Document 6RKp97e6DQBav01rngYX9ELLd

RICE UNIVERSITY GEORGE R. BROWN SCHOOL OF ENGINEERING HOUSTON, TEXAS 77001 March 26, 1976 DEPARTMENT OF CHEMICAL ENGINEERING Mr. W, P. Anderson Tenneco Chemicals Park 80 Plaza West-One Saddle Brook, N. J. 07662 Dear Mr. Anderson; At your request I have reviewed the document "Quantitative Risk Assessment For Community Exposure To Vinyl Chloride", by Robt. E. McGaughy of the U. S. Environmental Protection Agency. I have also considered the atmospheric dispersion modeling and air sampling conducted by your staff to assess the vinyl chloride exposure at your Deer Park site. The following com ments pertain to the general risk calculations of McGaughy, my review of your modeling results, and my estimate of the particular risk associated with the Deer Park facility. The essential elements of an exposure evaluation for a given source are the following: (1) Estimating the strength and geometry of the pollutant emission. (2) Modeling the atmospheric dispersion, which combines the source data and meteorological con ditions of the site to predict average pollutant concentration as a function of location relative to the center of the source. (3) Combining this concentration function with population distribution data to obtain human exposure. The estimate of risk to human health follows from (3) and clinica epidemiological health observations. In the ^ase of vinyl COLORITE 009031 Mr. w. P. Anderson March 26 Page 2 chloride exposure at the concentration levels predicted to exist near PVC plants, the health data must be extrapolated downward more than three orders of magnitude (50 ppm to 17 ppb). As McGaughy points out, the difference between the more conservative linear model and the more often accepted log-probit model for extrapolation is about a factor of 100 at the 17 ppb level. His figure B-l shows the linear model.to be more conservative (predicts a higher risk) by a factor of about 10,000 at the 1 ppb level. These elements are next considered as they apply to the specific problem at hand - the risk to human health of an expanded Tenneco facility at Deer Park. Using data acquired by your staff, I have made approximate corrections to apply McGaughy's methodology to this site. The important points are: (1) Your estimate of vinyl chloride monomer emission rate (stack emissions plus fugitive losses) is in the range of 0.1 percent of throughput, contrasted with 4 percent assumed by McGaughy for the typical PVC plant. Accounting for your larger production rate, this still represents a lower emission rate by a factor of about 30. (2) Your dispersion modeling is directly comparable to McGaughy's since the same mathematical models were used. It is probable, in fact, that the same basic computer program package was used in both studies. (3) The dispersion model used to prediction concen trations from source rate data is intrinsically linear in source strength. Hence, direct comparison can be made between the concentration plots in McGaughy (Appendix A) and the plots generated by your study. Considering the differences in meteorological conditions between McGaughy's average PVC plant case and your site, I find the two results to be quite equivalent. That is, your concentrations at the same distances from the plant center are approximately 30 times lower than McGaughy's, a_s expected. COLORITE 009032 Mr. W. P: Anderson March 26, 1976 Page 3 (4) Your inability to detect vinyl chloride at locations more than 1/2 mile down-wind of the present plant site are also consistent with the predictions of the dispersion model, considering the lower limit of your analytical procedures (about 10 ppb). (5) There are factors ignored in the dispersion modeling which would further decrease the predicted concentrations if accounted for. These include homogeneous decay reactions and reactions with surfaces involving vinyl chloride. The latter may be quite important inside dwellings, etc. (6) Using the population numbers given by McGaughy N in his Table A-l- (a) for the large PVC plant in Deer Park and the decreased concentrations predicted by the i dispersion model, I calculate an average exposure of ^ ,yabout 0.45 ppb to 131,000 people. The overall effect rate can then be calculated using McGaughy's linear ' model of 0.071 cases of angiosaj:oojua per ppo per mirlion population per year. The result for the Deer Park site is .131 x .45 x 0.071 .004 cases per year^ cr^Che case in 250 years. (7) Considering the difference between the linear model and the log-probit model discusses above, this estimate may be as much as a factor of 10,000 too high. Considering the uncertainties in the basic health risk data, I do not consider this result to be very reliable. If it is in fact an upper limit, the risk in my opinion is ne Sincerely, HAD:Id H. A. Deans Professor of Chemical Engineering COLORITE 009033 by A. Kuzmao- "llul '< Mc(,..