United States
                   Environmental Protection
                   Agency
Hazardous Waste Engineering
Research Laboratory
Cincinnati OH 45268
                   Research and Development
EPA/600/S2-87/004 Apr. 1987
v>EPA         Project  Summary
                   Characterization of  PCB
                   Transformer/Capacitor Fluids
                   and  Correlation  with  PCDDs
                   and  PCDFs  in  Soot
                   Beverly Campbell and Anthony Lee
                     Dielectric fluids in transformers and
                   capacitors often contain polychlori-
                   nated biphenyls (PCBs) or chloroben-
                   zenes. These substances may generate
                   polychlorinated dibenzofurans (PCDFs)
                   and polychlorinated dibenzo-p-dioxins
                   (PCDDs) under certain conditions of
                   combustion/pyrolysis. When electrical
                   equipment containing these fluids is in-
                   volved in an accidental fire, the result-
                   ing smoke, soot, and residues may be
                   contaminated with PCDDs, PCDFs, and
                   other chlorinated hydrocarbons.
                     The full report contains a review of
                   several laboratory studies investigating
                   the sources of PCDDs and PCDFs as
                   well as the conditions under which they
                   are formed. In addition, some data from
                   sites of actual fire incidents are avail-
                   able and are discussed. Chloroben-
                   zenes and PCBs do not form PCDDs and
                   PCDFs when heated in the absence of
                   oxygen. During fires, the dielectric fluid
                   of transformers or capacitors may be
                   leaked or vented from ruptured cas-
                   ings. With exposure to oxygen, PCBs
                   can produce PCDFs and chloro-
                   benzenes can produce PCDDs. The par-
                   ticular isomers of  PCDDs and PCDFs
                   formed are related to the number of
                   chlorine substituents in  the reacting
                   material.
                     This Project Summary was devel-
                   oped by EPA's Hazardous Waste Engi-
                   neering Research Laboratory, Cincin-
                   nati, OH, to announce key findings of
                   the research project that  is fully docu-
                   mented in a separate report of the same
                   title (see Project Report ordering infor-
                   mation at back).
Introduction

  In August 1982, the U.S. Environmen-
tal Protection Agency (EPA) decided to
permit the continued use of electrical
transformers containing polychlori-
nated biphenyls (PCBs) based on the re-
ported low frequency of leaks and spills
of PCBs from this equipment relative to
the high costs of replacing or securing
these  transformers. Under Section
6(e)(2)(B) of the Toxic Substances Con-
trol Act (TSCA), EPA can authorize a use
of PCBs provided that the use "will not
present an unreasonable risk of injury to
health or the environment." EPA deter-
mined that the continued use of PCBs-
contaminated transformers (50-500
ppm PCBs) and non-PCB transformers
(<50 ppm PCBs) did not present unrea-
sonable risks to public health.
  A closer evaluation of the fire-related
risks posed by the continued use of PCB
transformers, and the costs and bene-
fits of actions designed to reduce those
risks followed the 1982 determination.
EPA issued a Proposed Rule on October
11,1984, concerning PCB transformers.
EPA determined that fires involving
transformers containing >500 ppm
PCBs present risks to human health and
the environment. The extreme toxicity
of materials which can be formed dur-
ing  fires  involving PCB  transformers,
and the potential for human and envi-
ronmental exposures to these com-
pounds, contributed to EPA's proposed
rule.
  Considering the extensive comments
received during the public comment pe-

-------
riod for the Proposed Rule, EPA modi-
fied the Final Rule concerning:
   Evaluation of the use of PCB trans-
    formers in or near industrial build-
    ings separately from the use of PCB
    transformers in or near commercial
    buildings.
   Relative probabilities of failures
    and fires in different types of PCB
    transformers installations, placing
    more stringent controls  on those
    transformers that EPA  believes
    pose higher risks of failures and
    fires.
   Increased emphasis on the preven-
    tion  of PCB transformer fires
    through increased electrical protec-
    tion,  and decreased emphasis on
    the use of isolation measures to
    minimize  the spread of already
    formed and/or released  contami-
    nants.

  On July 9,1985, EPA promulgated its
Final Rule on PCBs in electrical trans-
formers, the culmination of a  long fact-
finding and rule-making process that
began shortly after the transformer fire
at the State Office Building in Bingham-
ton, NY, February 5,  1981.

