POWER REMOTE THERMAL PERFORMANCE MONITORING AND DIAGNOSTICS TURNING DATA INTO KNOWLEDGE

Transcription

1 Proceedings of the ASME 2013 Power Conference Power2013 July 29 August 1, 2013, Boston, Massachusetts, USA POWER REMOTE THERMAL PERFORMANCE MONITORING AND DIAGNOSTICS TURNING DATA INTO KNOWLEDGE Xiaomo Jiang General Electric Company, Power & Water, Monitoring and Diagnostics Center Atlanta, GA 30339, USA Craig Foster General Electric Company, Power & Water, Monitoring and Diagnostics Center Atlanta, GA 30339, USA ABSTRACT Gas turbine simple or combined cycle plants are built and operated with higher availability, reliability, and performance in order to provide the customer with sufficient operating revenues and reduced fuel costs meanwhile enhancing customer dispatch competitiveness. A tremendous amount of operational data is usually collected from the everyday operation of a power plant. It has become an increasingly important but challenging issue about how to turn this data into knowledge and further solutions via developing advanced state-of-the-art analytics. This paper presents an integrated system and methodology to pursue this purpose by automating multi-level, multi-paradigm, multi-facet performance monitoring and anomaly detection for heavy duty gas turbines. The system provides an intelligent platform to drive site-specific performance improvements, mitigate outage risk, rationalize operational pattern, and enhance maintenance schedule and service offerings via taking appropriate proactive actions. In addition, the paper also presents the components in the system, including data sensing, hardware, and operational anomaly detection, expertise proactive act of company, site specific degradation assessment, and water wash effectiveness monitoring and analytics. As demonstrated in two examples, this remote performance monitoring aims to improve equipment efficiency by converting data into knowledge and solutions in order to drive value for customers including lowering operating fuel cost and increasing customer power sales and life cycle value. INTRODUCTION Gas turbine simple or combined cycle plants are built and operated with higher availability, reliability, and performance in order to provide the customer with sufficient operating revenues and reduced fuel costs meanwhile enhancing customer dispatch competitiveness. The availability of heavy duty gas turbines in the plant can be increased through increasing the turbine reliability by maintenance enhancement and recovering performance degradation by remote efficiency monitoring to provide timely corrective recommendations (Balevic et al. 2010, Brooks 2000, and Johnston 2000). In addition, increasing fuel costs requires maintaining the higher efficiency in a gas turbine system. For example, a combined cycle plant with the engineered capacity of 900MW power output can have an annual fuel bill of over 200 million dollars (Meher-homji et al. 2001). Therefore, remote performance monitoring of power generation equipment like heavy duty gas turbines has become increasingly important and popular in the energy industry since its introduction in the 90 s. Remote health monitoring of a power generation asset involves the full spectrum of data collection, data processing, mechanical condition monitoring algorithms, and alarm disposition, as well as issue analysis and diagnostics and recommendations for improvement. For instance, every day of the year, the remote Monitoring & Diagnostics (M&D) center of General Electric Company (GE) in Atlanta, Georgia collects more than 30,000 operating hours of data from a fleet of more than 1,600 globally deployed gas turbine, steam turbine, and generator assets. Therefore, it has become an increasingly important but challenging issue about how to turn this data into knowledge and further solution via developing advanced stateof-the-art analytics. In the past decade, GE s M&D center has deployed over 100 proprietary algorithms for real time monitoring of the fleet. Its purpose is to monitor the operation status of each asset by using the time series data collected and transmitted from the customer site to detect abnormality at nearly real-time. Figure 1 shows the conceptual definition of two types of abnormality targeted by the remote monitoring. The pattern indicates the relationship between two monitoring parameters. Generally, there are two types of abnormality: (1) the parameter exceeds a predefined limit, which could cause damage to the equipment if operation continues under such a condition, and (2) the anomaly pattern is away from the normal 1 Copyright 2013 by ASME