-iu<. ny The authors have considered a number of very important problems concerned with the assessment of the fish of cancer, especially liver angiosarcoma, and have taken as quantitative art approach os possible in the analysis Of the avail able data. It can be fairly stated that the authors haVc taken a reasonable approach, ,their conclusions follow from the assumptions made and, insofar as possible, they have used actual data that bear on the calculation of risk. In any problem of this typo, there arc bound to be .numerous unknowns, data from small samples that are variable in outcome and problems in explaining interspecies differences. Average or median values have been used and weighting facto; applied (that themselves are subject to variability) to obtain answers, ^nn cor.not fault the approach since it was reasonable and as quantitative as possible, However, in choosing figures from which to estimate risks, the authors have nearl} always taken the figure that would lead to a higher estimate of risk of liver angiosarcoma (or cancer). They have evaluated the various possible errors in their estimates as well as possible, however their figures do tend to lend to upper limits of the number of cases to be expected. One example of this is the choice of a linear dose-response curve rather than a curve linear in log dose for calculating the proportion of animals that will develop liver angiosarcoma following exposure to different VC concentrations. Assuming the log dose rela tionship yields estimates of the number of cases of liver angiosarcoma .01 - .1 times that obtained from the linear dose-response curve. This study by the authors points up the need further studies to determine the risk of exposure to VC more precisely. Tor example, it is certainly important to carry out surveys of liver cancer and other liver related diseases in indivi duals living near VC or PVC plants to compare their incidence with that i:i some control areas not close to VC or PVC plants. Secondly, some animal studies should be carried out to estimate the dose-response relationship to VC concentration. For example, the lowest non-zero dose of VC in Maltoni's study was 50 ppm and this dose led to a 2t incidence of liver angiosarcomas. This dose is nearly 30C0 time# larger than the 17 ppb average concentration exposure for people living within 5 miles of FVC or VC plants. One can only speculate on what the incidence of l.j.vor angiosarcoma (and other tumors) is for VC concentrations in the neighborhood of 10 - 100 ppb. The key results in the authors' study arc that: an estimated 4.C' million people arc living within 5 miles of FVC or VC plants; the average exposure to VC for these individuals is 17 ppb and the estimated number of cancers to bo observed in those individuals is 5.5 cases of liver angiosarcoma per year and 11 cases of cancer.. Assuming a log promt model reduces the estimated numbers of cases by some factor between 10 and 100. It is evident that if one accepts the log probit. model, VC exposure docs uoi present a major risk of cancer, crons the linear dose response curve indicates a tr-iior risk. There is no available data for choosing between those models, r.o the authors give the results for both models and take the prudent approach in assuming that Lhcro is >o threshold dose of VC, below which no cases of liver angiosarcoma would bo expected to occur. COLOR!TE 009034  s q( VC or PVC plants, tin; autivu c a on t u 11 y used Live resui Ls <;L a i 1/ by the APHA of census tracts based un<>n 1970 census population figures* Thorn arc known flaws,with the estimate (since no adjustment is made for mobility or increase in the population since 