As noted in Section 7.2 above, this statistical approach (use of the upper tolerance limit) for calculation of the concentration in normal residue is based on the assumption that the concentration data are distributed normally. The department is aware that concentration data of this type may not always be distributed normally, particularly when concentrations are near the detection limits. There are a number of procedures that can be used to test the distribution of a data set. For example, the Shapiro-Wilk test, examination of a histogram or plot of the data on normal probability paper, and examination of the coefficient of skewness are methods that may be applicable, depending on the nature of the data (References 1 and 2).
If the concentration data are not adequately represented by a normal distribution, the data may be transformed to attain a near normal distribution. The department has found that concentration data, especially when near detection levels, often exhibit a lognormal distribution. The assumption of a lognormal distribution has been used in various programs at EPA, such as in the Office of Solid Waste Land Disposal Restrictions program for determination of BDAT treatment standards. The transformed data may be tested for normality using the procedures identified above. If the transformed data are better represented by a normal distribution than the untransformed data, the transformed data should be used in determining the upper tolerance limit using the procedures in Section 7.2 above.
In all cases where the owner or operator wishes to use other than an assumption of normally distributed data or believes that use of an alternate statistical approach is appropriate to the specific data set, the owner or operator shall provide supporting rationale in the operating record that demonstrates that the data treatment is based upon sound statistical practice.
7.4 Nondetect Values
The department is developing guidance regarding the treatment of nondetect values (data where the concentration of the constituent being measured is below the lowest concentration for which the analytical method is valid) in carrying out the statistical determination described above. Until the guidance information is available, facilities may present their own approach to the handling of nondetect data points, but shall provide supporting rationale in the operating record for consideration by the department.
-
See PDF for table 
7.5 References
1. Shapiro, S.S. and Wilk, M.B. (1965), “An Analysis of Variance Test for Normality (complete samples)," Biometrika, 52,591-611.
2. Bhattacharyya, G.K. and R.A. Johnson (1977), Statistical Concepts and Methods, John Wiley and Sons, New York.
Section 8.0—
Procedures for Determining Default Values for Air Pollution Control System Removal Efficiencies
During interim license, owners or operators of boilers and industrial furnaces burning hazardous waste shall submit documentation to the department that certifies that emissions of HCl, Cl
2, metals, and particulate matter (PM) are not likely to exceed allowable emission rates. See certification of precompliance under s.
NR 666.103(2). This documentation also establishes interim license feed rate and operating limits for the facility. For the initial certification, estimates of emissions and system removal efficiencies (SREs) can be made to establish the operating limits. Subsequently, owners or operators shall use emissions testing to demonstrate that emissions do not exceed allowable levels, and to establish operating limits (see s.
NR 666.103(3)). However, initial estimates of emissions for certification of precompliance can be based on estimated or established SREs.
The SRE combines the effect of partitioning of the chorine, metals, or PM and the air pollution control system removal efficiency (APCS RE) for these pollutants. The SRE is defined as:
SRE = (species input – species emitted) / species input
The SRE can be calculated from the partitioning factor (PF) and APCS RE by the following formula:
SRE=1 – [(PF/l00) X (1 – APCS RE/100)]
where:
PF = percentage of the pollutant partitioned to the combustion gas
Estimates of the PF and/or the APCS RE can be based on either EPA's default values or engineering judgement. EPA's `default values for the APCS RE for metals, HCl, Cl2, and PM are described in this section. EPA's default values for partitioning of these pollutants are described in section 9.0.
Guidelines for the use of engineering judgement to estimate APCS REs or PFs are described in section 9.4.
8.1 APCS RE Default Values for Metals
EPA's default assumptions for APCS RE for metals are shown in Table 8.1-1. The default values in the table are conservative estimates of the removal efficiencies for metals in BIFs, depending on the volatility of the metal and the type of APCS.
