Results of Activated Carbon Injection Upstream of Electrostatic Precipitators for Mercury Control Paper No. Travis Starns, Jean Bustard, Michael Durham Ph.D., Cam Martin, Richard Schlager, Sharon Sjostrom, Charles Lindsey, Brian Donnelly ADA Environmental Solutions, LLC 8100 SouthPark Way, Suite B-2, Littleton, CO 80120 303-734-1727, 303-734-0330 (Fax) Rui Afonso Energy and Environmental Strategies 50 Old Faith Road, Shrewsbury, MA 01545 Ramsay Chang, Ph.D. EPRI 3412 Hillview Ave., Palo Alto, CA 94304-1395 Scott Renninger U.S. Department of Energy National Energy Technology Laboratory 3610 Collins Ferry Rd., P.O. Box 880, Morgantown, WV 26507-0880 ABSTRACT Under a cooperative agreement with the Department of Energy National Technology Laboratory (NETL), ADA-ES has conducted a series of tests evaluating the performance of activated carbon injected (ACI) upstream of electrostatic precipitators (ESPs) for mercury control and the effect of changing operations such as temperature and LOI on natural mercury control. Host site configurations included units burning both Powder River Basin and low sulfur bituminous coals and sites that had high and low natural mercury removal. Results from this test program have shown the viability and limitations of using ACI for mercury control with ESPs. This paper will include results from Wisconsin Electric Pleasant Prairie, PG&E NEG Brayton Point, and the recently completed results from PG&E National Energy Groups Salem Harbor Station. This site provided key information on the importance of temperature and LOI in mercury removal with and without ACI. INTRODUCTION In December 2000 EPA announced their intent to regulate mercury emissions from the nations coal-fired power plants. Draft legislation indicates that new regulations may require removal efficiencies as low as 50% or as high as 90% from existing sources. Estimates for the cost of meeting mercury regulations at the level of 90% removal efficiency range from $2 to $5 billion

2

per year. With mercury regulations imminent, mercury control technologies need to be proven at full scale to document performance and costs. The most mature, retrofit technology available today is the injection of sorbents such as powdered activated carbon (PAC) into the flue gas upstream of the particle control equipment. The gas-phase mercury in the flue gas contacts the sorbent and attaches to its surface. Existing particle control equipment collects the sorbent with mercury attached along with the fly ash. Under a DOE/NETL cooperative agreement, ADA-ES is working in partnership with PG&E National Energy Group (NEG), Wisconsin Electric, a subsidiary of Wisconsin Energy Corp., Alabama Power Company, a subsidiary of Southern Company, and EPRI on a field test program of sorbent injection upstream of existing particle control devices for mercury control1. Other cost share partners include: Tennessee Valley Authority, Ontario Power Generation, Kennecott Energy Company, Hamon Research-Cottrell and Arch Coal. The test program, which took place at four different sites during 2001 and 2002, is described in detail in the July 2001 EM Journal2. Testing at all four sites has been completed and this paper focuses on the three sites that utilize electrostatic precipitators for their particulate control devices. Results for the full-scale evaluation at Alabama Power’s Gaston Unit 3 are documented in the A&WMA Special Edition Journal, published in June 2002. Results from the full-scale evaluation conducted at Wisconsin Energies Pleasant Prairie was detailed in a paper presented at AWMA’s 2002 annual meeting. This paper presents detailed results from testing at PG&E NEG’s Brayton Point and Salem Harbor Stations. Semi-Continuous Mercury Analyzer Near real-time vapor phase mercury measurements were made using a Semi-Continuous Emissions Monitor (S-CEM) designed and operated by Apogee Scientific. This instrument was developed with EPRI funding to facilitate EPRI research and development efforts. Two analyzers were dedicated to the program. They were set up at the inlet and outlet of the PCD. The S-CEMs operate continuously throughout the testing period, providing speciated, vapor phase mercury concentrations. This particular analyzer employs a wet-chemical system to either convert all vapor-phase mercury species to elemental mercury or to remove the oxidized portion to measure the elemental mercury fraction. The sample is then measured for mercury content by utilizing a cold vapor atomic absorption spectrometer (CVAAS).3 A further detailed description of the mercury S-CEM can be found in previous publications.

3

Figure 1. Process Sketch of Mercury Semi-Continuous Analyzer

Chilled Impingers

Flue Gas

Waste

Carbon Trap

CVAA

Mass Flow Controller

Gold Trap

Waste

Timed12V, 5A

Micro controller with Display

Purge Air

PG&E NEG Brayton Point & Salem Harbor Test Objectives The primary testing objective at these sites was to determine the cost and impacts of sorbent injection into the cold ESP for mercury control. The specific objectives for Brayton Point included documenting performance at several injection concentrations, the performance of sorbents supplied by various manufacturers, the ability of activated carbon to capture mercury in-flight, and the effect of different injection lances designed to increase the effective spray coverage in the duct. The test objectives for Salem Harbor included sorbent injection and it’s impacts on the ESP for mercury control as well as modification to process variables and documenting their role on mercury capture. These process variables were flue gas temperature, LOI, and SNCR.

