Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
4 Accuracy of Toxic Chemical Release Estimates Congress was aware that the toxic chemi- cal release estimates reported under SARA Section 313 might not accurately reflect the amounts actually released from reporting facilities (U.S. Congress, House, 1986~. This potential inaccuracy is based on the provi- sion that quantities of chemical releases can be obtained from theoretical calculations, engineering estimates, or by subtracting mass balance quantities (e.g., chemical quantity purchased minus the quantity contained in the product) rather than from measurements of actual releases. This chapter discusses the applicability of EMB and MA in providing a reference value to assess the accuracy of a toxic chemical re- lease estimate. EMB provides measurement data of releases that have been evaluated through a check for closure. When releases have not been measured directly, the reference value is calculated by taking the difference between the sum of measured inputs and the sum of measured outputs. The committee also considered the use of mass balance data for assessing the nature of reported releases in the context of the way manufacturing facilities produce and use toxic chemicals. Release data from facilities 31 that produce and use TRI-listed chemicals are used to demonstrate how mass balance data could provide information on release estimate reasonableness. ENGINEERING MASS BALANCE The use of EMB at industrial facilities is similar to the use of a rigorous, double-entry recordkeeping system. Figure 4.1 is an ex- ample of a ledger sheet for mass balance components. The ledger has a credit column for materials transported into or produced at a facility and a debit column for materials consumed at or in some other way subtracted from a facility. The handling of arsenic (a TRI-listed chemical) at a copper smelter is an example of how mass balance components would be entered on the ledger sheet for a smelter facility. The credit column of the ledger sheet would list arsenic transported into the smelter facility as a contaminant of copper ore. The debit column would include the quantity of arsenic transported from the facility as a product (blister copper) con- taminant and released to the environment via
32 Credit 1. Transported into facility 2. Produced at facility as: a) product b) byproduct c) raw material contaminant the furnace slag, flue dust' and scrubber sludge generated during the smelter operation. Although it is feasible to perform EMB for individual processing steps or simple production processes, it is almost impossible to obtain complete closure in EMBs for complex industrial facilities. As discussed in Chapter 2, EMB analysis or MA practice is limited by measurement error and the assumptions or judgments necessary to implement the analysis. When the quantities of mass balance components are very dif- ferent (e.g., thousands of tons versus hundreds of pounds), imbalance between the inputs and outputs likely would be greater than the smallest component quantity. Therefore, to work within the limits inherent in EMB, the smallest component must be at least several times greater than the largest uncertainty in the other mass balance com- ponents. This condition often results when the quantity of each mass balance component is approximately equivalent in order of mag- nitude. EMB analysis performed on a decreasing operation within a manufacturing facility that does not recover any of its used solvent is an example of input and output quantities being within approximately equivalent orders of magnitude. In this case, the extent to which closure is obtained would provide a good assessment of the validity of the release estimates. The imbalance resulting from ~4SS BAL4NCE INFORMATION Debit 1. Consumed at facility (to produce a product or, e.g., to control pollution) 2. Used at facility 3. Accumulated at fatality 4. Released from facility, via: a) air b) water c) solid waste 5. Transported from facility FIGURE 4.1 Ledger Sheet for Quantities in an EMB EMB likely would be less than any of the amounts quantified purchases, environmental releases, and spent solvent shipped off site. Alternatively, consider a facility that produces 5 million lb per day of ethylene as an example of a mass balance analysis on in- put and output quantities with great dif- ferences in orders of magnitude. Even though EMB at this facility achieved closure to within Folio, an imbalance of +50,000 lb of ethylene per day remained. Emissions of ethylene from this facility were estimated to be 191 Ib per day by recognized calculation techniques~onfirmed by monitoring data (Chlapek, lamb. In this case, even a small- percentage imbalance overwhelms the emis- sions estimate because of the great disparity between the quantity of material feti into the facility and the amount released. This ex- ample illustrates that even a high degree of closure is of little use in evaluating the ac- curacy of release estimates if the orders of magnitude in mass balance components dif- fer greatly. The committee constructed the 2 x 2 matrix in Fig. 4.2 to illustrate the potential utility of EMB for determining the accuracy of information on toxic chemical releases to the environment. For cases A and B. accuracy of the reported release estimate is assessed by-comparing it with measured release data obtained through an EMB. When all EMB components are measured and