The volatility of a metal depends on the temperature, the thermal input, the chlorine content of the waste, and the identity and concentration of the metal. Metals that do not vaporize at combustion zone temperatures are classified as “nonvolatile". Such metals typically enter the APCS in the form of large particles that are removed relatively easily. Metals that vaporize in the combustion zone and condense before entering the APCS are classified as “volatile". Such metals typically enter the APCS in the form of very fine, submicron particles that are rather inefficiently removed in many APCSs. Metals that vaporize in the combustion zone and do not condense before entering the APCS are classified as “very volatile". Such metals enter the APCS in the form of a vapor that is very inefficiently removed in many APCSs.
Typically, BIFs have combustion zone temperatures high enough to vaporize any hazardous metal at concentrations sufficient to exceed risk-based emission limits. For this reason, the default assumption is that there are no nonvolatile metals. Tables 8.1-2 and 8.1-3 are used to determine whether metals are classified as “volatile" or “very volatile" depending on the temperature entering the APCS, the thermal input, and whether the waste is chlorinated or nonchlorinated.
-
See PDF for table
WS = Wet Scrubber including: Sieve Tray Tower, Packed Tower, Bubble Cap Tower
VS-20 = Venturi Scrubber, ca. 20-30 in W.G. Ä p
VS-60 = Venturi Scrubber, ca. >60 in W.G. Ä p
ESP-l = Electrostatic Precipitator; 1 stage
ESP-2 = Electrostatic Precipitator; 2 stage
ESP-4 = Electrostatic Precipitator; 4 stage
IWS = Ionizing Wet Scrubber
DS = Dry Scrubber
FF = Fabric Filter (Baghouse)
SD = Spray Dryer (Wet/Dry Scrubber)
1 Interpolation of thermal input is not allowed. If a BIF fires between 2 ranges, the APCS temperature under the higher thermal input shall be used. Example: For a BIF firing 10-100 MMBtu/hr, Mercury is considered very volatile at APCS temperatures above 260 F and volatile at APCS temperatures of 260 F and below.
- See PDF for table
1 Interpolation of thermal input is not allowed. If a BIF fires between two ranges, the APCS temperature under the higher thermal input shall be used. Example: For a BIF firing 10-100 MMBtu/hr, Mercury is considered very volatile at APCS temperatures above 260 F and volatile at APCS temperatures of 260 F and below.
A waste is considered chlorinated if chlorine is present in concentrations greater than 0.1% by weight. In the EPA guidance document “Guidance for Metals and Hydrogen Chloride Controls for Hazardous Waste Incinerators, Volume IV of the Hazardous Waste Incineration Guidance Series,"(1) one percent is used for the chlorinated/nonchlorinated cutoff. However, best engineering judgement, based on examination of pilot-scale data reported by Carroll et al. (2) on the effects of waste chlorine content on metals emissions, suggests that the one percent cutoff may not be sufficiently conservative.
Tables 8.1-2 and 8.1-3 were compiled based on equilibrium calculations. Metals are classified as very volatile at all temperatures above the temperature at which the vapor pressure of the metal is greater than 10% of the vapor pressure that results in emissions exceeding the most conservative risk-based emissions limits.
8.2 APCS RE Default Values for HCl and Cl2
Default assumptions for APCS RE for HCl in BIFs are shown in Table 8.2-1. This table is identical to the column for other BIFs except that cement kilns have a minimum HCl removal efficiency of 83%. Because of the alkaline nature of the raw materials in cement kilns, most of the chlorine is converted to chloride salts. Thus, the minimum APCS RE for HCl for cement kilns is independent of the APCS train.
Removal efficiency of Cl2 for most types of APCS is generally minimal. Therefore, the default assumption for APCS RE for Cl2 for all APCSs is 0%. This is applicable to all BIFs, including cement kilns.
8.3 APCS RE Default Values for Ash
Default assumptions for APCS RE for PM are also shown in Table 8.1-4. These figures are conservative estimates of PM removal efficiencies for different types of APCSs. They are identical to the figures in the Nonvolatile APCS RE column for hazardous metals presented in Table 8.1-1 because the same collection mechanisms and collection efficiencies that apply to nonvolatile metals also apply to PM.
