• Energy Research

Pumps are one of the most critical machines in the petrochemical process. Condition monitoring of such parts and detecting faults at an early stage are crucial for reducing downtime in the production line and improving plant safety, efficiency and reliability. This paper develops a fault detection and isolation scheme based on an unsupervised machine learning method, sparse autoencoder (SAE), and evaluates the model on industrial multivariate data. The Mahalanobis distance (MD) is employed to calculate the statistical difference of the residual outputs between monitoring and normal states and is used as a system-wide health indicator. Furthermore, fault isolation is achieved by a reconstruction-based two-dimensional contribution map, in which the variables with larger contributions are responsible for the detected fault. To demonstrate the effectiveness of the proposed scheme, two case studies are carried out based on a multivariate data set from a pump system in an oil and petrochemical factory. The classical principal component analysis (PCA) method is compared with the proposed method and results show that SAE performs better in terms of fault detection than PCA, and can effectively isolate the abnormal variables, which can hence help effectively trace the root cause of the detected fault.

Este documento presenta un enfoque de control efectivo para la vibración estructural de las turbinas eólicas terrestres en la dirección del borde. Actualmente se desarrollan enormes turbinas eólicas de varios megavatios para cosechar grandes cantidades de energía del viento. Tales diseños requieren la construcción de enormes palas y torres delgadas que, en consecuencia, conducen a deformaciones estructurales indeseables que dificultan la producción de energía y reducen la vida útil de la turbina eólica. Muchos investigadores han trabajado en el control estructural de las turbinas eólicas. Sin embargo, estos esfuerzos no han dado como resultado una mitigación efectiva y confiable de la deformación de los elementos estructurales, ni han logrado una solución económica en términos de explotación de los actuadores. El trabajo presentado en este trabajo, sin embargo, introduce un controlador semi-activo basado en la optimización del enjambre de partículas que aprovecha los amortiguadores magnetorreológicos para mitigar los desplazamientos de las palas en el borde. Los amortiguadores se modelan utilizando redes neuronales, ya que son capaces de predecir fuerzas futuras y eliminar el retraso de control. El controlador desarrollado se prueba en varias configuraciones de colocación de actuadores en un aerogenerador de referencia de 5 MW. El enfoque propuesto, de hecho, mostró una reducción significativa de más del 80% en las respuestas de pico y alrededor del 77% de la respuesta de pico a pico de las palas contra sistemas incontrolados y pasivos, lo que conduce a promover la longevidad de las turbinas eólicas. Cet article présente une approche de contrôle efficace pour les vibrations structurelles des éoliennes terrestres dans la direction du bord. D'énormes éoliennes de plusieurs mégawatts sont actuellement développées pour récolter de grandes quantités d'énergie éolienne. De telles conceptions nécessitent la construction d'énormes pales et tours minces qui entraînent par conséquent des déformations structurelles indésirables qui entravent la production d'énergie et réduisent la durée de vie de l'éolienne. De nombreux chercheurs ont travaillé sur le contrôle structurel des éoliennes. Cependant, ces efforts n'ont ni abouti à une atténuation efficace et fiable de la déformation des éléments structurels, ni à une solution économique en termes d'exploitation des actionneurs. Le travail présenté dans cet article, cependant, présente un contrôleur semi-actif basé sur l'optimisation de l'essaim de particules qui exploite les amortisseurs magnétorhéologiques pour atténuer les déplacements des lames sur le bord. Les amortisseurs sont modélisés à l'aide de réseaux de neurones car ils sont capables de prédire les forces futures et d'éliminer le retard de contrôle. Le contrôleur développé est testé dans plusieurs configurations de positionnement des actionneurs sur une éolienne de référence de 5 MW. L'approche proposée, en effet, a montré une réduction significative de plus de 80% des réponses de crête et d'environ 77% de la réponse de crête à crête des pales contre des systèmes non contrôlés et passifs, ce qui conduit à promouvoir la longévité des éoliennes. This paper presents an effective control approach for structural vibration of onshore wind turbines in the edgewise direction. Huge multi mega-watt wind turbines are currently developed to harvest large amounts of energy from the wind. Such designs require the construction of huge slender blades and towers which consequently lead to undesirable structural deformations that hinder the power production and reduce life span of the wind turbine. Many researchers have worked on structural control of wind turbines. However, these efforts neither have resulted in an effective reliable mitigation for deformation of structural elements, nor they have achieved an economical solution in terms of actuators exploitation. The work presented in this paper, however, introduces a particle swarm optimisation-based semi-active controller which exploits magnetorheological dampers to mitigate edgewise blade displacements. Dampers are modelled using neural networks for they are capable of predicting future forces and eliminating control lag. The developed controller is tested at several configurations of actuators placement on a benchmark 5-MW wind turbine. The proposed approach, indeed, showed a significant reduction of over 80% in the peak responses and about 77% of peak-to-peak response of blades against uncontrolled and passive systems which leads to promoting longevity of wind turbines. تقدم هذه الورقة نهج تحكم فعال للاهتزاز الهيكلي لتوربينات الرياح البرية في اتجاه الحافة. يتم حاليًا تطوير توربينات رياح ضخمة متعددة الميجاوات لحصاد كميات كبيرة من الطاقة من الرياح. تتطلب مثل هذه التصاميم بناء شفرات وأبراج نحيلة ضخمة مما يؤدي بالتالي إلى تشوهات هيكلية غير مرغوب فيها تعيق إنتاج الطاقة وتقلل من العمر الافتراضي لتوربينات الرياح. عمل العديد من الباحثين على التحكم الهيكلي في توربينات الرياح. ومع ذلك، لم تسفر هذه الجهود عن تخفيف فعال وموثوق به لتشوه العناصر الهيكلية، كما أنها لم تحقق حلاً اقتصاديًا من حيث استغلال المشغلات. ومع ذلك، يقدم العمل المقدم في هذه الورقة وحدة تحكم شبه نشطة تعتمد على تحسين سرب الجسيمات والتي تستغل مخمدات الانسيابية المغناطيسية للتخفيف من إزاحة الشفرات الطرفية. يتم نمذجة المخمدات باستخدام الشبكات العصبية لأنها قادرة على التنبؤ بالقوى المستقبلية والقضاء على تأخر التحكم. يتم اختبار وحدة التحكم المطورة في العديد من تكوينات وضع المشغلات على توربين رياح قياسي بقدرة 5 ميجاوات. أظهر النهج المقترح، في الواقع، انخفاضًا كبيرًا بأكثر من 80 ٪ في استجابات الذروة وحوالي 77 ٪ من استجابة الشفرات من الذروة إلى الذروة ضد الأنظمة غير المنضبطة والسلبية مما يؤدي إلى تعزيز طول عمر توربينات الرياح.