PG&E NEG Brayton Point Site Description PG&E National Energy Group owns and operates Brayton Point Station located in Somerset, Massachusetts. There are four fossil fuel fired units at the facility designated as Units 1, 2, 3, and 4. In 1982, three of the four units, (Units 1, 2, and 3) were converted from oil to coal. The units fire a low sulfur, bituminous coal. Unit 1, which is scheduled to be the test unit, has a tangentially fired boiler rated at 245 net MW. The primary particulate control equipment consists of two cold-side ESP’s in series, with an EPRICON flue gas conditioning system that provides SO3 for fly ash resistivity control. The EPRICON system is not used continuously, but on an as-needed basis. The first ESP (Old ESP) in this particular configuration was designed and manufactured by Koppers. The Koppers ESP has a weighted wire design and a specific collection area (SCA) of 156 ft2/1000 acfm. The second ESP (New ESP) in the series configuration was designed and manufactured by Research-Cottrell. The second ESP has a rigid electrode design and an SCA of 403 ft2/1000 acfm. Total SCA for the unit is 559 ft2/1000 acfm. The precipitator inlet gas temperature is nominally about 280°F at full load.

4

The first precipitator consists of four parallel chambers each with 28 gas passages 24’ long at 10” centers. Each chamber is further divided into three collecting surface fields. The first ESP has a total of 12 T/R sets. The second precipitator consists of two parallel chambers. Each chamber is subdivided into 38 gas passages 54’ long at 12” centers. The chambers are then divided into six collecting surface fields. The second ESP contains a total of 24 T/R sets. Hopper ash is combined between both precipitators in the dry ash-pull system. The ash is processed by an on-site STI carbon separation system, to reduce the carbon content to approximately 2%. This processed ash is sold as base for concrete and is considered a valuable product for the Brayton Point Station. The remainder of the higher carbon ash is a disposable waste. One precipitator’s ash can be isolated from the balance of the unit, however this is a labor-intensive procedure. A summary of important descriptive parameters for Brayton Point Unit 1 is presented in Table 1. Table 1. Site Description Summary, Brayton Point Unit 1

PARAMETER IDENTIFICATION DESCRIPTION Process Boiler Manufacturer C-E tangential, twin furnace Burner Type C-E LNCFS III (32 burners) Low NOx Burners Yes Steam Coils Yes Over Fire Air Yes NOx Control (Post Combustion) None Temperature (APH Outlet) 280oF Coal Type Eastern Bituminous Heating Value (Btu/lb) 12,319 Moisture (%) 6.6 Sulfur (%) 0.72 Ash (%) 11.32 Hg (µg/g) 0.05 Cl (%) 0.08 Control Device Type Cold-Side ESPs in series ESP #1 Manufacturer Koppers Design Weighted Wire Specific Collection Area (ft2/1000afcm) 156 ESP #2 Manufacturer Research Cottrell Design Rigid Electrode Specific Collection Area (ft2/1000afcm) 403 Flue Gas Conditioning SO3 Injection, EPRICON

5

Sorbent for mercury control was injected into the ductwork in between the two electrostatic precipitators. Only one of the two inlet precipitator ducts was treated, nominally 125 MW. This met DOE’s requirement to evaluate units up to 150 MW. This particular ESP configuration allowed the Hg team to measure in-flight mercury removal efficiency between the two ESP’s and mercury removal efficiency across each ESP.

Figure 2 presents a diagram of the particulate control equipment at Brayton Point. This figure shows that each unit has two ESP’s in series. Figure 2. Isometric View of Precipitator Arrangement at Brayton Point Unit 1

Hg S-CEM

6

Brayton Point Test Results The field-tests were broken down into four distinct phases during the full-scale evaluation at Brayton Point:

• Sorbent Screening; • Baseline; • Parametric; and • Long-Term.

Each of these phases is described in the subsequent paragraphs below. Sorbent Screening The first phase of field-testing was the sorbent screening process conducted by URS Corporation in February 2002. Sorbents are screened using a fixed bed of carbon to measure the sorbents mercury adsorption capacity. The mercury adsorption tests were carried out on a slipstream of flue gas extracted from upstream of the first precipitator, with and without SO3 injection. Eight coal derived sorbents, two fly ash, one tire derived sorbent, and one zeolite based sorbent were each tested at a temperature of 275°F. The major conclusions from the fixed-bed tests were:

• Carbons are capable of achieving high mercury capacities in Brayton Point Unit 1 flue gas;

• SO3 appears to inhibit carbon adsorption and with certain sorbents decreased the adsorption capacity to zero. With the activated carbon products, the presence of SO3 in the flue gas decreased the adsorption capacity in some cases by a factor of six, however the measured adsorption capacity was still above the threshold capacity (nominally 150 µg/g for an ESP). Therefore performance of these sorbents should not be impacted;

• Only one of the fly ash based sorbents tested showed an adsorption capacity greater than 150 µg/g;

• The zeolite based sorbent showed a low adsorption capacity in the Brayton Point flue gas, thus, this particular sorbent was not chosen for full-scale testing.