-
See PDF for table
WS = Wet Scrubber including: Sieve Tray Tower, Packed Tower, Bubble Cap Tower
PS = Proprietary Wet Scrubber Design (A number of proprietary wet scrubbers have come on the market in recent years that are highly efficient on both particulates and corrosive gases. Two such units are offered by Calvert Environmental Equipment Co. and by Hydro-Sonic Systems, Inc.).
VS-20 = Venturi Scrubber, ca. 20-30 in W.G. Ä p
VS-60 = Venturi Scrubber, ca. >60 in W.G. Ä p
ESP-l = Electrostatic Precipitator; 1 stage
ESP-2 = Electrostatic Precipitator; 2 stage
ESP-4 = Electrostatic Precipitator; 4 stage
IWS = Ionizing Wet Scrubber
DS = Dry Scrubber
FF = Fabric Filter (Baghouse)
SD = Spray Dryer (Wet/Dry Scrubber)
8.4 References
1. U.S. Environmental Protection Agency. “Guidance on Metals and Hydrogen Chloride Controls for Hazardous Waste Incinerators," Office of Solid Waste, Washington, DC, August 1989.
2. Carroll, G.J., R.C. Thurnau, R.E. Maurnighan, L.R. Waterland, J.W. Lee, and D.J. Fournier. The Partitioning of Metals in Rotary Kiln Incineration. Proceedings of the Third International Conference on New Frontiers for Hazardous Waste Management. NTIS Document No. EPA/600/9-89/072, p. 555 (1989).
Section 9.0—Procedures for Determining
Default Values
for Partitioning of
Metals, Ash,
and Total Chloride/Chlorine
Pollutant partitioning factor estimates can come from 2 sources: default assumptions or engineering judgement. The department's default assumptions are discussed below for metals, HCl2, Cl, and PM. The default assumptions are used to conservatively predict the partitioning factor for several types of BIFs. Engineering judgement-based partitioning factor estimates are discussed in section 9.4.
9.1 Partitioning Default Value for Metals
To be conservative, the department is assuming that 100% of each metal in each feed stream is partitioned to the combustion gas. Owners/operators may use this default value or a supportable, site-specific value developed following the general guidelines provided in section 9.4.
9.2 Special Procedures for Chlorine, HCl, and Cl2
The department has established the special procedures presented below for chlorine because the emission limits are based on the pollutants HCl and Cl2 formed from chlorine fed to the combustor. Therefore, the owner/operator shall estimate the controlled emission rate of both HCl and Cl2 and show that they do not exceed allowable levels.
1. The default partitioning value for the fraction of chlorine in the total feed streams that is partitioned to combustion gas is 100%. Owners/operators may use this default value or a supportable, site-specific value developed following the general guidelines provided in section 9.4.
2. To determine the partitioning of chlorine in the combustion gas to HCl versus Cl2, either use the default values below or use supportable site-specific values developed following the general guidelines provided in section 9.4.
• For BIFs excluding halogen acid furnaces (HAFs), with a total feed stream chlorine/hydrogen ratio =0.95, the default partitioning factor is 20% Cl2, 80% HCl.
• For HAFs and for BIFs with a total feed stream chlorine/hydrogen ratio >0.95, the default partitioning factor is 100% Cl2.
3. To determine the uncontrolled (i.e., prior to acid gas APCS) emission rate of HCl and Cl2, multiply the feed rate of chlorine times the partitioning factor for each pollutant. Then, for HCl, convert the chlorine emission rate to HCl by multiplying it by the ratio of the molecular weight of HCl to the molecular weight of Cl (i.e., 36.5/35.5). No conversion is needed for Cl2.
9.3 Special Procedures for Ash
This section: (1) Explains why ash feed rate limits are not applicable to cement and light-weight aggregate kilns; (2) presents the default partitioning values for ash; and (3) explains how to convert the 0.08 gr/dscf, corrected to 7% O2, PM emission limit to a PM emission rate.
