Copyright 1994 by ASME SOLAR TURBINES INCORPORATED "TAURUS 60" GAS TURBINE DEVELOPMENT

Transcription

1 THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS y^ 345 E. 47th St., Ne York, N.Y C Yi C The Society shall not be responsible for statements or opinions advanced in l ^J 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 in an ASME Journal. Papers are available from ASME for 15 months after the meeting. Printed in U.S.A. Copyright 1994 by ASME 94-GT-115 SOLAR TURBINES INCORPORATED "TAURUS 60" GAS TURBINE DEVELOPMENT Vern Van Leuven Department of Mechanical Design Engineering Solar Turbines Incorporated San Diego, California ABSTRACT The Taurus gas turbine as first introduced in 1989 ith ratings of 6200 HP for single shaft and 6500 HP for tin shaft configurations. A ne design of the single shaft third stage turbine rotor and exhaust diffuser brought its poer to 6500 HP in A program as initiated early in 1992 to identify opportunities to further optimize performance of the Taurus. Thorough investigation of performance sensitivity to thermodynamic cycle parameters has resulted in significant improvement over the original design ith no change in firing temperature. Aerodynamic and mechanical design changes ere implemented in 1993 hich raised Taurus performance to 7000 HP and 32% thermal efficiency. Selection of the final design configuration as the outcome of performance maximization versus cost increase, durability risk and loss of commonality ith previous engines. This paper details these changes and the design selection process. INTRODUCTION History The Taurus 60 gas turbine (Figures 1 and 2) as originally introduced to the market in Table 1 summarizes original ISO performance and significant features of the single and to shaft versions. The engine configuration is similar to the Solar Centaur 50 ith the most important changes being an additional stage at the front of the compressor and a ne design to stage poer turbine for the to shaft version. The compressor stage is a scale of the Solar Mars first stage blade hile the to shaft poer turbine as a ne design. In 1991 the third stage turbine and exhaust diffuser for the single shaft version as redesigned. Engine performance improved by 390 HP and -534 BTU/HP-hr. Quantities of engines produced from 6/90 to 10/93 ere (22) 6200 HP single shaft, (27) 6500 HP single shaft and (39) 6500 HP to shaft engines. Program Objectives Industrial gas turbine designs in all size ranges are continually responding to evolving application demands for higher engine performance and the Taurus 60 is no exception. Competitive challenges and the potential for design improvement using advances in design and analysis technology motivated Solar to initiate a development program for the Taurus 60. Goals of the program ere to deliver a product to the market in 18 months, maintain existing cost normalized by poer, incur no sacrifice in durability and produce performance shon in Table 2. Program Summary The program objectives provided a tough challenge to the ne product introduction (NPl) teaming approach nely implemented at Solar. The basis of NPI teaming is to assign responsibility for product development to a cross functional team so that communication and participation throughout all disciplines are maximized. An NPI team as assigned by management and the team's first task as to complete conceptual product design. Fundamentally there are to ays to improve the performance of a gas turbine. The cycle (pressure ratio and firing temperature) can be changed or the component efficiency can be increased (this includes reducing cooling air flo and leakage). Brainstorming sessions ere held to create a list of options to achieve the program objectives ith the criteria used in the evaluation being performance, durability risk, cost, development time and product commonality. Once preliminary design as completed and the optimum configuration determined detail design ork as initiated. A timeline as prepared summarizing the program, and tasks ere organized ith consideration to component lead times, manpoer and test rig availability. More refined performance, temperature and stress evaluations ere begun and definition for necessary testing as created. Previous experience from Centaur 50 and original Taurus 60 ere also heavily utilized in this program. Presented at the International Gas Turbine and Aeroengine Congress and Exposition The Hague, Netherlands June 13-16, 1994

