Allied Environmental Technologies (ALENTEC) has recently installed SNCR systems on four gas- and coal-fired boilers. These boilers are owned by PERMENERGO, and are located 800 miles east of Moscow in Tchaikovsky, Russia. These sister units have a maximum continuous rating of 420 tonnes steam per hour (882,000 pounds per hour).

The SNCR systems were designed and fabricated in the United States by WAHLCO, Inc. The designs were based on HVT and gaseous emissions measurements made on site at the initiation of the project. The design was subsequently fine-tuned by computational fluid dynamics (CFD) modeling performed by Reaction Engineering International. The client dictated that only existing penetrations could be used for injection ports, which limited the final SNCR system design flexibility.

This paper presents the SNCR system design specific, modeling predictions and field test results. Also discussed are the logistics and challenges of performing complex engineering projects in the former Soviet Union.

Allied Environmental Technologies, Inc. (ALENTEC) was awarded a contract to install urea-based Selective Non-Catalytic Reduction (SNCR) systems on four gas- and coal-fired boilers located in Russia. These boilers are owned by PERMENERGO and are located 800 miles east of Moscow in Tchaikovsky, Russia. The project will comprise four tasks, as follows, upon completion:

To date, Tasks 1 through 3 have been completed, as well as the initial portion of Task 4.

Tchaikovsky Units 1 through 4 are sister units having a maximum continuous rating of 420 tonnes steam per hour (882,000 pounds per hour). The units are opposed-wall-fired units, each having one row of six burners on both the front and back furnace walls. The boilers are a balanced draft design. The units fire natural gas as their primary fuel and coal as a back-up. Baseline full load NO_x emissions varied from 139 to 152 ppm_c (ppm, dry at 6% O₂) when firing natural gas. When firing coal, baseline NO_x emissions were 558 ppm_c at a load of 398 TPH.

The SNCR systems installed on Tchaikovsky Units 1 and 2 are of the "high energy" type. These systems utilize low-pressure steam as the carrier fluid. The design steam flow is 4,400 kg/hour at full load, which represents approximately 1.5 percent of the total combustion products flow. Each unit incorporates three (3) injection levels on the front wall, in addition to two (2) levels of sidewall injectors. Figure 1 shows a schematic view of the boiler and Figure 2 shows the location of the injectors at each level.

A nominal 50% urea solution is prepared on-site from bulk urea solids. After mixing, the solution is pumped to a storage tank in the boiler house. Each of the four individual SNCR systems is supplied by this tank. Urea flow is set according to a flow versus load curve generated during the start-up/optimization testing.

Because the client dictated that only existing penetrations could be used for injection ports, the final system design flexibility was limited. To overcome these limitations, the final system design incorporated two individual injectors in each available front wall port. The design of these injectors allows them to be independently rotated in order to provide the best possible mixing between the injected urea and the combustion products.

The test results presented below include the HVT tests, the CFD tests and the field start-up and optimization tests. Each of these test series is discussed below.

The field testing tasks of this program required the measurement of both temperature and gaseous emissions. Temperature measurements were made using a high velocity thermocouple (HVT) probe. This HVT was of a standard water-cooled design utilizing a single radiation shield. Suction power was provided by an air-powered vacuum eductor. The HVT probe also incorporated the capability of providing a flue gas sample from the aspirated thermocouple location. During the temperature measurement task, gaseous emissions of NO, CO and O₂ were measured using a NOVA Model 2000 portable combustion analyzer. This analyzer utilizes electrochemical "fuel cell" type sensors to measure species concentrations.

The HVT tests were conducted over a nominal load range of 210-400 tones/hour (TPH) when firing natural gas. The nominal test load range for coal firing was 300-400 TPH. Each unit was operated in a "normal" firing configuration during the HVT testing; no attempts were made to balance fuel and/or air flows, or to otherwise alter unit operation.

The desired temperature range for urea injection is nominally 930C to 1150C, with the maximum NO_x reduction performance achieved at a temperature of 1010C ¹. However, operation at below-optimum temperatures results in high NH₃ slip levels. For this reason, it is desirable to operate at or above the optimum injection temperature. Thus, the desired injection temperature range for this project is between 1010 and 1150C.