Using the results from the fixed-bed tests as one of the selection criteria, five sorbents were selected for the full-scale test. Mercury adsorption capacity measured during the screening tests are presented in Table 2.

7

Table 2. Results of fixed bed screening tests by URS at Brayton Point

Sorbent/Supplier Base

Ads. Cap SO3 Off @ 50 µg/Nm3

Ads. Cap SO3 On @ 50 µg/Nm3

FGD/Norit Americas Lignite 4314 1380 HOK300/Donau Carbon Lignite 4786 CC/CarboChem Bituminous 1948 SAI-B/Superior Adsorbents Activated Carbon 1799 EPRI – LAC * Activated Carbon 2196** * Data Supplied by URS Corporation ** Number calculated using a different flue gas composition Baseline Testing Series After equipment installation and checkout, a set of baseline tests was conducted the week of June 6-7, 2002. During these tests, Unit 1 boiler load was held steady at “full-load” conditions during testing hours, nominally 8:00 am to 6:00 pm. Both the S-CEMs and the modified Ontario Hydro Method were used to measure mercury across both ESPs. The S-CEMs were strategically placed throughout the system, and the Ontario Hydro tests were conducted at the inlet to the first ESP and the outlet of the second ESP. Previous Ontario Hydro measurements conducted at Brayton Point’s Unit 1 have shown native mercury removal efficiencies ranging from 29%- 75%. Results from the Ontario Hydro tests conducted during the baseline testing series indicate an average native mercury removal efficiency of 90.8%. These results can be seen in Table 3. Table 3. Results from Baseline Testing Series – Brayton Point Unit 1 June, 2002 Location Particle Bound Oxidized Hg2+ Elemental Hg0 Total, Hg Inlet - Loc 1 (µg/dncm) 3.5 0.20 <0.16 3.9 Outlet - Loc 4 (µg/dncm) <0.005 0.14 0.23 0.36 RE (%) 99.8 30.0 43.8 90.8 The mercury removal efficiency of 90.8% is high when compared to historical data for Brayton Point Unit 1. Parametric Testing Series A series of parametric tests was conducted to determine the optimum operating conditions for several levels of mercury control. During this particular series, the primary variables that were tested included injection concentration, sorbent type, and SO3 flue gas conditioning on/off. In all, 20 different parametric conditions were tested. A summary of the parametric tests is presented in Table 4. Standard testing conditions were with Unit 1 boiler at full load operation and the EPRICON flue gas conditioning system on. Exceptions to the standard conditions are

8

noted in the table. Each condition was tested for a minimum of six hours, or until a state of equilibrium had been reached. Table 4. Summary of Parametric Testing Conditions – Brayton Point

Test Series Carbon Name

Target Injection Concentration(lbs/MMacf) Testing Conditions

1 - 4 FGD 1, 3, 10, 20 Standard 5 - 6 SAI-B 3, 10 Standard 7 - 8 SAI-B 10, 20 Multiple Nozzle Lance 9 - 11 CC 3, 10, 20 Standard 12 - 14 HOK300 3, 10, 20 Standard 15 - 16 FGD 10, 20 EPRICON Off 17 - 18 FGD 7, 15 Standard 19 - 20 LAC 3, 10 EPRICON Off

A summary of results from all the parametric tests is presented on Figure 3. This figure plots mercury removal efficiency as a function of sorbent injection concentration. It is important to emphasize that this graph represents the mercury capture across the second ESP. This is incremental mercury capture that is being measured independent of the baseline mercury capture that is happening across the first ESP. The different symbols represent different test conditions including carbon type and SO3 off (EPRICON). This graph shows that there was a direct relationship between the Hg removal efficiency and sorbent injection concentration. Hg removal efficiencies ranged from 2-35% at the lower sorbent injection concentration of 3 lbs/MMacf to 75-93% at the highest sorbent injection concentration of 20 lbs/MMacf. Figure 3. Mercury Removal Trends Across Second ESP – Parametric Testing Summary 0

10

20

30

40

50

60

70

80

90

100

0 5 10 15 20 25

Injection Concentration (lb/MMacf)

Hg

Rem

ova

l (%

) FGD, New ESPSAI, New ESPCC, New ESPHOK, New ESPFGD no SO3, New ESPFGD repeat, newLAC, New ESP