2 Monitoring variable 2 operational pattern even below the predefined threshold, which could be a symptom of potential failure or improper operation. Therefore, early detection of these anomalies becomes of paramount importance in proactively avoiding forced outage or liberation of components, thus reducing property loss and saving customers maintenance costs. Over limit (Alarm Possible damage or improper operation) Threshold level Normal pattern Abnormal pattern (Alarm Potential damage or improper operation) compressor via regular water washes and parts replacement or upgrade during major inspection. Usually most of the degradation resulted from the equipment operation can be recovered through proper offline water wash of its compressor, while the degradation resulted from the mechanical deterioration (e.g., hot gas path component wear/damage) or parts malfunction (e.g., compressor bleed valve open) can be recovered from an overhaul. On the other hand, the nonrecoverable degradation becomes permanent deterioration on the gas turbine even after a major overhaul. The remote performance monitoring of the gas turbine is able to monitor both recoverable and non-recoverable performance degradation. With an allowable degradation level predefined, an alarm can be triggered to recommend corrective actions to the site such that the optimal maintenance can be taken to restore the recoverable performance degradation, as explained in the subsequent sections. Monitoring variable 1 Figure 1. Remote monitoring of two anomalies Recently thermodynamic performance monitoring has been developed in industry to continuously track performance changes of the power generation assets. This helps to monitor that generating facilities are appropriately operated and maintained in order to obtain optimal performance from the machinery and its operation. These analytics and processes have made thermal performance monitoring a critical service in enabling customers to increase operating revenues, reduce fuel costs and enhance customer dispatch competitiveness. This paper intends to present the system, methodology and processes developed for remote performance monitoring of gas turbines. PERFORMANCE MONITORING BACKGROUND Gas turbine performance is a function of many factors including, but not limited to, turbine design, component technology upgrade, operating mode, and site ambient conditions. The turbine degradation accumulates as its operating hour increases. It is usually divided into two categories: mechanical degradation and performance degradation. The mechanical degradation is caused by a variety of factors such as wear in bearings and seals, coupling issues, excessive vibration and noise, or problems in the lube oil system, which is beyond the scope of this paper. Refer to Meher-Homji et al. (2000) for more information about the mechanical degradation. This paper is focused on remote monitoring of performance degradation for a gas turbine. The performance degradation can be categorized into two types: recoverable and non-recoverable, as illustrated in Fig. 2. The recoverable degradation can be recovered by proper maintenance actions such as cleaning Figure 2. Performance degradation components There are a number of time-varying site conditions impacting the turbine performance (Meher-Homji et al. 2000). These conditions include, but not limited to, ambient humidity, ambient pressure, ambient temperature, inlet filter pressure losses, exhaust system pressure losses, fuel heating value, fuel flow, and fuel temperature. It should be pointed out that usually the change of site conditions is not the cause of performance degradation, but will result in variability of performance results. In order to assess the performance of a gas turbine over a given operational period (e.g., two year), or to compare the performance over a fleet of gas turbines over a certain time, the effect of site conditions on performance degradation needs to be eliminated from the instrumented performance indicators (e.g., power output) via proper techniques. In this paper, the measured performance output along with the calculated performance heat rate and compressor efficiency is calibrated to remove the effects of site conditions. The obtained performance indicators are referred to as the corrected ones and used as monitoring indicators. Figure 3 shows the example data of corrected performance output used for remote monitoring. The performance degradation and its recovery after offline water wash are also demonstrated in Fig Copyright 2013 by ASME

3 Figure 3. Example of corrected power output, degradation, and performance recovery after offline water wash MONITORING FRAMEWORK Figure 4 shows the high-level process and framework for performance monitoring of a gas turbine. The monitoring system gathers data from the sites on a near real-time basis, monitoring data quality, conducting performance calculations. It triggers alarms based on the predefined multilayer logic and escalating issues to provide proactive recommendations for performance recovery after conducting in-detail diagnostics. Recommendations regarding performance degradation may be projected to next planned outage for facilitating business decision making such as resource allocation and part procurement. These projection analytics are usually called performance degradation prognostics, which utilize historical performance data, degradation trends, and physics-based simulation. The financial value of this process can be directly related to improving the efficiency of the turbine s thermodynamic conversion (i.e., the consumption of fuel per unit of generated output), thus lowering the operating cost