Authors & Contributors: Anthony P. Schneider, Patrick M. Cook, Jonathan A. Cross, Andrew M. Mertz, Denis Motovilov, Alejandra Ng-Feng, Weston P. Moyer, Alexandra Sipershteyn, Erik M. Tobin, Joseph Siperstein.
Sorbent trap measurements were essential for coal-fired utilities leading up to and through the implementation of Mercury and Air Toxics Standards (MATS), and continue to be essential for a wide variety of applications. They are widely used through EPA M30B1 and PS12B2 methodologies for regulatory compliance as well as engineering studies with the goal of studying and reducing mercury emissions. More recently, innovative sorbent technology has opened the door to easy, high quality measurements at both stack and upstream sampling locations for a variety of analytes which include Hg, SO3/H2SO4, NH3, HCl/HBr, As, and Se. These analytes are of interest because they are or are expected to be regulated in various ways and can represent or otherwise impact control technology performance.
Sorbent trap measurements are advantageous because Sorbent Trap Sampling Systems (STSS) are lightweight, versatile, and easy to use. A configured STSS is capable of measuring any of these analytes at almost any sampling location. EPA reference methods have served as inspiration for the development of sampling and analytical methodologies. Proper temperature control and flow rate are vital for successful measurements. Isokinetic sampling may be required for certain analytes/sampling locations and can be easily performed using nozzles designed for typical sorbent trap flow rates. Measurement quality is verified by EPA M30B/OTM-403 performance criteria which include Breakthrough (%BT), Pair Agreement (%RD), Spike Recovery (%R), and various sampling and analytical checks.
Since many experiments have been performed with these sorbent traps, two case studies are presented:
1. Measurement of Arsenic reduction by limestone addition for extended catalyst lifetime
2. Testing in-situ SO3 filtration techniques around the SCR
Experimental methods and results from these case studies will be presented and discussed.
Many of the EPA Emissions Test Methods, developed well before the turn of the century, were based on sample trains consisting of Greenburg-Smith impingers in series which contain solutions designed to absorb targeted pollutants out of the sample gas stream. Sampling impinger trains typically requires a team of technicians and significant mobilization time. Sampling locations are often difficult to access and unforgiving to the specialty glassware used in these manual methods. Impinger solutions may consist of hazardous materials which must be transferred and containerized on-site. Impinger train methods are prescribed, and if the procedures are not followed correctly, it is difficult to know that a deviation occurred without observation by a third party. The methods are cumbersome, but are well-tested and provide high quality results at stacks. Many of the method features such as filtration technique and temperature control were incorporated into the corresponding sorbent trap methods during their development.
Instrumental measurements in the form of both compliance CEMS and process monitors provide the advantage of instantaneous data which are useful for immediate detection of pollution control failures so that rapid corrections may be made. These data are valuable, but come with significant costs to cover installation, maintenance, and troubleshooting. Once installed, it is difficult to move the instrument to a different sampling location. Sorbent traps should not be used to replace instrumental monitors where necessary. Instrumentation installed in problematic sampling environments should be expected to require more maintenance and yield lower data availability. Some measurements require that the sample gas is extracted and transported through heated line to a distant instrument. These measurements are susceptible to scrubbing and loss of analyte during the delivery process, notably if the sample gas contains significant particulate and condensable particulate matter. Instrument performance should be verified by an appropriate technique which could include a gas-phase matrix spike, an accepted EPA reference method, or a sorbent trap method.
Innovative sorbent trap methods have been developed within the last few years with the primary objective of simplifying stack testing as well as engineering testing. Sorbent traps, tubes packed with sorbents for absorbing analytes of interest, are designed to be sampled in pairs in a probe that can be operated by a single technician. A lightweight sampling system is preferred so that it can easily be hauled to discrete locations and even moved between locations throughout the day which is especially useful for engineering tests with tight schedules. Sorbent traps are small, non-hazardous, disposable, and directly billable. All of the traps are designed for meeting the performance criteria set by EPA M30B, so that if a deviation from procedure occurs during either sampling or analysis, it may be easily detected by a QA/QC failure.
