New Technique for Online Washing of Large Mechanical-Drive Condensing Steam Turbines

NEW TECHNIQUE FOR ONLINE WASHING OF LARGE MECHANICAL-DRIVE CONDENSING STEAM TURBINES

by Gampa I. Bhat Chief Machinery Engineer ExxonMobil Chemical Company Baytown, Texas Satoshi Hata Manager, Turbine Design Section Kyoichi Ikeno Mechanical Engineer, Turbine Design Section and Yuzo Tsurusaki Mechanical Engineer, Turbine Design Section Mitsubishi Heavy Industries, Ltd. Hiroshima, Japan

ABSTRACT Gampa I. Bhat is Chief Machinery To compete in today’s economic climate, petrochemical plants Engineer for ExxonMobil Chemical Compa- are strategizing on continuous long-term operation to reduce main- ny, in Baytown, Texas. As Lead Specialist, tenance costs and increase productivity. This strategy has led some he acts as the focal point for the plants to go from eight years between turnarounds to 10 years. For ExxonMobil Chemical Worldwide Machin- rotating machinery such as mechanical-drive steam turbines, one ery Network and is involved with the factor that affects this strategy is heavy deposition on steam turbine development of machinery strategies for internals, caused by impurities in the steam. These impurities result new and upgrade projects. He is also in fouling on the blade and nozzle path surfaces due to contami- involved in the selection, operation, main- nated materials such as silica and sodium in the steam. As a result, tenance, and troubleshooting of machinery turbine performance tends to deteriorate gradually. systems. This paper introduces an innovative online washing technique to Mr. Bhat received his B.S. degree (Mechanical Engineering) minimize the impact caused by fouling of the steam path for large from Karnataka University in India, and an M.S. degree from West multistage condensing steam turbines. This technique, although Virginia College of Graduate Studies. He is a member of ASME. applied here to extracting-condensing turbines, is also applicable to large condensing turbines. The new technology has water Satoshi Hata is a Manager of the Turbine injection nozzles located in the steam chest of the extraction valve Design Section in the Turbomachinery rack. The injection nozzles are manifolded to a water supply Engineering Department, Mitsubishi Heavy source, which controls a setpoint temperature, by controlling the Industries, Ltd., in Hiroshima, Japan. He water injection rate. The objective is to directly wash off deposits has had experience with R&D for nuclear adhering to the blades and nozzles on the low-pressure side with uranium centrifuges, turbomolecular pumps, minimal power turndown, and without impacting the turbine’s and heavy-duty and aero engine derivative long-term performance. Erosion damage and thermal stress of type gas turbines and steam turbines for 21 internal parts such as chest valves and blades due to the injected years. water had to be taken into consideration. Mr. Hata has B.S. and M.S. degrees To properly achieve this objective, and considering the potential (Mechanical Engineering) from Kyushu for damage during the online wash, a new extraction valve box had Institute of Technology. to be designed. The new design had to consider the effects of optimizing the mixing zone of the steam and water injection to generate a specific Kyoichi Ikeno is the Mechanical particle size, moisture propagation through the condensing section, Engineer of the Turbine Design Section in mechanical deflection of stationary components, and the overall the Turbomachinery Engineering Depart- thermodynamic analysis of each stage during the online wash. A ment, Mitsubishi Heavy Industries, Ltd., in prototype model was built and several experiments carried out Hiroshima, Japan. He is a blade design based on the practical operating condition of actual steam turbines. specialist and has seven years of experience This paper discusses the evaluations made from the model, by pre- with R&D for synthesis gas compressor senting the thermodynamic analysis results, and the finite element steam turbines and gas turbines. analysis (FEA) that was used to evaluate the strength of the internal Mr. Ikeno has a B.S. degree from Miyazaki parts during actual online washing. University, and an M.S. degree (Mechanical The final design was a compact extraction box that could replace Engineering) from Kyushu University. existing models without any machining of the casing. To date there are two such installations worldwide. These are operating effectively 57 58 PROCEEDINGS OF THE THIRTY-THIRD TURBOMACHINERY SYMPOSIUM ¥ 2004 without incident. This paper also discusses online washing test results, COMPOSITION % CONCENTRATION % which were obtained using an actual steam chest with the special 100% 10% injected nozzle and a risk assessment of online washing in general. Si INTRODUCTION NaCl During steam turbine continuous long-term operation, steam con- SiO2 taminants, such as silicate and sodium, deposit on the internals as 50% 5% solids. This occurs under certain operating conditions related to steam Na pressure and temperature for each of the stages. These contaminants foul the surfaces of nozzles and blades and gradually build up during steam turbine operation. Figure 1 shows the typical steam turbine operating condition for an ethylene plant application. Figure 2 shows NaOH Na2SiO3 the fouling condition after seven years of continuous operation. 0% HP PART MP PART LP PART

