Advancing industrial gas turbine field performance testing: a review of procedures and key considerations with emerging technologies

1. Introduction

In the past four decades, Industrial Gas Turbines (IGTs) have been employed in power generation, mechanical drives, and marine applications [1, 2]. The performance and reliability of industrial gas turbines are determined by taking measurement of critical parameters such as efficiency, fuel flow, and power output. Traditional methods for testing and evaluating these parameters have followed well-defined procedures and standards [1]. However, the emergence of artificial intelligence and other advanced technologies has necessitated the need to employ these new technologies, in addition to existing traditional methods, to optimize testing methodologies and improve the decision-making process.

The industrial gas turbine performance test procedure provides a structured approach for accurately assessing gas turbine performance while considering the operational environment [1]. Performance testing is essential for newly manufactured, repaired, or overhauled gas turbines [3]. It ensures the optimization of the IGT’s performance and helps minimize unscheduled shutdowns and repairs, as turbine component failures can lead to significant financial losses [1]. Although the testing procedure may involve considerable costs for operators and manufacturers, it serves as a critical foundation for operational decisions, including modifications, performance monitoring, and plant extensions. It also provides valuable complementary data to factory testing results. Established standard testing codes, such as those from the American Society of Mechanical Engineers (ASME), Verband Deutscher Ingenieure (VDI), and the International Organization for Standardization (ISO), have traditionally been used to assess the performance of industrial gas turbines. These codes provide comprehensive guidelines for site preparation, instrumentation requirements, thermodynamic calculations, and test result reporting [4]. They ensure consistency and accuracy across various testing scenarios, facilitating fair comparisons and informed operational decision-making [1].

The efficiency and operational stability of the gas turbine are determined by conducting performance testing. Performance testing provides information about power output, thermal efficiency, and fuel consumption under standardized and varying operational conditions [5]. Affonso et al. [6] stated that obtaining the heat rate is crucial for determining the overall performance of a gas turbine and is most commonly considered when conducting an acceptance test. The guidelines for accurate and repeatable tests, which are provided by international standards such as ASME PTC 22 [4] and ISO 2314 [7], are required to be adhered to when conducting performance testing. Comprehensive methods for evaluating turbine power output and efficiency are provided by the American Society of Mechanical Engineers (ASME) Performance Test Code [4]. Test requirements such as instrumentation setup, environmental correction factors, and test duration are outlined in ISO 2022[8]. A detailed framework for determining performance based on direct measurement and thermodynamic principles is provided in the ASME PTC 22 standard [4].

This review explores the possibility of enhancing the efficiency and accuracy of these tests by critically assessing the traditional methods, their limitations, and how modern technologies, including AI, can be deployed in addition to traditional methods to achieve enhanced testing methodologies and procedures in industrial gas turbines. The focus of this study is on using AI to complement the existing traditional testing approaches, optimize data acquisition, and predict operational failures.

2. Materials and methods

The methodology adopted in this study involved search strategies, inclusion/exclusion criteria, data extraction, and analysis procedures. The strategies adopted in conducting this study include examining current procedures and best practices in industrial gas turbine performance testing, identifying key challenges associated with performance testing, and exploring the role and impact of emerging technologies, including AI, in enhancing gas turbine performance testing. Fig. 1 shows the flow chart of the study.

To identify relevant and important literature in the area of study, a systematic and comprehensive search strategy was employed. Academic and industry databases were pivotal in conducting the study. The academic databases covered in the study include articles from Elsevier, Springer Nature, ASME Journal of Engineering for Gas Turbines and Power, PhD Thesis from reputable universities etc. The industry databases include the ASME Digital Collection, Gasturbine World, ASME Turbo Expo, and communications from leading gas turbine manufacturers such as General Electric, Siemens, Solar Turbines, and Mitsubishi. Grey literature, including technical reports, industry white papers, and conference proceedings, were also reviewed during the literature search. In addition, key search word combinations such as industrial gas turbine performance testing, gas turbine testing procedures, key considerations in turbine testing were used. Other key word combinations included application emerging technologies in gas turbine performance diagnostics, artificial intelligence in performance testing, and AI-based turbine diagnostics.

The literature was examined through a thematic analysis to uncover recurring patterns and emerging trends. The primary themes identified were as follows: traditional performance testing procedures, including benchmarking methods and their limitations; challenges in performance testing, such as instrumentation errors, environmental variability, and operational constraints; and the integration of emerging technologies, particularly the role of AI and machine learning in optimizing performance.