EPA has recently approved ALT-129 which allows using OTM-40 for quarterly compliance testing and RATA testing at sources without entrained water droplets under 40 CFR part 63 Subpart UUUUU. EPA’s publication of OTM-40 (based on EPA M30B) for HCl measurement using sorbent traps provided additional framework for ensuring measurement quality where Ion Chromatography (IC) is used for analysis. IC is used for HCl, HBr, SO3/H2SO4, and NH3 trap analysis. Hg sorbent traps are analyzed by Thermal Desorption Atomic Absorption Spectrometry (TD-AA).4 Arsenic and Selenium traps are digested and can be analyzed by any standard metals instrument including Graphite Furnace Atomic Absorption Spectrometry (GF-AA), Inductively Coupled Plasma Mass Spectrometry (ICP-MS), and Hydride Generation Atomic Fluorescence Spectrometry (HG-AF). HG-AF has been utilized most extensively thus far due to the low operating cost and minimal effect from matrix interferences.
Significant field testing has shown that sorbent traps are not only useful for various engineering studies, but also that they are likely to outperform other methods in terms of ease of use, data quality, and cost.
Table 1. Summarization of Sorbent Trap Method Characteristics
*Corresponds to the temperature below which an isokinetic sampling approach should be used, due to the expectation that the analyte will condense into an aerosol or other particulate form
**For each analyzed trap section
Selection of appropriate sampling locations is critical to any experimental test design. It is well-known that stratification can affect measurements of various flue gas constituents, and this should be kept in mind when selecting ports for sorbent trap sampling (it is best to avoid ports at the far edges of ductwork, and ports that immediately follow a sharp turn in ductwork).
If some sort of mitigation injection (activated carbon, DSI, etc.) is being used at the test site, it is important to know how far upstream that injection is, and how it may affect measurements, whether those measurements be done via sorbent traps or some other sampling method. Our field technicians have observed on more than one occasion experimental tests where an injection lance was placed a few feet upstream of a test method. It was not surprising when all measurement methods (sorbent traps and otherwise) were inundated with the injected material and unable to measure the analyte of interest.
Generally, experiments are designed to measure one or more flue gas constituents repeatedly, with each sample run occurring during some set of stable and unique sample conditions. These conditions are often changed between runs or between test days, in order to determine the effects these conditions have on the measured concentration of the analyte(s) of interest. The conditions that can be changed often include injection/addition rate of a material that is designed to adsorb, mitigate, or otherwise react with the flue gas constituent being measured. The experiment could also compare analyte concentrations at different unit loads, or different variants (manufacturers) of similar injection materials.
There are a variety of sorbent trap sampling systems (STSS) available on the market, but the model used for the experiments described in this paper is the OLM30B STSS (Figure 1). The sampling system consists of four primary components: sorbent trap probe, sample lines, desiccant canisters, and pump console. The probe (in its standard configuration) is six feet long, made of stainless steel, contains two sorbent trap liners designed for 10mm OD sorbent traps, houses a thermocouple for the measurement of stack gas temperature, is equipped with a 750W heater, and can be cooled via a cooling air inlet. Desiccant canisters contain calcium sulfate and sodium carbonate for the removal of water and acid gases, respectively. The pump console contains all electronics and power systems required to run the system, two mass flow controllers, and two pumps that can operate from 0-2 LPM. For many applications, an in-line thermocouple may be placed inside one of the trap liners to get an extremely accurate trap temperature measurement. Oxygen sensors are included in some OLM30B samplers for specific applications that require oxygen corrections.
Figure 1. OLM30B Sorbent Trap Sampling System
Inspiration from EPA Methods
Each of the sorbent trap methodologies developed by the investigators has been inspired by the Method 30B sorbent trap design (Figure 2).
Figure 2. Standard Method 30B Sorbent Trap
Method 30B includes a set of quality control criteria to validate measurements. These criteria include pass/fail thresholds on the calculated field spike recovery, pair agreement, and breakthrough for each run, in addition to a variety of other controls. In essence, the design of the method is such that when a measurement is obtained and all QC validations have passed, the user of the method can have great certainty that the measurement is accurate. This method design has been adapted to the variety of other sorbent trap methods that followed Method 30B, including HCl/HBr, SO3/H2SO4, NH3, As, and Se. Corresponding EPA methods for these analytes were used when considering sampling procedures including isokinetic sampling, filtration technique, filtration temperature, and more.
