Mercury control options16 November 2001
Last year EPA made the long awaited announcement that mercury emissions from US power plants are to be regulated. EPRI has been investigating the problem of mercury from power plants for several years, and the potential solutions. Ramsay Chang and George Offen, EPRI, Palo Alto, CA, USA
Mercury emissions from US coal-fired power plants vary widely, from around 0.01 to 56 lb/TBtu (1012 Btu). One reason is the wide range of mercury concentrations in the coal being used, from about 2 to 65 lb/TBtu.
Another factor is the wide variation in the effectiveness of existing flue gas emissions control technology in capturing mercury in the flue gas. For the half of the US boiler population equipped with just a cold-side ESP, EPRI estimates that reductions average around 20 per cent (ranging from 0 to 54 per cent). Units with fabric filters tend to capture greater amounts of mercury, but show a larger variability (35 to 99 per cent). Boilers firing bituminous coals with moderate-to-high chlorine content and equipped with an FGD or spray dryer are likely to capture significant amounts of mercury - 65 per cent on average (with a range of 42 to 84 per cent) for sites with a wet FGD, and 98 per cent for those with spray dryers and fabric filters.
One of the reasons the mercury capture effectiveness of existing controls varies so much is that it depends on the speciation of the mercury - ie whether it is elemental (Hg0), oxidised (Hg++), or particulate-bound. The Hg++ is captured effectively by SO2 controls and also appears to be captured more readily by ESPs. But virtually no Hg0 is captured by SO2 controls. Coal chlorine content appears to be the major determinant of mercury speciation; generally, the more chlorine, the more oxidised mercury. On the other hand, SO2 appears to impede the capture of mercury by ESPs. Particulate-bound mercury may be associated with unburned carbon in the fly ash, and is generally captured in the particulate control device.
Near term options
If additional mercury control is required, over and above capture by existing particulate and SO2 controls, a likely contender is activated carbon injection (ACI) upstream of a particulate control device. If the carbon achieves good contact with the gaseous mercury for sufficient time, it adsorbs the mercury. The mercury-laden carbon is then collected by the downstream particulate control.
Table 1 shows EPRI assessments of several technologies likely to be available in the near term, subject to the proviso that these results come only from small pilot-scale test devices. There is, as yet, only one full-scale example of carbon injection at a power plant, the DOE/EPRI-sponsored test at Alabama Power's Gaston plant. However, we believe the technologies are applicable to large-scale plants.
The performance figures in Table 1 are total mercury reduction; that is, they include capture by both the existing air pollution device and the mercury control.
Several factors may affect these performance numbers in full-scale applications. Besides the exogenous parameters of residence time (duct length between the air heater and ESP) and sorbent particle mean size that can affect performance of the carbon, other factors are the degree of mixing between the carbon and flue gas, the particle size distribution (the presence of a disproportionate fraction of fine particles would improve mercury capture, and vice versa), and the capture of mercury by carbon that deposits on the plates of the ESP along with the fly ash. Table 1 does not include this supplemental capture because we have not had the opportunity to test for it at full-scale; to date we tested ACI ahead of an ESP only in our pilot test units, where we have observed supplemental capture of 20-50 per cent. The tests at Wisconsin Electric's Pleasant Prairie power plant in late 2001 should provide the first indication of whether this effect actually occurs in full-scale applications.
The Table 1 cost figures also do not account for the potential adverse impacts of sorbent injection. If the added carbon degrades the ESP's performance to the point where the operator would need to add an additional field, the control cost would increase $10-15/kW. The injected carbon could also prevent a plant from selling its ash. This would change a revenue stream (eg, $10-20/ton ash) to a disposal cost (which can be as high as $20/ton); an average differential cost between sale and disposal is $10-20/ton ash. If the mercury-laden carbon collected in the baghouse is deemed to be hazardous (although results obtained to date do not show any mercury release to the environment from this material), it could cost $200/ton to dispose of.
Table 2 presents examples of the costs of achieving different levels of mercury emission reductions for the most common air pollution control configurations and both bituminous and subbituminous coals. Together, the configurations in this table account for about three-quarters of the US coal-fired boiler population.
The table shows that EPRI-patented TOXECON, which is a compact baghouse (COHPAC), added downstream of an existing ESP, with active carbon injection between them, is a potentially attractive technology for greater than 50 per cent total mercury reduction. However, the technology is new and has been tested only once, briefly, at full-scale. The short term test showed that carbon injection can impact baghouse pressure drop and may limit the amount of carbon that can be used. The long term impacts on bag life, pressure drop, and emissions need to be determined.