ACCURACY OF TOXIC CHEMICAL RELEASE ESTIMATES WAYS TO ASSESS ACCURACY OF REPORTED RELEASES 33 Measure all EMB data quantities; check degree of closure Determine releases by calculating difference in measured quantities 1 Order-of-magnitude equivalence for all quantities Major disparity in quantities High potential value A Low potential value Limited value; (no potential value when reported releases obtained_ by difference) C - No potential value D FIGURE 4.2 Potential value of an EMB approach in determining the accuracy of information on toxic chemical releases to the environment. involve equivalent order-of-magnitude quantities, EMB might have significant util- ity for determining the accuracy of release estimates (A in Fig. 4.2~. When all com- ponents are measured but the magnitude of those components is disparate, the potential value of EMB is low (B in Fig. 4.2~. When EMB data are not available on environmental releases, they sometimes are estimated by assigning the release as the value of the difference between the sum of the inputs and sum of the outputs other than releases. Congress acknowledged that this approach might be used to estimate mass balance components: "quantities [to apply in the mass balance] . . . will be determined by a variety of methods including . . . estimates derived from differences between measure- ments . . .~ (U.S. Congress, House, 1986~. For cases C and D, accuracy is assessed by comparing the release estimate with the release determined by calculating the dif- ference in measured quantities. However, the accuracy of a release estimate cannot be checked by comparing it with the value obtained by taking the difference between inputs and outputs other than the release if the reported estimate was obtained in the same way. This approach to estimating releases by difference should be applied only to facili- ties where the input and output streams containing the chemical of interest are readi- ly identified, and the mass balance com- ponents are of similar orders of magnitude. The quantity of releases from the facility with the decreasing operation described above probably represents the difference be
34 tween the solvent purchased and the spent solvent shipped off site (C in Fig. 4.2~. However, this approach is likely to be a less reliable check on accuracy than the measure- ment of all EMB data. Estimating releases by difference can produce misleading information or lead to misapplications, even if the mass balance components are judged to be of similar orders of magnitude. For example, two EMBs were performed for two different facilities that produce methyl chloride and methylchlorosilane (Supple, 1988~. To obtain a methyl material balance over both facili- ties, all components were independently de- termined, except for emissions to the atmos- phere. Closure to within 1% was achieved. Methyl emissions were calculated as the difference between the sum of inputs (1S,167,000 lb) and outputs (1S,356,000 lb) for a fixed period. Emissions were assigned a value of -189,000 lb of methyl per period, implying that methyl compounds were being recovered from the air, which is obviously false. Calculating the emissions by using factors provided by EPA resulted in an emissions value of approximately 37,000 lb of methyl for the same measurement period (D in Fig. 4.2~. By comparing the magnitude of uncer- tainty in mass balance data with the mag- nitude of releases, the committee determined that, in the context of this study, mass balance data are not generally adequate to assess the accuracy of release estimates. Although this approach can be considered a rudimentary error analysis, it was sufficient to answer the question, and more rigorous quantitative methods (e.g., probability density functions) were not necessary. MATERIALS ACCOUNTING As discussed in Chapter 2, MA is in- herently less accurate and precise than EMB. Therefore, MA practice generally precludes obtaining useful information on the accuracy of chemical release estimates, such as those reported to the TRI. However, MA infor- mation could provide improved public insight for understanding the nature of cur- rent releases, if the program to collect MA information were well designed and properly conducted. The usefulness of MA, sup- ported by technical assistance and expert MASS BALANCE INFORMATION analysis, is discussed in detail in Chapter 7. Although EMB could also provide these benefits, the accuracy, precision, and ex- pense of EMB are not necessary for achieving them. The examples provided below describe the circumstances under which MA informa- tion could improve understanding of chemi- cal releases. The illustrations also point out some of the knowledge useful for the analysis and interpretation of MA informa- tion and the need for a clear plan for ul- timate use and application of information before reporting is implemented. For pur- poses of discussion in these examples, MA information is the following: · Quantity of a listed chemical pro- duced at a facility. · Quantity of a listed chemical brought into a facility. · Quantity of a listed chemical con- sumed at a facility. · Quantity of a listed chemical shipped from a facility as product or in products. MA: Illustrative Applications The following scenarios are based on use of a single chemical, although actual condi- tions probably would be more complex than those presented. · An adhesive manufacturer uses a chlorinated solvent 1,1,1 -trichioroethane (TCA)- in adhesive formulations. Very small amounts of TCA are emitted to the atmosphere during the blending process, and only a small amount of adhesive waste-con- taining TCA is generated. Waste-water re- leases are also small; most of the TCA is part of the final product. · In the same city, a lock manufacturer uses TCA in a vapor decreasing operation to remove cutting fluid from locks. Ap- proximately 93% of the TCA is emitted to the atmosphere, and 6.7% is contaminated waste solvent. · A third facility manufactures TCA. A small amount of the chemical is lost in the process as fugitive emissions. A waste prod- uct containing TCA at a low concentration, unreacted precursor chemicals, and other unwanted byproducts is sent off site for dis- posal.
