m in an ASME Journal. Authorization to photocopy material for internal or personal use under circumstance not

Transcription

1 THE AMERICA SOCIETY OF MECHAICAL EGIEERS 345 E. 47th St., ew York,.Y. 117 S The Society shall not be responsible for statements or opinions advanced in papers or discussion at meetings of the Society or of its Divisions or Sections, or printed in its publications. Discussion is printed only if the paper is published 95-GT-255 m in an ASME Journal. Authorization to photocopy material for internal or personal use under circumstance not falling within the fair use provisions of the Copyright Act is granted by ASME to libraries and other users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service provided that the base fee of $.3 per page is paid directly to the CCC, 27 Congress Street, Salem MA 197. Requests for special permission or bulk reproduction should be addressed to the ASME Technical Publishing Department. Copyright 1995 by ASME All Rights Reserved Printed in.s.a. DEVELOPMET OF A SECOD GEERATIO DRY LOW Ox COMBSTOR FOR 1.5MW GAS TRBIE J. Kitajima, T. Kimura, T. Sasaki and A. Okuto Akashi Technical Institute S. Kajita, S. Ohga and M. Ogata Industrial Gas Turbine Division Kawasaki Heavy Industries, Ltd. Akashi, Japan ABSTRACT Development of a second generation dry low Ox combustor for KHTs 1.5MW industrial gas turbine M1A-13A is described A lean pre-mix and multiple burners design was tested on a benchscale can combustor using methane gas for fuel. Effects of key design variables on combustion characteristics are evaluated. The key design variables include the air flow distribution within the combustor, swirler angle, injector position relative to the swirler, and air/fuel distribution. Emission results obtained by the optimization of these design parameters were Ox < 1 ppm (at 15 % 2), CO < 15 ppm (15 % 2), and negligible TC at the rated load in the engine test. ITRODCTIO Concern for serious environmental problems and legislation against harmful emissions are becoming more and more stringent worldwide. To meet this tendency, III executed a development program for a dry low Ox combustion system for the M1A-13A gas turbine in co-operation with Tokyo Gas Co., Osaka Gas Co., and Toho Gas Co., in As a part of this program, the first generation combustor which can achieve Ox level lower than 42.8 ppm had already been developed (Kajita et al., 1992). The development program of a second generation combustor has been in progress since The development goal is less than 2 ppm of Ox emissions and over 99.5 % combustion efficiency (equivalent to CO emission lower than about 25 ppm) between 75 and 1 % load on the engine by Lean pre-mix and parallel fuel staging with multiple burners were applied to the combustor design in this program. Development efforts made to reduce Ox emissions without increase of CO and THC emissions were as follows. 1.To achieve a lean fuel/air mixture and to reduce pressure loss, the diameter of the swirler and combustor were enlarged. 2.To obtain uniform fuel/air distribution, the burner shape such as fuel injector and swirler angle was optimized. 3.Air flow rate for eight main burners was held equal without much increase of pressure loss. 4.To increase combustion air and to decrease the cold region, liner and scroll cooling were reduced. By conducting a combustion rig test at midium pressure (.3 MPa), an air flow test at atmospheric pressure, mixing tests using a single burner, and a multiple burner combustor and engine test, the effects of following factors on combustion characteristics were evaluated. 1)Relative position of injector from swirler 2) Fuel distribution 3) Swirler angle of main burners 4) Air flow distribution within the combustor Presented at the International Gas Turbine and Aeroengine Congress and Exposition Houston, Texas - June 5-8, 1995 Downloaded From: on 12/23/218 Terms of se:

