Proceedings of the ASME Power Conference POWER2010 July 13-15, 2010, Chicago, Illinois, USA POWER2010-27248 TURBINE STEAM CHEST LIFE ASSESSMENT Daniel T. Peters, PE Structural Integrity Associates Uniontown, OH, USA dpeters@structint.com Eric Jones Structural Integrity Associates Uniontown, OH, USA ejones@structint.com Sean Hastings Structural Integrity Associates Uniontown, OH, USA shastings@structint.com Steven Greco We Energies Milwaukee, WI, USA Steve.Greco@we-energies.com ABSTRACT The optimal approach to condition assessment, regardless of the component involved, is to use a programmatic approach, and steam chest condition assessment is no exception. Steam chests typically vary significantly from one to the next in shape and complexity; consequently, stress distributions vary and damage occurs first and is most advanced at the high stress regions, accordingly. One of the most significant cost drivers in an overall program is an ongoing implementation of NDE that has little technical justification, i.e., implementing NDE as the means of identifying the high stress locations via flaw detection. Keep in mind that flaws can manifest themselves at both macro and micro levels. Therefore, inspection typically includes surface inspection using liquid dye penetrant and/or magnetic particle inspection for macro damage and metallographic replication for micro damage, plus ultrasonic inspection for volumetric inspection of subsurface flaws and flaws at otherwise inaccessible surfaces. In a programmatic approach, the first step is to accurately understand the stresses of the steam chest to determine the appropriate areas requiring examination and monitoring. Then, only after identifying the critical areas on the steam chest, attention turns to defining the optimal techniques and procedures to examine the areas identified. By implementing a focused inspection that concentrates on the critical areas, as opposed to a shotgun approach, the scope, cost, and the frequency of the inspection is significantly reduced. The programmatic approach identifies these critical areas up front and helps to determine the best method for their inspection. The best method is most often dictated by access constraints and limitations at the region of interest. In recent years, significant strides have been made in the use of advanced UT techniques such as linear phased array (LPA) and annular phased array (APA) ultrasonic inspection for sizing cracks in some of the least accessible areas. In many cases, once identified, the damage can subsequently be monitored periodically with only the local removal of insulation. The disassembly of the valve is not required on an on-going basis, nor is full insulation removal in most cases. Finally, once damage has been identified and characterized, be it early form cavitation through to defined cracks, the model used initially to identify the inspection locations is then used to assess the damage in terms of growth rates and failure potential. This information is utilized for a complete Fitness for Service Assessment of the unit. This would include definition of re-inspection intervals, monitoring requirements, and possibly to assess repair/replace options and schedules. These assessments meet the requirements of current Standards in Fitness for Service Assessment. The robust life assessment program presented here includes: 1. upfront analysis of the steam chest to identify problem areas including modeling of the valve, 2. focused baseline inspection of identified potential problem areas, 3. Fitness for Service Analysis utilizing focused baseline inspection results, 4. continued monitoring of critical areas of the valve. This programmatic approach results in a focused, optimized integrity assessment program at minimized cost. 1 Copyright 2010 by ASME
INTRODUCTION The objective of this paper is to present an overview of the challenges in valve and steam chest management. This paper will also include the overview of a programmatic approach to steam chest management and discuss an integrated approach between periodic NDE inspection of these components and analytical assessment. The most cost effective management of these components long term is to implement an inspection program with a comprehensive technical basis. This includes utilization of the most appropriate NDE techniques, including state-of-the-art technologies such as advanced ultrasonic techniques. This paper will also present a summary of some case studies to show practical application of the programmatic approach in real world applications, to demonstrate the benefits of the approach. NOMENCLATURE APA Annular Phased Array Ultrasonics API American Petroleum Institute BFP Boiler Feed Pump EPRI Electric Power Research Institute ESV Emergency Stop Valve FAC Flow Assisted Corrosion FE Finite Element FFS Fitness for Service LPA Linear Phased Array Ultrasonics NDE Non-Destructive Evaluation UT Ultrasonic Testing ASSESSMENT CHALLENGES