9

One of the significant observations made during the parametric tests was that mercury removal increased with higher injection concentrations. This was in contrast with what was observed at Pleasant Prairie where increasing the sorbent injection concentration above nominally 10 lb/MMacf resulted in almost no increase in mercury removal. Researchers have observed that very low concentrations of HCl in the flue gas is required for standard activated carbon to effectively remove elemental mercury.6 The activated carbon sorbent is designed to adsorb contaminants in the flue gas whether they be vapor phase mercury, sulfur dioxides, or gaseous HCl. At Pleasant Prairie, where gaseous HCl concentrations are less than 1ppm, once all of the HCl in the flue gas was adsorbed by the activated carbon, the effectiveness of activated carbon to capture elemental mercury was greatly reduced. This could help explain the ceiling phenomenon seen at Pleasant Prairie where the mercury removal efficiencies did not increase when sorbent injection concentrations increased above 10 lbs/MMacf. Theory suggests that oxidized mercury adsorption is not as sensitive to the presence of HCl in the flue gas. At Brayton Point the predominant species of mercury is in the oxidized form, in contrast to Pleasant Prairie where the majority of vapor phase mercury was in the elemental form, and there is a significant amount of HCl present in the flue gas. These two factors create an environment in the flue gas for activated carbon to capture both forms of mercury; oxidized and elemental, at all injection concentrations. Thus, as can be seen in Figure 4 increasing activated carbon injection increases the amount of mercury capture. Activated Carbon Injection Summary for PPPP and Brayton Point The benchmark sorbent, Norit Americas Darco FGD activated carbon, was tested at all three sites at the same injection concentrations. Figure 4 shows mercury removal trends seen at two of the three sites described herein. Data from the Salem Harbor evaluation was omitted because activated carbon injection showed no additional increase in the capture of mercury. Salem Harbor currently experiences a >90% native mercury removal efficiency. In Figure 4 one can see that below an injection concentration of 10 lbs/MMacf, that activated carbon injection performs better in terms of mercury capture at Pleasant Prairie when compared to Brayton Point. This difference is likely attributed to the significant amount of HCl present in the Brayton Point flue gas. The HCl competes with the mercury for surface vacancies on the activated carbon particle, thus reducing the carbon’s ability to capture gaseous mercury. It is commonly accepted that once activated carbon injection concentration requirements are greater than 10 lbs/MMacf, the economics show that it is better to install a baghouse to increase the mercury capture. The injection concentration, and sorbent costs, to achieve 90% mercury control with a baghouse is nearly a tenth of what would be required for an ESP.

10

Figure 4. Mercury Removal Trends for Pleasant Prairie And Brayton Point

Another important observation made during the parametric testing series was the sorbents ability to capture the mercury “in-flight” in a relatively short amount of time. During testing, activated carbon was injected into the duct just upstream of the Hg S-CEM location 3, which can be seen in Figure 2. The distance between the sorbent injection location and location 3 was approximately 24 feet. This allowed for a residence time of < 0.5 seconds between the activated carbon particle and the gaseous mercury. Measurements from location 3 indicate that a vast majority of the mercury captured across the second precipitator was in-flight. The data presented in Table 5 shows in-flight removal efficiencies versus removal efficiencies measured across the second precipitator. Table 5. In-Flight Mercury Capture – Brayton Point Parametric Testing Series

Sorbent

Injection Concentration (lbs/MMacf)

Mercury Removal Efficiency (%) In-Flight

Mercury Removal Efficiency (%) Across Second ESP

FGD 3 22 26 FGD 10 68 73 FGD 20 87 93 SAI 10 58 62 SAI 20 85 85 CC 10 41 55 CC 20 68 86 HOK300 10 45 58 HOK300 20 61 75

0

10

20

30

40

50

60

70

80

90

100

0 5 10 15 20 25 30

Injection Concentration (lbs/Mmacf)

% H

g R

emo

val

ESP Bit

ESP PRB

11

EPRI LAC 10 61 72 Long-Term Testing Series The long-term testing series was conducted over a 10 day period in which two different sorbent injection concentrations were tested. The first five days of testing called for injecting the DARCO FGD product continuously at an injection concentration of 10 lbs/MMacf. Upon completion of this test, the injection concentration was immediately increased to 20 lbs/MMacf. This test condition was maintained throughout the rest of the long-term testing series. During both periods of testing, vapor phase mercury was measured with the Hg S-CEMs. To verify results, mercury was also measured by the draft Ontario Hydro method conducted by TRC Environmental Corporation. In addition to the mercury measurements, metals in the second ESP exhaust were sampled using an EPA Method 29 sampling train. The results from the Ontario Hydros conducted during the long-term period in which sorbent was continuously injected at 10 lbs/MMacf are presented in Table 6. Table 6. Results from the Ontario Hydros 10 lb/MMacf – Brayton Point Location Particle Bound Oxidized Hg2+ Elemental Hg0 Total, Hg Inlet - Loc 1 (µg/dncm) 5.6 0.60 <0.10 6.5 Outlet - Loc 4 (µg/dncm) <0.03 0.19 0.50 0.36 RE (%) 99.5 68.1 -396 94.5 Ash Characterization Coal and fly ash samples were collected daily. Ultimate and Proximate analyses of selected coal samples were performed by Microbeam Technologies. Mercury content from the coal was used to predict total gaseous mercury content in the flue gas. During the long-term testing series, the total gaseous mercury content in the flue gas was estimated to be approximately 7 µg/dncm, which corresponds to the amount measured by the Ontario Hydros as seen in Table 6 above. Ash analyses performed on the samples collected included:

