https://www.mdu.se/

The flue gas associated with hydrogen fueled gas turbines has enhanced steam content and different thermophysical properties as compared to that of natural gas fuel case. The enhanced steam might lead to a rigorous corrosion degradation in the hot gas path components of the gas turbines such as turbine blades. In addition to this, high heat transfer rate can contribute to erosion, thermal fatigue, and creep damages. Hydrogen fueled gas turbines are also susceptible to some common routine faults such as fouling in the compressor section. Consequently, the health and performance of a hydrogen fueled gas turbines are degraded. Therefore, health monitoring in terms of remaining useful life (RUL) estimation of such turbines is of greater interest for the gas turbines OEMs and operators to ensure an enhanced availability and reliability in line with industry 4.0. The current study, therefore, develops a performance-based RUL estimation model for a 100-kW micro gas turbine that was recently retrofitted with hydrogen compliant FLOX burner. The validated performance model was further utilized for synthesizing run to failure data for fault diagnosis and RUL estimation. The study further incorporated linear and polynomial regression approaches and compared the end of life of gas turbines running on natural gas and hydrogen fuels. It became evident from the study that RUL of a gas turbine running on hydrogen fuel is approximately 6.47% lower than that of natural gas fueled gas turbines. These findings underline the necessity of using strong prediction models, as well as targeted maintenance actions, to limit the consequences of turbine corrosion in hydrogen powered micro gas turbines. The findings of the present study further provide new horizons for design modification and effective health monitoring of hydrogen fueled gas turbines.

In manufacturing, and particularly in manually driven processes, diagnostics and decision support tools that utilize data-driven methods are key factors for reliable production processes. The investment casting manufacturing process relies on quality assessment through microscope examinations of cross-sections (cutups) of produced pieces, traditionally depending on operator judgment to manually approve or reject parts, which may introduce bias. This work focuses on identifying and addressing the need for reliability and efficiency in the investment casting manufacturing process by proposing a decision support tool to assist the operator in defect detection and fault identification in a semi-automated way. Initially, we explore the machine learning classifier Random Forest and then propose the use of a convolutional neural network, a deep learning method, for improving binary classification accuracy when predicting the presence of a defect in a microscope-derived image. The model presents classification accuracy between faulty and non-faulty images at 98% as a key finding and also tested on new, never-before-seen images from the production process. The results demonstrate the transformative potential of introducing data-driven methods such as convolutional neural networks into manual manufacturing processes, paving the path for more reliable production methods in the investment casting manufacturing industry.

Global energy transition efforts towards decarbonization requires significant advances within the energy sector. In this regard, hydrogen is envisioned as a long-term alternative fuel for gas turbines. Accordingly, the gas turbine industry has expedited their efforts in developing 100% hydrogen compliant burners and associated auxiliary components for retrofitting the existing gas turbines. The utilization of hydrogen in gas turbines has some underlying challenges such as corrosion mainly originating from increased steam content in the hot gas path. In addition to corrosion, the gas turbine compressor is vulnerable to fouling which is the most commonly occurring fault in gas turbine operating over certain time window. Both faults are susceptible to performance and health degradation. To avoid expensive asset loss caused by unexpected downtimes and shutdowns, timely maintenance decision making is required. Therefore, simple, accurate and computationally efficient fault detection and diagnostics models become crucial for timely assessment of health status of the gas turbines. The present study encompassed development of a physics-based performance model of a 100-kWe micro gas turbine running on 100% hydrogen fuel. The model is validated with experimental data acquired from test campaigns at the University of Stavanger. Data synthesized from experimentally validated performance model are utilized further for training machine learning algorithms. To identify an accurate algorithm, various algorithms such as support vector machine, decision tree, random forest algorithm, k-nearest neighbors, and artificial neural network were tested. The findings from fault diagnostics process (classification) revealed that ANN outperformed its counterpart algorithm by giving accuracy of 94.55%. Similarly, ANN also showed higher accuracy in performance degradation estimation process (regression) by showing the MSE of training loss as low as ~0.14. The comparative analysis of all the chosen algorithms in the present study revealed ANN as the most accurate algorithm for fault diagnostics of hydrogen fueled gas turbines. However, there is need to further implement the ensemble machine learning models or deep learning model to explore and expedite the real time fault diagnostic accuracy to avoid false alarms and missed detections in context of hydrogen fuel.