Maintenance budgets today are under continuous pressure to reduce costs and expenses. Many times, inspections of key components have little technical justification. In certain cases, examination may be randomly used to look for areas where flaws may be present as a means of detecting the high stress areas in a component. This is exactly opposite of a programmatic approach with a technical basis. Flaws can manifest themselves at both macro and micro levels with steam chests and valves. The flaws may be due to a number of issues in these types of components. Many of these components are made from cast materials, which have flaws inherent to the casting process. Some are due to creep cavitation caused by the extreme temperatures present in these components. Others are stress related due to fatigue caused by the pressure, thermal, or other loading, and some are a combination of effects such as creep and fatigue interaction. Typical inspections of these components may include dimensional readings of the valves looking for distortion in the measurements of the valves indicative of creep progression, and/or surface types of inspections such as liquid penetrant, magnetic particle, or metallographic replication. Sometimes, a volumetric inspection will be used such as ultrasonic inspection, and often hardness readings may be taken. There are many challenges in getting accurate meaningful information from these inspections. Run-out measurements in valves can be challenging and can result in misleading information if not properly performed. Replication is a very old technique for assessment of creep. However, this is often difficult to implement in most areas of steam chests, due to tight internal dimensions. Often, replication is performed on the outer surface of valves and chests based on ease of accessibility, though these may not be the most likely places for creep to occur. Hardness checking is an effective screening tool for the evaluation of the initial fabrication of the valve. This technique is not effective as a monitoring technique long term for determination of material degradation. It also is difficult to implement on inner surfaces and in locations where some of the highest stressed areas may occur. PROGRAMMATIC APPROACH The recommended approach to life management of these components is to use inspection techniques optimized for critical areas in the unit based on an in-depth analytical assessment. The slight increase in upfront costs of this type of approach versus a shotgun type of approach can be offset over time by resulting in a reduced inspection scope for each inspection, reducing long term inspection costs, and reducing down time by minimizing the need for disassembly of the valves and optimizing the inspection frequency. The technically based approach should include a complete upfront analysis to identify the problem areas, a focused baseline inspection of the identified potential problem areas, an evaluation of any issues found during the baseline inspection utilizing the most advanced Fitness for Service techniques, and the establishment of continued monitoring and inspection interval of the critical areas. The upfront analysis is used to determine the critical areas in the valve. This includes performing finite element stress analysis, including any thermal or transient effects and effects due to external influences such as piping loads. By determining the actual high stressed areas in the components, NDE methods used for assessments of these areas can be optimized. This stress analysis will be required for the subsequent life assessment which the inspection interval should be based. The second part of the programmatic approach would be to perform a focused baseline inspection. This would be a complete inspection of all identified problem areas using optimized NDE methods. This may include use of advanced 2 Copyright 2010 by ASME
ultrasonic inspection techniques such as linear phased array (LPA), annular phased array (APA), or other techniques which can identify creep issues at the earliest stages. This also may include replication, but of identified accessible problem areas and not random areas. Replication of random areas may be misleading to the extent of damage in critical areas. The third part of the assessment would be to perform a complete Fitness for Service assessment for the valve or chest. This type of analysis will determine the remaining life of the valve based on the current condition and actual expected operating conditions. This type of remaining life analysis is not generally considered during the original design and construction. Probabilistic life analyses may also be utilized to get an evaluation of the remaining life which can be used to get realistic timing needed for future repair and / or replacement of the components. This commonly includes utilizing computational fluid dynamics modeling for evaluation of issues such as Flow Assisted Corrosion (FAC) damage, or