Summary and Conclusions
  An estimated 74,000 tons of PCBs are
still in use in U.S. transformers and ca-
pacitors. On July 9, 1985, the EPA pro-
mulgated its Final Rule on PCBs in elec-
trical transformers. This rule specifies
that PCBs at any concentration may be
used in transformers (other than in rail-
road locomotives and self-propelled
railroad cars) subject to the  following
conditions:

  1} the use of higher secondary volt-
    age  (>480 volts) network PCB
    transformers in or near  commer-
    cial  buildings  after October 1,
    1990, is prohibited.
  2) the installation of enhanced elec-
    trical protection on lower second-
    ary voltage network PCB trans-
    formers  and higher secondary
    voltage radial PCB transformers in
    use in or  near commercial build-
    ings is required by October 1,
    1990.
  3) further installation of PCB trans-
    formers  in or near  commercial
    buildings is prohibited after Octo-
    ber 1, 1985.
  4) the registration of all PCB trans-
    formers with fire-response person-
    nel and  building  owners is  re-
    quired by December 1, 1985.
  5) the markings of the exterior of all
    PCB transformer locations is re-
    quired by December 1, 1985.
  6) the removal of  stored combusti-
     bles located near PCB transform-
     ers is required  by December 1,
     1985.
  There are still  many uncertainties of
the scope and nature of the hazards
created in PCB transformer and capaci-
tor fires. One potential hazard is the
generation  of highly  toxic substances
such as PCDDs  and  PCDFs  from the
pyrolysis of PCBs and chlorobenzenes.
This report identifies 30 fire  incidents
involving PCB transformers and capaci-
tors in the  United States  and western
Europe that occurred from September
1978 through February 1985. The fol-
lowing questions are  addressed in this
study:
   Are PCDDs and PCDFs formed in
    PCB transformers under normal op-
    erating conditions?
   How are the constituents of the
    transformer  fluids related to the
    type and amount of PCDDs and
    PCDFs formed?
   What are the temperature and
    other reaction conditions that favor
    the formation of the PCB combus-
    tion products?
  The July  9, 1985 rule provides for a
gradual phaseout of some PCB trans-
formers while recognizing  the potential
for additional fire accidents in the in-
terim. Moreover,  PCB capacitors are not
covered under the July 9,1985 rule and,
since they have also  been involved in
fire incidents, capacitors are  potential
release sources for PCBs and combus-
tion by-products  into the environment.
  The full report presents and evaluates
the available  literature and published
data on analyses of transformer fluids
and soot generated in  fires. Even
though there  have been at least 30 re-
ported incidents, a wide variety of prob-
lems that hinder the analysis and evalu-
ation of the data  remain, including:
   Limitation of Analytical Data. Very
    few analytical data have been gen-
    erated for each PCB fire  incident.
    Because of the high cost of isomer
    analysis and the  large number of
    isomers that  characterize the PCBs,
    PCDFs, and  PCDDs, few  analyses
    are actually performed  for any
    specific isomer in the aftermath of a
    PCB fire incident.
   Differences in Sampling  and Ana-
    lytical  Protocols. Sampling and
    analysis  protocols for contami-
    nants generated  in transformer
    fires are not yet fully standardized,
    and thus a wide variety of sampling
    and analytical  methodologies are
    often employed. Some data are
    based on the analysis of soot and
    are reported on a weight/weight
    basis. Other data are based on the
    analysis of wipe samples  and re-
    ported on a weight/area basis.
    Thus, it is very difficult to compare
    one fire incident with another, or to
    evaluate the significance of the
    data in one incident relative to that
    of another.

   Lack of Background Data. Very few
    background data are available on:
    (a) composition of transformer
    fluids and, (b) composition and
    levels of PCBs, PCDFs, chloroben-
    zenes, and PCDDs in the  environ-
    ment.
  Despite these problems, the following
conclusions may be made from analysis
of the data from the literature on PCB
transformer fires:
  1)  PCDFs and PCDDs are not formed
     in transformers containing PCBs
     under normal operating condi-
     tions.  Their formation requires
     thermally stressful conditions and
     the presence of oxygen.
  2)  Electrical arcings  in transformers
     do not lead to the formation of
     PCDFs and PCDDs.
  3)  A temperature  zone between
     600C and 680C may be regarded
     as optimal for the formation of
     PCDFs.
  4)  The amount and the specific PCDF
     isomers formed are related to the
     concentration  of and type of PCB
     homologs in the transformer fluid.
  5)  Chlorobenzene diluents in the
     transformer fluids are required for
     the formation  of PCDDs.
  The Binghamton, NY transformer fire
accident was the first to capture major
media  and scientific attention. In that
fire, both PCDDs and PCDFs were found
in the  generated soot, leading to the
concern that these compounds  were be-
ing formed in situ in PCB transformers
and capacitors under normal operating
conditions. The available evidence does
not support this concern. Analyses con-
ducted by the Electric  Power Research
Institute (EPRI) and EPA of samples of
dielectric fluids taken from in-service
transformers and capacitors and those
involved in fire accidents showed no ap-
preciable difference in PCDF values
from stock material. No PCDDs  were de-