• LOI; • Mercury; • Leaching (TCLP and SGLP); • Chlorine; and • Alkali and alkaline earth elements.

Results from the analyses indicate:

12

• The amount of LOI in the ash appears to correlate with the mercury or chlorine contents of the ash for the second ESP ash.

• Results from both the leaching protocols (TCLP and SGLP) indicate the amount of mercury leached from the samples were about 100 times lower than the primary drinking water standard.

• As expected, the mercury content in the ash appears to increase with increased activated carbon injection.

• Preliminary analysis indicates there may be a relationship between the alkali/alkaline earth content and the chlorine content in the ash collected from the first ESP. Additional samples need to be collected and analyzed to support a solid conclusion.

ESP Performance ESP performance during carbon injection was closely monitored. Power levels and stack opacity were the primary performance parameters. Figure 5 shows the total power level of the second ESP including a baseline value three days before the start of the long-term testing series. No significant changes were observed in either the power levels, Figure 5, or -in stack opacity. Figure 5. Total Power of the Second ESP – Brayton Point Unit 1 PG&E NEG Salem Harbor Site Description PG&E National Energy Group owns and operates Salem Harbor Station located in Salem, Massachusetts. There are four fossil fuel fired units at the facility designated as Units 1, 2, 3, and 4. Units 1-3 fire a low sulfur, bituminous coal and use oil for startup. Unit 4 fires #6 fuel oil. Unit 1, which is scheduled to be the test unit, is a B&W single-wall-fired unit with twelve DB Riley CCV-90 burners. It is rated at 88 gross MW. The particulate control equipment consists of a two-chamber, cold-side ESP (chambers designated 1-1 and 1-2), which provides two separate gas flow paths from the outlet of the tubular air heaters to the ID fan inlets. This Environmental Elements ESP has a rigid electrode design and a specific collection area (SCA) of 474 ft2/1000 acfm. The precipitator inlet gas temperature is nominally 295ºF at full load.

0

50

100

150

200

250

300

7/12/02 0:00 7/14/02 12:00 7/17/02 0:00 7/19/02 12:00 7/22/02 0:00

ES

P T

ota

l P

ow

er

0

10

20

30

40

Inje

ctio

n C

on

cen

trat

ion

(lb

s/M

Mac

f)

13

There are eight electrical fields in the direction of flow, and two across. The discharge electrodes are 44.5 feet in length and are spaced 18” apart in the direction of gas flow. There are eight precipitator ash hoppers on Unit 1, four in the direction of flow and two across. A pneumatic conveying system ties into each hopper and blows dry ash into the fly ash storage silo, where it is combined with flyash from the ESPs, economizer hoppers, and air preheater hoppers from Units 1, 2 and 3. Both wet and dry unloading systems are available to feed the ash from the fly ash storage silo into a truck. Typical LOI / carbon content of the Unit 1 ash is about 25%. This ash is land filled. A summary of important descriptive parameters for Salem Harbor Unit 1 is presented in Table 7. Table 7. Site Description Summary, Salem Harbor Unit 1

PARAMETER IDENTIFICATION DESCRIPTION Process Boiler Manufacturer

B&W 85 MW Radiant Boiler

Burner Type DB Riley CCV-90 Low NOx Burners Yes Steam Coils Yes Over Fire Air No NOx Control (Post Combustion) SNCR Temperature (APH Outlet) 295 Coal Type South American Bituminous Heating Value (Btu/lb) 12701 Moisture (%) 9.64 Sulfur (%) 0.63 Ash (%) 3.92 Hg (µg/g) 0.03 Cl (µg/g) 206 Control Device Type Cold-Side ESP ESP Manufacturer Environmental Elements Design Cold-Side, Rigid-Electrode Specific Collection Area (ft2/1000afcm) 474 Flue Gas Conditioning None

Four test locations were used for mercury measurements on Salem Harbor Unit 1: the economizer outlet, air preheater exit, the ESP inlet, and the ESP outlet (ID fan inlet). This allowed for mercury measurements at various locations throughout the system, including mercury removal across the ESP.