2 FIGURE 1. TAURUS 60 SINGLE SHAFT ENGINE CROSS SECTION RE93133 FIGURE 2. TAURUS 60 TWO SHAFT ENGINE CROSS SECTION Both the single and to shaft designs ere completed on schedule and the cost target as achieved. Performance of engines tested to date (Table 2) sho that performance goals ere also exceeded. DISCUSSION Conceptual Design A substantial period of time as spent identifying all potential design changes hich might be made to achieve the performance goal. A list of options as compiled and evaluated using the established criteria as illustrated in Table 3. Several different combinations of the potential modifications ould satisfy the performance goal but most involved compromises in cost, schedule and risk. A configuration hich increased firing temperature and pressure ratio as seriously considered in the beginning of the conceptual stage. This required additional stages added to the aft end of the compressor and a ne gas producer turbine design. The compressor stages ould be scaled form existing Solar Mars engine stages and ould not require a lengthy development. Hoever, the ne GP as to have ne nozzles and ould require a directionally solidified first stage GP turbine blade. The third stage nozzle ould also have to be redesigned to provide the connect flo characteristics. Although this configuration met the performance goals, the increased engine cost of about 15%, PA

3 TABLE 1. ORIGINAL PERFORMANCE OF TAURUS 60 ENGINES Parameter Original Single Shaft Taurus 60 Original To Shaft Taurus 60 Poer (1) 6200 HP (Pre 1992) 6450 HP Cycle Efficiency (1) 29.6% (Pre 1992) 30.3% 6500 HP 31.0% Mass Flo 46.0 Lbm/s 46.0 Lbm/s TR1T 1850 F 1850 F Pressure Ratio Exhaust Temperature 901 F 910 F Speed RPM RPM GP RPM PT (max) Compressor Stages Variable Comp Stages 4 4 Turbine Stages 3 4 Note: 1. All values represent no duct losses, natural gas fuel and 59 F, sea level, 60% relative humidity air. TABLE 2. PERFORMANCE OF UPRATED TAURUS 60 ENGINES Engine Poer Efficiency Single Shaft Taurus HP 32.0% Program Goal To Shaft Taurus HP 32.0% Program Goal Single Shaft Taurus HP 32.0% Average - 8 Engines To Shaft Taurus HP 31.9% - 1 Engine Note: 1. All values represent no duct losses, natural gas fuel and 59F, sea level, 60% relative humidity air. longer program development time, and loss of commonalty ith the current fleet of Taurus engines led to closer evaluation of other options. The most attractive configuration as found by improving component efficiencies. This could be accomplished in a shorter time hile maintaining fleet commonalty, lo engine cost, and ithout increased durability risk. Efficiency improvements ere made for both compressor and turbine components in the ne Taurus 60. In the compressor, the 0-stage blade as modified, tip clearances ere reduced, improvements ere made in surface finish and the guide vane settings ere optimized. The turbine also had the tip clearances reduced as ell as leakage and cooling flo reduction. One option that as in the concept initially as a longer exhaust diffuser for both single and to shaft engines. This option as eliminated because the performance gain did not outeigh the significant package changes required and loss of opportunity for customers ishing to uprate their units. In this instance, the early design conceptualization involving Solar Package Engineering and Customer Services groups facilitated by the NPI process as of great benefit.