of fuel for customers. As such, the system helps to ensure that the customer receives the contracted levels of power output and/or efficiency. In addition, it provides a platform for engineers to drive performance improvements, mitigate outage risk, and enhance service offerings by taking appropriate actions proactively. For instance, in 2012, over 100 performance escalations had been created from this process, averaging kwh projected savings to the customer from each case (i.e., the estimated loss if the unit was kept operating to the next scheduled outage). Assuming an average sale price for unit electricity is $0.05kWh, each escalation may save up to $500,000 for the plant owner from the improved turbine operating efficiency. ON-SITE MONITORING AND DATA ACQUISITION In remote performance monitoring of the gas turbine or asset s health, the on-site monitor (OSM) data is continuously collected in real time via the sensors installed on the equipment and control system on the power site, usually at specified intervals (e.g., one-minute or one-second). The data is then transmitted to the monitoring center via a multi-layer secured network, as shown in Fig. 5. The OSM data used for performance calculation includes key ambient condition parameters such as temperature, humidity, and pressures, and turbine operational parameters such as power output, fuel temperature, fuel heating value, turbine shaft speed, and compressor inlet and exhaust pressure drop. In addition, static constants about the asset may be required for performance calculation and modeling, including turbine configuration profile, configured condition as well as certain accessory options (e.g., steam injection flow). Both static constants and time-dependent dynamic data are merged in the monitoring central database and then read as inputs to the performance calculation and, further, to performance anomaly detection in the rule calculation engine. Before performance anomaly detection analytics are implemented on an asset in the central system, the quality of all input data and the existence of all critical performance parameters, such as compressor inlet temperature, compressor discharge temperature, generator gross power output, and turbine shaft speed, will be checked for sufficient quality. If any of critical input variables are unavailable or invalid, an error code will be output to the central data base. In addition, parameter anomaly detection analytics trigger alarms for the corresponding variable with unavailable or invalid data in order to take corrective actions. For other vital tags such as ambient pressure and compressor inlet pressure drop, ISO (International Standard Organization) or configured condition values are applied to impute the missing or invalid data point to ensure continuous performance monitoring. PERFORMANCE CALCULATION AND MODELING Performance monitoring involves comparing and tracking the asset performance indicators consistently over time. These performance indicators typically include power output, heat rate, and compressor efficiency. For comparison and assessment purpose, these performance parameters must be continuously corrected to desired conditions, such as configured or ISO conditions, as mentioned earlier. Two sets of performance calculation methodologies have been integrated in the system for performance calculation and correction: thermodynamic modeling and site-specific condition correction based. The thermodynamic modeling approach calculates the performance for a gas turbine unit using thermodynamic cycle matching. Inputs into this approach include three parts: dynamic OSM data, static condition data, and equipment configuration data. The time-dependent dynamic inputs are first merged with the static data from the database. These static inputs contain the performance of the gas turbine at ISO condition, configured condition and certain accessory options available for the asset. All these inputs are merged with the unit configuration data to generate an input file for performance calculation. Then a data reduction technique is employed to satisfy continuity and conserve energy of the gas turbine using OSM data. Finally, the performance of the gas turbine is calculated and corrected via thermodynamic cycle matching at baseload and ISO conditions. 3 Copyright 2013 by ASME

4 Figure 4. Performance monitoring framework A site-specific calculation algorithm, as documented in ASME test procedure ( ASME PTC ) for field test of gas turbine performance, is employed to calculate the corrected output and heat rate using site-specific correction factors, raw OSM data, performance test methodology and baseline test information. This approach has the potential to simulate and validate, if not replace, the field performance test, particularly if OSM data is collected from well calibrated sensors on site. This approach can be run for an individual site or for all the sites of interest for every critical operating condition, such as baseload (i.e., inlet guide vane full open and turbine shaft full speed). The results are generated on a specified interval corresponding to the inputs, often every 5 