BFW Si<0.5ppb WATER BOILER TREATMENT NaPO4 Figure 3. Distribution of Compounds along Steam Path. INLET STEAM CONDITION 1500 PSIG 910 DEG.F 800 KLB/H the application of new technologies in the design of the extraction AFTER EXT.PRESS. valve rack. The injection nozzles are manifolded to a water supply INLET Paext TEMP. Tin source, which controls a setpoint temperature, by controlling the water injection rate. This new technology places water injection nozzles in the SPEED 3800 RPM AXIAL FORCE. extraction steam chest with a controlled water supply source. The BL:8600kgf HP:4300kgf LP:13000kg MAX POWER 49500 BHP Fa f POWER RECOVER 2000 BHP 4688 LBS(MAX FLOW) AFTER WASHING advantage of this compact design is its ability to directly wash off deposits adhering to blades and nozzles in the low-pressure zones, by controlling the amount of moisture at each stage. Consideration for this new design factored in erosion damage and thermal stresses EXT. STEAM CONDITION EXHAUST STEAM 650 PSIG CONDITION 615 KLB/H 2 IN HG ABS of internal parts, such as chest valves and blades, induced by the 185 KLB/H water injected. Figure 1. Typical Operation Condition. A prototype model was built and several experiments carried out based on the practical operating condition of actual steam turbines. This paper discusses the evaluations made from the model, by pre- senting the thermodynamic analysis results and the finite element analysis (FEA) that was used to evaluate the strength of the internal parts during actual online washing. In addition, online washing test results were obtained by using actual steam chest conditions equipped with these special injection nozzles, and a risk assessment produced. CONVENTIONAL STEAM TURBINE ONLINE WASHING PROCEDURE Figure 4 shows the trend over time of the relationship between condensing flow and after-extraction pressure, comparing design pressure and actual pressure measured during operation. The after- extraction pressure increases due to the decrease in throat area caused by the deposition of chemical materials.

LCT01 Fouling Monitoring Figure 2. Fouling Condition after Seven Years’ Continuous

Operation. 350

3㨪4 MONTHS

The composition and characteristics of the fouling materials are 5㨪6 MONTHS different along the steam path as we move from the high-pressure 300 stages to the low-pressure side, as shown in Figure 3. Under these fouling conditions, the pressure profiles across the nozzles, blades, 5㨪6 MONTHS and throat areas are increased. These profiles in turn result in dete- 250 Aug - Sep 2000 97WASH㨪3Q99 Jan - Mar 2001 rioration of turbine performance, and will continue to do so over Jul - Sep 2001 After Extraction Pressue (psig)Pressue Extraction After Jan - Mar 2002 Apri - Jun 2002 time if left unattended. 200 Clean Turbine per MHI Current online washing techniques involve dropping the steam Maximum Allowable Pressure inlet temperature to increase the moisture at the condensing stages for the purpose of generating enough moisture to wash the soluble 150 0 50 100 150 200 250 300 salts. This procedure cannot eliminate insoluble contaminants, Condensing Flow (klb/hr) especially in the high-pressure stages. By reducing the steam inlet temperature, enthalpy across the turbine drops significantly—this Figure 4. Internal Deposition Trending. can result in production losses. In order to resolve this fouling problem, an innovative online As mentioned previously, current online washing techniques washing technique was developed. This new technique incorporates involve dropping the steam inlet temperature to increase the NEW TECHNIQUE FOR ONLINE WASHING OF LARGE MECHANICAL-DRIVE CONDENSING STEAM TURBINES 59 moisture at the condensing stages for the purpose of generating a clean turbine when inlet temperature was changed. In addition, it enough moisture to wash the soluble salts. This procedure cannot was found that the extraction stage and after-extraction stage eliminate insoluble contaminants, especially in the high-pressure remained in the superheated zone. (HP) stages. By reducing the steam inlet temperature, enthalpy across the turbine drops significantly, which can result in produc- THERMODYNAMICS ANALYSIS RESULTS tion losses. FOR A FOULED CONDITION The effects of this conventional procedure can be seen from the It is well known that when fouling occurs, it increases the deposition and pressure profile in Figure 5. The inlet steam tem- pressure drop across the particular stage. This increase then perature is decreased thereby increasing the moisture within the cascades through the steam path causing backpressure. This back- low-pressure (LP) stages in order to wash several kinds of fouled pressure can be monitored at the stage after extraction, as materials off the nozzles and blades. The concentration and com- previously shown in Figure 4. Knowing the after-extraction position of these chemicals under dry and wet conditions are pressure, the thickness of the cumulative deposits can be deter- shown in Figure 3. mined both before and after washing. Figure 7 illustrates the fouling profiles generated from these findings. This chart shows FOULING WASHING OFF that a change in nozzle area for the stages after extraction has a NOZZLE AREA INCREASE SODIUM & SILICA PRESSURE DROP CONCENTRATION ppb definite effect on pressure rise resulting from deposition thickness. The deposition thickness up to the maximum operating pressure 900 degF Tin 1000 limit for the fouled stages can range from 0.5 mm to 0.6 mm (0.02 DEG.F 350 PSIG SODIUM inch to 0.024 inch). Figure 7 also illustrates that stages downstream NOZZLE THROAT AREA of the extraction stage have minimal impact on pressure rise. APPR. 10% RECOVER Paext PSIG  270 PSIG 500