4 TABLE 3. MATRIX OF PERFORMANCE IMPROVEMENT OPTIONS Modification Performance Cost Risk Duration Sensitivity Raise TRIT High High Med 24 M Ne 1 st, 2nd GPT Blade Mat. Raise Pressure High High Med 18 M Ratio Add Comp Stages Redesign GP High High High 30 M Turbine Redesign PT High High High 30 M Improve Diffuser Med High Lo 24 M Ne Package Reduce Cooling, High Lo Lo 12 M Leakage Flos Optimize 0 Stg Med Lo Lo 6 M Comp Blade Ne Air Inlet Unknon Med Lo 12 M Reduce Comp High Lo Lo 4 M Tip CL Med /Rub Coat Reduce Turb High Lo Lo 4 M Tip CL Improve Comp Lo Lo Lo 4 M Surface Finish Optimize Comp Med Lo Lo 2 M Guide Vanes KWl^zmm Zero Stage Blade. The zero stage blade as a scale of the Mars first stage compressor stage hich as originally designed in Recent mechanical analysis of the blade ith finite element methods revealed that the deflections due to centrifugal and aerodynamic loading ere slightly different than hat as originally calculated. Today's capability of finite element softare to account for large deformations here the load vector relative to the position of the structure changes significantly ith deflection (Figure 3) as not available at the time of original design. It is extremely difficult to predict by analysis alone compressor performance sensitivity to changes in blade incidence thus rig or engine testing is essential in the evaluation process. Rather than just correcting a knon problem, testing as used to further optimize the stage incidence. The objective as to increase flo as much as possible ithout sacrificing efficiency. The airfoil incidence angle as modified and back to back engine testing as performed to compare against the original design. One compressor case half as removed, the blades ere changed out and the engine as rerun all in the same day. By changing out blades in the test stand a minimum amount of uncertainty as introduced in the evaluation. The result of several tests as a blade incidence angle hich produced additional airflo and no detriment to compressor efficiency. Increase in poer is summarized in Table 4. There ere no additional material costs associated ith the change in ne production and no mechanical risks ere introduced. Compressor Tip Clearance Reduction Compressor performance is knon to be very sensitive to tip clearances. An optimum value of running clearance is generally believed to be beteen % of chord length. Finite element 4

5 TABLE 4. RESULTS OF PERFORMANCE ENHANCEMENTS Modification Zero Stage Blade Tist Compressor Tip Clearance Reduction Compressor Guide Vane Optimization GP Turbine Tip Clearance Reduction Single Shaft PT Cooling Reduction To Shaft PT Cooling Reduction Piston Seal Leakage Reduction 114 HP 0.0% Efficiency 180 HP 0.7% Efficiency 50 HP 0.4% Efficiency 77 HP 0.37% Efficiency 92 HP 0.29% Efficiency 157 HP 0.45% Efficiency 110 HP 0.19% Efficiency Performance Improvement U 1 - U2 = Difference in orientation of structure to load vector after loading. REM1 FIGURE 3. ZERO STAGE COMPRESSOR BLADE DEFLECTION UNDER LOAD modelling of the compressor rotor and a component tolerance stackup study shoed that there as potential for optimization of tip clearances. Adequate margin of safety must be included in the design because heavy nibs in the compressor cannot be tolerated. A rub can result in material build up at the contact area hich may not clear aay. If this happens, the rubbed material buildup can continue to gro resulting in damage to the blading. Also, if a stationary airfoil rubs against the rotor heating from friction can cause distortion, further tubbing, loss of material strength and possible component failure. A tub tolerant coating could be introduced to permit occasional rubs by design, hoever the expense of a rub coating and the risk of it not adhering sufficiently outeighed the benefit of tighter clearances. It as decided to reduce tip clearances as near as possible to optimum ithout permitting the possibility of tub. Finite element methods ere used to calculate the rotor radial groth due to thermal and centrifugal loads and static structure thermal displacements ere calculated. Positional tolerance of the rotor relative to the cases involved a stackup of four features on the front end and three features on the aft end hile diametral tolerance consisted of to features. Clearance values ere then set so that rubs ould never occur. To confirm proper machining and assembly, clearances ere measured during build of each engine by alternately removing each compressor case half and measuring gaps ith feeler gages. A development test engine as run to verify analytically predicted minimum clearances by brazing in small diameter instrumentation tubing to act as rub pins. Worst case thermal transient condition clearance as measured by the pins and compared favorably ith design predictions. Performance improvement due to tip clearance reduction as determined both by comparison of test engine performance built ith the tighter clearances against data from previous production engines, as ell as a back-to-back test ith the compressor returned to original clearances. The measured performance sensitivity to tip clearance compared ell ith analytical predictions. No cost and minimal durability risk are associated ith this change. Air Inlet Evaluation Another area for potential performance improvement analyzed as the air inlet collector and duct. Previous testing ith traversing probes shoed a potential flo distortion in the air entering the compressor inlet guide vane. Computational fluid dynamics (CFD) modeling ith the ACE (copyright CFD 5