2 LOW Ox COMBSTOR The low Ox combustor consists of a can-type combustor liner and a burner module is as shown in Figure 1. The liner is made of sheet metal, double-wall construction and incorporates impingement and film cooling. The diameter of the combustion zone is enlarged to ensure sufficient residence time for complete combustion, but the exit geometry of the liner is the same as that of the current production design for retrofitting. The liner head contains one pilot diffusion burner at the center and eight main pre-mix burners surrounding it. The pilot burner has a conventional multiholenozzle. A small amount of pilot fuel is injected directly into the combustion zone. The pilot burner holds a stable diffusion flame to enhance combustion stability during low-load operation, but fuel is not provided near the rated load operation. The main burner consists of a swirler and a series of radial tubes with many fuel injection holes. The top end of the main burner is cooled by internal fuel flow. The main fuel is mixed with combustion air prior to entering the combustion zone. burner is stopped (See Figure 2). The fuel flow into the combustor is controlled by a single regulator valve. Each group of burners can be controlled by operating a solenoid valve. EXPERIMETAL EQIPMET Mixing test A mixing test was carried out for two purposes: one was to make a pre-mix burner with uniform fuel distribution, and the other was to evaluate the effects of mixture uniformity on combustion characteristics. The test facility is shown in Figure 3. The test was conducted at the atmospheric pressure, using 98.5 % of methane as a fuel The fuel distribution in the pre-mix burner was obtained by measuring the methane concentration at 153 points located 1 nun downstream of the burner exit. Fuel distribution in the combustor was obtained in the same way. The pilot burner was located downstream of the main burner exit (refer to Figure 1), so that the measuring plane was set 1 mm downstream of the pilot nozzle and the number of sampling points was 67. Air Compressor 2D-Traverser " " ' Fuel (Gt)Cylinlers FIQAnalyzer Fig.3 Schematic diagram of mixing test facility. Fig.1 Cross section of Low Ox Combustor. o o i 4, i i i i ioi L oho ff off P+2M P+4M P+6M 8M Fig.2 Burner groups. The eight main burners are devided into four groups of two. The pilot and one group of main burners are operated in idle condition and successive groups of main burners are operated as engine output increases. Finally, all the burners are operated and then the pilot Air flow test It is difficult to have uniform air flow distribution within the combustor for a single can type gas turbine owing to the tangential arrangement of the combustor (Refer to Figure 9). In the multipleburner combustor, biased air flow may cause an uneven air flow distnbutionfor each burner. If so, it is difficult to maintain ultimate lean pre-mix combustion in each burner. Consequently, a uniform airflow within the combustor is necessary to reduce the Ox emissions in the parallel fuel staging lean pre-mix combustor. On the other hand, a large amount of pressure loss due to arrangement of airflow causes a deterioration of gas turbine efficiency. Therefore, a flow guide which can provide an almost uniform combustor inlet air flow with little pressure loss was developed. Downloaded From: on 12/23/218 Terms of se: 2

3 To evaluate the performance of the flow guide, an air flow test was carried out. The air flow test facility was almost the same as that of a combustor rig test with scroll illustrated in Figure 4. The airflow velocity distribution was measured with a pitot tube around the combustor casing under atmospheric conditions (Refer to Figure 9-A). Furthermore, the combustor rig test with scroll was executed to evaluate the effect of the flow guide on combustion characteristics. Combustor rig test To estimate the effects of the above key design variables, a combustor rig test was carried out The test facility is shown in Figure 4. The test was completed, using 985 % of methane at.3 MPa air pressure. The air temperature and velocity were adjusted to simulate the engine operating conditions. Most of the rig tests were carried out in a uniform airflow condition without scroll. Test rig air flow was different from the flow condition in an actual engine, but results could be appreciated simply and clearly. The test results were corrected as follows. First, Ox emissions were converted to those of the engine operating conditions by using the conventional square root dependence on pressure of the thermally generated Ox. The validity of this conversion is rather doubtful, because there are some differences between the conditions of the rig test and those of the engine test, including the difference of fuel (98.5 % of methane was used in the rig test, while natural gas was used in the engine test) and flow conditions (for example, existence of a turbine and a compressor). But an approximate prediction of Ox emissions is possible by using it. Test Section Gas Analyzer (niform flow condition) u Cn Cylinders Test Section (with Scroll) Fig.4 Schematic diagram of combustor rig test facility TEST RESLTS Position of fuel injector To avoid a bum down of the swirler caused by flashback, the fuel injector was moved from the upstream to the downstream of the swirler. To assess this modification, Ox and CO emission data of the two combustor rig tests are shown as a function of the fuel/air ratio in Figure 5 (one is for fuel injected downstream of the swirler, and the other for upstream of it). The swirler angle of both burners was 49 degrees and the fuel distribution was almost uniform. The emission data show a saw-tooth pattern because the number of firing burners was changed according to the engine load. It can be said that emissions were very similar regardless of whether the fuel injector was located upstream of the swirler or downstream in "P+6M" and "8M" conditions. In low load operating condition ("P+2M" and "P-+4M"), the upstream injector type emitted more Ox and less CO than the downstream type. This result showed that the upstream type had a higher local fuel/air ratio at the burner exit at the same total fuel/air ratio. This can be explained from pressure loss data as follws. The upstream type had a higher pressure loss caused by combustion, so that the air flow rate of the operated burner was less than that of the downstream type in low load operating condition. Yri 1 5 x Z 1 Y i n 5 P1=.3 Pa p+2w T1=32 C -- Wa=1.8 kg/s r p+4w ' P+6Y Ox goal I Downstream pstream Fig.5 Effect of fuel injector position. Downloaded From: on 12/23/218 Terms of se: 3