-------
tected. Under normal use conditions, it
does not appear that PCDFs are gener-
ated to any significant extent in the
transformer.
  Correspondingly,  a second concern
was whether  PCDDs and PCDFs are
formed during electrical discharges as-
sociated with transformer performance.
To  investigate this issue, experiments
involving the arcing of electrical energy
through various transformer fluids were
performed. There was no appreciable
difference in PCDF  levels before and
after electrical arcing. The supposition
is that the level of oxygen is very low in
the transformer and, thus,  does not
offer the environment for combustion
resulting in the formation of PCDFs and
PCDDs.
  The  amount and  the specific PCDD
and PCDF  isomers formed  in a PCB
transformer fire appear to be related to
the concentration of and type of PCB
homologs  in  the transformer fluids.
This supposition is supported by recent
estimations on the  boiling points for
dioxins and furans that indicate that the
boiling temperatures for tri-CDDs and
tri-CDFs and  the higher chlorinated
PCDDs and PCDFs range from 375C to
537C. A first approximation of the ther-
modynamic conditions would favor the
formation of tri- and higher chlorinated
PCDFs and PCDDs. Combustion studies
conducted by  EPA and EPRI also pro-
vide evidence that the concentration of
and type of PCB homologs in the trans-
former fluids are probably related to the
amount  and the specific PCDD and
PCDF isomers formed in a PCB trans-
former fire. The EPA study  indicated
that the optimal conditions for PCDF for-
mation from PCBs are a temperature
near 675C and a residence time of 0.8
second or longer.  The EPRI study
demonstrated that tetra- and penta-CDF
yields are roughly proportional to PCB
concentrations in the starting material,
but it indicated that significant dibenzo-
furan destruction begins to occur at ap-
proximately 550C.
  A final  issue pertains to the question
of the  use  of  diluents in transformer
fluids. Of the 30 reported fire incidents
involving PCB transformers and capaci-
tors, unequivocal evidence of PCDDs
formation was found  only in the
Binghamton, NY fire. The Binghamton,
NY transformer contained chloroben-
zenes as a  diluent, adding to the evi-
dence that the pyrolysis of chloroben-
zenes leads to  the formation of PCDDs.
Such evidence has also been found in
chemical manufacturing processes and
in metal  recovery sites involving PCB
transformers. Chlorobenzenes should
be carefully evaluated for use as trans-
former fluids or diluents.

Fire Incidents Involving  PCB
Transformers and Capacitors
  PCBs have been used extensively as
dielectric fluid in capacitors and trans-
formers since the 1950s. An  estimated
74,000 tons of PCBs are still used in U.S.
transformers and capacitors. At least 30
fire incidents involving PCB transform-
ers and capacitors have occurred in the
past 7 years as identified in Table 1.
  There are no Federal guidelines to de-
fine acceptable cleanup levels for toxic
releases from PCB transformer and ca-
pacitor fires. Current regulations state,
however, that all spills and leaks of
PCBs or dioxins-contaminated material
should be cleaned  up to preexisting
background levels whenever  there is a
threat of  contamination to water, food,
feed, or humans. NIOSH  has detected
background levels in urban areas of up
to 0.5  meg PCBs/100 cm2 of  surface
area. Following the occurrence of PCBs-
related fire incidents, several states and
other countries have established con-
tamination cleanup criteria. These crite-
ria are presented in Table 2.

Correlation of Combustion/
Pyrolysis Products Generated
and Constituents of
Transformer Fluids
  The ability to predict the type and the
quantity of toxic contaminants that may
form in a PCB transformer/capacitor fire
is of prime importance in the develop-
ment of prevention and control  meas-
ures. Because of the scarcity and gener-
ally poor quality of data obtained from
PCB transformer fire incidents, pyrolytic
studies under laboratory-controlled
conditions have been employed.
  EPA, through a contract with Midwest
Research  Institute, Kansas City, MO,
conducted a study to evaluate thermal
degradation products using  a bench-
scale thermal destruction system. The
results indicated that  both temperature
and oxygen significantly affected PCDF
yield. Statistical analysis showed a lin-
ear relationship for PCDFs formed ver-
sus the concentration of PCBs.
 Table 1.   Fire Incidents Involving PCB Transformers or Capacitors Since 1978