250 PSIG SILICA  DESIGN PRESS. 275PSIG  LP FLOW @185KLB/HR INCREASE TO MAX  0

 Figure 5. Deposition and Pressure Profile for Conventional Online  Washing.  %CUG6JGHQWNKPIQEEWTTGFVQVJUVCIG #HVGTGZVUVCIGRTGUUWTG 25+) THERMODYNAMICS ANALYSIS RESULTS  %CUG6JGHQWNKPIQEEWTTGFVQVJUVCIG FOR A CLEAN CONDITION %CUG6JGHQWNKPIQEEWTTGFVQVJUVCIG  %CUG6JGHQWNKPIQEEWTTGFVQVJVJCPFVJUVCIGU When applying conventional online washing, the turbine power  and after-extraction pressure do not recover to their original levels,              but rather to some lower level. The extent of this recovery will now (QWNKPI6JKEMPGUU OO be related to the lowering of the inlet steam temperature. Changes in moisture distribution for each stage during washing have to be Figure 7. Deposition Profile Across Steam Path for Fouled Condition. analyzed in order to determine how much moisture is needed for effective washing. Figure 6 shows stage exit moisture levels in METHODOLOGY TO DESIGN relation to inlet temperature changes (910¡F to 800¡F). ONLINE WASH NOZZLES As mentioned previously, conventional online water washing  involves decreasing the inlet temperature thereby increasing the moisture content of each stage, resulting in power losses and ulti- 6*56#)'  674$+0'+0.'66'/2 mately affecting production. 6*56#)' &')( In the new online washing system, pressurized water is injected  &')( directly into the extraction valve chest. Figure 8 shows a schematic &')(  6*56#)' of the new water injection valve box installation. Table 1 shows the &')( procedure implemented to design and evaluate the new injection  6*56#)' &')( &')( nozzles. The effects of mass flow, erosion damage, and structural  6*56#)' &')( integrity were some of the criteria used in this design evaluation. This evaluation took into account that the after-extraction pressure

56#)'':+6/1+5674'   6*56#)' can get close to the operating pressure limit during a wash.  6*56#)' Heat Balance Analysis  Figure 9 shows the method used for calculating the injected           water mass flow using the energy conservation method around the 24'557+4'#(6'4'#%*56#)'5 25+# extraction valve chest.

Figure 6. Moisture Distribution Across Each Stage for Clean Evaluation of Erosion Damage Due to Water Droplets Condition. Evaluation of erosion damage caused by water droplets is another factor considered in the methodology. To understand this Knowing the moisture distribution with respect to changes in phenomenon, it is necessary to evaluate the size of droplets along temperature, other thermodynamic properties, such as specific the steam path, from the valve chest through the nozzles and blades volume, delta-P across diaphragms, and nozzle exit velocities can for each LP stage. Specifically, we need to know how these be determined. Except for moisture contents, the calculation of droplets are scattered, how they flow within the main steam stream, these thermodynamic properties showed no significant changes for and what damage they cause to the nozzles and blades. 60 PROCEEDINGS OF THE THIRTY-THIRD TURBOMACHINERY SYMPOSIUM ¥ 2004

APPLICATION OF Figure 10 shows a flow diagram to determine the maximum HYBRID SPRAY NOZZLE SUPPLYING STEAM & WATER droplet size, using Reynolds number to calculate the drag force and equating this to the inertia force. According to this analysis, the droplet diameter is expected to range between 10 μm to 1 mm. STEAM & WATER SUPPLY WHAT SIZE OF DROPLET AND HOW DROPLET ARE SCATTERED ? CONTROL FOR DROPLET SIZE CONSTANT CAN THESE DROPLETS FLOW WITH MAIN STEAM STREAM ? RATIO OF STEAM/WATER CASE STUDY : CALCULATION OF MAX.DIAMETER OF DROPLET

RANGE OF 20ata 210͠ Steam & Saturated Water DRAG COEFFICIENT DROPLET Steam Velocity 500m/sec 10Ǵm to 1mm 1000 Reynolds Number 100 Droplet Drag Force By Steam Flow = Droplet Inertia Force 10

1 Droplet Diameter 1 mm

10-1 Steam Density & Velocity is Large Enough. 10-1 10-0 101 102 103 104 105 106 Large Droplets are Scattered by Shear Force. REYNOLDS NUMBER Re ω Flow Pattern of Steam & Water is Same. BASED ON DROPLET DIAMETER Figure 8. Schematic of New Water Injection System. Water can be Scattered Equally.

Table 1. Evaluation Procedure to Design Injection Nozzles. Figure 10. Water Droplet Size and Distribution.

ITEM DESCRIPTION The next step is to evaluate the mechanical damage caused by erosion. Erosion caused by water droplets is a fatigue phenomenon HEAT BALANCE ANALYSIS ٨ REQUIRED MASS FLOW EVALUATION FOR MASS FLOW OF resulting from impact compression pressure on the surface of a ٨ FOR WATER INJECTION WATER INJECTION given profile. EROSION DAMAGE Erosion index is a parameter used to quantify the severity of the WATER EVAPORATION DURATION TIME ANALYSIS ٨ EVALUATION PREDICTION OF IMPACT PRESSURE OF DROPLET damage caused by erosion. This index relates to the cyclic impact ٨ OF OPTIMUM DROPLET DIAMETER ٨ CASING compression stress divided by the fatigue endurance limit. The FOR EROSION PREVENTION NOZZLE AND BLADE erosion index can be obtained by calculating the Mach number of a droplet and the fatigue impact pressure on the profile surface. NOZZLE TYPE SELECTION ٨ SELECTION OF When the erosion index parameter is less than one, erosion damage NOZZLE SIZE NOZZLE ATTACHMENT DESIGN will not occur. A flow diagram illustrating the basic evaluation ٨ AND STRUCTURE DESIGN procedure is shown in Figure 11. THERMAL STRESS ANALYSIS OF CASING PROVISION ٨

