Driven by increasingly stringent environmental regulations, Mitsui Babcock has developed and demonstrated a novel NOx removal system. The new system is typically only one-third of the capital cost of an SCR system, and can also bring generators other advantages, such as operational flexibility.
Keith Morris and Scott Affelt, Mitsui Babcock, Atlanta, USA
During the spring of 2002, Mitsui Babcock demonstrated a novel post-combustion NOx removal system at the Tennessee Valley Authority (TVA) Kingston 9 power plant in the USA. The name of the demonstration system is NOxStar; it was developed initially from diesel engines and has since made the leap to power stations.
The NOxStar process is, in essence, a post-combustion process that continuously injects controlled quantities of an ammonia-based reagent with relatively minute amounts of hydrocarbon (typically either natural gas or propane) into the superheater/reheater pass of an operating boiler. The injection grid comprises a permanent array of feed lances attached to the adjacent heat transfer assemblies in the upper furnace pass. This arrangement optimizes reagent distribution throughout the relatively high temperature NOx-bearing flue gas mass. At these existing elevated temperatures, the hydrocarbon auto-ignites to form a plasma of free radicals which autocatalyses the reaction of ammonia and NOx to form harmless nitrogen (N2) and water vapour (H2O).
The process was originally developed by Mitsui Babcock to treat the exhaust gases of large stationary diesel engines and was demonstrated successfully in this application at Santa Catalina Island in California, USA. In 2001, Mitsui Babcock embarked on an extensive multi-phase technology development programme to test the concept for pulverized coal boiler applications.
The first validation was generated at Mitsui Babcock’s 160 kWt NOx Reduction Test Facility (NRTF) which comprises a single, low-NOx burner firing vertically downwards into a cylindrical refractory lined furnace. Early testing demonstrated that the process would work in coal-fired systems at conditions pertinent to those experienced in large-scale utility plants. An extended period of parametric testing followed in order to generate the basic process design information. The relative effects of ammonia molar ratio, flue gas temperature, residence time, injector geometry, hydrocarbon type, fuel blend and other related process parameters were investigated.
Testing demonstrated that the process could achieve high NOx reduction levels of up to 90 per cent, with low outlet concentrations of 60 ppm at six per cent O2 while maintaining acceptable ammonia slip below 5 ppm. The process was shown to be effective over a significantly wide temperature window of flue gas from 850°C to 1100°C. Minimal quantities of hydrocarbon were required to achieve optimal reduction rates and the presence of cooling surface in the reaction zone did not compromise the process.
The NOxStar process was further validated in Mitsui Babcock’s Multi-fuel Burner Test Facility (MBTF) while firing coal at a rate of 40 MWt. The objective of this phase of testing was, firstly, to show the scalability and controllability of the process and secondly, the mechanical design and integrity of the feed lances in the high temperature and high particulate environment of a coal-fired combustion chamber.
Figure 1. The NOxStar process has been developed and demonstrated by Mitsui Babcock
A series of tests were conducted using eight, 5 m long, vertically arranged lances, containing multiple spray nozzles. Reagent and hydrocarbon flows to the lances were independently controlled to six individual zones in the lance grid to effect maximum intermixing with a flue gas stream that typically possesses a non-homogeneous flow and temperature distribution across the grid plane.
The test results were very positive and tracked the earlier performance achieved in the 160 kWt test facility. The ability to control, by zone, the reagent and hydrocarbon flow rates to maximize NOx reduction and maintain acceptable ammonia slip levels was also shown. The injection lances, virtually identical to those required for a full-scale installation, proved to be capable of reliably functioning in the high temperature flue gas stream.
The process was then demonstrated at TVA’s Kingston 9 power plant in 2002. This coal-fired boiler is a 200 MWe tangentially fired unit with twin furnaces (superheat and reheat), each of which has three elevations of burners supplied by six pulverizers.
The NOxStar injection grid comprised 22 lances subdivided into nine control zones in each of the superheater and reheater passes. These nine zones allowed NOx reduction and ammonia slip to be optimized on a zone-by-zone basis to account for variability in the flue gas temperature and flow profiles at the reagent injection plane. Each water-cooled lance was fed with a controlled stream of reagent, steam and propane to satisfy the flue gas conditions at the time. Reagent rates were controlled to keep ammonia slip below 5 ppm.