finite element analysis for evaluation of stresses. There are many published standards available today to assist with these evaluations such as ASME Risk Based Inspection Planning (PCC-3)[5], EPRI Life Management of Creep Strength Enhanced Ferritic Steels Project Solutions for the Performance of Grade 91 and other Steels [6], and API 579-1/ASME FFS-1Fitness for Service Standard 2007 [3]. The last part of the programmatic management approach is to continue monitoring critical areas of the components using an inspection frequency based on the analytical assessment. This approach results in the longest justifiable inspection frequencies to minimize long term costs using risk based planning. The NDE techniques used are fine tuned to the specific component being evaluated and often to minimize insulation removal and disassembly costs. Other instrumentation and monitoring packages may also be used for long term monitoring of the components such as high temperature strain gages and EPRI s Creep-FatiguePRO [7] for monitoring of creep-crack growth accumulation. UPFRONT ANALYSIS The upfront analysis is used to determine the critical areas in the valve. The more accurate the information going into this analysis, the more useful the results of the analysis will be. Therefore, accurate information regarding the material of the valve is important for the assessment. This includes material physical property information and documentation of any past repairs made. Equally as important is having good information regarding the number of unit startups, shutdowns, and unit trip information, as well as operational data such as pressure, temperatures, and flow rates experienced in service. One of the most difficult aspects of this can be generation of accurate geometry for input into the analysis. Finite element stress analysis is generally used for this type of analysis, and accurate geometric model is critical. Advanced techniques are now available for this purpose. Scanned images of scale drawings, digital pictures of components, and threedimensional point clouds can all directly be imported into modeling packages for this purpose today. Figure 1 is an example where a poor quality drawing was able to be scanned and imported directly into a solid modeling package for this purpose. Only one dimension was legible on the original print, but the valve model was able to be accurately created on that basis. This was verified by actual measurements once the baseline assessment occurred. OPTIMIZATION OF INSPECTION METHODS The methods used in inspection frequently need to be tailored to the constraints on the access to critical areas. This can result in compromises made in the flaw sizes that can be found in certain areas. For this reason, the use of the most sensitive and advanced techniques for inspection are critical to optimizing the overall program and minimizing costs. Significant advancements have been made recently in the use of ultrasonic techniques such as linear phased array (LPA) and annular phased array (APA) in these types of applications. Thick section cast components which previously had been thought not able to be inspected with ultrasonic techniques due to inherent fabrication flaws, thick sections, and poor grain structure are now being performed on a regular basis. This can allow for access to difficult critical areas that were otherwise considered inaccessible, and in some cases allowing for monitoring of flaws without disassembly as shown in Figure 2. Also, by leveraging the technology of solid modeling, ultrasonic techniques can be optimized on complex geometry prior to the inspection to reduce issues once the outage starts. LIFE ASSESSMENT/FITNESS FOR SERVICE ASSESSMENT The finite element modeling can be used directly for the Fitness for Service (FFS) Assessment discussed previously. FE analysis is the only realistic method in most steam chest evaluations for determination of stresses. The result of the FFS assessment is a complete assessment of any problem areas, including evaluation of the monitoring requirements, assessment of repair/replacement options, and determination of the reinspection interval for the component. This may include selectively monitoring specific areas of the component, based on either existing issues or expected problems in the chest, minimizing the insulation removal required and the need for disassembly. Types of FFS assessments may include either general or local metal loss, pitting and corrosion effects, cracks and cracklike flaw growth, creep and creep-crack interaction, and potentially FAC issues in localized areas. Several case studies are presented here to demonstrate the use of this type of programmatic approach. 3 Copyright 2010 by ASME
