the Port of Mobile", September 19, 1984.

EPA, At-Sea  Incineration  of.  Herbicide Orange Onboard  £hja  M/I
     Vulcanus,  Office of Research and Development, EPA 600/2-78-
     086, 1978.

EPA, At-Sea Incineration of PCB-Containing Wastes Onboard the M/T
     Vulcanus. Office of Water, EPA-600/7-83-024, 1983.

-------
EPA, Background  Document fin The Tentative  Determination IQ  Issue
     Incineration-At-Sea Permits. Office  of Water  Regulations  and
     Standards, 1983.

EPA, Environmental Assessment; At-Sea and. Land-Based  Incineration
     QL   Organochlorine   W_afi£fifi,   Office   of    Research    and
     Development, EPA-600/2-78-087, 1978.
EPA,  Environmental     Transport    and    Transformation
     Polychlorinated  Biphenyls,  Office  of Pesticides  and  Toxic
     Substances, EPA 560/5-83-025, 1983.

EPA, Health Asssessment Document for 1 r 2-Dichloroethane (Ethylene
     Dichloride) .    Review   D_r_afJt,   Office   of    Health    and
     Environmental Assessment, EPA-600/8-84-006A, 1984.

EPA, Health  Assesssment Document for  Polychlorinated  Dibenzo-p-
     Dioxinsr  Review  Hr_af_t,  Office  of  Health and  Environmental
     Assessment, EPA-600/8-84-014A, 1984.

EPA,  The  Magnitude and Nature of £hj£ Air Toxics Problem  in   the
     United States. D_££f_fc Report,. Office  of Air and  Radiation and
     Office of Policy, Planning and Evaluation, 1984.
EPA, The  Potential Atmospheric Impact  of  Chemicals   Released
     the Environment.  Office of Toxic  Substances,   EPA  560/5-80-
     001, 1981.

EPA, Workbook of Atmospheric Dispersion Estimates,   Office  of  Air
     Programs, revised 1970.

EPA Region  V,   "Approval  to  Incinerate  PCBs:    SCA   Chemical
     Services", no date.

EPA Region  VI,  Energy Systems Company,   "Incineration  of   PCBs:
     Summary of Approval Actions", 1981.

EPA  Region VI,  Rollins Environmental  Services,  "Incineration of
     PCBs:  Summary of Approval Actions",  1981.

Franklin Associates,  Ltd.,  "Composition  and  Management of Used
     Oil Generated in the United States",  1983.

Giam, Dr.  C.S.,  Presentation  to EPA's Hearing  in   Brownsville,
     Texas  regarding  permit  to  burn PCBs  by  M/T   Vulcanus,
     November 21, 1983.

-------
GKSS-Forschungszentrum   Geesthacht   GmbH,   Hydrogen   Chloride
     Distribution  in  Plumes of Incinerator  Ships;  Preliminary
     Results of a Campaign with R/V Tabasis in Summer 1982, 1984.

Hagopian,  J.  (ADD,  Memorandum to J. Ehrenfeld and L. Bendixen
     entitled   "Fugitive  Emissions  from  Incinerator   Ships",
     September 19, 1984.

Harless,  Robert  (EPA),  Memorandum to Merrill Jackson  entitled
     "Summary Report - Vulcanus Project", January 20, 1983.

JRB Associates, Expanded Modeling of Jncineration at Sea Impacts;
     Gulf of Mexicof  Draft Report, 1984.

JRB Associates,   Permissible  Metal,   PCS,  and' TCDD  (Dioxin)
     Concentrations in Incineration Waste Material, 1983.

JRB Associates,  Worst Case Scenarios for Spillage of PCBs at Sea
     and Tranport of Stack Gas Over Land, 1983.

Kolb, Jeff,  Memorandum  to Michael Huguenin entitled "Deposition
     of PCBs from Ambient Air into the Gulf of  Mexico",  October
     10, 1984.

Rolb, Jeff, Memorandum to Mike Huguenin entitled  "Distribution of
     Release Fractions for Truck Spills", September 24, 1984.

Kolb, Jeff,  Memorandum  to Mike Huguenin entitled "Estimation of
     Vessel Accident Rates in Mobile Bay", September 24, 1984.

Kolb, Jeff,  Memorandum to Mike Huguenin entitled "Information on
     Vessel Casualties", September 12, 1984.

Kolb, Jeff,  Memorandum  to  Mike  Huguenin  entitled   "Operating
     Record of Incineration Ships in Europe", September 11, 1984.

Kolb, Jeff, Memorandum to Mike Huguenin entitled  "Reference Risks
     — Continuous Emissions from Ocean Incineration",  September
     11, 1984.

Kolb, Jeff, Memorandum to Mike Huguenin entitled  "Reference Risks
     - Spills", September 11, 1984.

-------
Kolb, Jeff,  Memorandum to Mike Huguenin entitled "Transportation
     Data  from  Companies  in  the  Hazardous  Waste  Business",
     September 24, 1984.

Life Systems,  Inc.,  A  Comparison  af. PCS  Exposures  £0.  Human
     Populations from Oceanic and Terrestrial Incineration,, 1984.

Lusk, Admiral  Clyde T.,  Jr.,  Chief,  Office of Merchant Marine
     Safety, Statement before the House Subcommittee on Fisheries
     and Wildlife Conservation and the Environment,  Subcommittee
     on Oceanography, December 7, 1983.

Midwest  Research  Institute,   Inhalation  Exposure  Methodology
     Modeling of Hazardous Waste Incinerators, 1983.

Midwest Research Institute,  Performance Evaluation of Full-Scale
     Hazardous Waste Incinerators, 1984.

Mitre Corporation,   Composition   of  Hazardous  Haste   Streams
     Currently Incinerated. 1983.

Mitre Corporation,    Hazardous    Haste   Stream   Trace   H&tal
     Concentrations and Emissions,. 1983.

National  Oceanic and Atmospheric  Administration,  Environmental
     Guide for the. ilS Gulf Coast. 1972.

Netherlands   Organization   of  Applied   Scientific   Research,
     Monitoring of Combustion Efficiency.  Destruction Efficiency
     and Safety During the Test Incineration af PCB  Waste.  Part
     Is Combustion and Destruction. 1982.

Netherlands   Organization   of  Applied   Scientific   Research,
     Monitoring af. Combustion Efficiency During tllfi Certification
     Voyage af. the Jncineration Vessel Vulcanus IX, 1983.

Oak Ridge  National Laboratory,  The  Atmospheric Transport  Modfil
     fat Toxic Substances  (ATM-TQX).  1983.

Oak Ridge National Laboratory, inhalation  Pathway filsJt Assessment
     of Hazardous Waste Incineration  Facilities. 1984.

Oak Ridge National Laboratory,  Multiple-Pathways  Screening-Level
     Assessment af a Hazardous Waste  Incineration  Facility. 1984.

-------
Poten & Partners,   Inc. ,   Liquified Gas  Ship Safety ;   Analysis qf
                               , 1982.
TRW, Inc. ,  At- Sea   Incineration of, Organochlorine Wastes Onboard
     the  M/T Vulcanus,.  1977.

TRW, Inc. , History  of  Env i ronment al Testing  of  the Chemical Waste
     Incinerator  Ships M/T Vulcanus and  I/V  Vulcanus IIf 1983.

TRW, Inc. ,  Incineration of Volatile Organic Compounds on the I/V
     Vulcanus  II. 1983.

Wilson,   R. ,   "Analyzing  the  Daily Risks of  Life,"  Technology
     Review, pp.  41-46, 1979.
                                          •v.s. Gonmamrr PRINTING omen  1985-526-781/30377

-------