سیاست‌گذاری نگهداری و تعمیرات پیشگویانه در مراکز فرآوری نفت و گاز

نوع مقاله : پژوهشی

نویسندگان

1 دانش آموخته دکتری مدیریت صنعتی- گرایش استراتژی صنعتی، دانشکده مدیریت دانشگاه ازاد تهران مرکز،تهران، ایران

2 استادیار مهندسی صنایع، دانشگاه شهید بهشتی، تهران، ایران

3 استادیار گروه مدیریت صنعتی، واحد تهران مرکزی، دانشگاه آزاد اسلامی، تهران، ایران

چکیده

سیاست مناسب در نگهداری و تعمیرات می‌گوید تعمیر زمانی باید انجام شود که نیاز باشد. اگرچه انجام تعمیرات پیشگیرانه می‌تواند تعمیرات ناگهانی و غیرمترقبه را کاهش دهد اما همچنان باعث کاهش در دستـرس بودن و افزایش هزینه‌های تعمیرات می‌گردد. شرکت‌ها می‌بایست استراتژی‌های نگهداری آنلاین و پیشگویانه را گسترش دهند که بتوانند تصور کنند که هر خرابی در هر زمانی ممکن است به‌طور تصادفی رخ دهد و این نیاز را از علائم و نشانه‌هایی که تجهیزات از خود بروز می‌دهند تشخیص دهند که به آن نگهداری و تعمیرات پیشگویانه یا نگهداری مبتنی بر شرایط می‌گویند. در این مقاله سعی شده است یک مدل پشتیبان تصمیم‌گیر برای نگهداری و تعمیرات پیشگویانه مبتنی بر شرایط بر اساس تکنیک‌های داده‌کاوی طراحی شود. این پروژه در یکی از شرکت‌های بهره‌برداری نفت و گاز جنوب کشور انجام شده و تجهیز انتخابی برای این پروژه توربین‌های گازی است که یکی از اساسی‌ترین و بحرانی‌ترین تجهیزات در کارخانه‌های فرآوری نفتی است. در این پروژه تکنیک داده‌کاوی شبکه عصبی MPL برای پیش‌بینی رخداد خرابی در تجهیز مورد نظر به کار گرفته شده است. در انتها پیشنهادهایی از جمله توسعه این مدل برای سایر تجهیزات، کنترل مدت زمان مشاهده صورت وضعیت تجهیزات و تعیین زمان بهینه تعمیر و نگهداری برای آینده ارائه شده است.
در این پروژه تکنیک داده کاوی شبکه عصبی MPL برای پیش بینی رخداد خرابی در تجهیز مورد نظر به کار گرفته شده است. در انتها پیشنهادهایی از جمله توسعه این مدل برای سایر تجهیزات، کنترل مدت زمان مشاهده صورت وضعیت تجهیزات و تعیین زمان بهینه تعمیر و نگهداری برای آینده ارائه شده است.

کلیدواژه‌ها


عنوان مقاله [English]

Predictive maintenance and repair policy in oil and gas processing centers

نویسندگان [English]

  • Saman Sorbi 1
  • Sajad Shokouhyar 2
  • Jalal Haghighat monfared 3
1 PhD student in Industrial Management - Industrial Strategy, Islamic Azad University Central Tehran Branch, Tehran, Iran
2 Assistant Professor of Industrial Engineering, Shahid Beheshti University, Tehran, Iran
3 Assistant Professor, Department of Industrial Management, Islamic Azad University Central Tehran Branch, Tehran, Iran
چکیده [English]