CASE STUDY 1 STEAM CHEST UPFRONT AND SENSITIVITY ANALYSIS The first case study is the evaluation of two steam chests shown in Figure 3. Figure 3 shows the solid models of two steam chests. The Figure shows a ½ section symmetric model for each of the steam chests. The model on the left is a chest for the turbine with two control valves in the center and emergency stop valves (ESV) on either end. The one on the right is a chest from a boiler feed pump (BFP) including the control valve on the left and ESV on the right. This one shows a cutaway to show the internal of the valve. The upfront modeling of the valves proved very successful in this case study. Each FE model was able to successfully predict the location of cracking in each of the steam chests. It was a number of years prior that either of these valves were inspected. The BFP valve was modeled to help understand the stresses in the valve during operation (see Figure 4). The original inspection plan was to inspect only accessible areas of the valve, specifically the upper bowl on the ESV. The FE analysis successfully predicted cracking in the lower bowl at the high stressed areas. Once predicted, UT and video probe evaluation of the lower bowl was added to the inspection plan, which successfully identified the cracking. Cracking was also identified in the upper bowl of the steam chest on the left. The decision was to machine out the cracking. The FE model was then used to evaluate the best technique for the removal of the cracks. The proposed machining scheme is shown in Figure 5. Typically, the larger the radius used results in a lower stress concentration. However, in this case, the loss of material in the wall due to the larger radius tool would have resulted in higher stresses in the remaining material. The critical crack size with a four inch radius tool was predicted to be 4.25 inches, while the two inch radius tool would result in a 5 inch critical size. Similarly, the larger tool would result in lowering the remaining life of the valve. This demonstrates the use of the model from the upfront analysis in supporting the FFS assessment later in the process. CASE STUDY 2 DEFORMATION IN A STEAM CHEST The second case study is to show the benefits of using analytical approaches in understanding the results of an evaluation on a steam chest. The steam chest in question is shown in Figure 6. The valve in question was a direct replacement of the steam chest original to the unit which was in operation for over 40 years. The valves of the steam chest were developing performance reliability issues within 2 years after replacement. A root cause evaluation was begun to determine the potential issues. This included review of any potential distortion in the valve case, replication of the material in the chest, hardness of the steam chest, as well as review of other issues with the valve stems and operators. The thrust of the initial assessment was to start by evaluating the valve directly. This included measurements of run-out dimensions in the valves, replication on the valve outer surfaces, and hardness of the valve materials. Several issues were found with conflicting information that had been gathered through the initial physical measurements taken. The FE model (see Figure 7) was generated in this case to help with the understanding of the readings taken during process. The loading on the valve included both pressure and thermal loadings, as well as loads from the external piping. The use of the model provided clarity in the evaluation of the physical measurements taken of the distortion in the valve chest casting. The displacements from the model were used to evaluate the initial readings and aided in optimizing the measurement techniques being used. The final results were that no unreasonable deformation was found in the valve casing, and that while there were some localized stresses near opening locations, no stresses were high enough to result in significant distortion of the chest casting. This insight into the behavior of the casting, allowed for the investigation to then concentrate on other areas for a root cause. CASE STUDY 3 VALVE CASE STUDY UTILIZING PROGRAMMATIC APPROACH The third case study is for the assessment of a valve with known cracking. The valve in question is a 24 inch diameter valve approximately 39 inches long, shown in Figure 8. The internal of the valve is shown in the cross-section shown in Figure 1. The valve was known to have a crack prior to implementation of the programmatic approach to managing the valve life. The flaw found in the valve is shown in Figure 2 of this report. It was identified as a service crack in an identical valve on a sister unit during a maintenance cycle in a prior outage. The valves were extremely long lead items, so management in repair and replacement decisions were warranted. The Fitness for Service analysis was performed to confirm the cause of the cracking, the expected mode of failure, and evaluate the expected life of the valve. Following the programmatic approach described previously, the model of the valve was generated, as discussed earlier in Figure 1. A FE model was then constructed for the evaluation of both full operating pressure and thermal transient stresses induced in the valve during a unit trip. The critical crack sizes were able to be evaluated based on this analysis, and a leak-before-break mode of failure was predicted. The valve was inspected, utilizing techniques specifically developed for the valve based on the 3-D solid model. This inspection included both utilization of LPA sizing of the crack 4 Copyright 2010 by ASME