4 In addition, almost the same results were obtained from the test with a 35- and a 55-degree swirler. Incidentally, it became difficult for the downstream injector type to control the fuel distribution, because the air distribution was changed by the swirler angle or shape. Fuel distribution To clarify the effect of fuel/air mixing, three burners (A, B, and C) were tested : rich mixture in the center(a), almost uniform distribution(b), and rich mixture around the outer side(c). Combustion test results of the multiple-burner combustor setting with eight each of these burners are shown with the fuel/air distribution data of mixing test for only one burner in Figure 6. The swirler angle of each burner was 49 degrees and the fuel injector was located downstream of the swirler. From these results, Ox emissions of burner B (uniform fuel distribution) were lowest, but it was unstable near the lean blowout limit. On the other hand, CO emission was almost similar irrespective of flame stability. The reason for this difference is that Ox is mainly formed in the hot gas zone generated by a locally fuel-rich mixture and flame is stabilized by this hot gas, while oxidation of CO is quenched in the cold gas region. The hot gas zone is mainly 3 (mm) 3 3 (mm) 3 3 (MM) 3 Burner A 3 (11111) 3 Burner B Conditions 6 P1 =.1 MPa 4 T1 =25 C 2 FAR=. 25 `'' FueI=CH4 3 (nn) 3 P1=.3 YPa P+2Y T1=32 c Wa=1.8 kg/s l.+4y :/ j Volume % 3 (MM) 3 Burner C formed by the burner, while the cold gas region is made not only by E a AL P+6 the burner but also by combustor wall cooling. Since it can be n 5 considered that fuel distribution has a strong effect on Ox emissions and lean blowout limit, it does not affect unburned gas emis- y c sions. After all, it is necessary for reducing Ox emissions to remove locally fuel-rich mixture and it is necessary for reducing CO Z x Ox goal emission to eliminate locally fuel-lean mixture, but localy fuel-rich 1 mixture is good for flame stability. Burner A `,l fv Moreover, the same test was carried out to clarify the cause of the Burner B slight difference in fuel distribution as shown in Figure 7. From this, emission results were almost similar in "8M" condition but the lean blowout limit was different by fuel/air mixing as in the results shown in Figure 6. Emission results were not the same in 'P+6M",'P+4M", and "P+2M" conditions, because there was a difference in quenching by no-firing burners. It can be said that the flame stability was changed delicately by the fuel distribution. 1 e 5 a n Burner C Fig.6 Effect of fuel/air concentration (1). 4 Downloaded From: on 12/23/218 Terms of se:

5 .- 1 Yin n 5 X z 1 E 5 a 3 (MM) P1=.3 MPa P+2M T1=32 C. 1 a=1 kg/s - Ox g I I P+4M P+6Y 8Y Burner D 1 Burner B ':: I Burner E Fig.7 Effect of fuel/air concentration (2). 3 (MM) Ke (m) 3 3 (m) 3 Burner D Burner E Swirler angle Figure 8 shows the effect of swirler angle of the eight main burners on emissions. Three lands of swirlers were tested: 35 degrees, 49 degrees and 55 degrees. All the burners had almost uniform fuel distribution and fuel injection upstream of the swirler. Ox emissions of the 35-degree swirler were less than those of the others in "8M" and "P+6M" conditions, but it had instability near the lean blowout limit_ The air flow rate of main burners was a little more than the rest of the burners, because it had a little wider air passage at the swirler. To the contrary, the local fuel/air ratio at the firing E 1 n 5 x Z 1 Yi E 5 P1=.3 MPa P+2P+4M T1=32 C a=1.8 k /s i f II I / - i Ir P+6Y I Ox goal f -d 55 swirler 49 swirler 35 swirler OL Fig.8 Effect of swirler angle. burner exit was higher in "P+4M" and 'P+2M" conditions, because the pressure loss caused by combustion affected the burner inlet air flow rate more in the case of large air passage at the swirler. Considering this, as for the parallel fuel staging combustor, it seemed that emissions did not depend on the swirler angle of main burners very much. In addition, combustion noise level of the combustor with the 55-degree swirler was higher than that of the others. It was thought, therefore, that the 49-degree swirler was most suitable. Air flow within the combustor Previously, the flow guide vane shown in Figure 9-A was applied to distribute the air flow evenly within the combustor. For the sake of low Ox combustion, scroll cooling was reduced and the air flow rate of the main burners was increased in this development program. Consequently, pressure loss by this flow guide increased. So a new flow guide body was devised (See Figure 9-B). Figure 9 shows the results of an air flow test and a schematic diagram of both flow guides. The results of the combustor rig test with scroll and of the engine test are shown together. This flow guide body had a good performance in temps of air flow with the lowest pressure loss. Ox emissions with the im- Downloaded From: on 12/23/218 Terms of se: 5