ACCURACY OF TOXIC CHEMICAL RELEASE ESTIMATES Each firm must report TCA releases to the TRI, because each exceeds the SARA threshold use or production level for the chemical. A summary of the MA data for each of the three facilities is listed in Table 4.1. (It was assumed that none of the TCA is inventoried; that is, in a given year, all TCA that is purchased or produced is not accumulated at the facility but leaves the facility as a product or is otherwise released.) The adhesive manufacturer and the lock manufacturer almost certainly would es- timate their values for releases to air by dif- ference. Both facilities know the amounts of TCA brought on site to the level of accuracy recorded on their purchase records. The ad- hesive manufacturer also knows the amount of TCA transported off site as a product with reasonably good accuracy. Both facili- ties are required to report the amount of TCA sent off site as waste. The adhesive manufacturer probably knows the amount of TCA discharged to water because of sewer regulations. The remaining term in both cases the amount of TCA released to air was determined by difference. Interpreting Release Data The chemical manufacturer, who has low emissions, makes TCA as a product. When it is sold to firms, such as adhesive manu- facturers and lock manufacturers, and used in processing, emissions occur as shown in Table 4.1. Emissions from adhesive and lock production are approximately the same (about 2S,000 lb), but they occur at different times and locations. The people in the community who are un- familiar with the specifics of production processes might compare the three facilities releasing the same chemical. Why are the releases from the lock manufacturer so high? If the adhesive and chemical manufacturers have such low releases, why can't the lock manufacturer reduce his releases? In the case of the adhesive manufacturer, most 27,000 lb-of the 30,000 lb of TCA purchased was sent off site in the product. This adhesive product will be purchased by construction facilities or consumers who use the adhesives in homes and buildings. When the adhesive is used at these thousands of different sites, all of the TCA will be emit 35 ted into the atmosphere. The adhesive facil- ity's direct releases to the environment are small: 500 lb, or 1.7% of its purchases, is released to the air and water; and 2,000 lb, or 6.7% of purchases, is waste. After ac- counting for releases resulting from con- sumer use, however, 27,500 lb is released into the atmosphere. The lock manufacturer has a very dif- ferent operation. The facility uses the TCA to clean the locks, and none of the solvent is incorporated into the final product. Produc- tion-site air releases are very high-2S,000 lb, or 93% of purchases reflecting the vola- tility of the chemical, a characteristic that makes it a desirable solvent. TCA waste amounts to 2,000 lb, or 6.7% of purchases. The chemical manufacturer has yet a dif- ferent operation. The facility's releases are small (5% of production). Most of the re- leases occur to the air-1,000 lb, or 3% of production. The remaining losses to the water and as hazardous waste are 0.7Yo and 1% of production, respectively. MA Characteristics Estimates of releases to air and water and waste sent off site reported to the TRI are presented in the three bottom rows in Table 4.1. Additional data that might be required in MA reporting are found in the four top rows of values in Table 4.1. These data might be used to explain measures taken to protect against chemical releases that communities might perceive as posing a threat to human health and the environment. The additional data in the top four rows of Table 4.1 should be examined to determine whether they contribute to a better understanding of TRI data. Additional MA information could be collected on chemicals of special interest. If it is assumed that TCA is one of these sub- stances, then the data in the four top rows of Table 4.1 are useful for comparing opera- tions between different industries using the same listed chemical and for helping to identify large procedural errors in reported release estimates, particularly overestimates. First, the additional data would allow better comparisons to be made among manu- facturers in a particular industry. For in- stance, two adhesive manufacturers might purchase 30,000 lb of TCA. If one manu