Waiver for Cement and Light-Weight Aggregate Kilns. For cement kilns and light-weight aggregate kilns, raw material feed streams contain the vast majority of the ash input, and a significant amount of the ash in the feed stream is entrained into the kiln exhaust gas. For these devices, the ash content of the hazardous waste stream is expected to have a negligible effect on total ash emissions. For this reason, there is no ash feed rate compliance limit for cement kilns or light-weight aggregate kilns. Nonetheless, cement kilns and light-weight aggregate kilns are required to initially certify that PM emissions are not likely to exceed the PM limit, and subsequently, certify through compliance testing that the PM limit is not exceeded.
Default Partitioning Value for Ash. The default assumption for partitioning of ash depends on the feed stream firing system. There are 2 methods by which materials may be fired into BIFs: Suspension-firing and bed-firing.
The suspension category includes atomized and lanced pumpable liquids and suspension-fired pulverized solids. The default partitioning assumption for materials fired by these systems is that 100% of the ash partitions to the combustion gas.
The bed-fired category consists principally of stoker boilers and raw materials (and in some cases containerized hazardous waste) fed into cement and light-weight aggregate kilns. The default partitioning assumption for materials fired on a bed is that 5% of the ash partitions to the combustion gas.
Converting the PM Concentration-Based Standard to a PM Mass Emission Rate. The emission limit for BIFs is 0.08 gr/dscf, corrected to 7% O2, unless a more stringent standard applies [for example, a New Source Performance Standard (NSPS) or a State standard implemented under the State Implementation Plan (SIP)]. To convert the 0.08 gr/dscf standard to a PM mass emission rate:
1. Determine the flue gas O2 concentration (% by volume, dry) and flue gas flow rate (dry standard cubic feet per minute); and
2. Calculate the allowable PM mass emission rate by multiplying the concentration- based PM emission standard times the flue gas flow rate times a dilution correction factor equal to [(21-O2 concentration from step 1)/(21-7)].
9.4 Use of Engineering Judgement To Estimate
Partitioning and APCS RE Values
Engineering judgement may be used in place of the department's conservative default assumptions to estimate partitioning and APCS RE values if the engineering judgement is defensible and properly documented. To properly document engineering judgement, the owner/operator shall keep a written record of all assumptions and calculations necessary to justify the APCS RE used. The owner/operator shall provide this record to the department upon request and shall be prepared to defend the assumptions and calculations used.
If the engineering judgement is based on emissions testing, the testing will often document the emission rate of a pollutant relative to the feed rate of that pollutant rather than the partitioning factor or APCS RE.
Examples of situations where the use of engineering judgement may be supportable to estimate a partitioning factor, APCS RE, or SRE include:
• Using emissions testing data from the facility to support an SRE, even though the testing may not meet full QA/QC procedures (e.g., triplicate test runs). The closer the test results conform with full QA/QC procedures and the closer the operating conditions during the test conform with the established operating conditions for the facility, the more supportable the engineering judgement will be.
• Applying emissions testing data documenting an SRE for one metal, including nonhazardous surrogate metals to another less volatile metal.
• Applying emissions testing data documenting an SRE from one facility to a similar facility.
• Using APCS vendor guarantees of removal efficiency.
9.5 Restrictions on Use of Test Data
The measurement of an SRE or an APCS RE may be limited by the detection limits of the measurement technique. If the emission of a pollutant is undetectable, then the calculation of SRE or APCS RE should be based on the lower limit of detectability. An SRE or APCS RE of 100% is not acceptable.
Further, mass balance data of facility inputs, emissions, and products/residues may not be used to support a partitioning factor, given the inherent uncertainties of such procedures. Partitioning factors other than the default values may be supported based on engineering judgement, considering, for example, process chemistry. Emissions test data may be used to support an engineering judgement-based SRE, which includes both partitioning and APCS RE.
9.5 References
1. Barton, R.G., W.D. Clark, and W.R. Seeker. (1990) “Fate of Metals in Waste Combustion Systems". Combustion Science and Technology. 74, 1-6, p. 327