Steam Condition Nozzles & Blades Design Ƕmax =Function of INJECTION WATER Surface Tension of Water Max. Droplet Diameter Critical Weber Number Ƕmax G2, h2 Reaching to Blades Droplet Velocity AFTER INLET SIDE 3 RD Steam Velocity 4 TH Steam Density STAGE STAGE Impact Mach Number Mo=V1/Co EXTRACTION of Max. Droplet Mo = Function of HP VALVE LP Velocity of Sound in Water CHEST Impact Velocity of Max. Droplet Steam Impact Pressure P Steam P = Function of G , h Impact Mach Number of Max. Droplet G1, h1 3 3 Erosion Parameter CǶP/䕫e G : flow [KLB/H] 䕫e = Fatigue Limit of Material G2 = G1 (h1 -h3 ) / (h3 -h2 ) h : specific enthalpy [J/KLB] Comparison Of C = f(Ƕmax) influence function of Field Observation Ƕޓ Subscript 1:Condition of Upstream Droplet Diameter G =13[KLB/H] of Extraction Valve Chest 2 Erosion IndexޓCriteria (danger/Safety) Subscript 2:Water Injection Condition Described by Distribution Subscript 3 :Condition of Downstream FIG.11 Along blade height & Axial apace of Extraction Valve Chest Figure 11. Erosion Evaluation Procedure. Figure 9. Heat Balance Analysis. An evaluation of erosion damage to valve, nozzles, and blades is The basic theory to determine water droplet size and distribution shown in Figure 12 for different velocities along the steam path. is the balancing of drag and inertia forces. The drag force acts to This figure shows the correlation between droplet size and relation slow down the droplets, whereas the inertia force is the force steam velocity, as well as the calculated erosion damage along the required to move the droplets along the steam path. The relative steam path. For example, a low velocity inside the valve chest results difference between the drag force and the inertia force is the shear in larger droplet size, with a corresponding higher erosion index. force, which determines the size of the droplet. When the drag and However, downstream of the valve chest, erosion is influenced more inertia forces are equal, and the shear force is zero, the droplet size by droplet size than relative velocity. Therefore, since the droplet will be maintained and will flow with the main steam stream. Drag sizes are smaller downstream the erosion condition is milder. force is a function of Reynolds number calculated by steam In each case, the erosion index parameter is much lower than velocity and droplet size. Since inertia force is also a function of 1.0, and therefore the authors can conclude that erosion damage droplet size, the balancing of these two forces will determine will not occur. However, there is a potential for cavitation damage droplet size. when injected water is mixed with steam in a narrow space such as NEW TECHNIQUE FOR ONLINE WASHING OF LARGE MECHANICAL-DRIVE CONDENSING STEAM TURBINES 61

㪈㪅㪇㪇㪇㪇 STRENGTH EVALUATION OF 㪴 Erosion Evaluation

㫄 Valve room

㫄 ONLINE WASHING NOZZLE 㪲

㩷 㪇㪅㪈㪇㪇㪇 㫉 㪼

㫋 Strength evaluation using FEA was done on a component basis; 㪼

㫄 namely the injection nozzles, the extraction steam chest, and the

㪸 㪇㪅㪇㪈㪇㪇 Nozzle, Blade inner lift bar. 㪛㫀 㩷 㫋 㪼 㫃 㪇㪅㪇㪇㪈㪇 Wash Nozzles 㫆㫇

㪛㫉 Wash nozzles were evaluated in transient and steady-state con- 㪇㪅㪇㪇㪇㪈 ditions to determine temperature distribution and associated 㪇 㪈㪇㪇 㪉㪇㪇 㪊㪇㪇 㪋㪇㪇 㪌㪇㪇 stresses during water injection. Figure 15 shows the temperature 㪩㪼㫃㪸㫋㫀㫍㪼㩷㪭㪼㫃㫆㪺㫀㫋㫐㩷㩷㪲㫄㪆㫊㪴 distribution in both states at low temperature water supply. As can be seen in the transient state, there is a large temperature gradient Valve Chest Nozzle Blade at the inlet flange of the water injection nozzle, compared to the Relative velocity m/s 14.3 496.0 496.0 steady-state where the temperature gradient is minimal. Impact velocity m/s 14.3 496.0 291.0 Conversely, the temperature gradient for the transient condition Droplet Diameter mm 0.427 0.0004 0.0004 along the welded pipe between the valve chest and the inlet flange Impact Pressure MPa 62 2381 1266 is minimal compared to the steady-state condition. The associated Erosion Parameter 0.320 below 0.001 below 0.001 thermal stress distribution is detailed in Figure 16. Note, in the 䋭 detail design process, stress can be decreased especially in the Erosion Parameter < 1 then OK transient case by changing boss size, pipe length, groove size, groove location, and water supply temperature. Figure 12. Erosion Evaluation Results. [㷄][㷄] a valve chest. This mixture of water and steam is critical and a Body potential risk, requiring that the effects be quantified in the labora- tory. The effects were quantified by conducting a shop test using an actual valve chest. Selection of Injection Nozzles A wide spray-angle type water injection nozzle was selected Gasket based on mass flow rate: spray angle, pressure, and temperature application ranges. As previously mentioned, the mass flow rate was determined based on heat balance of the steam and injected water. The new design extraction valve with the selected water Nozzle injection nozzles is presented in Figure 13. Figure 14 shows the After 60sec Steady detail of the selected nozzle as installed in the extraction box. The mechanical integrity of the welded nozzle arrangement was Figure 15. FEA Temperature Distribution—Nozzle Box. evaluated using FEA as explained below. [kgf/mm2]

After 60sec Steady

Figure 16. FEA Thermal Stress—Nozzle Box.

Figure 17 shows the peak stress point for both transient and FIG.13 EM-MHI-33 TURBO steady-state. Based on these stress analyses, low cycle fatigue eval- uation was carried out to determine the safety factor of the design Figure 13. Injection Nozzle Arrangement. during periodical online washing. Figure 18, in accordance with ASME Code (Section VIII), shows that for the strain calculated, there is an inherent safety factor of 2. This corresponds to a life factor of 20 times the design criterion of 12 washes per year over 30 years.