Figure 3 shows average furnace gas temperatures plotted versus load. Data are included for both levels D and E from Units 1 and 2 while firing natural gas. Average temperatures at Level E varied from 1023C to 802C, while average temperatures at Level D varied from 1152C to 865C over the load range tested while firing natural gas. The difference between maximum and minimum temperatures at a given load ranged from 118C to 283C, and varied with the measurement location and load. The degree of stratification can also be characterized by calculating the standard deviation of measurements made at a given level and load. For natural gas firing, the standard deviations ranged between 32C and 67C. These data indicate that Level D may provide the preferred temperature zone from loads of about 270-370 TPH. At loads in excess of 370 TPH, Level E appears to be in the preferred temperature region.

Figure 3 also shows average temperature plotted versus load for the coal firing tests performed on Unit 4. These test data are divided into two groups; tests performed on September 13 and 14 and tests performed September 18 through 20. Temperatures at Level E averaged 870C and 956C at a nominal load of 350 TPH for the two test series. The corresponding temperatures at Level D averaged 1003C and 1074C. Temperature stratification was more pronounced when firing coal, ranging from 45C to 133C.

The data show that temperatures increased during this five day time period between the two sets of measurements. Coal was initially fired in Unit 4 during the first week of September, allowing about one week of seasoning before the initial testing. The second set of coal tests was performed five days after the initial coal tests were completed. During this time period, average gas temperatures increased by about 65C at Level D to 85C at Level E. It appears that the measured increases in gas temperatures were due to the gradual build-up of ash deposits on the heat transfer surfaces, since soot blowers are not used (or needed) on the Tchaikovsky boilers. These variations in temperature with time may adversely impact SNCR performance if coal is fired for extended periods of time.

A typical temperature contour plot for the gas-fired tests is shown in Figure 4. This shows a characteristic saddle-shaped profile for these units. The profile showed two temperature peaks near the furnace center of 1070 to 1090C at a load of 370 TPH. From the center, gas temperatures dropped as either furnace sidewall was approached.

Emissions measurements made during the HVT testing showed that NO emissions varied from 113 ppm_c to 152 ppm_c across the load range at Level E when firing natural gas. (Note that in this paper, ppm_c is equal to ppm, dry, corrected to 6% O₂, which is the measurement standard at the utility site.) When firing coal, NO_x emissions at Level E varied from 290 ppm_c to 558 ppm_c. CO emissions at Level E were less than 40 ppm when firing natural gas and less than 90 ppm when firing coal.

Computational fluid dynamics (CFD) modeling was performed by Reaction Engineering International (REI) to evaluate reagent mixing and SNCR performance. REI developed a reduced mechanism for gas-phase SNCR chemistry, which was used to quantify NO_x reductions. The reduced mechanism of seven (7) reactions and individual rate constants were developed so that the mechanism could be incorporated into a CFD code ². This model accurately describes the SNCR chemistry as indicated by comparison of process performance relative to predictions obtained using a complete chemical mechanism. The SNCR submodel was incorporated into GLACIER, a computer code which solves the governing fluid mechanics and reaction equations in an Eulerian framework. Reference 3 provides further details regarding this model.

Prior to completing the system design, four cases were modeled for full boiler load as follows:

Table 1 provides the performance estimates for each of these cases. These data show that Case 3 presented the most flexibility in terms of NO_x reduction and NH₃ slip.

The CFD model was divided into two parts; a lower furnace model and an upper furnace model. The initial modeling of the lower furnace showed that thermal boundary conditions had to be varied between units in order to replicate the field test results. The boundary conditions modified included the furnace wall emissivity. In one case, the furnace wall emissivity corresponded to that expected for a gas-fires boiler, while the other case required a furnace wall emissivity that corresponded to a dirty wall environment that could have resulted from previous coal firing. This work provided good agreement between the field measurements and the CFD lower furnace model.