14

The parametric test series at Salem Harbor is unique in its focus on process variables, in addition to injection of sorbent. Process variables of interest were temperature, SNCR on/off, and LOI/carbon content in the ash. These variables were successfully tested to further broaden the knowledge base for mercury control using sorbent injection and to help determine the cause for a high native mercury removal capture experienced at Salem Harbor. Figure 6. Flue Gas Travel and Equipment Arrangement at Salem Harbor

15

Figure 7. Plan View of Mercury Measurement Locations – Salem Harbor

Salem Harbor Test Results The field-tests at Salem Harbor were broken down into three different phases during the full-scale evaluation:

• Baseline • Parametric • Long-Term

Each of these phases is described in the subsequent sections below. Baseline Testing Series During the baseline, there was no sorbent injection and Unit 1 was held at full-load (~ 86 MW) under normal operating conditions. Mercury S-CEM measurements were made at locations 1, 2, and 3, which can be seen in Figure 7. In addition to the Hg S-CEM, TRC Environmental Corporation also conducted manual measurements of mercury following the draft Ontario Hydro method. The triplicate runs of the Ontario Hydro testing method were conducted at locations 1 and 3. During the baseline series, vapor phase Hg levels at location #1 ranged from approximately 2 µg/dNm3 to 6 µg/dNm3. Coal samples were taken daily and sent to a lab for analysis, including Hg content. At the outlet location (location #3) vapor phase Hg levels were in the approximate range of 0.3 – 1.2 µg/dNm3. These mercury measurements from the Hg S-CEMs were compared to results from the Ontario Hydros. The data collected from the Ontario Hydros are presented in Table 1. Inlet mercury levels measured by the two methods were significantly different. This difference is due to a large percentage of the particulate mercury at the inlet; the S-CEM is unable to measure particulate mercury. The two measurement methods showed similar mercury

16

levels at the outlet. This also makes sense since nearly all of the particulate is removed in the ESP and only vapor phase mercury is present at this location. The average Hg removal efficiency as measured by the Ontario Hydro testing method was 90.8%. Table 8. Mercury Emissions and Removal Efficiencies – Baseline Salem Harbor Location Particle Bound Oxidized Hg2+ Elemental Hg0 Total, Hg Inlet - Loc 1 (µg/dncm) 9.3 0.08 <0.23 9.6 Outlet - Loc 3 (µg/dncm) <0.23 0.27 <0.38 <0.88 RE (%) 97.5 -238 -65.2 90.8 The inlet oxidized and elemental mercury numbers in Table 8 above are relatively low compared to the corresponding data collected by the Hg S-CEMs. This difference can be attributed to the measurement artifact of vapor phase mercury as seen by the Ontario Hydro testing method. The artifact is imposed by the sampling device used by the Ontario Hydro method, which essentially converts a fraction of the vapor phase mercury to particle bound especially at high particulate loading locations such as the inlet measurement location. Historical data has shown mercury removal efficiencies for Unit 1 ranged from 87 – 94%. Parametric Testing Series The two primary objectives of the parametric testing series were to evaluate certain process variables and determine their roles for mercury capture and to evaluate the effect of activated carbon injection on mercury removal. The configuration of plant equipment for Salem Harbor Unit 1 allowed the Hg team to evaluate the following parameters:

• SNCR • Unburned carbon in the fly ash (LOI) • Flue Gas Temperature • Activated Carbon Injection (ACI)

SNCR Minimal data are available to assess the affect of SNCR on mercury capture, and there is some debate in the industry as to its potential effectiveness. Salem Harbor’s Unit 1 utilizes a urea based SNCR system to help reduce NOx emissions. With permission from Massachusetts Department of Environmental Protection (MADEP) along with plant personnel, Salem Harbor’s Unit 1 operated at full load (~ 86 MW) without the SNCR system upon start-up from a week long outage. This would ensure the system was free of any residual ammonia and help quantify the impacts of SNCR on mercury capture.

17

During the period in which Unit 1 operated without the SNCR system, vapor phase mercury measurements were made throughout the system with the S-CEM. The Hg S-CEM only has the ability to measure total and elemental mercury content (vapor phase). With the SNCR system out of service, vapor phase mercury removal efficiencies ranged from 80-95%. This particular data are seen in Figure 8 below. Figure 8. Vapor Phase Mercury Concentrations During SNCR Testing

Mercury removal efficiencies were consistently high throughout this particular test and there was no decrease in mercury removal when SNCR was turned on. If anything, the data show slightly lower and more variable mercury removal with SNCR on; however, mercury levels were changing during this period and the variation in mercury removal seen in Figure 2 during this transition period (80 to 90% over 2 ½ days) should be considered insignificant because of the variation in the coal mercury content. LOI Laboratory data suggest that unburned carbon in the native fly ash (LOI), does play a role in terms of mercury capture. Prior to the Hg field evaluation, the Hg test team looked at several different operating conditions that could significantly reduce the normal LOI levels. Typical LOI levels ranged from 25-35% as measured by the on-line LOI analyzer placed upstream of the air preheater. It was determined that by reducing boiler load from 86 MW to approximately 65 MW and keeping all mills in service, LOI levels were significantly reduced to 15 –20%. This phenomenon can be seen in Figure 9.

