Developing a machine learning tool to predict discharge temperatures of gas compressor

  • Ngoc Trung Tran Bien Dong Petroleum Operating Company (Bien Dong POC)
  • Thanh Trung Nguyen Bien Dong Petroleum Operating Company (Bien Dong POC)
  • Duy Minh Nguyen Bien Dong Petroleum Operating Company (Bien Dong POC)
  • Quang Khoa Dao Bien Dong Petroleum Operating Company (Bien Dong POC)
  • Vu Tung Tran Bien Dong Petroleum Operating Company (Bien Dong POC)
  • Ky Son Hoang Bien Dong Petroleum Operating Company (Bien Dong POC)
Keywords: Machine learning, gas compressor, discharge temperature prediction


Gas compressors are important equipment on the central processing platform PQP-HT. After dehydration and ensuring the dew point temperature in accordance with inlet conditions and specifications of the Nam Con Son gas pipeline (NCSP), natural gas is transferred to a gas compression system consisting of 2 compressor lines. Optimizing operating conditions by reducing the inlet pressure of the natural gas processing system is normally used to extend production time of a gas well. However, alterations in inlet operating conditions will directly affect the gas compressor system, potentially causing the discharge temperature to exceed safe operating thresholds.

Commercial thermodynamic simulation software (such as Hysys, ProII) is typically employed to assess the effect of changing gas compressor operating conditions on the outlet temperature of each stage. This allows simulation and selection of optimal working conditions to ensure safety within the natural gas processing system. Nevertheless, the cost of licensing and maintaining commercial software is substantial. Nowadays, machine learning algorithms are proven to be able to predict operating parameters based on historical data. Many studies have been devoted to accurately predicting compressor performance to improve operational efficiency. Machine learning algorithms have the advantage of highly precise prediction results and the ability to operate continuously and re-train automatically upon any operational condition change. Therefore, they can be used as a viable alternative to commercial thermodynamic simulation software.


Johm M. Campbell, Larry L. Lilly, and Robert N. Maddox, Gas conditioning and processing - Volume 2: The equipment modules, 6th edition. Campbell Petroleum Series, 1984.

Maurice Stewart Ken Arnold, Surface production operations: Volume 2: Design of gas-handling systems and facilities, 2nd edition. Elsevier, 1999.

John M. Schultz, "The polytropic analysis of centrifugal compressors", Journal of Engineering for Power, Volume 84, Issue 1, pp. 69 - 82, 1962. DOI: 10.1115/1.3673381.

Chunzhi Wei, "Predicting discharge temperature of a centrifugal compressor in a realistic way based on ASME PTC 10", 2017. [Online]. Available: https:// -temperature-of-a-Centrifugal-a-Wei/ddecfafacbc9608b b1fd8ca34203418fd2f11382.

Yu Zhang, Miguel Martinez-Garcia, and Anthony Latimer, "Estimating gas turbine compressor discharge temperature using Bayesian neuro-fuzzy modelling", IEEE International Conference on Systems, Man, and Cybernetics (SMC), 5 - 8 October 2017. DOI: 10.1109/SMC.2017.8123194.

Jingzhou Fei, Ningbo Zhao, Yong Shi, Yongming Feng, and Zhongwei Wang, "Compressor performance prediction using a novel feed-forward neural network based on Gaussian kernel function", Advances in Mechanical Engineering, Volume 8, Issue 1, 2016. DOI: 10.1177/1687814016628396.

Hongsheng Jiang, Sujun Dong, Zheng Liu, Yue He, and Fengming Ai, "Performance prediction of the centrifugal compressor based on a limited number of sample data", Mathematical Problems in Engineering, 2019. DOI: 10.1155/2019/5954128.

Xavier Glorot and Yoshua Bengio, “Understanding the difficulty of training deep feedforward neural networks”, Proceedings of the 13th International Conference on Artificial Intelligence and Statistics (AISTATS), Sardinia, Italy, 13 - 15 May 2010.

C. Olah, "Understanding LSTM networks". [Online]. Available:

Sepp Hochreiter and Jürgen Schmidhuber, "Long short-term memory", Neural Computation, Volume 9, Issue 8, pp. 1735 - 1780, 1997. DOI: 10.1162/neco.1997.9.8.1735.

How to Cite
Tran, N. T., Nguyen, T. T., Nguyen, D. M., Dao, Q. K., Tran, V. T., & Hoang, K. S. (2024). Developing a machine learning tool to predict discharge temperatures of gas compressor. Petrovietnam Journal, (1), 67-77.

Most read articles by the same author(s)