                  Location                                     Date
         Norrtalje, Sweden
         Cincinnati, Ohio
         Binghamton, New York
         Stockholm, Sweden
         Danviken, Sweden
         Boston, Massachusetts
         Skovde, Sweden
         Miami, Florida
         Arvika, Sweden
         St. Paul, Minnesota
         Imatra, Finland
         Helsinki, Finland
         Surahammar, Sweden
         Hallstahammar, Sweden
         Railway Locomotive, Sweden
         Kaukopaa, Sweden
         Kisa, Sweden
         San Francisco, California
         Halmstad, Sweden
         Chicago, Illinois
         Bofors, Sweden
         Columbus, Ohio
         Sodertalje, Sweden
         Finspang, Sweden
         Motors, Sweden
         Vetlanda, Sweden
         Reims, France
         Oslo Lysverker, Norway
         Sandnes, Norway
         Raufoss, Norway
           September 25, 1978
           Decembers, 1980
           February 5, 1981
           August 25, 1981
           1981
           January 1982
           March 19, 1982
           April 13, 1982
           May 1982
           June 22, 1982
           Augusts, 1982
           August 1982
           September 23, 1982
           Novembers, 1982
           Winter 1982/83
           1982
           April 25, 1983
           May 15, 1983
           August 15, 1983
           September 28, 1983
           December 21, 1983
           March 1984
           April 27, 1984
           May 24, 1984
           September 13, 1984
           October 10, 1984
           January 14, 1985
           January 1985
           February 1985
           February 1985

-------
Table 2.    Contamination Cleanup Criteria

       Location                 Contaminant
                                 Air
                                                                   Surface
Binghamton, NY Building

  Inside Vault


San.Francisco, CA

  Inside Vault


Sante Fe, NM

Finland

Sweden
PCDDs/PCDFs
PCBs
PCDDs/PCDFs
PCBs

PCDDs/PCDFs
PCBs
PCDDs/PCDFs
PCBs

PCDDs/PCDFs

PCDDs/PCDFs

PCDDs/PCDFs
10 pg/m3 of 2,3,7,8-TCDD/TCDF
200 ng/m3 of PCBs
80 pg/m3 of 2,3,7,8-TCDD/TCDF
1 mcg/m3 of PCBs

10 pg/m3 of 2,3,7,8-TCDD/TCDF
200 ng/m3 of PCBs
80 pg/m3 of 2,3,7,8-TCDD/TCDF
1 mcg/m3 of PCBs
3 pg/m2 of 2,3,7,8-TCDD/TCDF
60 mcg/m2 of PCBs
24 ng/m2 of 2,3,7,8-TCDD/TCDF
1 mg/m2 of PCBs

