| OCR Text |
Show manner considering plant equipment inconsistencies and economic dispatch strategies. Therefore, statistical analysis methods for long-term data have been developed that can be used to detennine the achievable emissions limit or emission tonnage of an emissions control technology. These analysis methods have been developed over the past fifteen years by the Control Technology Committee of the Utility Air Regulatory Group (UARG). Because the uncertainty in the analysis methods is reduced with increasing data set size, UARG recommends that acceptable results can be achieved with data sets of at least 51 days with each day containing at least 18 valid hourly averages. v. RESUL TS DISCUSSION Wall-Fired Boiler Baseline Characterization. In November 1989. baseline testing at the wall-fired site began with a series of thirtysix diagnostic tests conducted at four load conditions. The matrix for these tests focused on the parameters that most directly affect NOx production (excess oxygen. load. and mill patterns). During the one- to three-hour tests. manual data were collected from the control room. automated boiler operational data were recorded on the DAS, furnace backpass ash samples were collected from the automatic samplers. coal samples were collected from the individual feeders. and economizer exit and air pre heater exit species and temperatures were recorded utilizing the sample distribution manifold. Following completion of the diagnostic tests, perfonnance tests were performed in December 1989. For these tests, additional subcontractors were retained to conduct fueUair and particulate emissions measurements. Boiler efficiency testing consistent with the ASME PTC 4.1 Heat-Loss Method was also conducted. Each test covered a period lasting ten to twelve hours during which the manual and automated boiler operation data were recorded, the fuel samples acquired, gaseous and solid emissions measurements made, and the engineering perfonnance tests conducted. Immediately preceding the AOFA retrofit outage. a sequence of short-tenn verification tests were performed. These tests were designed to identify any changes in boiler operating trends which might have occurred during the long-term test phase. An analysis of the tests results indicated no significant changes in boiler operating conditions over the long-tenn test period. All tests during both the diagnostic. perfonnance. and verification portions of the short-tenn test effort were conducted within the nonnal operating parameters for the unit with the exception of excess oxygen. Excess oxygen was exercised well above and below the nonnal operating range commonly encountered for the unit All major boiler components as well as auxiliary equipment were in nonnal "as-found" operating conditions. A summary of the NOx emissions from Phase I short-tenn testing is presented in Figure 5. Although not evident in this figure, extensive scatter existed in NOx emissions even for seemingly identical test conditions and it was evident that a number of uncontrolled and unidentified variables greatly influence NOx production: however. their influence could not be quantified. These influencing variables are believed to be mill operating conditions flows. grind, and condition) and secondary air non-unifonnity (air register settings). Phase I long-tenn testing was perfonned from late December 1989 through early April 1990. During the fourteenweek long-term testing, continuous emissions data were collected except during periods of unit outages. CEM calibrations, and CEM maintenance. yielding 52 valid days of data. The re ults f the long-term testing is summarized in Table 2 and Figure 6. Advanced Overfire Air Retrofit. During April 1990, the AOFA system was installed at the demonstration site. The majority of the work was performed during the scheduled four-week outage with only insulation. handrails, and controls setup work left for on-line completion. During start-up and initial operation. the AOFA flow control dampers sustained minor damage due to overheating. The warping caused the dampers to bind in the closed position. A brief outage was required to return the dampers to the fully open position. These dampers were repaired during the LNB outage discussed later in this paper. 8 |