1970, or adjustment for age, sox, or direction from plant), however the' estimate is unlikely to be in serious error. Concerning the estimate of the exposure of 17 ppb for the 4.6 million people the estimated ambient concentrations were based upon studies by EPA and Teknckron These studies wore reasonably consistent with one another and so this result can not be criticized severely. Further, adjustments were made for the number of plants, the size'Of the plants, and meteorological conditions. The weighting factors used for meteorological conditions were 0.56 for low, 1.00 for averago, 1.55 for high, and 2.55 for very high. The basis for choosing those figures is not given in this report; reference is made to the Teknekron results. It is of interest that when the figures for weighted population exposed are given by distance from plant (Table A-2) , the .major populations exposed are those 1-5 miles from rvc plants. If one wore to do a study of incidence of liver angro* sarcoma and other tumors in the largo cities (about 100,000 or more), then a relatively small number of cities have largo populations near PVC plants. From Table A-l these arc: Carson City, Louisville, Burlington, I'assaic, S. Kearny, Hicksville, Fitchburg,Springfield, Fedrickton, Williamsvilie, Oklahoma City, Deer Park, and South Charleston. Incidence of disease studies might be under taken in those areas. The major possibility for error in this report (and this is recognized by the authors) is when one attempts to predict the-' number of cancer (and liver angiosarcomas) based on given levels of exposure to VC. The authors use animal data to predict the results in humans, and then use data from humans insofar as possible to confirm the animal data. Results from the animal and human data agree reasonably well so this gives more confidence in the methodology utilized. When considering the data from humans, the authors refer to reports by Tabcrshaw-Gaffey, Nicholson, Heath and Falk, and Wagoner. For each of these studios, data are given concerning the number of individuals studied, the number of cares of liver angiosarcoma, and the approximate length of exposure. However, m calculating the incidence rate, defined as a number of cases of liver angio sarcoma per person per year of exposure, the estimate of .00475 per person year was made only from the Tabcrshaw-Gaffey report (Pago D-4). It is suggested that tiie data from the other studies might also be used to obtain this typo of estimate since it appears that the various studies would lead to estimates that were not very different from one another. One could make reasonable assumptions regarding the average exposure for the workers in each reference. The authors indicate a preference for calculating incidence rates using Liio distribution of exposure dural ion and this is available only from the Tabershaw-C.a f fey study; this prefer ence may be justified, however when one has a series of several studies that would appear to lead to relatively consistent estimates of incidence rates, it suggests Uiat these studies bo used. COLOR!TE 009035 v.'.'ie page / , I_n,: .Uiuiois indicate Uiat when ail the uncot taint ies arc rnti- oidered, their judgment is that the nu1 her of eases of liver nngiofmrrow | reduced per year of exposure in people residing near VC plants is somevdiere betc-''-" 1 and 10 cases, with these eases not to be diagnosed until 15 - 20 years lion n The estimated number of cases of primary cancer at other sites is also between < 1 and 10 cases. It is not clear why the upper limit is not 20 cases in the latter case, since at other places in the manuscript the authors refer to an expectation of twice os many cases of cancer at other sites as cases of liver anoiosurcoma. Also, an estimated 1 - 300 cases of serious liver daingo would lie predicted (though the number could possibly bo loss). Based on careful review of the methodology used in the report, there are no good reasons to seriously question these figures. - / Edmund A. Gehan, Ph.D. / COLORITE 009036 Case 1 2 3 4 5 6 7 8 9 10 12 13 15 16 TABLE I SUMMARY OF RESULTS FROM INITIAL SCREENING4 Precondensation VCM Recovery of Stripper Gas Method No i'- No No No Yes Yes Yes Yes No No Carbon Bed Carbon Bed Carbon Bed Carbon Bed"'' Carbon Bed Carbon Bed Carbon Bed Carbon Bed"^ None None Yes None Yes None Yes Solvent Absorption Yes Solvent Absorption Emission Control Net Revenue Total $/year Capital, $ Incinerator - 19,250 Boiler + 8,2 60 None + 9,290 Incinerator - 35,600 Incinerator + 21,180 Boiler + 48,710 None + 49,730 Incinerator + 22,608 Incinerator - 122,830 Inc.