Extraction Steam Chest A 3-D computer-aided design (CAD) model was used as input to perform the FEA, and the stress and displacement were calcu- lated based on actual operating conditions. Results are shown in Figure 14. Injection Nozzle Details. Figures 19 and 20. 62 PROCEEDINGS OF THE THIRTY-THIRD TURBOMACHINERY SYMPOSIUM ¥ 2004 yp BOSS FOR STRESS RELEASE

PEAK STRESS AT STEADY CONDITION PEAK STRESS unit : mm AT UNSTEADY CONDITION

IN CASE OF STRESS RELEASE MODIFICATION Figure 17. Thermal Stress Relief Modification—Nozzle Box. FIG 20 Figure 20. Analysis Results of Valve Chest Displacement. TOTAL STRAIN Inner Lift Bar (%) During online washing, the inner lift bar is exposed to a temper- ature gradient caused by low temperature water injected on one side, while the other side is exposed to a higher steam temperature. This results in deformation and thermal stress occurring in both transient and steady-states. S.F 2 The lift bar was modeled using FEA, and was based on a tem- 0.43% perature distribution calculated over a two hour wash period with an inlet temperature of 20¡C (68¡F). Based on this inlet tempera- Based on S.F 20 ture, it was estimated that the temperature difference between the 2 85kgf/mm water injection spot area and the opposite side was about 250¡C Unsteady (482¡F) for the model. Actual shop testing showed that the tem- Peak Stress CYCLE NUMBER perature of the injected spot area was close to the mixture of Nozzle temperature of the water and steam. However, during real Flange 360 CYCLES (12/YEAR˜30YEARS) operating conditions, because the actual temperature difference is not large between the water injection spot area side and the Figure 18. Low Cycle Fatigue Evaluation. opposite side, the thermal stress is not particularly severe. Using 250¡C (482¡F) delta-T for the model, the thermal stress distribution was calculated for a two hour wash period as mentioned above. The maximum stress in tension was approxi- mately 60 kgf/mm2 (85,340.1 lbf/in2) and compression stress was 20 kgf/mm2 (28,446.7 lbf/in2). A low cycle fatigue evaluation was also carried out. This confirmed a safety factor of 2 for a total stress 2 2 unit : kgf/cm^2 of 60 kgf/mm (85,340.1 lbf/in ), resulting in a life factor of 20 times the design criterion of 12 washes per year over 30 years. Figure 21 and Figure 22 show the contours of deformation and stress for the inner bar. The maximum deformation is 1.2 mm (.045 inch) and equivalent stress is 58 kgf/mm2 (82,495.4 lbf/in2). Although the model used a larger than normal delta-T of 250¡C (482¡F), the stress and deformation calculated were well within the acceptable range. Expected actual stress and deformation will be about 1/10 of the calculated values. SHOP TESTING EVALUATION AND RESULTS

FIG.19 In order to verify the results of the design calculation, a static EM-MHI-33 TURBO shop test was conducted on the newly designed extraction steam chest. Figure 19. Analysis Results of Valve Chest Stresses. Actual steam conditions were simulated. Injected flow rates were adjusted during the test to determine whether or not the The results of this analysis of the extraction box, which is made injected water steam mixture was completely vaporized. from chromium molybdenum (CrMo), show the stress to range Nonvaporized mixture could cause damage such as breakage or from 2.3 to 4.6 kgf/mm2 (3271.4 to 6542.7 lbf/in2). This range deformation of internal parts in the valve box or turbine casing due provides adequate margins when compared to the allowable stress to heavy drain erosion under the condition of insufficient vapor- of 9.8 kgf/mm2 (13,938.9 lbf/in2) based on creep rupture. ization of water. NEW TECHNIQUE FOR ONLINE WASHING OF LARGE MECHANICAL-DRIVE CONDENSING STEAM TURBINES 63

-.089 .055 .199 .343 .487 .631 775. ٨ XX .919 U 1.063 Z MN 1.207 Z MX 㧔ᢿ㕙 A㧕

UZ㧔mm㧕

MX

-.144 MN -.116 -.088 -.061 Y -.033 -.005 UY .023 MX 05. ٨ X .078 .106 A MN 111 141 171 200 Y 230 X 259 Z 289 319 348 378

T㧔͠㧕 X MX

Z MN 㧕 SECTION A

Figure 21. Deformation Analysis for Inner Lift Bar During Online Washing. Figure 24. Test Setup for Online Washing. MN

o p MN q MX MX Valve Chest Sch

Y Y X X Z Z

MN MN X X Point of measuring MX MX steam temperature Z Z 0 0 6.4 4.4 12.9 8.9 Point of measuring (kgf/mm2) 2 ǻeq 19.3 ǻeq (kgf/mm ) 13.3 25.8 17.8 internal parts temperature 32.2 22.2 SECTION A 38.7 SECTION A 26.6 45.1 31.1 51.5 35.5 r f 58 max 40 max d

Figure 22. Stress Analysis for Inner Lift Bar During Online b e Washing. Water injection c a,v,w

Description of Test Setup g The flow diagram in Figure 23 illustrates the shop test setup for h the valve chest. Figure 24 shows a photograph of the actual test i j facility. As can be seen, the testing arrangement can simulate the Inlet steam Outlet s,t,u k(equivalent position x actual online washing condition by varying measured water to nozzle) injection flow rates and steam temperature distribution along the steam path internals including the valve box.

z n Valve chest T P m l m FI 3 3 3 VW1 Water TURBO Figure 25. Valve Chest Schematic. Steam VS1 T1 P1 m1 T2 P2 VS2 Boiler Silencer Table 2. Test Conditions.