Fig. 1Flow chart of study

3. Studies on industrial gas turbine performance testing using traditional approaches

Diakunchak [9] conducted a fully factory-loaded test to verify the performance and mechanical integrity of a new engine model. To obtain measurement data on the engine and its components' performance, a large number of specialized instruments were employed. At each measurement point, nearly 500 individual readings were recorded. The results showed that the power output exceeded 47 MW when measurements were conducted under ISO conditions using natural gas as the fuel. Also, the author reported that metal temperature measurements in the combustor and turbine confirmed the engine's mechanical integrity. Bustos et al. [10] stated that during individual test campaign, some major gas turbine components can be verified to ascertain their mechanical integrity and performance. The authors further added that vibration levels in the power turbine are assessed through a mechanical run test (MRT), while the gas generator is expected undergo performance testing to determine whether its operating parameters are within acceptable limits. Lee et al. [11] conducted a performance test on a micro gas turbine consisting of a single-stage centrifugal compressor, a radial turbine, and an annular combustor. Various instrumentations were used to measure temperature, pressure, and flow rate at different sections of the engine as shown in Fig. 2. The compressor inlet and turbine exit temperatures were measured using T-type and K-type thermocouples, respectively, installed within the engine. Meanwhile, the fuel flow rate was measured using a thermal mass flow meter, and the compressor exit pressure was obtained using a pressure probe. Based on the measurement data, the authors calculated the shaft power and thermal efficiency.

Decker and Pathak [12] conducted an Engineer, Procure and Construct (EPC) performance test on Combustion Turbine Generators under base load condition to ascertain the facility-wide unfired net electrical output and heat rate. It was stated that during the test, the Chiller was switch off while the HRSG remained unfired. The study revealed that the corrected unfired net electrical output was greater than the guaranteed value, while the corrected unfired net heat rate was less than the guarantee value. Purvis [13] conducted the measurement of turbine inlet temperature using a single three-point stagnation thermocouple rake at the outlet from each of the six combustion chambers to provide a rough guide to temperature. The authors ensured that single unit readings were never accepted as representing a true mean outlet temperature. Six thermocouple units are usually used to measure the exhaust temperature. Of these, three are installed in the exhaust duct elbows in pairs [30]. Mathioudakis [14] presented methods for correcting data from gas turbine acceptance testing. The study focused on addressing issues that were not sufficiently covered by existing standards. The author presented a procedure for verifying guarantee data at specific operating points, and methods were proposed for correcting performance test data. Kurz et al. [15] discussed field testing of gas turbine-driven compressors and measurement uncertainties when test guidelines were judiciously followed. The authors outlined a compressor field testing procedure that reduces measurement inaccuracies and maintains cost efficiency. In addition, the authors addressed issues related to the planning and organization of field tests and necessary instrumentation. Other areas covered in the study include data reduction, data correction, test uncertainty, and the interpretation of test data. Furthermore, the study reviewed necessary test codes and their relevance to field testing.

Fig. 2Micro gas turbine test facility [11]

Purvis [13] conducted tests using standard procedures and rules similar to those outlined in the ASME Power Test Code '22 or the CIMAC code of acceptance requirements. The tests were performed on normal production units that were being prepared for delivery or on-site acceptance. During the measurement exercise, attempts were made to provide an alternative for measuring parameters where doubt existed regarding the accuracy of the recorded measurement data. The author used a measuring device that functions by sensing the distortion of the magnetic flux induced by the shaft torque when it was subjected to torsional strain to obtain the measurement of the output torque produced. Performance testing of an APU GTCP was conducted by MSc Thermal Power class of 2010. The performance test was carried out on an APU GTCP 30-92 Avon Test Facility located at Cranfield University, UK, to provide a first-hand insight on the operation of a gas turbine. The APU been investigated was an old gas turbine engine which performs multiple functions in an aircraft (see Fig. 3).