Aim and Introduction: In recent years, what have attracted the attention of various industry managers in today's competitive world is the reduction of production costs and, consequently, the reduction of the final price of products. Among these, repair costs are one of the most controllable costs in the industry, and it is natural that reducing them should be on the agenda of industry managers. One of the most important tools available to the authorities to achieve this goal is the use of new maintenance methods based on monitoring the condition of the devices, which is especially important in the industries of continuous production, such as oil, gas and petrochemicals. Existential philosophy of repair methods such as maintenance and preventive maintenance and maintenance and predictive repairs is to provide solutions to reduce repair costs and thus increase the efficiency of production units [12].
Maintenance and repairs are usually performed either at specified and pre-determined times, or whenever a failure occurs, depending on the type of failure. However, it also reduces the availability and increases the cost of repairs. Sometimes preventive repairs are performed on the equipment while the equipment is working well and does not need to be repaired or stopped. Accessibility, maintenance and safety of systems, traditional maintenance strategies are losing their effectiveness and are becoming obsolete [14].
But because maintenance is pre-planned for a certain period of time, it should be done according to your routine. An appropriate maintenance policy states that repairs should be performed when needed [12].
Industrial plants should no longer assume that equipment failures occur only after a fixed period of time in operation. They need to develop online and predictive maintenance strategies so that they can imagine that any breakdown could happen at any time. The onset of equipment failure may manifest itself in data generated by various methods. The equipment shows signs and indications that the equipment must be repaired, replaced or abandoned in order to continue working [13]. The conditional maintenance hypothesis is a regular observation for the actual condition of the equipment based on their important , salient features and the performance efficiency of the process systems, ensuring that the distance between repairs is maximized, reducing the cost of unplanned repairs due to machine failure and improve comprehensive access to the performance of industrial units. One of the most important and cost-effective maintenance techniques is condition-based maintenance (CBM) [19].
The use of predictive maintenance or condition-based maintenance can lead to major improvements in maintenance costs, reduce unplanned machine breakdowns, reduce downtime due to equipment repairs, and improve spare parts inventory. Be [13]. Predictive maintenance is one of the maintenance and repair strategies based on which a number of equipment parameters are measured at regular intervals or continuously, and based on this data, decisions are made to repair and replace parts and equipment.
In recent years, artificial neural networks have been successfully used in pattern recognition and troubleshooting tasks. One of the main advantages of using artificial neural networks is the ability to detect patterns that are inaccurate. This research seeks to implement MPL networks in predictive maintenance.
Discussion and Conclusion: In this article, with the help of data mining methods, a policy on failure prediction to design and develop turbine equipment was designed and developed. The use of experimental, manual and physical methods could not significantly help the maintenance team and provide timely warnings and accurate predictions of equipment failure in the future. Therefore, it was decided to design a model with the help of data-based methods and data mining methods in order to be able to accurately and with a high probability of accurate prediction of future failure. It is clear that predicting a breakdown will help the maintenance team in advance and prepare them in advance for the intended operation to rectify the breakdown. Plans for repairs, prepares workshops and equipment, and orders parts and goods if needed.
This can have a huge impact on reducing maintenance costs. Prevents unexpected repairs that cause unplanned shutdown of equipment and factory and reduces the negative impact on the production process and productivity.
For this purpose, by using continuous and scheduled observation and equipment status and conditions, as well as by examining failure conditions with the help of previous data, an accurate and quality data set was obtained, then the relationship between status data and failure data with the help PCA data analysis method was discovered to use related data sets as input for model design. In the next step, with the help of MPL neural network data mining method, the model designed to predict the occurrence of failure at different time intervals and with the aim of determining the output label between the three clusters A, B, C was tested and then information, statistics. The output results were presented accurately, completely and separately in the form of graphs, tables and various indicators to determine the measurement accuracy of the method. It was observed that the MLP neural network method with high accuracy has the ability to predict failure in different time periods has in the future.