Large rotating machinery, such as centrifugal gas compressors and pumps, have been widely applied and acted as crucial components in the oil and gas industries. Breakdowns or deteriorated performance of these rotating machines can bring significant economic loss to the companies. In order to conduct effective maintenance and avoid unplanned downtime, a system-wide health indicator is proposed in this paper. The health indicator not only uses a dynamic risk profile, but also considers financial loss and the fault probability based on condition monitoring data. This methodology is carried out by four steps: fault detection, probability of fault calculation, consequence of fault calculation and dynamic risk assessment. In our methodology, the fault probability is calculated by robust Mahalanobis distance, presenting as a system-wide feature from a sparse autoencoder fault detection model enabled early fault detection. The value of the health indicator is presented in financial loss, which assists in effective operational decision-making in a process system. To evaluate the performance of the proposed indicator, two case studies were carried out—one case tested on multivariate industrial data obtained from a pump, and another one tested on an industrial data set from a compressor. Results prove that the integrated health indicator can detect the faults at their incipient stages, indicate the degradation of the system with dynamically updated process risk at each sampling instant, and suggest an appropriate shutdown time before the system suffers severe damage. In addition, this methodology can be adapted to other machines’ health assessments, such as those of turbines and motors. The presented method of processing the industrial data set can benefit relevant readers.

The oil lubrication system is a critical part of the helicopter main gearbox (MGB) and this is evident in the many accidents and incidents over the last 30 years. On a category A rotorcraft, a regulatory requirement mandates the MGB to sustain operation for at least 30 minutes following the loss of the primary oil lubrication pressure. The aim of this study was to undertake a comparative investigation into the performance of mist lubrication, using commercially available thioether (MCS-293™), on a category A helicopter MGB under loss of oil conditions. Experimental observations highlighted that the high-speed input module of the MGB attained the highest temperature and was a limiting factor to continued gearbox operation under loss of oil conditions. Results showed that by routing thioether mist through existing galleries within the MGB a lower rate of temperature increase was achieved, in comparison with a dry-run condition.

Toward automobile electrification and automation, a smart scenario of DC-charging plug-in electric vehicles (PEVs) at any parking lot equipped with chargers is proposed. In this paper, this scenario is composed of four main stages; In the first stage, an investigation of the temperature or/and relative humidity impact on the charging process of the PEVs is implemented using the constant current-constant voltage (CC-CV) protocol. This was followed by a novel PEV classification model under the impacts of various ambient circumstances. Then an estimation of the charging characteristic parameters at the corresponding conditions is obtained. Finally, the model identification of the battery dynamic behaviour is sufficiently proposed. The feedforward backpropagation neural network (FFBP-NN) as a supervised classification algorithm supported by the statistical analysis of an instant charging current sample is used, which achieves an accuracy of 83.2%. In addition, the FFBP-NN perfectly estimated the charging current, terminal voltage, and charging interval time with a maximum error of 1%. Eventually, a sufficient identification model of the battery dynamic behaviour based on the Hammerstein-Wiener (HW) model is introduced with the best fit of 89.62% and an error of 1.1876%. The experimental and simulated results are within 1%error with the preceding research literature.

Stochastic fast-charging of electric vehicles (EVs) affect the security and economic operation of the distribution power network. Aggregator awareness in the electric power industry is fast growing in tandem with the growing number of EVs. This paper proposes a novel smart techno-economic operation of the electric vehicle charging station (EVCS) in Egypt controlled by the aggregator based on a hierarchal model. The deterministic charging scheduling of the EVs is the upper stage of the model to balance the generated and consumed power of the station and flat the surplus power supplied to the utility grid. Mixed-integer linear programming (MILP) is used to solve the first stage where the peak demand value is reduced to 48.17% (4.5 kW) without using any extra battery storage systems. The second challenging stage is to maximize the charging station profit whilst minimizing the EV charging tariff, which needs a trade-off. In this stage, MILP and Markov Decision Process Reinforcement Learning (MDP-RL) resulted an increase in EVCS revenue by 28.88% and 20.10%, respectively. However, the EVs charging tariff is increased by 21.19%, and 15.03%, respectively. Hence, MDP-RL is an adequate algorithm for such a complex model. The outcomes reveal a sufficient techno-economic hierarchal model concerning the normal operation stated in the literature.