According to Pachidis [16] the measurements were obtained after starting and running until it is stabilized before readings were taken. The fuel flow was measured using a ball flow meter located just before the injection port into the combustion chamber. The flow meter was calibrated for the density of the fuel used, and the corresponding chart between the graduations and fuel flow were used to obtain the fuel flow in kg/s. While the pressure values were measured at the compressor exit and the exhaust using five tubes connected to a digital meter (DP101). The compressor intake total pressure was measured using three of the Pitot probes provided. While the remaining tubes, which consist of a rail of three tubes were used to measure static pressure at the exhaust, whose mean value provides the static exhaust pressure. The compressor delivery temperature and exhaust temperature were measured through thermocouples installed at the exhaust tubes. The K-type thermocouples made of Chromel-Alumel material, which are poor conductors and are well-suited for high-temperature measurements due to their low electromotive force, were used for obtaining the temperature measurements, as shown in Fig. 4. Due to the non-homogeneous gas flow in the exhaust duct, three thermocouple probes were installed at strategic positions according to the British Standard for the measurement of fluid flow.

Fig. 3Main components of the GTCP 30-92 engine [16]

Cong et al. [17] conducted experiments to mimic varying environmental conditions by controlling intake humidity using ultrasonic atomizers. The intake pressure was adjusted with a wind deflector, and intake temperature was regulated by selecting different room temperatures. The measurements collected during the experiment included flow, pressure, temperature, humidity, and smoke composition. Flow and pressure data were obtained using the micro gas turbine’s pre-installed system, while temperature and humidity were measured with an AS847 split-type meter. An infrared smoke analyzer (MGA5) was employed to measure the combustion-generated smoke components. The results showed that operating parameters of the micro gas turbine fluctuated slightly with changes in intake humidity, but the magnitude of these fluctuations was small, indicating that the turbine can operate normally even in high humidity environments. Also, power generation efficiency and compressor power consumption exhibited opposite trends when the intake temperature changed. Changes in intake temperature also affected the concentrations of NOx and CO emissions.

Fig. 4Exhaust instrumentations [16]

Sitanggang et al. [18] conducted an experimental study to analyse the performance of Gas Turbine Units in Jambi, Indonesia. The performance testing was conducted by collecting data accordance ASME Performance Test Code 22. Data obtained included parameters such as inlet air temperature, air filter pressure differential, fuel flow rate, exhaust gas temperature, self-use electrical energy, and electricity production. The kWh-meter transactions in the Generator Auxiliary Control (GAC) room was used to obtain the Net energy production data, while gross energy production data was read from the gross kWh-meter in the Local Control Room. The heat rate and compressor efficiency were calculated from data obtained from the measured parameters.

Syverud et al. [19] conducted an investigation to determine the effect of saltwater ingestion on the performance of a General Electric J85-13 turbojet engine. Salt ingestion was achieved by spraying atomized droplets of saltwater into the engine intake, leading to the fouling of salt deposits on the axial compressor. To obtain more detailed data than what was available from the existing test instrumentation, additional sensors were deployed. Figures 5 and 6 show the General Electric J85-13 turbojet engine and test equipment mounted at the inlet screen of the engine respectively. Four sensors positioned at the engine inlet screen measured the compressor inlet temperature. The compressor inlet total pressure was measured using three pitot tubes located in the bellmouth, while the static pressure at the same location was measured using three static ports. Static pressure at stage 5 was recorded via a 1 mm tap located at a single point on the circumference within the bleed channel. Gas path temperatures were measured using unshielded Resistance Temperature Detectors (RTDs) installed at stator rows 1, 3, 4, 6, and 8. Ambient temperature and relative humidity were manually recorded outside the test cell at the same location. The compressor inlet temperature was observed to be higher than the ambient temperature and varied with engine load, attributed to test cell recirculation. A minimum of 60 data points were collected at each setting, using a 2 Hz sampling rate, to reduce data scatter. The average of these readings was taken as the steady-state data point. All performance data were corrected to standard ISO reference conditions (15°C ambient temperature and 101,325 kPa barometric pressures). The authors reported that the largest shift in mean value occurred in the compressor discharge pressure, which resulted in partial overlap of the measurement uncertainties.

Fig. 5Single spool turbojet engine [19]

Fig. 6GE J85-13 with test equipment mounted at the inlet screen [19]

It is evident from the foregoing that numerous studies have been conducted using traditional methods of industrial gas turbine performance testing. Table 1 provides a summary of the key references reviewed for this research, highlighting the key performance parameters measured, the equipment used, and the objectives achieved in the respective tests. From the reviewed literature, it is evident that field performance testing requires considerable human effort for data collection and analysis. Also, accurate calibration of instruments is also required to reduce errors in the measured results. In addition, the personnel conducting the test must have a strong understanding of gas turbine performance calculations to estimate parameters that cannot be measured directly. In light of these challenges associated with traditional performance testing, it has become necessary to integrate emerging technologies, such as artificial intelligence, to address these issues and optimize the testing process for improved accuracy and efficiency.