^Acid Rec. - 29,790 Incinerator - 51,660 Inc. /Atjid Recov. + 22,040 Boiler + 145,120 3,605,000 3,270,000 3,215,000 3,965,000 3,500,000 3,220,000 3,160,000 3,620,000 2,125,000 2,960,000 1,625,000 2,115,000 1,010,000 Incinerator + 119,530 1,400,000 3 Total Net Present Value, M$ - 2,387 - 2,073 - 2,034 - 2,676 - 2,226 - 1,896 - 1,854 - 2,247 - 1,805 - 2,009 - 1,229 - 1,281 131 473 Economic Rank 13 10 9 14 11 7 6 12 5 8 3 4 1 2 COLORITE 009037 1. In cases 4 and 8 the carbon bed recovers VCM only from the vent condenser stream; gas from the dispersion strippers goes directly to the incinerator. 2. In cases 10 and 13 the HC1 from the combustion gases is recovered as 20% hydrochloric acid and a credit is taken for it. 3. Calculated using a 25% discount rate, 10-year life and no salvage value. 4. Cases 11 and 14, omitted here, involved recovery of 36% hydrochloric acid from incinerator combustion gases. The cost estimates from Trane Thermal made it obvious that the cases were not worth comple TABLE II COMPARISON OF ALTERNATIVE APPROACHES CAPITAL UTILITIES - DEMAND 'Case Ope rat in ; Cost $/Vr. Operating Reve nue (Rec'd VCM) $/Yr. Wet Operating Process Income Equip. $/Yr. $ Methyl Chloride S ya tem s ''PesBimiatlc" C irbon Bed Alone 7A 378,920 170,970 -207,950 2,321,000 100,000 Incineration A1 jne ? ~ . 12A 198,020 35,880 -162,140 1,475,000 - Solvent Aba. /in :ineratlon ? 1 16A 195,790 169,600 - 26,190 1,800,000 _ "Optimiatic" Ca :bon Bed/incineration 17 295,010 170,970 -124,040 2,868,000 - "Optimistic11 Ca :bon Bed Alone 18 * 212,260 170,970 - 41,290 1,848,000 - Utilities Total $$ KVA S team (Approx) #/Hr. 545,000 2,966,000 435 150,000 1,625,000 61 4,200 _ 150,000 1,950,000 50 600 295,000 3,163,000 196 1,050 295,000 2,143,000 190 1,050 Cooling Water Brine GMP Tons 920 7 .7 50 10 240 7 240 7 Chilled Total Net Water Present Tons Va lue 76 -2,649 M 18 -1,624 M 18 -1,347 M 52 -2,477 M 52 -1,525 M Economic 5 (,* Vear/' <J(, c 3 I 4* 2 ,, c -c a * ^ .? * ., -' - COLOR!TE 009038 ii j., . f " L ; / = `,t d . / , .,r 3 COLORITE 0 0 9 0 3 9 Case S team Use #/Dsy Cost $/Yr, 7A 79,470 95,960 12A 0 0 16A 12,010 16,500 17 13,900 16,790 18 13,900 16,790 TABLE III DETAILED OPERATING COSTS FOR ALTERNATE APPROACHES Power Use KW Hr./ Day Cost $/Yr. 2,380 22,990 1,091 10,540 1,170 11,300 2,201 21,260 2,093 20,220 Fuel on Use Cos t GPD $/Yr. 00 55 7,700 no 15,400 no 15,400 00 Caustic $/Yr, Chilled Water $/Yr. 0 59,200 370 10 0 12,690 9,070 9,070 11,220 11,220 Brine $/Yr, Cooling Water $/ Yr. Nitrogen $/Yr. 5,880 5,880 8,400 5,880 5,880 42,840 - 8,750 11,110 11,110 2,280 - 5,610 5,610 Carbon/ Solvent Make-Up $/Yr. 5,500 - 4,500 3,300 3,300 Maintenance Taxes, Ins, $/Yr. 190,780 105,630 123,500 204,430 138,130 Total Operating Cost S/Yr. 378,920 198,020 195,790 295,010 212,260 WATER STRIPPER 77777-, ! j 1 2 tI I, 1 II 4 I I) [ v_y. _y v_w_y PLASTISOL STRIPPERS ( EXISTING) --n c- n-- fSPARE) --Oc-ibV]-- VACUUM PUMP C-3a ('spare) c-3b COMPRESSOR FROM EX1ST IMG COLORITE 009040 _,Ai_P_ ^|jE L C , / ~ ^ f -n MPT SCALE-- DRAWN 0V- JCJ&f CHECKED-- TENNECO CHEMICALS, INC. 5nnec^ file. DRAWING rm-E- INCINERATION AUOUE COLORITE 009041 WATER. t,-Tp,pcEt( !!! X " A T ' _L_. ' 11 1 Z 1 1 3 1 ; A, I I! V_y- /V_ PLASTISOL STRIPPERS ( CXtSTlMS) COLORITE 009042 r COLORITE 009043 l;." WATER STRIPPER % i is; cSi % & C AC 7 TO RECOVERED MONOMER STORAGE om.scaleDRAWN r~ CHECKED-- ' TENNECO CHEMICALS, INC. 4 Tint-' OPT! MISTIC ~ CARBON BED/t UClNERATlON FILEDMWIIW HUM KM V WATER STRIPPER n t;x" t1 : 2=^ PL ASTI SOL. STfcroprf.; r T M', l -TK-'.b---- ] SPAP rf;HT~ I s t=* r e' <AZj >' oij>'X I------ f I PROM EXISTING, RECOVERV S /STEM I CASE +L__. t-----1 !E-2| CONOSNSEft I I I V!:^~T ; ( CONOENSEK i , ^EXISTING) U-. -i--t- L?SZ ' L'Ay \8 j- j .4 / TO RECOVERED ^ MONOMER STORAGE ! COLORITE 009044 1 E 1, ! ___ L H EC,", C RE O \ C M DEPT.SCALE-- OS*-' DRAWN B CWHCIU TENNECO CHEM1CALS, 1NC. :%7 title- OPTIMISTIC CARBON OED SYSTEM FILE- .--if 0MWIM6 MUMMER 1 0\