6 Research Corp.) computer code also shoed some potential flo field distortion (Figure 4). Hoever, the flo velocities in this area are lo and the influence of this distortion on compressor performance as not knon. A back-to-back test as designed to sho the potential for compressor performance improvement before changing the production design. Static pressure taps along ith Kiel Probes ere used to sho the flo distortion, and engine performance as tested, first ith the inlet collector then ith the collector removed (see Figures 1, 2). There is adequate space on the engine package so that negligible flo distortion occurs entering the compressor ith the collector removed. This simulates the loest flo distortion expected from a collector redesign. The test revealed that any flo distortion created in the air-inlet as not reducing engine performance. Therefore, no change to the existing air intake as made for the ne Taurus 60. the results are summarized in Table 5. The 5380 DP coating unfortunately did not meet its roughness objective. Still, the finish as better than previous coating so evaluation continued. TABLE 5. SURFACE ROUGHNESS TEST RESULTS Component Seimatel 725 Sermatel 5380DP Compressor Blade 65 pin 60 on Compessor Vane 51 pin 36 pin Note: Roughness measure ment made using 0030.in cutoff length profilometer. Back to back performance testing ithout introducing variables more significant than the predicted benefit from the coating could not be performed. To be conservative, no performance increase as predicted from the ne coating. Since cost and durability risk ere lo and there may also be a benefit of improved dirt fouling resistance the ne coating as implemented in the design. RE93135 FIGURE 4. STATIC PRESSURE DISTRIBUTION IN COMPRESOR INLET Compressor Surface Finish The compressor flopath materials are stainless steel and ductile iron hich require coating for optimum corrosion protection. The fact that the surface finish of the compressor has a significant effect on performance is obvious to anyone that has ever ashed a dirty gas turbine in service. Hoever, the relationship of uniform surface roughness measured ith a profilometer to compressor performance is not ell quantified. Taurus 60 compressors had previously been coated ith Sermatel 725 coating ith a typical surface roughness of 65 pin. Sermatel also offers a coating ith a surface roughness objective of 10 pin (0.010 in cutoff) called Sermatel 5380 DP ([Mosser, 1988]. Since the difference in cost as not great and the corrosion protection as expected to be as good as the 725 coating, the team chose to pursue the 5380 DP coating. Surface roughness testing as performed on a Solar airfoil uncoated and ith both coatings and Compressor Guide Vane Optimization Recent expertise gained by the aerodynamic group in the use of a poerful multivariable optimization routine hich is part of the RS1/Discover (copyright BBN Softare Products Corp.) computer program permitted fast and accurate optimization of the guide vane settings. Optimization variables ere individual positions of the four guide vane stages and the objective variables ere engine horsepoer and specific fuel consumption. A matrix of the variables as first generated using 17 different combinations of guide vane settings. The computer program fits a regression to the data and then solves for the vane settings at maximized objective variables. Some iteration around the solution is required as the regression fit improves ith more data. Figure 5 displays the Taurus 60 compressor efficiency as a function of flo ith isometric lines shon for poer and cycle thermal efficiency. Points ere plotted on this curve during the optimization to graphically sho ho the compressor flo and efficiency are interelated and hat tradeoffs must be made hen choosing cycle poer or efficiency as an objective. Optimization as performed using to different production engines to improve accuracy of the result as compared to the average of production engines. Results from this testing is shon in Table 4. No durability impact or cost increase is incurred ith the change of guide vane settings. Turbine Tip Clearances In the attempt to reduce turbine tip clearances, teaming beteen the design and manufacturing engineers proved beneficial. Manufacturing suggested an improved machining process that ould reduce the tolerance on the first and second stage turbine nozzle tip shoes. Previously the rub tolerant coating