کلیدواژه‌ها [English]

  • Maintenance & Repairs
  • Conditional Predictive Repairs
  • MPL Neural Network Technique
  • prediction
  1. Baptista, M., Sankararaman, S., De Medeiros, I.P., Nascimento Jr, C., Prendinger, H., & Henriques, E.M. (2018). Forecasting fault events for predictive maintenance using data-driven techniques and ARMA modeling, Computers & Industrial Engineering, 115, 41-53.
  2. Biswal, S., & Sabareesh, G. (2015). Design and development of a wind turbine test rig for condition monitoring studies, Proceeding of International Conference on Industrial Instrumentation and Control (ICIC), IEEE, 891-896.
  3. Chang, C.-C., & Lin, C.-J. (2004). LIBSVM: a library for support vector machines.
  4. Felix Salfner, M. L., & Mirosla, W. (2010). A survey of online failure prediction methods, CM Computing Surveys, 42, 42.
  5. Garga, A. K., & Byington, C. S. (2011). Data fusion for developing predictive diagnostics for electromechanical systems. Handbook of. Multisensor Data Fusion, 10, 701 –737.
  6. Giuseppe Curcur, G., & Alberto, L. (2010). A predictive maintenance policy with imperfect monitoring, Reliability EngineeringandSystemSafety, 95, 9.
  7. Ignat, S.S., & Stoian, I. (2012). Support/or condition based maintenance operating equipment performances monitoring, Automation Quality and Testing Robotics (AQTR), 2012, IEEE International Conference on, 6.
  8. Jardine, A.K., Lin, D., & Banjevic, D. (2006). A review on machinery diagnostics and prognostics implementing condition-based maintenance, Mechanical systems and signal processing, 20, 1483-1510.
  9. Jezzini, A., Ayache, M., Elkhansa, L., Makki, B., & Zein, M. (2013). Effects of predictive maintenance (PdM), Proactive maintenace (PoM) & Preventive maintenance (PM) on minimizing the faults in medical instruments, Proceeding of. 2013 2nd International Conference on Advances in Biomedical Engineering, IEEE, 53-56.
  10. Katakey, R. (2016) Energy giant shell says oil demand could peak in just five years, Bloomberg, http://www.bloomberg.com/news/ articles/2016-11-02/europe-s-biggest-oil-company-thinksdemand-may-peak-in-5-years
  11. Khoshairat, M., Khashei, M., & Molaverdi, N. (2018). A review of machine learning models in the field of predictive maintenance and repairs, 16th International Conference on Industrial Engineering, Tehran
  12. Mark Last, A. S., & Halasya Siva, S. (2010). Predictive maintenance with multi-target classification models, Intelligent information and database systems, 10.
  13. Mustakerov, I., & Borissova, D. (2013). An intelligent approach to optimal predictive maintenance strategy defining, Innovations in Intelligent Systems and Applications (INISTA), 2013 IEEE International Symposium on, 10-22.
  14. Khanh T.P., & Kamal, M. (2019). A new dynamic predictive maintenance framework using deep learning for failure prognostics, Reliability Engineering & System Safety, 188, 251-262  
  15. Peres, R.S., Rocha, A.D., Leitao, P., & Barata, J. (2018). IDARTS–Towards intelligent data analysis and realtime supervision for industry 4.0, Computers in Industry, 101, 138-146.
  16. Shannon, K. (2012). Asset life cycle management optimizes performance. The American oil and gas reporters. November 2012 Exclusive Story. http://www.aogr.com/web-exclusives/exclusivestory/asset-life-cycle-management-optimizes-performance.
  17. Shin, J.-H., Jun, H.-B., & Kim, J.-G. (2018). Dynamic control of intelligent parking guidance using neural network predictive control, Computers & Industrial Engineering, 120, 15-30.
  18. Susto, G.A., McLoone, S., Pagano, D., Schirru, A., Pampuri, S., & Beghi, A. (2013). Prediction of integral type failures in semiconductor manufacturing through classification methods, Proceeding of. 2013 IEEE 18th Conference on Emerging Technologies & Factory Automation (ETFA), IEEE, 1-4.
  19. Yeo, B. (2010). Shale Gas versus CSG - Shale the New Flavour, May/June, Aspermount Limited, 22-25.