36 TABLE 4.1 Approximate Annual MA Quantities of TCA (lb) Facility Type AL4SS BALANCE INFORMATION MA Components AdhesiveLockChemical ManufacturerManufacturerManufacturera Transported into, produced on site, or purchased 30,00030,00030,000 Consumed or used 30,00030,000 Accumulated OOO Transported off site as product 27,000028,500 Released to am 500b28 oooc1,OOO Released to water 5000200 Sent off site es waste 2,000&2,000300 Note: Although the three cases in the table are hypothetical, an attempt has been made to estimate the emissions realis- tically using information contained in EPA documents. Specific documents are cited in other footnotes to this table bMa~nmum losses of TCA based on production process described in Key et al. (1980~. Assumes small fraction of emissions occur in facility. CAssumes about 94% of solvent is emitted during use; the balance is solid waste. See EPA (1985b,c) and Pandullo et al. i1985) for estimates of atmospheric emissions of other, similar chlorinated solvents in degreasing applications. Assumes that waste losses are those reported as "solid waste" plus "storage/waste disposal/destn~ction in end product," plus one-third of the TCA recovered and sold as waste solvent. facturer ships 27,000 Ib off site in the prod- uct and another ships only 20,000 lb, the lat- ter manufacturer might be presumed to have a relatively inefficient operation. If only release data were collected, total releases to the environment in the first case would be known to be 3,000 lb; it would also be known that releases were much larger (10,000 lb) in the second case. With only the release data, it would not be obvious that the manufacturers purchased the same amount of TCA. Indeed, because the losses in the second instance are so much larger, it could be assumed that the facility purchased much more TCA. Second, the additional data demonstrate differences among types of industry in which identical chemicals are involved. Table 4.1 shows that none of the solvent leaves the lock manufacturer's facility in the product. In fact, since the TCA is used only to clean locks, all of it is lost through re leases, and the additional data demonstrate that none of the TCA is put into the prod- uct. This information contrasts with that for adhesive manufacturing, in which nearly all of the solvent ends up in the product. Third, the additional data can be used to enhance efforts to validate reported releases by helping to identify large overestimates. For example, the adhesive manufacturer could mistakenly report the total mass of water and contaminants discharged, instead of reporting the mass of TCA contained in the water. In this case, the total mass of waste water would significantly exceed the TCA mass transported into the facility and the amount transported off site as product, thus appearing as an obvious reporting er- ror. However, without the additional infor- mation shown in Table 4.1, it would be much more difficult to detect this over- estimate.
ACCURACY OF TOMC CHEMICAL RELEASE ESTIMATES CONCLUSIONS Neither EMB nor MA is generally ade- quate, in the context of this study, to assess the accuracy of estimates of toxic chemical releases. Although direct measurement of releases is the best approach, this might not always be feasible because of the potential lack of satisfactory analytical procedures for some TRI-listed chemicals. EMB approaches have only limited applicability for assessing the accuracy of toxic chemical releases. Only in circum- stances in which all quantities can be in- dependently measured and are within an equivalent quantitative order of magnitude can EMB serve as a check on the accuracy of methods of estimating toxic chemical releases. 37 MA is even less precise and less accurate than EMB. MA also does not have the potential to assess the accuracy of estimates of toxic chemical releases. However, EMB or MA information might be helpful for detecting gross errors in reported estimates of environmental releases and for improving understanding of chemi- cal use patterns and environmental releases, provided that data collection is accompanied by expert technical knowledge, data valida- tion, and analysis. MA has a better potential for these applications than EMB, because MA typically is less difficult and less expen- sive to obtain. Therefore, further considera- tion of MA information is warranted.