The utilization of hydrogen fuel in gas turbines brings significant changes to the thermophysical properties of flue gas, including higher specific heat capacities and an enhanced steam content. Therefore, hydrogen-fueled gas turbines are susceptible to health degradation in the form of steam-induced corrosion and erosion in the hot gas path. In this context, the fault diagnosis of hydrogen-fueled gas turbines becomes indispensable. To the authors' knowledge, there is a scarcity of fault diagnosis studies for retrofitted gas turbines considering hydrogen as a potential fuel. The present study, however, develops an artificial neural network (ANN)-based fault diagnosis model using the MATLAB environment. Prior to the fault detection, isolation, and identification modules, physics-based performance data of a 100 kW micro gas turbine (MGT) were synthesized using the GasTurb tool. An ANN-based classification algorithm showed a 96.2% classification accuracy for the fault detection and isolation. Moreover, the feedforward neural network-based regression algorithm showed quite good training, testing, and validation accuracies in terms of the root mean square error (RMSE). The study revealed that the presence of hydrogen-induced corrosion faults (both as a single corrosion fault or as simultaneous fouling and corrosion) led to false alarms, thereby prompting other incorrect faults during the fault detection and isolation modules. Additionally, the performance of the fault identification module for the hydrogen fuel scenario was found to be marginally lower than that of the natural gas case due to assumption of small magnitudes of faults arising from hydrogen-induced corrosion.

Depending on the data available, gas turbine diagnostics may be performed in various ways. For model based diagnostics methods, the most common data to use is from steady state operation. Regarding the amount of data, the smallest dataset comprise of a single operating condition for a specific gas turbine. The other end of the scale considers a fleet of engines, where data can be utilized for cross comparisons and identification of deviations in engine operation. In-between these two extremes, twin-engine airplanes can be utilized to obtain additional diagnostic information. In this paper, a multi-point diagnostic method for a twin-engine airplane is developed and evaluated. It is based on the assumption that one engine can be operated individually for data collection, by varying the bleed flow extraction, while the other engine supply the airplane subsystems with the required bleed flow during the data collection time. The collected data then goes into a multi-point optimization which minimizes the difference between the health parameter estimations. From the health parameters, measurements corresponding to a reference operating condition are obtained and used by a data-driven classifier for fault identification and isolation. The method has proven to be able to detect both single and double component faults with high accuracy for the evaluated dataset.

Gas turbines play a vital role in various industries. Timely and accurately predicting their degradation is essential for efficient operation and optimal maintenance planning. Diagnostic and prognostic outcomes aid in determining the optimal compressor washing intervals. Diagnostics detects compressor fouling and estimates the trend up to the current time. If the forecast indicates fast progress in the fouling trend, scheduling offline washing during the next inspection event or earlier may be crucial to address the fouling deposit comprehensively. This approach ensures that compressor cleaning is performed based on its actual health status, leading to improved operation and maintenance costs. This paper presents a novel prognostic method for gas turbine degradation forecasting through a time-series analysis. The proposed approach uses the Temporal Fusion Transformer model capable of capturing time-series relationships at different scales. It combines encoder and decoder layers to capture temporal dependencies and temporal-attention layers to capture long-range dependencies across the encoded degradation trends. Temporal attention is a self-attention mechanism that enables the model to consider the importance of each time step degradation in the context of the entire degradation profile of the given health parameter. Performance data from multiple two-spool turbofan engines is employed to train and test the method. The test results show promising forecasting ability of the proposed method multiple flight cycles into the future. By leveraging the insights provided by the method, maintenance events and activities can be scheduled in a proactive manner. Future work is to extend the method to estimate remaining useful life.