V S3 Item Testing machine Actual turbine T: Temperature Silencer P: Pressure Steam pressure[MPa] 4.6 4.6 m: Flow rate Steam temperature[K] 651 651 Figure 23. Testing Flow Diagram. Steam flow rate[kg/s] 4.7 23 Water pressure[MPa] 4.9 4.8 Test Conditions Water temperature[K] 373 373 Figure 25 is a schematic showing the points of measurement for Water flow rate[kg/s] 0.38 1.9 the various parameters. Actual operation conditions are used in the simulation, including steam velocity. Since only one injection * Same steam velocity as the actual turbine valve is used compared to four in a real turbine, the flow rate is a quarter of the actual steam flow. Table 2 shows a comparative Results summary of the parameters used in the tests versus the conditions The temperature profile along the steam path downstream of point for a real turbine. Note: Point is equivalent to the actual position of “k” at varying water flow rates is shown in Figure 26. Under increas- the first LP stage nozzle, i.e., the extraction stage. ing flow rates, steam temperature gradually drops to saturation 64 PROCEEDINGS OF THE THIRTY-THIRD TURBOMACHINERY SYMPOSIUM ¥ 2004

temperature (250¡C, 482¡F). As can be seen, for flow rates below 㪋㪉㪌 㪠㫅㫅㪼㫉㩷㪹㪸㫉㩷㫋㪼㫄㫇㪅䇸㫆䇹 㪋㪌 18 l/min (4.8 g/min), the temperature profile is fairly constant 㪠㫅㫅㪼㫉㩷㪹㪸㫉㩷㫋㪼㫄㫇㪅䇸㫇䇹 㪋㪇㪇 㪠㫅㫅㪼㫉㩷㪹㪸㫉㩷㫋㪼㫄㫇㪅䇸㫈䇹 㪋㪇 along the flow path downstream of “k.” Whereas, flow rates above 㪭㪸㫃㫍㪼㩷㪻㫀㫊㫂㩷㫋㪼㫄㫇㪅䇸㫉䇹 18 l/min (4.8 g/min) result in temperature increases along the same 㪊㪎㪌 㪮㪸㫋㪼㫉㩷㪽㫃㫆㫎䇸㪸䇹 㪊㪌 flow path, leading the authors to conclude that water is therefore not vaporized completely. 㪊㪌㪇 㪊㪇 㪊㪉㪌 㪉㪌 㪋㪉㪌 㪪㫋㪼㪸㫄㩷㫋㪼㫄㫇㪅䇸㫀䇹 㪋㪌 㪪㫋㪼㪸㫄㩷㫋㪼㫄㫇㪅䇸㫁䇹 㪪㫋㪼㪸㫄㩷㫋㪼㫄㫇㪅䇸㫂䇹 㪊㪇㪇 㪉㪇 㪋㪇㪇 㪋㪇 㪪㫋㪼㪸㫄㩷㫋㪼㫄㫇㪅䇸㫄䇹 㪪㫋㪼㪸㫄㩷㫋㪼㫄㫇㪅䇸㫅䇹 㪉㪎㪌 㪈㪌 㪊㪎㪌 㪮㪸㫋㪼㫉㩷㪽㫃㫆㫎䇸㪸䇹 㪊㪌 㪫㪼㫄㫇㪼㫉㪸㫋㫌㫉㪼䇭㪲㷄㪴 㪮㪸㫋㪼㫉㩷㪽㫃㫆㫎㩷㪲㫃㪆㫄㫀㫅㪴 㪉㪌㪇 㪈㪇 㪊㪌㪇 㪊㪇 㪉㪉㪌 㪌 㪊㪉㪌 㪉㪌 㪉㪇㪇 㪇 㪊㪇㪇 㪉㪇 㪋㪌㪇㪇 㪌㪇㪇㪇 㪌㪌㪇㪇 㪍㪇㪇㪇 㪍㪌㪇㪇 㪎㪇㪇㪇 㪎㪌㪇㪇 㪏㪇㪇㪇 㪮㪸㫋㪼㫉㩷㪽㫃㫆㫎㩷㪲㫃㪆㫄㫀㫅㪴 㪫㪼㫄㫇㪼㫉㪸㫋㫌㫉㪼䇭㪲㷄㪴 㪉㪎㪌 㪈㪌 㪫㫀㫄㪼㩷㪲㫊㪼㪺㪴

㪉㪌㪇 㪈㪇 Figure 28. Test Results for Inner Lift Bar Temperature Profile.

㪉㪉㪌 㪌 ) FOR MIXING TEMPERATURE

㷄

(

㪉㪇㪇 㪇

㪋㪌㪇㪇 㪌㪇㪇㪇 㪌㪌㪇㪇 㪍㪇㪇㪇 㪍㪌㪇㪇 㪎㪇㪇㪇 㪎㪌㪇㪇 㪏㪇㪇㪇 E 㪈࿁⋡ -33 TURBO 㪫㫀㫄㪼㩷㪲㫊㪼㪺㪴 FIG.26 R 㪋㪇㪇

U

Figure 26. Test Results for Steam Temperature Profile. T 㪊㪏㪇 㫂ὐK

A

R 㫅ὐN

Figure 27 shows the temperature profile in the upper portion of E 㪊㪍㪇 the valve box at varying water injection rates. Temperature in this P ⸘▚୯CAL.

