|Back to HWF Library Page||Gossman Consulting, Inc.|
GCI wishes to gratefully acknowledge the support provided by Lone Star Industries in the preparation of this month's HWF Notes.
A number of EPA regional authorities have interpreted the BIF regulations to require that each batch of hazardous waste fuel used in a cement kiln must be analyzed for all 10 BIF metals prior to use as a fuel. Presumably, this batch by batch analysis would be performed using EPA SW-846 methods. This interpretation appears to contradict EPA guidance provided in Section 6.1 of the Technical Implementation Manual,
"Until an operating permit is issued, all fuels, raw material feedstocks, and waste materials fed to the combustion device must be routinely analyzed (as often as necessary)..... to ensure that the BIF operates within these limits...."
"At a minimum, the analyses must be repeated: .....(2) for facilities that accept hazardous waste generated off site, whenever the results of a "fingerprint" analysis indicate that the incoming hazardous waste shipment does not match the description on the manifest or shipping paper."
The clear implication of these citations is that analyses of every batch of HWF prior to burning is not required, nor was it anticipated by EPA.
An alternative methodology for compliance with BIF metal input limits has been used by many facilities. This approach consists of periodic analyses of composite and/or grab samples coupled with statistical analyses of the data along with more thorough and accurate methods of analysis.
Batch by Batch Analysis - The Problem
Generally speaking, EPA regional as well as state officials rely heavily on SW-846 as the sole source for waste analysis methods. The BIF regulations specifically require that metal testing of hazardous waste be performed to the detection limits available using SW-846 methods. In addition, the Technical Implementation Document says "The waste sampling and analysis methods should be SW-846 methods where applicable. Alternate methods may be used provided the methods meet or exceed the comparable SW-846 performance criteria." All this despite the fact that SW-846 methods were not designed to meet these regulatory objectives, nor to analyze these sample matrices. Specifically:
"SW-846 performance criteria" do not exist for metal determinations on sample matrices found as inputs to cement kilns.
SW-846 sample preparation methods are not vigorous enough to digest many sample matrices and therefore do not provide accurate determinations of total metals. SW-846 methods were designed primarily for aqueous matrices and are derived directly from previously established EPA waste water methods.
EPA in the preamble to the 3rd edition of SW-846 acknowledges that "Sample preparation methods are not currently available in SW-846 to render non-aqueous liquids in a form that can be analyzed by the atomic absorption or inductively coupled plasma atomic emissions (ICP) type analytical methods for six important elements. These elements are: mercury, arsenic, selenium, lead, barium, and silver."
The apparent lack of familiarity with SW-846 by those who developed the BIF regulations as well as regional and state authorities has provided a very confusing and difficult environment in which industry must attempt to comply. For example, since EPA requires metal testing for only those metals which SW-846 methods can detect and validated methods do not exist in SW-846, perhaps it follows that metal testing is not required until SW-846 contains validated methods for the organic matrix. Or perhaps it is acceptable to use SW-846 methods knowing full well that quantities of metals in excess of BIF limits could be put into the kilns because SW-846 methods will not detect them. (Use of SW-846 has been the common past practice at certain commercial hazardous waste incinerators. None of the commercial hazardous waste incinerators contacted perform batch by batch metals analyses on waste incinerated.)
Many in the industry have chosen not to follow either of these strict interpretations, but have worked to develop and utilize techniques that determine true total metals in the input matrices of cement kilns. As examples:
Many facilities have developed and used microwave techniques to enhance the digestion potential while still utilizing reagents called for by EPA SW-846.
At least nine compliance tests nation-wide utilized ASTM E926 Method A for digestion to ensure the accurate determination of total metals. This digestion technique utilizes nitric, hydrofluoric and perchloric acids; and requires 24-48 hours to perform. Some facilities, despite the cost and hazards inherent in this procedure, continue to use it for compliance purposes.
Systech's Bruce Pederson has developed and published at last spring's AWMA BIF conference an enhanced microwave digestion technique. GCI and General Engineering Labs have studied and shown this procedure to be comparable to the previously mentioned ASTM technique.
A number of facilities have experimented with x-ray fluorescence (XRF) technology. XRF has proven to be much faster and more accurate than SW-846 methods. XRF is particularly attractive for those facilities choosing to screen all incoming shipments for selected BIF metals.
Problems with SW-846 methods are not limited to digestion techniques. Many labs have experienced difficulty using ICP for determining selected BIF metals. Thallium, arsenic and lead have produced false highs from apparent spectral interferences. The extremely high background concentrations of calcium, silica, aluminum and iron in kiln feed matrices produces numerous spectral interferences which can affect detection limits or require the use of alternate procedures.
Facilities have been faced with a significant dilemma in trying to deal with a confusing and self-contradictory set of requirements. One option was to equip for and perform metal analysis knowing full well that these determinations were inaccurate and that sooner or later EPA would realize the inaccuracy of the existing SW-846 methods and require more accurate methods to be utilized.
The second option was to utilize more thorough and accurate techniques of analysis with corporate and/or commercial labs coupled with statistical data analysis and comfortably high operating limits. This option then allows those companies to more carefully evaluate existing and developing technologies prior to implementation at the plant level without compromising compliance objectives and criteria.
The Case for the Statistical Approach
Comparisons of SW-846 methodology with the previously described ASTM E926 method show that SW-846 recoveries of BIF metals varies from 10% to 70% depending on the metal and matrix. This produces relative errors of 50% to 1000%.
The statistical approach to compliance involves the step-wise development of compliance limits and periodic verification of continual compliance. Initially, a statistical analysis of BIF metal concentrations in raw materials and non-waste fuels is performed. The mean plus two standard deviations is used to establish minimum input levels from these sources. Then either Tier IA or best engineering judgment for Tier III metals is used to back calculate maximum HWF metal input levels with adjustments made to prevent TCLP problems in CKD. These maximum HWF metal input levels provide the basis for metal spiking during the compliance test. Actual metal input levels relative to this procedure for setting maximum input limits are generally one or more orders of magnitude below limits. This provides a significant safety margin even when metal concentrations vary. Periodic testing of raw materials, non-waste fuel and HWF is used to demonstrate continued compliance with a high degree of confidence. Another one or more codes of magnitude safety margin exists between actual emission rates and emission limits as demonstrated in recent compliance testing.
In contrast with using SW-846 procedures on every batch of HWF, the statistical procedure is designed to err on the high side. Clearly use of the EPA procedure will err on the low side and could result in undetected excursions above input limits.
If comfortably high compliance limits are established, either of the previously discussed compliance procedures will likely prevent excursions above metal input limits. Nevertheless, the statistical approach produces more accurate values for individual analyses and is designed to err on the high side in demonstrating compliance. The statistical approach therefore provides the more conservative (technically) approach to compliance as well as providing better data for the long term development of understanding and better compliance procedures.