6 [JI Flay B A Pitot Tube ' - 7 Measuring Point(16) Fli f Flow Velocity Vector This length moans sectional average velocity + B B m c A B l 2 4 G 78 Circumference Circumference Figure 9 A Air flow test result Figure 9 B Air flow test result for Flow Guide Vane. for Flow Guide Body. m & 1 1 E5O j / C- >< 5 1 -, _ 5 C4, C Figure 9 C Rig test results (with scroll). k\, 2!3 2 s' LOAD (%) Figure 9 D Engine test results. 6 Downloaded From: on 12/23/218 Terms of se:

7 ae LO proved flow guide body were less than those with the flow guide vane. On the other hand, their CO emissions were almost similar in the rig test with scroll. In addition, total pressure loss with the flow guide body was less than that with the flow guide vane by almost 15 %. Ox emissions were changed by the increase of the main burner inlet air flow, but CO emission was not changed because of the trade-off relationship between the amount of combustion air and cooling air. (Reducing pressure loss caused a decrease of cooling air.) Moreover, this improvement in pressure loss had a favorable effect on engine efficiency, so that further lean pre-mix combustion became possible in engine operating condition. As a result, low Ox emissions were observed in the engine test by using the improved flow guide body, but CO emission was increased, accompanied by an increase of engine efficiency. Engine test The engine test was completed on the M 1A- 13A gas turbine using natural gas for fuel. Typical Ox and CO emission results of the engine test are shown in Figure 1. Burner D (refer to Figure 7) was used as the main burner because it had low Ox emissions and combustion stability near the lean blowout limit. Emissions are shown as a function of engine load. Ox emissions were less than 1 ppm and CO emission was less than 15 ppm at the rated load. The fuel distribution in this combustor is shown in Figure 11. There are many speckles, so that it is possible to reduce Ox and CO emissions further by improving the fuel distribution, but then the problem of the combustion instability must be cleared up. SMMARY AD COCLSIOS 1. 5 x Z Ox, CO i P+211 P+6Y P+4Y Ox goal..8m LOAD (%) 1 to 5 E a a All Burners Open (P+8M) / FAR=.2 14 (mm) ',, to Volume % Fig.1 1 Fuel/air concentration of combustor. The effects of the key design variables on combustion characteristics were evaluated by bench-scale rig tests of the parallel fuel staging lean pre-mix combustor at.3 MPa. As a result, the following conclusions were obtained. 1)Whether fuel was injected upstream of the swirler or down stream of it, emissions are almost similar. 2)Fuel distribution is closely related to Ox emissions and flame stability. 3)Swirler angle dose not have much influence on emissions, but it is necessary to be careful, because it has a close relation to the air flow rate of the burner and pressure loss. 4)It is good for low Ox combustion to distribute the combustor inlet air uniformly. By these optimizations, Ox emissions less than 1 ppm,co emission less than 15 ppm, and negligible THC emission were observed at the rated load in the engine test, but there is a problem in the turndown ratio with low unburned gas emissions. Expanding the range of low emissions with high combustion efficiency, we hope to commercialize this combustor as soon as possible. REFERECE 1. Kajita et al. "Development of a dry low Ox combustor for 1.5 MW Gas Turbine," ASME Paper o.93-gt-393 Fig.1 Emissions results of engine test. Downloaded From: on 12/23/218 Terms of se: 7