Hybrid Thermodynamic and Machine Learning Workflow for Forecasting Turbine Efficiency at Kamojang 3 Geothermal Power Plant

DOI

10.2991/978-94-6463-944-5_28How to use a DOI?

Keywords

Geothermal Power Plant; Efficiency; Turbine; Forecasting; Work Flow; Machine Learning; Predictive Maintenance

Abstract

Reliable operation of geothermal power plants depends on stable turbine performance and early detection of degradation. This study builds a thermodynamics-informed, data-driven workflow in Orange to forecast hourly isentropic turbine efficiency from operational logs. The dataset comprises 1,487 hourly records covering the steam, condenser, cooling-water, and generator subsystems. Missing entries were handled by interpolation at the hourly cadence. Thermodynamic consistency was enforced with IAPWS-97 to compute inlet and exhaust states and to form the efficiency target. The observed efficiency ranges from 0.774 to 0.853, approximately 77.4 to 85.3 percent, with a stable trajectory over the window. Five learners were benchmarked on a chronological training split using five-fold cross validation, and Gradient Boosting was selected based on the highest mean R2. Evaluation on the held-out test period achieved R2 0.926, RMSE 0.003, MAE 0.002, and MAPE 0.200 percent, which are small relative to the series spread. Combining first-principles thermodynamics with supervised learning in a reproducible Orange pipeline yields accurate and explainable efficiency forecasts that support condition monitoring and timely maintenance planning in geothermal power plants.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Meigi Rama
AU  - Sihana Sihana
AU  - Dwi Joko Suroso
PY  - 2025
DA  - 2025/12/26
TI  - Hybrid Thermodynamic and Machine Learning Workflow for Forecasting Turbine Efficiency at Kamojang 3 Geothermal Power Plant
BT  - Proceedings of the International Conference on Sustainable Energy: Toward Energy Transition and Net-Zero Emission (ICOSE 2025)
PB  - Atlantis Press
SP  - 432
EP  - 449
SN  - 3005-155X
UR  - https://doi.org/10.2991/978-94-6463-944-5_28
DO  - 10.2991/978-94-6463-944-5_28
ID  - Rama2025
ER  -