Prognosis, or the forecasting of remaining operational life of a component, is a fundamental step for predictive maintenance of turbomachines. While diagnostics gives important information on the current conditions of the engine, it is through prognostics that a suitable maintenance interval can be determined, which is critical to minimize costs. However, mature prognostic models are still lacking in industry, which still heavily relies on human experience or generic statistical quantifications. Predicting future conditions is very challenging due to many factors that introduce significant uncertainty, including unknown future machine operations, interaction between multiple faults, and inherent errors in diagnostic and prognostic models. Given the importance to quantify this uncertainty and its impact on operational decisions, this work presents an information fusion approach for gas turbine prognostics. Condition monitoring performed by a Bayesian network is fused with a particle filter for prognosis of gas turbine degradation, and the effect of diagnostic models uncertainty on the prognosis are estimated through probabilistic analysis. Gradual and rapid degradation are simulated on a gas turbine performance model and the impact of sensor noise and initial conditions for the particle filter estimation are assessed. This work demonstrates that the combination of Bayesian networks and particle filters can give good results for short-term prognosis.

In the forthcoming era of fully autonomous mining, spanning from drilling operations to port logistics, novel approaches will be essential to pre-empt hazardous situations in the absence of human intervention. The progression towards complete autonomy in mining operations must have meticulous approaches and uncompromised security. By ensuring a secure transition, the mining industry can navigate the transformative shift towards autonomy while upholding the highest standards of safety and operational reliability. Experiments involving autonomous pathways for mining machinery that utilize AI for route optimization demonstrate a higher speed capacity than manually operated approaches; this translates to enhanced productivity, subsequently fostering increased production capacity to meet the rising demand for metals. Nonetheless, accelerated wear on crucial elements like tires, brakes, and bearings on mining machines has been observed. Autonomous mining processes will require smarter machines without humans that guide and support actions prior to a hazardous situation occurring. This paper will delve into a comprehensive perspective on the safety of autonomous mining machines by using Bayesian networks (BN) to detect possible hazard fires. The BN is tuned with a combination of empirical field data and laboratory data. Various faults have been recognized, and their correlation with the measurements has been established.

The study focused on the development of -gas turbine full- and part-load operation diagnostics. The gas turbine performance model was developed using commercial software and validated using the engine manufacturer data. Upon the validation, fouling, erosion, and variable inlet guide vane drift were simulated to generate faulty data for the diagnostics development. Because the data from the model was noise-free, sensor noise was added to each of the diagnostic set parameters to reflect the actual scenario of the field operation. The data was normalized. In total, 13 single, and 61 double, classes, including 1 clean class, were prepared and used as input. The number of observations for single faults diagnostics were 1092, which was 84 for each class, and 20,496 for double faults diagnostics, which was 336 for each class. Twenty-eight machine learning techniques were investigated to select the one which outperformed the others, and further investigations were conducted with it. The diagnostics results show that the neural network group exhibited better diagnostic accuracy at both full- and part-load operations. The test results and its comparison with literature results demonstrated that the proposed method has a satisfactory and reliable accuracy in diagnosing the considered fault scenarios. The results are discussed, following the plots.

This study presents a comprehensive analysis of the impact of variable inlet guide vanes and physical faults on the performance of a three-shaft gas turbine engine operating at full load. By utilizing the input data provided by the engine manufacturer, the performance models for both the design point and off-design scenarios have been developed. To ensure the accuracy of our models, validation was conducted using the manufacturer’s data. Once the models were successfully validated, various degradation conditions, such as variable inlet guide vane drift, fouling, and erosion, were simulated. Three scenarios that cause gas turbine degradation have been considered and simulated: First, how would the variable inlet guide vane drift affect the gas turbine performance? Second, how would the combined effect of fouling and variable inlet guide vane drift cause the degradation of the engine performance? Third, how would the combined effect of erosion and variable inlet guide vane drift cause the degradation of the engine performance? The results revealed that up-VIGV drift, which is combined fouling and erosion, shows a small deviation because of offsetting the isentropic efficiency drop caused by fouling and erosion. It is clearly observed that fouling affects more upstream components, whereas erosion affects more downstream components. Furthermore, the deviation of performance and output parameters due to the combined faults has been discussed.