Controlling the temperature of a sorbent trap is of critical importance for all of the various sorbent trap methods, and the range of ideal sampling temperatures depends on the type of trap being sampled (refer to Table 1 in the introduction). In order to ensure accurate temperature measurement, we have designed an in-line thermocouple. This thermocouple sits directly behind one of the traps inside the trap liner, and is designed to not create any leaks in the sample train. Research has shown the standard trap temperature thermocouples built into most probes are unable to accurately measure true sorbent trap temperature when sampling extremely hot gas (>250°C) while simultaneously being exposed to cooling air. As all of our sorbent traps must be cooled in such environments, we found the only solution to be to expose a thermocouple directly to the sample gas immediately downstream of the sorbent trap. Sorbent traps often require a combination of heating (via probe heater) and cooling (via supplied air) in order to maintain a steady temperature. The supplied cooling air can come from a small electric pump for applications downstream of the air pre-heater or plant air for applications further upstream.
Some sorbent trap measurements may require the user to account for aerosols and other particulate-form or particulate-phase analytes. Isokinetic sampling is required to collect an amount of aerosols or particulate matter that is representative of what’s actually passing through the sample environment. This may be performed by using the isokinetic sampling technique described by EPA M55. Isokinetic nozzles are commercially available for typical sorbent trap flow rates.
For several sorbent trap sampling applications, it is necessary to prevent particulate matter from entering the sorbent trap. Particulate matter can often contain an interferent for the analyte being measured and would therefore bias the measurement high if included in the analysis. There are several options for particulate filtration, including sintered titanium filter tips, quartz tissue, static pre-filters, and inertial separators. The selection of an appropriate filtration technique is based on both sampling location and type of sorbent trap being sampled. For example, it would be acceptable to attach titanium filter tips to NH3 sorbent traps being sampled at an SCR outlet, but one should not use those same filter tips to sample SO3/H2SO4 sorbent traps, which have been found to cause excessive scrubbing in high particulate environments.
Oxygen and Moisture Corrections
Generally speaking, it is critical that both oxygen and moisture concentrations be accounted for when reporting measurements. This is especially true when comparing measurements acquired from multiple locations simultaneously. Sorbent trap results are always reported on a dry basis, so if comparing to a wet instrument method, a moisture correction factor must be applied. An oxygen sensor may be included in the OLM30B STSS described above, which operates via fluorescence quenching. Sorbent trap measurements are generally corrected to a 3% oxygen basis if required by the application.
RESULTS AND DISCUSSION
Case Study 1 – The Effects of Limestone Combustion on Arsenic
Arsenic concentration in flue gas plays a major role in determining the overall lifespan of a catalyst, as it is known to deactivate the catalyst by occupying active pore sites. More specifically, the As2O3 that is formed during the combustion of coal can react with oxygen on the vanadium compounds in the catalyst to form As2O5, which chemically bonds to the site.6
The addition of calcium to fuel (often via limestone feed) has been demonstrated to be an effective means of reducing the amount of As2O3 present in flue gas (via the formation of Ca3(AsO4)2), thereby slowing the rate of catalyst deactivation.7 This calcium-arsenic compound does not follow the same reaction pathway as As2O3 in the catalyst, but rather passes through the SCR to be captured downstream at the ESP and FGD.
Figure 3. Catalyst Life as a Function of CaO and Arsenic in Fuel
In this case study, arsenic sorbent traps were used to measure concentrations of total arsenic and vapor-phase arsenic at varying limestone feed rates in order to determine the effectiveness of limestone in reducing As2O3 concentrations and therefore extend catalyst lifespan. The correlation between limestone addition and vapor-phase arsenic removal is shown in Figure 4.
Figure 4. Limestone Addition vs. Vapor-phase Arsenic Removal
Method 29 is unable to reliably partition arsenic in the presence of high amounts of fly ash, such as upstream of ESPs where arsenic mitigation studies often take place. In this experiment, the investigators tested a novel sampling technique to differentiate between filterable and condensable arsenic.