minutes. ANOMALY DETECTION AND ALARMS Across the fleet, units can degrade much faster than expected for reasons such as significant compressor fouling, bleed valve open, lube oil leakage, and inlet filter ineffectiveness. One important goal of performance anomaly detection is to detect when this excessive degradation is occurring. When degradation is detected, a proactive solution such as an offline water wash, instrumentation calibration, or early maintenance outage is provided to the customer in order to restore the lost performance and minimize the total nonrecoverable degradation that can occur. Multiple rules are embedded in the monitoring system to detect performance anomalies related to either hardware or operational issues. Based on the site-specific performance calculation, the performance degradation, aggregated trend, and offline water wash analytics are developed. The obtained results are processed and sent to the system to trigger alarming and escalation. 4 Copyright 2013 by ASME

5 Figure 5. Thermal performance monitoring: Analytics and visualization In addition to performance degradation, corrected performance calculations can be used in conjunction with other information to detect potential hardware failures. Based on the ISO corrected results obtained from thermodynamic modeling, for example, compressor anomaly and performance degradation analytics are developed to detect the stator hardware damage and turbine operational issues. These analytics apply adaptive methodologies to continuously compare current performance to last known detected conditions, thus escalating abnormal changes for multi-tier diagnostics in order to recommend corrective actions and reduce risk of hardware damage. To successfully detect the performance anomaly issues and reduce false alarms, three-layer alarming logic is incorporated into the monitoring system including yellow (medium), orange (serious), and red (significant) alarms. This logic triggers an alarm based on the percent reduction of corrected power output and corrected heat rate relative to the corresponding values after the latest offline water wash was conducted. If no water wash information is available, the baseline performance after the unit was commissioned can be used as the reference. An alarm will also be triggered if the performance improvement after offline water wash is lower than a specified threshold (e.g., improvement is negative). These analytics can be easily extended to include other performance metrics such as corrected compressor efficiency or other indicators including speed of degradation (i.e., relative reduction over the operational time after the last offline water wash). The multilayer smart alarm system enables the detection of performance degradation issues and the reduction of false alarms. DIAGNOSTICS AND PROGNOSTICS Once an alarm is triggered from the monitoring system, detailed diagnostics are carried out by multi-level engineering personnel, including the monitoring operation team, product services, and hardware design engineers, depending on the severity of the problem. A wide variety of techniques are used to identify underlying causes of performance degradation, including statistical analytics, physics-based thermodynamic simulation, time series trend analysis, and ad-hoc empirical evaluation. Generally, the site manager is very involved in the diagnostics process in order to generate a proper, proactive strategy to recover performance for a specified site. In addition to the diagnostics, performance degradation may be projected to next planned outage for facilitating critical business decision making such as resource allocation, hardware upgrade, and part procurement. As shown in Figure 6, the projection analytics are usually called degradation prognostics, which utilize the historical performance data, degradation trend, and physics-based simulation to predict the degree of degradation at a given point in time (e.g., next planned outage). After proper actions are taken, the performance degradation is corrected to the target level. The diagnostics and prognostics information continues to be broadcast via multi-channels including telephone, , and web. Through performance prognostics from the current status of a given unit, the monitoring system enables the trade-off analytics to facilitate critical decision-making regarding hardware upgrades, maintenance scope, and resource allocation. For example, thermodynamics based simulation may be employed to quantify how much the performance improvement can be achieved for a certain hardware upgrade such as enhanced inlet filtration. Performance prognostics can also advise on the timing and frequency of offline water wash and predict the quality of next offline water wash based on historical performance degradation data. DEMONSTRATION EXAMPLE With multi-tier diagnostics, proactive recommendations are provided to customer sites in order to increase benefits and reduce losses. These recommendations may include maintenance actions, sensor calibrations, offline compressor water wash timing and assessments, maintenance schedules, hardware upgrades, and engineered improvements. In each case, a performance fishbone for a gas turbine is developed to enumerate possible root causes. This is used to facilitate the diagnostics and assessment of the gas turbine performance. For example, if performance diagnostics show that the continuously monitored degradation rate is attributed to the lack of compressor washes, lube oil leakage, inlet filter clogging or open bleed valve, maintenance earlier than planned may be needed to ensure that the gas turbine operates most efficiently. 5 Copyright 2013 by ASME

