906 PCBs AND ASSOCIATED AROMATICS
For example, if the unit should require servicing at some
time in the future it will have, instead, to be scrapped. If
tests showed that the transformer was about to fail it would
not be possible to retrofill it and service the unit as a PCB
contaminated or PCB free unit until it had been set back in
operation for 90 days after the retrofill. This may be nei-
ther possible nor feasible, depending upon the condition
of the transformer. If an owner makes the decision not to
continue to operate a unit as a PCB transformer he has then
to choose between replacement with a new unit (retrofit) or
retrofill of the old one. The age and performance history of
the transformer become factors as well as how long the unit
is likely to be needed.
TRANSFORMER RETROFILL OPERATIONS
The retrofill operation, as it is conventionally practiced today,
consists of multiple stages, each of which involves draining,
flushing and filling. The process can apply to two different
situations, viz. (1) contaminated mineral oil transformers
with 500 ppm PCB and (2) askarel filled transformers in
which the coils are impregnated with PCB.
(1) In the case of a PCB contaminated mineral oil
transformer it would be possible to reduce the
PCB concentration of the working fluid by first
draining the transformer and refilling it with non-
contaminated oil. Since the PCBs are contained
predominantly in the oil, and are not impregnated
into the porous insulation, it is possible to ret-
rofill the transformer to a required concentration
level by repeated washings, as necessary. The
effectiveness of the washings becomes increas-
ingly important as the concentration of the
original contamination increases since repeated
dilution of residual contamination produces sev-
eral times the original volume of the transformer
tank for disposal. In addition, the presence of
trapped, high concentration oil in the core/coil
assembly, which is not easily removed by con-
ventional flushing procedures, causes the bulk oil
to become re-contaminated.
(2) In the case of an askarel transformer which is
impregnated with PCBs, a simple flushing or
vapor cleaning process leaves about 2.5% of the
original askarel contained in the core/coil assem-
bly. For a typical 250 gal. askarel transformer the
impregnated fluid amounts to about 6 gal. A PCB
concentration of 500 ppm PCB in a 250 gal. trans-
former is equivalent to little more than about one
coffee mug in volume.
The equilibrium distribution of PCBs between transformer
solid materials and the liquid dielectric in a 500 KVA
transformer has been found to be about 97.5% in the bulk
oil and 2.5% distributed unevenly among the paper, core
steel and wire. The vast majority of the PCBs in the solid
materials of an impregnated transformer is contained in
the paper.
The diffusion of PCBs under the action of a strong
concentration gradient from the interior of the paper
towards the very low concentration in the bulk retrofill
fluid is a slow process governed by Fick’s laws. If the ret-
rofilled transformer is not put into operation and remains at
ambient temperature, then the viscosity of the askarel may
be high enough to prevent diffusion at a significant rate.
However, the EPA requires that the reclassification proce-
dure should put the transformer under a normal load or,
alternatively, that the bulk oil is heated to more than 50°C
for a period of 90 days following the completion of the
retrofill operation(s). The power losses of the transformer
under load conditions produce heat which, in turn, reduces
the viscosity of the impregnated askarel and increases
its diffusivity. If the concentration level of PCBs can be
reduced to such an extent that subsequent leaching does
not yield a concentration which is above the 500 ppm PCB
limit imposed by the EPA 90 days after the process has
TABLE 39
Major PCB fires in Scandinavia have involved mostly capacitors
Location Date
Stockholm, Sweden August 25, 1981
Stovde, Sweden March 19, 1982
Arvika, Sweden May 1982
Imatra, Finland August 2, 1982
Helsinki, Finland August 1982
Surahammar, Sweden September 23, 1982
Hallstahammar, Sweden November 8, 1982
Locmotive, Sweden Winter 1982/1983
Kisa, Sweden April 25, 1983
Halmstad, Sweden August 15, 1983
TABLE 38
Background level of PCBs found in swipe samples of surfaces in non-
manufacturing buildings
PCB (ng/m2)
Location Aroclor n/N1 Range Mean2
Cincinnati, OH 1242 31/60 0.5–4.5 1.3
St. Paul, MN 1260 14/18 0.5–5.0 2.0
Boston, MA 1254 30/32 0.1–3.2 0.7
Pine Bluff, AK 1254 4/4 3.5–7.2 5.1
TOTAL: 79/114 0.1/7.2 1.4
1 n number of samples above detection limit.
N total number of samples.
2 Mean mean of detected values only.
C016_003_r03.indd 906C016_003_r03.indd 906 11/18/2005 1:12:39 PM11/18/2005 1:12:39 PM