The experiment presented in this case study tested the ability to partition arsenic by sampling both isokinetic sorbent traps and non-isokinetic filtered traps (sintered titanium filter tips were attached to the ends of the traps) using a quadruplet sampling system. This was done over the course of several days while limestone feed rates increased over time, thereby reducing the amount of As2O3 (condensable or gaseous) in the flue gas. See figures 5 and 6.
Figure 5. Quadruplet Sampling System used to Compare Isokinetic vs. Non-isokinetic Sampling Techniques.
Figure 6. Vapor-phase Measurements Obtained via Isokinetic and Non-isokinetic (filtered) Methods
Both isokinetic and non-isokinetic sampling techniques display the same downward trend in vapor-phase arsenic concentration (measured as the mass collected in the first sorbent trap section), but the isokinetic sampling method yields lower concentrations. The investigators suspect this is due to the interaction between captured particulate matter and vapor-phase arsenic, which could prevent the arsenic from reaching the first sorbent section. Therefore, the investigators conclude the best method for partitioning of arsenic measurements is to use the isokinetic method for measurement of total arsenic, while using the non-isokinetic sorbent traps (filtered via titanium mesh filter tips) to determine the vapor-phase component of the flue gas. Following this approach, it is relatively straightforward to determine percent removal of As2O3 as a function of limestone feed rate.
Case Study 2 – Testing in-situ Filtration Techniques around the SCR
SO3 is an undesirable compound which causes corrosion and fouling in control technologies and also contributes to opacity emissions. It forms in the boiler as well as across SCR catalyst layers where it is routinely measured. SO3 measurements around the SCR are considered difficult and susceptible to biases from the heavy particulate matter present in the sample matrix.8 The industry standard for SO3 testing is the Controlled Condensate Method (EPA 8A/CTM-013)9 which employs a heated filter to separate particulate and particulate bound sulfates from a condensate sample. Gas phase SO3 is highly reactive and can be scrubbed easily when forced through particulate caught in the filter, especially when the flue gas contains basic components such as calcium oxide or ammonia.
The investigators sought to test in-situ filtration techniques, where the particulate is removed from sample gas prior to extraction from the duct. This filtration concept removes the need for external filter temperature control and sample gas upstream of the Air-Preheater (APH) is usually significantly higher than the EPA CTM-013 recommended filtration temperature which is > 260 °C (500 °F). The basis for this recommended temperature is that all of the SO3 will be present in the gas phase rather than an aerosol (H2SO4) which could be filtered. However, gas-phase SO3 is likely to react with any of a number of compounds present in the particulate at any temperature.
Several filtration techniques were tested at two different sampling locations which are described below:
A large low-density plug was created from quartz wool and placed into the sample end of the sorbent trap. It was designed such that particulate could collect on the end of the filter, but there was not space inside the trap for a significant layer to collect. The tangential flow of the sample gas in the duct naturally cleaned particulate off of the filter. The material had a nominal pore size of 8 µm.
A static pre-filter was created in such a way that a stamped circle of high purity quartz filter was “sandwiched” between two quartz wool plugs. The stamped quartz filter was fabricated from the same material as the thimble.
A high-purity, high-density quartz thimble was placed between ground glass joints designed for a controlled condensate sample train and connected to the end of each sorbent trap. The vendor advertised a nominal pore size of 3.0 µm and 99.95% aerosol retention of 0.30 µm DOP.
CTM-013 was used as a comparison and utilized a high purity quartz thimble (same as the thimble described above) to filter sample gas. Filtration was not performed in-situ like the other techniques. Analysis was performed using the barium-thorin titration method.
Gas was sampled through a bent nozzle pointed away from the direction of oncoming flue gas flow to reduce particulate collection.
Two different sampling locations were chosen to test SO3 concentrations for various filtration techniques:
Sampling Location 1
An SCR outlet with high levels of entrained particulate matter. The gas temperature exceeded 300 °C. Dual probes were used to sample two different ports within several feet of each other.
Sampling Location 2
An SCR outlet with low levels of entrained particulate matter (after a hot-side ESP). The gas temperature exceeded 300 °C. One port was used and controlled condensate was sampled in between quad probe runs.