through the over five inch wall thickness and replication of the crack tips to evaluate the extent of the creep effects in the valve. The crack was able to be accurately sized to be 1.2 inches deep with the orientation shown in Figure 2. EPRI s Creep-Fatigue Pro was then utilized for assessment of crack growth rate. The advantage of this technique is to evaluate the actual operational data on small intervals to account for all fluctuations in pressure and temperature that the unit experiences, as shown in Figure 9. The software is calibrated to account for the pressure and thermal transient stresses in the area of the crack utilizing the output of the FE model (see Figure 10). The geometry of the crack and material properties for the valve are also inputs to the crack growth rate routine. The result was that the prediction in crack growth rate showed significantly slow growth of the crack over time. The rate of growth was under 1/8 inch in a five year span as found in Figure 11. The valves were then setup on an 18 month inspection period. [6] EPRI Life Management of Creep Strength Enhanced Ferritic Steels Project Solutions for the Performance of Grade 91 and other Steels, Electric Power Research Institute, Ongoing. [7] EPRI Creep-FatiguePRO 3.0, Developed by Structural Integrity Associates, Electric Power Research Institute, 2000. The valves have been re-inspected once since the implementation of the program. Growth rates are very close to the predicted values. One of the benefits of the use of the advanced LPA techniques is that the cracks are sized without the need for disassembly of the valves. Insulation removal is the only requirement for the inspection, which greatly simplifies the outage planning. SUMMARY AND CONCLUSIONS A programmatic approach is greatly recommended for asset management of steam chests, valves, and other thick section components. The use of a thorough program gives a technical basis for the inspection plans used, while minimizing overall costs. This program includes the use of upfront analysis, performing a complete baseline inspection of the component, optimization of state-of-the-art NDE techniques for evaluation of critical areas, Fitness for Service assessment for determination of remaining life of the unit, and long term monitoring of the unit based on the latest risk based methods for inspection planning. REFERENCES [1] Peters, D. T., E. Jones, M. Berasi, Case Study on the Remaining Life Analysis of a Power Generation Valve, Proceedings of ASME Power 2008-60034, American Society of Mechanical Engineers, New York, NY, 2008. [2] Creep-Fatigue PRO 3.0, Electric Power Research Institute, Developed by Strutural Integrity Associate, 2000 [3] API 579-1 / ASME FFS-1, Fitness for Service, American Society of Mechanical Engineers, New York, NY, 2007. [4] Ansys Workbench Version 11.0 Release Documentation, Ansys, Inc. 2007. [5] Risk Based Inspection Planning PCC-3, American Society of Mechanical Engineers, New York, NY, 2008. 5 Copyright 2010 by ASME
Figure 1 Solid Modelling with Marginal Drawings Figure 2 Optimization of NDE Technique using Solid Model 6 Copyright 2010 by ASME
Turbine Steam Chest Boiler Feed Pump Steam Chest Emergency Stop Valve (ESV) Control Valve Figure 3 Solid Models of Steam Chests Turbine Steam Chest (Left) Boiler Feed Pump Chest (BFP) (Right) Figure 4 Finite Elements Analysis of Boiler Feed Pump Steam Chest Figure 5 Repair Machining Options for Steam Chest 7 Copyright 2010 by ASME
Figure 6 Steam Chest Model for Case Study 2 Figure 7 Equivalent Stresses in Steam Chest 8 Copyright 2010 by ASME
Figure 8 Valve with Known Flaw Temperature ( F) 1200 Avg. Temp. (IDOD-LR) Avg. Pressure (IDOD-LR) 1100 1000 900 800 700 600 500 400 300 200 100 Mar Apr May Jun Jul Aug Sep 2006 Figure 9 Pressure and Temperture Data used for Creep Crack Growth Analysis 2500 2000 1500 1000 500 0 Pressure (psi) 9 Copyright 2010 by ASME
Fatigue Stress (IDOD-LR) Fatigue Stress (IDOD-SR) 75 Creep Stress (IDOD-LR) Creep Stress (IDOD-SR) 1.50 Fatigue Stress (ksi) 50 25 0 1.25 1.00 0.75 0.50 0.25 Creep Stress (ksi) Mar Apr May Jun Jul Aug Sep 2006 Figure 10 Creep and Fatigue Stresses for the Valve Radial Crack Growth - Initial Size 1.24 inches 0.00 0.0125 Fatigue Crk Grw (IDOD-SR) Creep Crk Grwth (IDOD-SR) Crack Size (IDOD-SR) 1.2525 Crack Growth (in) 0.0100 0.0075 0.0050 0.0025 1.2500 1.2475 1.2450 1.2425 Crack Depth (in) 0.0000 Mar Apr May Jun Jul Aug Sep 2006 Figure 11 Predicted Crack Growth in Valve from Creep and Fatigue 1.2400 10 Copyright 2010 by ASME