M 㪊㪋㪇

area does not decrease even during water injection. On the other E

T hand, just below this area, the temperature falls to saturation. 㪊㪉㪇

Based on this observation, the authors found that the injected water M

does not vaporize completely in the extraction valve box and the A 㪊㪇㪇

E

T surface of the valve box gets wet at saturation temperature. ⫳᳇಴ญ᷷ᐲ䇭㷄

S 㪉㪏㪇

㪋㪉㪌 㪋㪌 T Evaluation point of outlet E 㪉㪍㪇 㪋㪇㪇 㪋㪇 L steam temperature T 㪉㪋㪇

U 㪊㪎㪌 㪊㪌 O 㪇㪅㪇 㪌㪅㪇 㪈㪇㪅㪇 㪈㪌㪅㪇 㪉㪇㪅㪇 㪉㪌㪅㪇

㪊㪌㪇 㪭㪸㫃㫍㪼 㩷㪹㫆 㫏㩷㫋㪼 㫄㫇㪅䇸 㪹䇹 㪊㪇 WATER INJECTION᳓ᵹ㊂䇭㫃㪆㫄㫀㫅 FLOW RATE (L/MIN) 㪭㪸㫃㫍㪼 㩷㪹㫆 㫏㩷㫋㪼 㫄㫇㪅䇸 㪺 䇹 㪊㪉㪌 㪭㪸㫃㫍㪼 㩷㪹㫆 㫏㩷㫋㪼 㫄㫇㪅䇸 㪻䇹 㪉㪌 㪭㪸㫃㫍㪼 㩷㪹㫆 㫏㩷㫋㪼 㫄㫇㪅䇸 㪼 䇹 Figure 29. Comparison Test Results for Mixing Temperature. 㪭㪸㫃㫍㪼 㩷㪹㫆 㫏㩷㫋㪼 㫄㫇㪅䇸 㪽䇹 㪊㪇㪇 㪭㪸㫃㫍㪼 㩷㪹㫆 㫏㩷㫋㪼 㫄㫇㪅䇸 㪾䇹 㪉㪇

㪭㪸㫃㫍㪼 㩷㪹㫆 㫏㩷㫋㪼 㫄㫇㪅䇸 㪿 䇹 Transient Temperature Comparison— 㫎㪸㫋㪼㫉㩷㪽㫃㫆㫎㩷㪲㫃㪆㫄㫀㫅㪴 㪫㪼㫄㫇㪼㫉㪸㫋㫌㫉㪼䇭㪲㷄㪴 㪉㪎㪌 㪮㪸㫋㪼㫉㩷㪽㫃㫆㫎䇸㪸䇹 㪈㪌 Calculated Versus Tested 㪉㪌㪇 㪈㪇 Figure 30 shows transient temperature during water injection, 㪉㪉㪌 㪌 comparing calculated and measured figures at different points along the valve chest. It is evident that the test results match very 㪉㪇㪇 㪇 well with the calculated values. 㪋㪌㪇㪇 㪌㪇㪇㪇 㪌㪌㪇㪇 㪍㪇㪇㪇 㪍㪌㪇㪇 㪎㪇㪇㪇 㪎㪌㪇㪇 㪏㪇㪇㪇 HI-33 TURBO 㪫㫀㫄㪼㩷㪲㫊㪼㪺㪴 FIG.27 Figure 27. Test Results for Valve Box Temperature Profile. h d Figure 28 shows the temperature profile on the surface of the g e inner bar and valve disc for varying water injection rates. The tem- perature gradually decreases to saturation temperature (250¡C, CALCULATED MEASURED

) 482¡F) with increasing water flow. The temperature of the valve 㪋㪇㪇 ) 㪋㪇㪇

㷄

㷄

(

(

disc is lower than that of the inner bar. Therefore from the test

E 㪊㪌㪇 㪻 E 㪊㪌㪇

R R 㪻

U results, the authors can see that the surfaces of these parts are wet 㪼 U

T T 㪼

A and the injected water will be heated to saturation temperature 㪊㪇㪇 㪾 A 㪊㪇㪇

R R 㪾

E ᷷ᐲ䇭㷄 ᷷ᐲ䇭㷄 before reaching the turbine internal parts. 㪿 E

P 㪉㪌㪇 P 㪉㪌㪇 㪿

M

M

E

E

T

T

Relation Between Water Injection

R 㪉㪇㪇 R 㪉㪇㪇

A

A

B Flow Rate and Mixing Temperature 㪇 㪉㪇㪋㪇㪍㪇㪏㪇㪈㪇㪇 B 㪇 㪉㪇㪋㪇㪍㪇㪏㪇㪈㪇㪇

R

R

E TIME AFTERᤨ㑆㩷㫊㪼㪺 INJECTION E TIME AFTERᤨ㑆㩷㫊㪼㪺 INJECTION

N Figure 29 shows the relationship between water injection flow (SEC) N

N N (SEC)

I rate and mixing temperature at point “k” in comparison to the cal- I culated values. As can be seen, the measured mixing temperature Figure 30. Temperature Comparison of Nozzle Box—Calculated is almost the same as that of the calculated value. Versus Tested. NEW TECHNIQUE FOR ONLINE WASHING OF LARGE MECHANICAL-DRIVE CONDENSING STEAM TURBINES 65

NDE Testing For inner bar, control valve, and water injection nozzle, Figure TIT 31 shows nondestructive evaluation (NDE) results, and no defects were found in any of these parts. As there is no appreciable thermal TO DCS deformation caused by water injection, the inner bar and valve will MONITOR ++ALARM L/H XV [LL] operate normally. E.C.V. (TI) (TAL / TAH)

730 psig˜721qF Design Condition of Extraction Control Valve 2&6 895 psig˜721qF 730 psig˜721qF (600 lb)