In-duct spray cooling is not included because its benefits appear to be marginal; in practice it can reduce temperatures only up to 30°F, and this is not likely to improve the performance of injected carbon unless the flue gas temperatures are very high. Spray cooling, however, can improve baseline capture by the fly ash if that is already occurring, especially in a baghouse. The addition of a spray tower to allow cooling down to spray dryer exit temperatures (< 200°F) has been suggested. While it is known that oxidised mercury will condense on solids, such as fly ash, as the temperature is reduced, the relationship between temperature and the amount condensed has not been quantified, nor have costs been developed for this approach.
Effect of NOx control
Also not included in Table 2 are any benefits from post-combustion NOx reduction technologies. Selective catalytic reduction (SCR) for NOx control may increase the level of oxidised mercury, which could lead to increased mercury removal in the particulate and SO2 controls (for sites with an SO2 control). Measurements reported for a small sample of German coal-fired power plants indicated that 60-70 per cent of the elemental mercury in the flue gas was converted to oxidised mercury across the SCR catalyst. In the selective non-catalytic reduction process (SNCR), the unused ammonia leaving the furnace (called ammonia slip) may cause some of the mercury to be absorbed onto the fly ash, where it would be captured by the particulate collector. EPRI's review of data from US sites with SCR and SNCR found that no consistent trends could be established from these results. However, because a large number of units are projected to install one of these processes by 2010, an understanding of their impact on mercury emissions is important in planning for future mercury and NOx controls.
As a first step to evaluate the potential effects of the SCR catalyst and ammonia injection on mercury speciation and removal, EPRI initiated a pilot-scale screening evaluation jointly with EPA, DOE, and various US and Canadian power producers. Four different coals were tested, including three eastern bituminous coals of varying sulphur and chlorine content, and a Powder River Basin coal - but the pilot results were not conclusive.
To overcome these uncertainties, EPRI, both in collaboration with its member power companies and in joint efforts with EPA and DOE, initiated a comprehensive programme to collect more data. Tests are being conducted in pilot slipstream devices (ie using actual flue gas from power plants) and on full-scale systems, in some cases both at the same site. The experience to date has been on sites burning PRB coals, which inherently produce virtually no oxidised mercury. Results from the first few test sites show that the oxidation rate across SCR catalysts is a strong function of space velocity (a measure of the quantity of catalyst in the SCR relative to the flue gas flow rate). The tests averaged around 10-30 per cent oxidation of elemental mercury with fresh catalyst at space velocities normally found in SCR systems and was virtually zero on a catalyst that had been in use for slightly over one year. For bituminous coals, with chlorine present in the flue gas and intrinsic oxidised mercury levels > 65 per cent, initial field results suggest that the SCR catalyst will perform better as a mercury oxidation enhancer, potentially allowing the entire system to reduce mercury emissions by over 80 per cent. Tests to be conducted in late 2001 and early-mid 2002 should provide answers to these questions.
Suppliers suggest that sorbent costs, assumed to be around $0.50/lb in our current estimates, could be reduced by moving to lower grade materials. Other approaches are being considered to reduce sorbent costs. The combined use of alkali (eg, calcium hydroxide) and ACI has been shown to reduce ACI demand considerably in a single, one-day test at a site firing low-sulphur eastern bituminous coal with sufficient chlorine to produce a high percentage of oxidised mercury. EPRI and DOE have also been seeking catalysts that could be used at low temperatures to oxidise mercury and enhance capture by SO2 controls. Palladium-based catalysts and some SCR-type catalysts can oxidise as much as 70-90 per cent of the elemental mercury in the flue gas.
EPRI is also beginning to assess the feasibility of injecting chemical additives to promote oxidation of mercury. Some form of chlorine, such as common salt or calcium chloride, are logical contenders, but most power plant operators are reluctant to add these corrosive materials to their boilers. This concern becomes more acute as the plants install deeply-staged low-NOx combustion systems to meet stringent NOx limits.
Taking another tack entirely, EPRI has patented the concept of adsorbing mercury on fixed structures coated with ultra-thin layers of materials that form amalgams with mercury (MerCAP, or Mercury Capture by Amalgamation Processes). These structures could be plates or banks of tubes placed in the duct between the ESP and stack (or SO2 control). Tests to date using gold-covered plates have been conducted only at a small pilot scale, but on actual flue gas from power plants, mostly PRB-fired. The results are promising from a performance perspective, but costs are, as yet, unknown.
PowerSpan's ElectroCatalytic Oxidation system is a recently announced processes that has shown 70-80 per cent mercury capture at the pilot-scale. A 50 MW demo of this combined NOx, SO2, mercury, and particulate control device is scheduled to start up in mid 2002.
EPRI believes that the technical community will have enough information on these emerging technologies to assess their capabilities and costs within the next three to five years, assuming continued support for their assessment and development.
TablesTable 1. Mercury control options - EPSI assesment Table 2. Representative costs to achieve various mercury emission reductions