7 o +2 >- C Z LL +1 4 U- xr0 0 0 U I m C -1 ^ 'Sv x o`yx?o0 moo G Thermal Efficiency % inches, ithout reducing the minimum clearance. Previous analytical ork, testing and field experience had already demonstrated that the minimum clearance should not be reduced. Tip clearances of the poer turbine stages had been fully optimized in the original development program. Turbine Cooling Flo Reduction The evaluation of reducing cooling flo began ith a cycle analysis shoing the sensitivity of cooling air use and leakage at each stage. Figure 7 and Table 6 sho the sensitivity of cycle poer and efficiency on cooling flos and leaks. With these results the team established a very high priority for evaluating flo reductions in the poer turbine. The analytical investigation began ith an extensive netork flo analysis using previous test data to correlate the model. Cooling and leakage changes could then be quickly evaluated. Table 3 summarizes the potential flo reductions that ere identified from this study AIRFLOW, Ibis FIGURE 5. TAURUS 60 COMPRESSOR PERFORMANCE MAP RE93143M had been applied to the integral tip shoes (Figure 6) ith dimensional control based on stackup of individual component tolerances. The.improved process is to machine the coating to a finish dimension referenced from the locating features of the nozzle. This alloed the reduction of the nominal tip clearance by RE93144M :OOLING FLOW SCHEMATIC AGE. It as suspected that there could be the piston ring seal on the first stage paint tests confirmed the high leakage ;ak paths inherent in the design. A ne h incorporated several ideas to reduce us leaks beteen the nozzle rails and ur hen the nozzle rail distorts under sized by trapping the rails in a rigid ised at the top of the rails to further ing is no assembled into the carrier ing around the circumference. The ne ;ing comprehensive thermocouple 31 paint. Back to back tests of the old )ns indicated that, as predicted, leakage Quantification of the performance accomplished ith the instrumented -rs felt that a field durability test ould iroduction implementation. Since the program could be met ithout the reed to delay implementation of this i 5000 hour field testing results are

8 TABLE 6. TURBINE COOLING FLOW PERFORMANCE SENSITIVITY Location Horsepoer Cost (HP) Cycle Efficiency Cost A B 0 0 C D E F G H I J Note: 1. See Figure 7 for location schematic. 2. Performance penalties are normalized to 1% compressor mass flo. 1st STAGE CARRIER PISTON NOZZLE RING SEAL C-SEAL HOUSING FIGURE 8. GAS PRODUCER TURBINE PISTON RING SEAL CONFIGURATION Poer Turbine Disk Impingement Flo. The performance sensitivity study directed the most in depth optimization of cooling flos to the poer turbine section. Analytical evaluations of both single and to shaft turbines and previous testing during RE93138M FIGURE 9. IMPROVED GAS PRODUCER TURBINE PISTON RING SEAL CONFIGURATION

9 TABLE 7. DISC METAL TEMPERATURE SUMMARY Disk Location Original Taunts 60 Single Shaft 3rd Turbine To Shaft (1) 3rd Turbine To Shaft (1) 4th Turbine Uprate Taunus 60 Change Rim 950 F 960F +16 F Bore 790 F 800 F +10 F Rim 975 F <1060 F + <85 F Bore 730 F <780 F + <50F Rim 960 F <1060 F + <100 F Bore 490 F <540 F + <50 F Note: 1. Upper limit temperatures are from preliminary tests. Results of final development testing ere not available at the time of this riting. the original design phase of the poer turbine indicated that impingement flos to the poer turbine disks could be completely eliminated ithout significant rise in disk temperatures. The main reason for the opportunity in the design is that originally the cooling system as sized for a potential thermal uprate. As previously mentioned, an increase in firing temperature as not part of this program. Testing of both the single shaft and to shaft turbines as performed ith instrumentation as shon in Figures 10 and 11. Disc metal temperature increases from complete elimination of cooling air is summarized in Table 7. Performance evaluation could not be performed ith the heavily instrumented test configurations so final confirmation of the improvements as taken from first production tests. From Table 4, it can be seen that the cooling air reductions in the turbine ere responsible for a significant portion of the performance improvement of the program. Excellent agreement of actual engine performance ith predictions thus validates the flo and sensitivity analyses. Elimination of impingement flos served to simplify both the nozzle machining and casting resulting in significant cost reduction. The cooling passage brazed covers, cast cooling cavity and impingement holes ere eliminated. Single Shaft Exhaust Diffuser Improvements Although the longer exhaust diffuser concept as determined to be impractical, other methods of improving the diffuser's performance ere also evaluated. The previous exhaust diffuser had 0.12 in thick sheet metal that protruded into the flo path. This causes a detriment to performance because of the upset flo along the alls and inhibited diffusion. In the quest to eliminate the flo-path step the manufacturing and design engineering team orked to improve the design. The shop floor orker ho personally manufactures the diffusers provided the most valuable ideas and a design change as realized that not only eliminated the step but reduced cost and manufacturing lead time. The success of this effort reveals the value of soliciting design input through our cross functional teams. Metal Temperature Air Temperature Pressure RE93139M FIGURE 10. SINGLE SHAFT TURBINE INSTRUMENTATION DIAGRAM 9