895 psig˜̪2qF Drain

2&%

̪1 psig˜̪2qF Setpoint (1 ECV inner pressure㧗75psig xv

Setpoint Figure 31. NDE Results after Water Injection Test. 550qF r 10 qF

Strainer RISK ASSESSMENTS ANALYSIS Condensate Water Based on hazard and operability (HazOp) and United States Condensate Water ̪1 psig ̪2 F Department of Defense military (MIL) standards shown in Tables Drain ˜ q F 3 and Figure 32, the authors designed the online wash system Figure 33. Water Injection Supply System. shown in Figure 33 for an actual field application. The main risks are draining carryover, thrust force increase, Table 3. Risk Assessment. drain-erosion, and thermal shock stress. The mixing temperature ASSESSMENT FACTOR HAZARD COUNTERMEASURES inside the valve chest is monitored and, if the system detects a low ELEMENT RISK SOURCE GRADE PROBABILITY RISK INDEX ALARM SAFETY LOADING WATER&STEAM NON-EFFECTIVE SHOP INJECTION TEST temperature close to saturated temperature, both the flow control 㸈 C 12 MIXING HEAT BALANCE TIT LL CONTROL INJECTION TRANSIENT TRANSIENT STRESS and solenoid valves are immediately closed. Other risks are 㸉 C 8 PROVISION THERMAL STRESS ANALYSIS & NDE T/A EXT. VALVE TRANSIENT TRANSIENT STRESS countered with appropriate alarms and relief valves. 㸉 C 8 INNER BAR THERMAL STRESS ANALYSIS & NDE T/A TRANSIENT TRANSIENT STRESS 㸉 C 8 THERMAL STRESS ANALYSIS & NDE T/A CONCLUSIONS INJECTION EMERGENCY SHUT XV DRAIN INVASION 㸈 C 12 NOZZLE PDCV DOUBLE SHUT CHOKE FOR FLUSHING & HEATING This paper has introduced an innovative online washing 㸉 C 8 SOLID PARTICLE DRYING DAMAGE INDEX NDE T/A technique to minimize the impact caused by fouling of the steam TURBINE DRAIN EROSION 㸉 D 6 COATING FOR PROTECT NOZZLE THRUST FORCE UP AFTER EXT. PRESS & AXIAL path in large multistage condensing steam turbines. PROFILE 㸉 D 6 IN CHOKE DISPLACEMENT MONITORING The methodology developed to effectively evaluate the design of DAMAGE INDEX NDE T/A DRAIN EROSION 㸉 D 6 COATING FOR PROTECT TURBINE the new online wash nozzles showed that the calculated parameters THRUST FORCE UP AXIAL DISPLACEMENT & METAL BLADE 㸉 C 8 IN CHOKE TEMPERATURE ARE MONITORED PROFILE compared well with shop test results. This methodology evaluated LP SECTION TIT IS CONTROLLED TO 㸉 D 6 MOISTURE UP MAINTAIN LP MOISTURE 14% the nozzle box on a component-by-component basis using FEA, THRUST FORCE UP AXIAL DISPLACEMENT & METAL TURBINE 㸉 C 8 IN CHOKE TEMPERATURE ARE MONITORED THRUST which confirmed the mechanical integrity of the overall system. AXIAL AXIAL DISPLACEMENT & METAL BEARING 㸉 C 8 DISPLACEMENT UP TEMPERATURE ARE MONITORED Life cycle calculations based on fatigue phenomenon showed the WATER FLOW EMERGENCY SHUT XV 㸉 D 6 UNCONTROL PDCV DOUBLE SHUT system to exceed the projected life by a factor of 20. Using HazOp OPERATION PRESSURE RAISE PRV IS INSTALLED IN 㸉 D 6 IN SUPPLY LINE WATER SUPPLY LINE and MIL risk assessment standards, the authors were able to design the supporting external facility. 1. HAZARD GRADE 3. RISK INDEX 䇭 The final design is a compact extraction box, which replaced an 䇭 Σ: CATASTROPHIC Σ Τ Υ Φ original design extraction box without machining of the turbine ޓΤ: CRITICAL A 24 18 12 6 casing. This design was shop tested and proved effective in ޓΥ: MARGINAL B 20 15 10 5 washing off deposits from LP blades and nozzles with minimal ޓΦ: NEGLIGIBLE C 16 12 8 4 D 12 9 6 3 power turndown, and without impacting the turbine’s long-term E 8 6 4 2 performance. This technique, although applied here to extraction- F 4 3 2 1 condensing turbines, is also applicable to large condensing turbines. 2. HAZARD PROBABILITY 4. CRITERIA FOR RISK INDEX 䇭 A: FREQUENT MORE THAN 13: NOT ACCEPTABLE B: REASONABLY PROBABLE STOP AND RE-DESIGN REQUIRED ACKNOWLEDGEMENT C: OCCUPATIONAL The authors gratefully wish to acknowledge the following indi- D: REMOTE 9 TO 12: BIG PROBLEM TEST AND RE-DESIGN REQUIRED viduals for their contribution and technical assistance in analyzing E: EXTREAMLY UNLIKELY F: IMPOSSIBLE 5 TO 8: SEVERAL PROBLEMS and reviewing the results: T. Hirano, T. Inoue, and M. Wakai of TECHNICAL MODIFICATION ONLY Mitsubishi Heavy Industries Ltd.; M. Fujimura of Mitsubishi Turbo Techno; H. G. Elliott of International Turbomachinery LESS THAN 4: ACCEPTABLE Consulting Services; and Joe Pachioli and Jin Yamada of MHI Figure 32. U.S. MIL Risk Assessment. America. 66 PROCEEDINGS OF THE THIRTY-THIRD TURBOMACHINERY SYMPOSIUM ¥ 2004