The lower furnace model conditions were then used as the inlet conditions for the upper furnace model. The initial upper furnace work (Cases 0-2) showed that over half of the gas at the furnace outlet plane (FOP) had molar N/NO ratios less than 0.5 or greater than 3.5. This led to the

Table 1

CFD Modeling Results

Case 3 configuration, in which the injectors were redistributed along the front and sidewalls using existing ports. The Case 3 configuration provided improved reagent mixing, as measured by the increase in the number of areas at the FOP having molar N/NO ratios between 0.5 and 3.5.

Figure 5 shows NO_x reduction and ammonia slip distributions in the outlet plane for Case 3. These data, presented as contour plots, show that NO_x reductions were predicted to be highest at the center of the boiler, and lower toward the edges with an increase right at the sidewalls due to the effects of the sidewall injection. The corresponding NH₃ slip data show peaks near the edges of the boiler, which correlate to low temperature regions in the furnace.

Another way to look at the CFD results is to plot NO_x reduction and NH₃ slip versus the N/NO ratio in each cell at the exit plane. This is shown in Figure 6 for Case 3. The degree of vertical scatter indicates the amount of temperature stratification present, while the range of N/NO ratios provides an indication of the "mixedness" at the exit plane. Note that the N/NO ratio ranges from 0 to about 3.5 at the exit plane. The NO_x reduction values fall in a fairly narrow band when plotted versus N/NO ratio. However, the NH₃ slip falls into two general bands. One group shows NH₃ slip less than 10 ppm for N/NO ratios up to 3.5, indicating a high temperature region. Conversely, there are a number of points showing high NH₃ slip levels, corresponding to temperatures on the low side of the SNCR temperature window. This figure shows that nearly the same NO_x reductions could be achieved with two different ammonia slip levels. This behavior may be characteristic of the two-level injection scheme, where the lower level provides lower NH₃ slip levels due to the higher gas temperatures encountered in that injection zone.

To date, tests have been performed on Units 1 and 2 while firing natural gas. Plans call for completion of the Unit 1 gas testing, as well as coal testing on another unit in the Fall of 1997. Logistical problems have resulted in the test program proceeding slower than originally planned. Table 2 shows the analyzers and analysis methods used during the field-testing.

Table 2

Emission Measurement Methods

Initial testing began on Unit 2 in the Spring of 1997. However, the wrong injectors were installed at the center ports on the front wall due to a misunderstanding between plant personnel and the on-site engineer. Due to this derivation from the system design, performance on this unit was not as expected, so the testing efforts were redirected to Unit 1. The Unit 1 installation included double injectors in all front wall ports as the final design had specified. Unit 1 data are limited, since only three days of testing were permitted due to a condensed test schedule resulting from an unscheduled unit outage.

The early Unit 2 testing was limited by the inability to accurately control urea flows to the in-service injectors. Figure 7 shows NO_x reduction and NH₃ slip plotted versus the molar N/NO ratio. These data show that NO_x reductions in excess of 35 percent were achieved at a molar N/NO ratio of 2.0. The corresponding NH₃ slip at these conditions was in the range of 25 to 32 ppm. The NO_x reduction profiles showed that the reductions were highest near the furnace sidewalls, while they were measurably lower in the center of the furnace. These data suggest that the NO_x removals may have been compromised by the installation of the single injectors on the front wall of Unit 2.

The Unit 1 tests were performed at a load of 360 TPH, which is 86% of the rated load of 420 TPH. NO_x removals on Unit 1 ranged from 39 to 47% with molar N/NO ratios of 1.5 to 2.1, as shown in Figure 8, when injecting at Level D. Tests conducted with urea injection at Level E showed NO_x reductions of 42% at a molar N/NO ratio of 2. Injection utilizing both Levels D and E resulted in NO_x reductions of 42 to 46% at a molar N/NO ratio of 2.0, depending on the specific injector pattern used.