10 12,920 11,850 10, O D = NpT, % of 14,300 rpm \SO NGP = 15,000 rpm 5380 [ ---fa T5 = 1400 F AMBIENT TEMPERATURE, OF RE93145M FIGURE 12. TWO SHAFT POWER TURBINE THRUST VERSUS ROTOR SPEED SHAFT CENTERLINE Metal Temperature Air Temperature A Pressure RE9314C 1 FIGURE 11. TWO-SHAFT TURBINE INSTRUMENTATION DIAGRAM To Shaft Version Poer Turbine Thrust Bearing The increase in pressure drop through the poer turbine of the to shaft version of the Taurus 60 resulting from the compressor performance enhancements mandated a revie of the thrust bearing load carrying capacity. Figure 12 shos the predicted thrust as a function of rotor speed. The thrust bearing design must analytically demonstrate a minimum oil film thickness of in and a maximum pressure-temperature severity ratio of 0.5 throughout the design speed range. Pressure-temperature severity ratio is a measure of the bearing babbitt material yield stress margin at a given temperature and pressure. The existing fixed geometry tapered land thrust bearing (Figure 13) did not satisfy the criteria so a tilting pad bearing ith a larger area as designed (Figure 14). Film thickness and pressure-temperature severity ratio are plotted against speed for several ambient temperatures in Figure 15. Engine testing ith load cells and temperature sensors in the thrust bearing verified analytical predictions. COLLAR 1-1,,n HOUSING RE93141 M FIGURE 13. ORIGINAL POWER TURBINE THRUST BEARING DESIGN CONCLUSION Thorough revie of the Taurus 60 design from a team representing all disciplines of Solar's industrial gas turbine business resulted in an improvement in product performance hich met the program performance goals. The program schedule as met and no increase in product cost or durability risk as introduced. The performance improvements have been verified by 9 production engine tests. The key to the success of the program as the NPI teaming concept because it insured continuing involvement and product onership from a ide area of expertise. 10

11 Responsibility to a focused group rather than as traditionally to a functional area contributes primarily to these motivations. As a result, critical evaluation of the design criteria ere applied to each concept in a systematic and unbiased fashion. REFERENCES Mosser, M. F., 1988, "An Improved Coating Process for Steel Compressor Components - SemieTel Process 5380 DP", SAE Technical Paper E 1.2 J LL 1 J Ta = 120 F Ta = 68 F ^Ta 0.2 SHAFT CENTERLINE I I I I I ' Q 0.5 SPEED, rpm (000) RE93146M THRUST I ILl PAD COLLAR THRUST BEARING RE93142M FIGURE 14. NEW POWER TURBINE THRUST BEARIN DESIGN 0.4 a U) 0.3 D 0.2 a 0.1 D 0 W a SPEED, rpm (000) RE93147M FIGURE 15. TWO SHAFT POWER TURBINE THRUST BEARING PERFORMANCE 11