40.0

50.0

60.0

70.0

80.0

90.0

100.0

9/17/2002 0:00 9/18/2002 0:00 9/19/2002 0:00 9/20/2002 0:00 9/21/2002 0:00

Date & Time

Hg

Rem

ova

l (%

)

SNCR ON

18

Figure 9. Plot of LOI vs. Boiler Load

As part of the the parametric series, boiler load on Unit 1 was reduced from 86 MW to approximately 65 MW and vapor phase mercury concentrations were made throughout the system. Reducing LOI levels did not reduce the high mercury capture across the system. Table 9 presents a summary of the parametric testing conditions at Salem Harbor. Table 9. Summary of Parametric Testing Conditions – Salem Harbor

Test Series Test Description SNCR Steam Coils (% Output)

Carbon Injection Concentration (lbs/MMacf)

1-2 SNCR Testing Off/On 0 0 3 Ontario Hydro/M29 On 0 0 4 Low Load/LOI On 0 0 5 Full Load/Standard On 0 0 6 Low Load/LOI/Temperature On 100 0 7-12 Full Load/Temperature On 100,50 0,10,20 13-16 Low Load/Standard On 0 0,5,10 17-22 Full Load/Standard On 0 0,5,10,20 23-25 Full Load/Temperature On 50 0,5,20

0

10

20

30

40

50

60

70

80

90

100

24-Sep 25-Sep 25-Sep 26-Sep 26-Sep 27-Sep

Date and Time

Bo

iler

Lo

ad (

MW

) an

d L

OI

%

Boiler Load (MW)LOI %

19

The data from Salem Harbor suggests that below certain temperatures the LOI carbon from this boiler effectively removes vapor-phase mercury. Temperature The final parameter evaluated was the role of flue gas temperature on mercury capture. Salem Harbor Unit 1 had the ability to increase ESP inlet flue gas temperatures 50°F by placing the steam coils in service. These steam coils were located just downstream of the exhaust side of the F.D. fan and upstream of the air preheater. Under normal operating conditions, ESP inlet temperatures averaged approximately 300°F. Placing the steam coils in service, the average ESP inlet temperature increased to 350°F. During the parametric series, the steam coils were placed into service while Unit 1 was held steady at full load (~ 86 MW). ESP inlet temperatures were increased from 300°F to 350°F. Increasing the flue gas temperature decreased the overall removal efficiency for the vapor phase mercury from ~ 90% to the 10-20% range. This can be seen in Figure 11 below. Figure 11 also identifies different LOI levels present in the flue gas. As the flue gas temperature increased, the mercury removal efficiency decreased independent of the LOI levels present in the flue gas. To illustrate the effect of temperature on the sorbents ability to adsorb mercury, Figure 10 represents the sorbents mercury adsorption capacity as a function of temperature. The adsorption capacity defines the sorbent’s capacity for impurity removal, in this case vapor phase mercury. Laboratory tests have also shown that, in general, a sorbent’s adsorption capacity decreases when the flue gas temperature increases. This can be seen from Figure 10 below, which shows data for activated carbon. A similar trend exists for LOI carbon. Figure 10. Sorbent Adsorption Capacity vs. Temperature

* Data Supplied by URS Corporation

Equilibrium Adsorption Capacity - Darco FGD

0

500

1000

1500

2000

2500

3000

200 225 250 275 300 325 350 375 400 425 450Adsorption Temperature (°F)

(µg

HgC

l 2/g

C)

20

The minor influence of LOI concentration shown on Figure 11 further supports the argument that when the temperature is appropriate for the Salem Harbor LOI carbon to be effective for mercury removal, even at the lowest LOI concentrations tested, they are high enough that it is difficult to see the incremental effects lowering LOI by 10%. Figure 11. Hg Removal Efficiency vs Temperature (No Sorbent Injection) Activated Carbon Injection Activated Carbon Injection Sorbent injection was tested to determine whether further removal beyond the 90% removal measured in baseline conditions was possible. At Salem Harbor just one sorbent was tested, the benchmark activated carbon sorbent, Norit America’s Darco FGD. A challenge with these parametric tests was the low mercury concentrations at the outlet and the difficulty accurately measuring at these levels. Mercury concentration at the outlet prior to sorbent injection generally was less than 0.5 µg/dncm and as low as 0.1 µg/dncm. Measuring at these baseline concentrations is difficult and accurately measuring a reduction from this level posed an additional challenge. During the parametric testing series, it should be noted that a different coal was being fired. This test coal was a low sulfur bituminous coal and generally showed a lower baseline mercury capture as compared to the standard coal fired during the baseline series. Figure 12 presents mercury removal efficiencies at various sorbent injection concentrations and different flue gas temperature ranges during the parametric tests. The results show that there were inconclusive trends when activated carbon was added at standard operating temperatures, 280 – 290oF. This is likely due to the relatively high LOI carbon concentrations at temperatures