6 Degradation This section presents two examples to demonstrate the proactive recommendations resulted from the remote performance monitoring system. Figure 6. Thermal performance monitoring: Escalation and prognostics Figure 7 illustrates an exemplary case in which nearly kwh was lost due to the lube oil leakage issue at the inlet system, a situation identified by multi-tier diagnostics and alarmed in accordance with the performance monitoring analytics rule. In this situation, the notification was reactively escalated through manual intervention (e.g., customer or customer contact creating a troubleshooting request case). With the developed performance monitoring system, the alarming and escalation of performance issues is automated to proactively initiate corrective actions for performance recovery. The system not only incorporates multiple alarm logic based on the degradation severity of the turbine, but also seamlessly integrates parameter data quality check and data anomaly alarming to enhance monitoring accuracy. Figure 8 illustrates another exemplary case in which about 11MW corrected performance output was lost due to the compressor fouling and sensor issues. This is another typical situation identified by multi-tier diagnostics and alarmed in accordance with the performance monitoring analytics rule. SUMMARY Since the performance monitoring system was implemented, this analytic has provided valuable identification and understanding of which units were suffering from rapid or severe degradation and required technical assistance to resolve the deficiencies. Newly developed analytics and processes have resulted in about 10 degradation cases per month. Each case generates significant savings to customers via improved performance with proactive recommendations. Two typical examples have been provided to show how a turbine gained more power output after the recommendation of offline water wash, sensor calibration, or correction of parts malfunction. In summary, the intelligent analytics are able to provide comprehensive thermal performance monitoring including data sensing, hardware/operational anomaly detection, proactive expert recommendations, site specific degradation assessment, water wash effectiveness monitoring. These performance activities have been validated to improve equipment efficiency and drive significant value for customers by lowering operating fuel cost, potentially increasing dispatch competitiveness. In particular, performance monitoring of gas turbine assets in a centralized and remote fashion provides reliable and repeatable business insight into unit performance with proactive real-time monitoring to capture performance issues sooner. This real-time monitoring system and methodology presented in this paper converts this data into knowledge and solutions, and enables business initiatives like degradation based maintenance. It has potential to replace the field performance testing via remote performance monitoring. These innovations have resulted in increased revenue to customers. For example, such monitoring systems have yielded over 100 performance degradation cases, resulting in an estimated saving for customers of over $50 million. More importantly, the performance monitoring capability and platform has been extended for combined cycle power plants, through the development of multiple sophisticated analytics and processes on both the thermodynamic performance calculation approach and data-driven adaptive methodology. REFERENCES ASME (2006), Performance Test Code on Gas Turbines. American Society of Mechanical Engineers (ASME), ASME PTC Balevic D, Hartman S and Youmans R (2010), Heavy-Duty Gas Turbine Operating and Maintenance Considerations. GER- 3620L.1, GE Energy, Atlanta, GA. Brooks FJ (2000), GE Gas Turbine Performance Characteristics. GER-3567H. GE Power Systems, Schenectady, NY. Della Villa SA and Koeneke C (2010), A Historical and Current Perspective of the Availability and Reliability Performance of Heavy Duty Gas Turbines: Benchmarking and Expectations, GT , ASME Turbo Expo, Glasgow, UK, June Johnston JR (2000), Performance and Reliability Improvements for Heavy-Duty Gas Turbines. GER-3571H, GE Power Systems, Schenectady, NY. Meher-Homji CB, Chaker MA and Motiwala HM (2001), Gas Turbine Performance Deterioration, Proceedings of the 30 th Turbomachinery Symposium, , Houston, TX. 6 Copyright 2013 by ASME

7 Figure 7. Thermal performance monitoring example 1: Escalation, Diagnostics and Recommendation Figure 8. Thermal performance monitoring example 2: Escalation, Diagnostics and Recommendation 7 Copyright 2013 by ASME