4. Findings

From the literature review conducted on industrial gas turbine performance testing, including the assessment of various test procedures, key considerations, and the transformative role of emerging technologies, a detailed combination of findings is presented. Traditional testing methods outline standard procedures for evaluating the performance of gas turbines. These methods include steady-state, transient, and emission testing. Steady-state testing involves measuring parameters such as temperature, pressure, and fuel flow under stable operating conditions, while transient testing entails monitoring dynamic responses to load changes or operational events. For emission testing, the exhaust gas composition is examined to ensure compliance with environmental regulations.

The study revealed that it is essential to monitor key parameters such as, efficiency, power output, exhaust temperature and gas flow rate, air temperature, pressure ratios, and airflow rates, to assess the performance of gas turbines. These parameters provide valuable insights into the turbine’s condition, enabling operators to identify issues related to efficiency degradation. With this information, operators can take appropriate measures to optimize the turbine’s performance. To optimize the operation and performance of IGTs, it is imperative to ensure that the efficiency, fuel flow, power output, capacity, and all control installations are adequately verified.

The study revealed that traditional methods provided a reliable foundation for performance testing. However, they are associated with numerous challenges, including being labor-intensive and prone to inaccuracies due to sensor limitations and variations in environmental conditions. These limitations highlight the need for integrating advanced technologies such as Artificial Intelligence, for a more automated and adaptive testing solutions to improve accuracy and efficiency.

It is evident from the reviewed studies that measured data were essential wherever emerging technologies were applied to assess gas turbine performance. Standard performance test data, obtained through established ASME or ISO procedures, provided key parameters such as turbine inlet temperature (TIT), compressor pressure ratio, exhaust temperature, fuel flow, shaft speed, and vibration. These parameters form the foundation upon which artificial intelligence (AI) methods rely to optimize performance testing. Data generated through traditional testing approaches also serves as training input for machine learning models. The studies indicate that emerging technologies like AI build upon these traditional methods-especially in data-rich, repeatable processes such as instrumentation, logging, and normalization. These features form the basis for AI technologies to make the entire testing cycle faster, smarter, and more adaptive.

The study further revealed that gas turbine performance testing can be significantly optimized by integrating artificial intelligence (AI) techniques such as machine learning, digital twins, deep learning, and artificial neural networks into traditional testing methods. AI enhances efficiency by enabling better planning, reducing unnecessary tests, and minimizing delays caused by weather or load conditions. This is possible because machine learning models, trained on historical operational data, can recommend optimal testing conditions and accurately detect faulty readings, data noise, and instrument drift. Also, AI can simulate ideal turbine behavior for comparison with actual test data, allowing for more precise evaluation. It can also track test results over time to identify trends that may indicate issues such as compressor fouling, turbine blade degradation, or seal leakage.

While the review provides valuable insights, several limitations must be acknowledged. The rapid development of AI and sensor technologies may render some findings out-dated. In addition, limited access to proprietary industry data constrained the analysis, and the focus on peer-reviewed literature may have excluded relevant grey literature.

Key findings from the study which highlight the significance of the integration emerging technologies such as Artificial Intelligence into traditional gas turbine performance testing methodologies are presented in Table 3.

5. Conclusions

This study provides a comprehensive review of established procedures, key considerations, and emerging trends in gas turbine performance assessment to improve testing practices. Traditional methodologies for measuring critical performance indicators, such as power output and efficiency, remain fundamental but face limitations in precision and adaptability under diverse operating conditions.

Key considerations, including accurate instrumentation, standardized testing protocols, and comprehensive data collection, are essential for reliable assessments. The integration of emerging technologies with traditional methods has shown great potential in advancing industrial gas turbine performance testing.

AI-driven solutions enable predictive analytics, automated anomaly detection, and optimization of operational parameters, thereby significantly reducing downtime and maintenance costs. Additionally, machine learning models enhance data processing, providing deeper insights into turbine performance trends.

To achieve optimized performance testing and further minimize downtime and maintenance costs, a hybrid approach that combines traditional methods with data-driven analytics and emerging sensor technologies must be adopted.