3 pg/m23
60 mcg/m2 of PCBs
24 ng/m2 of 2,3,7,8-TCDD/TCDF
1 mg/m2 of PCBs

1 ng/m2 of 2,3,7,8-TCDD'TCDF

5 ng/m2 of 2,3,7,8-TCDD/TCDF

50 ng/m2 of total TCDF
aSum of all PCDD/PCDF isomers CI4 - CI7 with Cl substitution in the 2,3,7, and 8 positions.
  A study showed that askarel fluid mix-
tures of 60% PCBs and 40% trichloro-
benzenes were combusted under vary-
ing flame temperatures. The results of
this study  indicated that the optimal
temperature for the formation of PCDFs
and PCDDs is approximately 600C. This
finding is in fair agreement with  the
work done  at the Midwest Research In-
stitute, where the optimal temperature
for PCDF formation from pyrolysis of
PCBs was approximately 675C.
  Under optimal  conditions, PCDFs  are
formed from mineral oil or  silicone oil
contaminated with  PCBs at >5 ppm.
PCDFs were also formed from a trichlo-
robenzene dielectric fluid that contained
no detectable PCBs. These results sup-
ported earlier laboratory work and ana-
lytical  results of soot material from
transformer and  capacitor fires, which
determined that chlorobenzenes are re-
quired for PCDD  formation.
  EPRI has also supported a  major
study of the thermal conversion of vari-
ous transformer fluid formulations to
PCDFs and PCDDs.  Fluids that have
been studied include mineral oil, tetra-
chloroethylene (TCE), and silicone  oil,
all spiked with Aroclor 1254. One hun-
dred mcl samples were either pyrolyzed
(heated in an oxygen-deficient environ-
ment) or combusted (injected  into a
flame or heated under conditions result-
ing in self-ignition). Pyrolyses were con-
ducted using a simple thermostatically
controlled apparatus, capable of accom-
modating  glass  or quartz tubes of di-
ameters up to 6 cm within its 9-cm-long
heated region. To simulate  more accu-
rately certain catastrophic incidents, py-
        rolyses were conducted at atmospheric
        pressure.
          Results of the EPRI study support the
        proposition that tetra- and penta-CDF
        yields are roughly proportional to PCB
        concentrations in the starting material.
        An interesting feature of the mineral oil/
        Aroclor 1254 data is the clear and repro-
        ducible differences between  the pat-
        terns of tetra-CDFs and  penta-CDFs
        formed by pyrolysis  of neat Aroclor
        1254 versus those formed  by pyrolysis
        of the 5,000-ppm mixture. For example,
        2,3,7,8-TCDF  and  co-eluters comprise
        16-21% of the tetra-CDF mixture formed
        from neat Aroclor, but they comprise
        45-55% of the mixture from mineral oil/
        Aroclor. Furthermore, combustion of
        biphenyl in TCE produced decreasing
        net dibenzofuran as the residence time
        was varied from 18 seconds to 6 sec-
        onds and  the wa)l temperature main-
        tained  at 450C. In contrast, at 550C,
        there is a net increase in dibenzofuran
        yield as residence time decreases. This
        suggests that at the higher temperature,
        significant dibenzofuran destruction  is
        occurring. There are also large effects
        on dibenzofuran yield in the presence of
        different solvents. Typically, yields ob-
        tained with combustion  in TCE are
        much higher than those obtained in sili-
        cone or mineral oil. Some sharp differ-
        ences on the effects of particular vari-
        able parameters on dibenzofuran yields
        have also been noted.
          Both EPA's and EPRI's pyrolytic and
        combustion studies support the propo-
        sition that the amount and the specific
        PCDF isomers are related to the concen-
        tration of and the PCB  homologs in the
                          transformer fluid. Additional supporting
                          data  include EPA's data on the trans-
                          former oil and the generated soot in the
                          Binghamton, NY fire incident. The data
                          on the transformer oil and soot from the
                          Binghamton fire  are presented in
                          Table 3.
                            Penta-CDF concentration was approx-
                          imately 7% of the total PCDFs in oil.
                          Penta-CDF concentration increased to
                          31% in the soot after the fire. This find-
                          ing corresponds to the 52% of pen-
                          tachlorobiphenyls  in the transformer
                          oil. Similar observations can be made
                          for the hexa- and hepta-CDF concentra-
                          tions in soots relative to the concentra-
                          tions of hexa-  and hepta-chlorinated
                          biphenyls.  The Binghamton,  NY fire
                          data  also  appear to  indicate that the
                          higher chlorinated  biphenyls are more
                          likely to convert to chlorinated dibenzo-
                          furans and the lower chlorinated PCBs
                          are more likely to decompose in a trans-
                          former fire.
                            The finding of PCDDs in the Bingham-
                          ton,  NY fire incident raised concerns
                          that PCDDs may be generated in  PCB
                          transformers. Analyses of  subsequent
                          PCB  transformer fire  incidents, how-
                          ever, identified only one other incident
                          where PCDDs  have been identified. It
                          appears that the presence of chloroben-
                          zene  is a requirement for the formation
                          of PCDDs,  and fires involving trans-
                          formers without chlorobenzenes do not
                          generate PCDDs.

-------
Table 3.    Correlation of Analytical Data on Transformer Oil and the Generated Soot from the
          Binghamton, NY Incident

                                  PCBs                             PCDFs
                            Isomer/Total PCBs                 Isomer/Total PCDFs
Tetrachloro-                        0.15                              0.013
Pentachloro-                        0.52                              0.31
Hexachloro-                        0.28                              0.45
Heptachloro-                        0.04                              0.21
Octachloro-                        0.001                             0.02
   BeverlyCampbellandAnthonyl.ee are with Technical Resources. Inc.. Rockville.
     MD 20852..
   Brian A. West fall is the EPA Project Officer (see below).
   The complete report,  entitled "Characterization of PCB Transformer/Capacitor
     Fluids and Correlation with PCDDs and PCDFs in Soot," (Order No. PB 87-
     145 785/AS; Cost: $18.95, subject to change) will be available only  from:
           National Technical Information Service
           5285 Port Royal Road
           Springfield, VA 22161
           Telephone: 703-487-4650
   The EPA Project Officer can be contacted at:
           Hazardous Waste Engineering Research Laboratory
           U.S. Environmental Protection Agency
           Cincinnati, OH 45268

-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
     BULK RATE
POSTAGE & FEES PA
        EPA
   PERMIT No  G-35
Official Business
Penalty for Private Use $300

 EPA/600/S2-87/004
                 0000329   PS

-------