Figure 7. Location 1 - SO3 Concentrations by Filtration Techniques at an SCR outlet with High Levels of Particulate
Figure 8. Location 2 - SO3 Concentrations by Filtration Technique at an SCR Outlet with Low Levels of Particulate
The traps were visually inspected after they were sampled. There were no signs of visible fly ash inside the traps except for the ones that were sampled via the anti-kinetic method. Some of the traps’ front plugs had a beige tint and some of the traps showed visible condensation on the inside of the tube.
All of the sorbent traps were sampled in pairs, which served as a measure of precision for each of the filtration techniques. The thimble filtration technique performed exceptionally poorly in terms of pair agreement.
A clear bias exists between the filtration techniques. There are two potential explanations for the bias:
Quartz wool did not effectively capture sulfate-containing particulate that was eventually captured in section 1 of the sorbent trap, where it was measured as SO3 and caused a high measurement bias
This implies that a very large amount of sulfate salts were present in particulate that was not visible to the naked eye
SO3 was scrubbed across the surface of high-purity quartz thimble material
The poor pair agreement (>10% RD) for the thimble filtered traps may have correlated to inconsistent scrubbing
The thimble filter may be likely to scrub since it is a rigid, 2-dimensional material that sample gas is forced through. In contrast, quartz wool is a 3-dimensional filter where particulate may be captured while still allowing micro-channels of sample gas to pass through unreacted. In the high particulate sampling environment, there was relatively poor precision for all of the filtration techniques except for the static pre-filter. The low biases, along with poor pair agreement exhibited by all of the other sampling methods, seem to indicate that SO3 interacts with fly ash (or other components). The physical characteristics of the static pre-filters seem to result in a weaker interaction between the fly ash and SO3.
Innovative sorbent traps have opened the door to economic and high quality data acquisition that can be used not only at the stack, but also upstream in traditionally difficult-to-test environments. Sampling methodology has been inspired by EPA and industry accepted methods, and is continuing to improve. These types of measurements have the potential to help coal-fired power plants operate more efficiently and reduce long-term operating costs.
U. S. Environmental Protection Agency Method 30B, Revised 8/2/2017: Determination of Total Vapor Phase Mercury Emissions from Coal-fired Combustion Sources using Carbon Sorbent Traps, Research Triangle Park, NC 2017.
U. S. Environmental Protection Agency Performance Specification 12B, Revised 8/7/2017: Specifications and Test Procedures for Monitoring Total Vapor Phase Mercury Emissions from Stationary Sources using a Sorbent Trap Monitoring System, Research Triangle Park, NC 2017.
U. S. Environmental Protection Agency. Other Test Method 40, Revised 5/31/2018: Determination of Hydrogen Chloride Emissions from Coal-fired Combustion Sources using Sorbent Traps, Research Triangle Park, NC 2018.
U.S. Environmental Protection Agency. 1998. "Method 7473 (SW-846): Mercury in Solids and Solutions by Thermal Decomposition, Amalgamation, and Atomic Absorption Spectrophotometry," Revision 0. Washington, DC.
U.S. Environmental Protection Agency, Revised 8/2/2017. . Determination of Particulate Matter Emissions from Stationary Sources, Research Triangle Park, NC 2017.
Staudt, J. (Andover Technology Partners), T. Engelmeyer (Orlando Utilities Commission), W. Weston (Alabama Power Company), and R. Sigling (Siemens Westinghouse Power Corporation). The Impact of Arsenic on Coal Fired Power Plants Equipped with SCR. Institute of Clean Air Companies (ICAC) Forum 2002. February 2002.
Ake, T., C. Erickson, W. Medeiros (Riley Power Inc.), L. Hutcheson (Duke Energy), and M. Barger and S. Rutherford (Cormetech, Inc.). Limestone Injection for Protection of SCR Catalyst. Mega Symposium. May 2003.
R.K. Srivastava , C.A. Miller , C. Erickson & R. Jambhekar (2004) Emissions of Sulfur Trioxide from Coal-Fired Power Plants, Journal of the Air & Waste Management Association, 54:6, 750-762, DOI: 10.1080/10473289.2004.10470943.
U. S. Environmental Protection Agency. Method 8A, Revised December 1996: Determination of Sulfuric Acid Vapor or Mist and Sulfur Dioxide Emissions from Kraft Recovery Furnaces, Research Triangle Park, NC 1997.