Predicting deep well pump performance with machine learning methods during hydraulic head changes

Area of Science:

• Engineering
• Computer Science
• Environmental Science

Background:

• Pumping plant efficiency is crucial for energy conservation and operational cost reduction.
• Accurate prediction of system efficiency under varying hydraulic conditions is essential for optimizing performance.
• Traditional methods for efficiency estimation may not capture complex system dynamics effectively.

Purpose of the Study:

• To employ machine learning techniques for estimating and predicting the system efficiency of a pumping plant.
• To evaluate the performance of different machine learning algorithms in modeling pumping plant efficiency.
• To identify the key parameters influencing system efficiency and their impact on prediction accuracy.

Main Methods:

• Utilized measured parameters including flow rate, outlet pressure, drawdown, and power.
• Implemented two approaches: Approach-I with additional parameters (hydraulic head, drawdown, flow, power, outlet pressure) and Approach-II with hydraulic head, outlet pressure, and power.
• Applied seven machine learning algorithms, including artificial neural networks, support vector machine regression, and lasso regression.

Main Results:

• A decrease in hydraulic head by 125 cm reduced pump system efficiency by 6.45%–13.8% at different flow rates.
• Artificial neural network, support vector machine regression, and lasso regression showed high performance in Approach-I (R² values up to 0.995).
• Artificial neural network demonstrated the best performance in both approaches, with R² values of 0.995 and 0.996.

Conclusions:

• Machine learning techniques, particularly artificial neural networks, are highly effective for predicting pumping plant system efficiency.
• The study demonstrates the potential of data-driven approaches for optimizing the operation and maintenance of pumping systems.
• Accurate efficiency prediction can lead to significant energy savings and improved operational management in pumping plants.