Ammonia (NH₃) slip levels are also plotted versus molar N/NO ratio in Figure 8. At a molar N/NO ratio of 2, NH₃ emissions varied from 32 ppm when injection at Level D to nearly 70 ppm when injection at Level E. The high ammonia slip encountered when injection at Level E may indicate that the E level is too cold for injection at the 360 TPH load. When injection at both Levels D and E, ammonia slip varied from 46 to 65 ppm at a molar N/NO ratio of 2.

NO_x reduction profiles were obtained by measuring both pre- and post-injection NO_x emissions at each of the 12 sample points at the economizer exit sample grid. Figure 9 shows NO_x reduction profiles for a test performed using Level D injection at a molar N/NO ratio of 1.5. The data show that the achievable NO_x reductions were highest on the outside walls and dropped significantly to the boiler center. For example, injection at Level D resulted in NO_x reductions at the walls averaging 51%, while NO_x reductions at the four innermost points averaged only 24%.

The results of the Case 3 CFD modeling and the Unit 1 field test are shown together in Figure 10. Both NO_x reduction and NH₃ slip are plotted versus N/NO ratio. These data show that the CFD modeling are field test results correlated well on an overall basis. A subsequent comparison of the CFD modeling and field test results shows that there are some differences in the details. It is hoped that the testing scheduled for Fall 1997 will provide additional data which can resolve these differences. In the interim, however, it appears that the CFD modeling provided a good approximation of the overall field test results.

As discussed previously, this project involved the design, fabrication and installation of four SNCR systems in Tchaikovsky, Russia. The design and fabrication of the SNCR skids was performed in Santa Ana, California by WAHLCO, Inc. These skids were fabricated so that two systems were installed in a single container. Each container contained the required pumps, liquid metering systems, control valves and piping for two SNCR systems. The SNCR injectors and associated mixing equipment were also designed and fabricated in California. ALENTEC designed the urea mixing and storage systems, and supervised their construction at the plant site. The plant was responsible for connecting the containers to the urea storage and water supply systems, and installing the piping from the container to the individual injection levels on each boiler.

Language was not the barrier originally thought, since the plant provided an experienced translator for the duration of the project. The primary delays in the project involved difficulties in getting the SNCR system through customs in Moscow. Once on site, the plant staff did an admirable job installing the SNCR systems. The primary on-site problem was a lack of spare parts. Spare parts were a week or more away, at best, due to the relatively remote location of the plant.

The authors wish to acknowledge the hospitality extended by Mr. Potapov, the Plant Manager. His cooperation allowed all parts of the project to proceed smoothly. Mr. Alexi Grebnev, the Chief Plant Engineer, also provided invaluable assistance in the installation and start-up of the SNCR systems. Finally, Mr. Nicolai Oschepkov did an excellent job of translating, which allowed the project to proceed smoothly. All members of the plant staff should also be commended for their efforts in completing a successful retrofit.

Figure 1. Tchaikovsky Units 1-4 Boiler Schematic Drawing

 

Figure 2. SNCR Injector Schematic

Figure 3. Furnace Gas Temperature versus Load. Tchaikovsky Units 1, 2 and 4

Figure 4. Measured Temperature Contour Plot, Tchaikovsky Unit 2,

370 TPH, Natural Gas Fuel

NO_x Reduction Profile

NH₃ Slip Distribution Profile

 

Figure 5. NO_x Reduction and NH₃ Slip Distribution Profiles.

CFD Modeling, Full Load, Natural Gas, Case 3

Figure 6. NO_x Reduction and NH₃ Slip versus Molar N/NO Ratio.

CFD Modeling, Full Load, Natural Gas, Case 3

Figure 7. NO_x Removal and NH₃ Slip versus Molar N/NO Ratio.

Tchaikovsky Unit 2, 400 TPH, Natural Gas

Figure 8. NO_x Removal and NH₃ Slip versus Molar N/NO Ratio.

Tchaikovsky Unit 1, 360 TPH, Natural Gas

Figure 9. NO_x Reduction Profile. Tchaikovsky Unit 1, Level D Injection

360 TPH, Natural Gas, N/NO = 2.0

Figure 10. Comparison of CFD and Field Test Results