Figure 2. NOxStar process flow diagram
The results of the these tests showed that NOx levels were reduced by 68 per cent from the baseline level while maintaining an ammonia slip of 4.2 ppm or less, as measured at the economizer outlet using wet chemistry techniques. Air staging in the combustion zone had contributed to this overall reduction. Alone, the NOxStar system had reduced NOx levels by 53 per cent.
This was somewhat less than had been anticipated based on the earlier test rig work but the reasons were well understood. The test results from both test rigs demonstrated that the NOx reduction potential was clearly a function of the flue gas temperature at the injection point, coupled with the retention time downstream of the lances. Results from both the 160 kWt test facility and the 40 MWt combustion test rig demonstrated this temperature dependence. Similarly, the time at temperature was also shown to influence the NOx reduction percentage.
The flue gas temperatures at Kingston in the region of the injection lances, were found to be lower than originally anticipated. Based on test results from the MBTF combustion rig facility the NOx reduction percentage would increase by 14 per cent, absolute, had these temperatures been at a more optimum range. Given the close correlation between the two facilities at the lower temperature levels, it is reasonable to presume that, had the injection lances been installed in a higher temperature environment, the improved level of performance would have been achieved at the plant.
The injection lances installed at Kingston were of a water-cooled, externally insulated design, having a multiplicity of nozzles along their length. They were divided into three separately controllable zones to facilitate the optimal delivery of reagent to the varied distribution of flue gas across the injection plane.
Figure 3. The NOxStar process was successfully demonstrated by Mitsui Babcock at the Tenessee Valley Authority’s Kingston 9 power plant in the USA in 2002
Many generators in the USA have recently updated their firing systems adding low-NOx burners and/or over-fire air (OFA) systems to reduce the NOx levels generated in the combustion zone. In addition, generators are also exploring various fuel blends and deeper staging to further reduce NOx levels. Depending on the fuel characteristics and the specific firing system, NOx outlet levels can range anywhere from 150 to 300 ppm and these modifications are usually the least expensive method of reducing NOx on a $/NOx t removed basis. However, it is difficult for most generators to then reach the impending regulatory limits, which can require NOx levels of 90 ppm or in some cases down to 60 ppm with more traditional technologies such as selective non-catalytic reduction systems (SNCR) which typically only realize a 25 to 30 per cent decrease in outlet NOx levels.
To reach these low NOx levels, many generators are initially considering selective catalytic reduction (SCR) technology. SCR retrofit systems have been installed in a number of coal-fired boilers in the USA during the past several years. Although industry estimates in the mid-1990s indicated that SCR retrofits would cost about $70/kW for 80 per cent removal in units rated from 400 to 900 MWe, the actual capital costs have grown from $80 to $160/kW, with the average retrofit costing more than $110/kW. The fully evaluated cost per NOx t removed (amortized capital plus operating and maintenance costs) over a ten-year period can be as high as $3000 to $4000/t. For many plants, this high initial capital cost along with the uncertainty around the longevity of plants that in some cases are 50 years old, have forced generators to explore more innovative and lower capital solutions for NOx reduction in order to remain competitive.
The NOxStar system is typically only one-third of the capital cost of an SCR but can achieve the NOx reduction levels required by many generators. In addition, a NOxStar system can be supplied within six to eight months from order and can be installed within a two-to-three week boiler shutdown period. This lead-time and outage duration are significantly shorter than an SCR installation. Once installed the NOxStar will allow full operational and fuel flexibility. There will also be no significant impact on the performance of the unit.
Figure 4. The NOxStar system was originally developed for small stationary diesel engines and has been scaled up for utility power plant applications
With the deregulating electric market and increased environmental regulations, generators need ingenious and low cost solutions. Whether alone or in combination with other combustion modifications or post-combustion polishing technologies, the NOxStar system can achieve most current and future NOx reduction requirements at a lower capital and on-going cost than most other available technologies.