0

10

20

30

40

50

60

70

80

90

100

270 290 310 330 350 370

Loc. 1 Temperature (F)

Hg

Re

mo

va

l (%

V)

17-19% LOI (45 lb/MMacf)

20-24% LOI (55 lb/MMacf)

25-29% LOI (68 lb/MMacf)

>30% LOI

30-35% LOI*

21-27% LOI*

* - Estimated values from previous tests.

21

where the LOI was effectively removing mercury. At the mid temperature range of 322-327°F, the LOI lost some of its ability to capture vapor phase Hg, however activated carbon performed relatively well. The addition of activated carbon did increase mercury removal and the trend showed the expected correlation between injection concentration and mercury removal. At the hotter temperature range of 343-347°F, ACI performance was severely impacted and maximum RE was nominally 45%. This is likely due to the higher temperatures where the LOI carbon was not as effective at removing mercury. Figure 12. Hg Removal Efficiency vs. Injection Concentration

Long-Term Testing Series Activated carbon was injected continuously from November 19 through November 22 at an injection concentration of 10 lbs/MMacf. Vapor phase mercury removal, as measured by the S-CEMs, was in the 90% range before and during injection. Data are still being reviewed and results from the Ontario Hydro tests are pending. As mentioned, measuring mercury at the outlet during sorbent injection was a challenge. During these tests, at least one hour was required to collect each data point with the analyzer located at the outlet as compared to less than 10 minutes at the inlet. CONCLUSIONS A full-scale evaluation of mercury control using activated carbon injection upstream of a cold-side ESP was conducted at three different coal fired power plants. These units fired different types of coal including Powder River Basin and low sulfur eastern bituminous. These full-scale evaluations answered many questions about the potential for mercury control using activated carbon injection and the role of certain operational parameters on mercury capture. The following is a list of summary conclusions the Hg team has developed from these tests:

0

10

20

30

40

50

60

70

80

90

0 5 10 15 20 25

Injection Concentration (lb/Mmacf)

Hg

Re

mo

va

l (%

V)

280-290 F

298-306F

322-327F

343-347F

* All high load data except 280-290F

22

• Activated carbon injection is effective at capturing both elemental and oxidized forms of

gaseous mercury. • Activated carbon injection is capable of controlling up to 70% of mercury emissions from

stack on both bituminous and sub-bituminous coals. • Activated carbon injection is effective at temperatures up to 340°F. • Spray cooling is required for temperatures > 350°F. • To date, activated carbon injection has shown no detrimental impacts to ESP

performance. • The mercury adsorption capacity for both activated carbon and unburned carbon (LOI) is

effected by flue gas conditions: o capacity decreases with an increase in flue gas temperature, o decreases with the presence of certain acid gases.

REFERENCES

1. Durham, M.D, C.J. Bustard, R. Schlager, C. Martin, S. Johnson, S. Renninger. “Field Test Program to Develop Comprehensive Design, Operating and Cost Data for Mercury Control Systems on Non Scrubbed Coal-Fired Boilers”. Presented at the Air & Waste Management Association 2001 Annual Conference and Exhibition, June 24-28, 2001 Orlando, FL.

2. Durham, MD, C.J. Bustard, R. Schlager, C. Martin, S. Johnson, S. Renninger.

“Controlling Mercury Emissions from Coal-Fired Utility Boilers: A Field Test” EM, Air & Waste Management Association’s Magazine for Environmental Managers, pp 27 – 33, July 2001.

3. T. Ley, R. Slye. “Continuous Real-Time Monitoring of Mercury in Flue Gas from Coal-

Fired Boilers: Field Experience”. Presented at the 19th Annual International Pittsburgh Coal Conference, September 23-27, 2002.

4. C. Senior. “Solid Sample Analysis from Long Term Testing at Brayton Point”.

December 12, 2002.

5. T. Starns, J. Bustard, R. Afonso, S. Sjostrom. “Field Evaluation Summary Report – Salem Harbor Unit 1”. Presented to Massachusetts Department of Environmental Protection, Jan. 17, 2003.

6. S. Sjostrom, T. Ebner, R. Slye, R. Chang, M. Strohfus, J. Pelerine, S. Smokey. “Full-

Scale Evaluation of Mercury Control at Great River Energy’s Stanton Generating Station Using Injected Sorbents and a Spray Dryer/Baghouse”. Presented at the 2002 Air Quality III Conference, Session A3b.