Scholarlink

<p>Mineral precipitation is a challenging subject in geothermal heat exchangers, directly impacting the heat transfer efficiency of vaporizers and consequently, the overall performance of geothermal power plants. This study investigates the characterization of a flocculation problem within a geothermal brine vaporizer utilized in a combined geothermal power plant located in Alaşehir, Turkey. The investigation identified fine particle migration originating from high production rates of geothermal wells and silicate mineralization as the primary contributors to flocculation within the shell-and-tube heat exchanger. To mitigate this issue, a two-pronged approach was implemented. Firstly, screen filters were employed to eliminate fine particles from the brine. Secondly, an Artificial Neural Network (ANN) was utilized to optimize flow rate through the vaporizer. The ANN model was trained on data encompassing flow rate, pressure drop, power generation and time series data across the vaporizer. The results demonstrated the effectiveness of ANN in optimizing heat transfer from the brine vaporizer, enabling estimation of power production from the heat exchanger, and facilitating the development of a predictive maintenance program for cleaning processes. The hydraulic resistance calculations were based on the relationship between pressure drop and the square of the flow rate (ΔP∼Q²). The resulting fouling/clogging indices exhibited logistic-growth band behavior due to operational variations and changing formation-particle transport conditions during geothermal production. A techno-economic analysis was performed by considering ANN predictions, maintenance downtime during cleaning operations, and operational costs associated with vaporizer performance degradation. The results indicated that an operating flow rate of approximately 1600&#xa0;m³/h was optimal for maximizing net power generation while maintaining acceptable cleaning frequency and operational sustainability under manpower and maintenance limitations. The developed ANN framework was additionally compared with benchmark machine-learning algorithms commonly used for nonlinear industrial prediction problems, including XGBoost and Random Forest. The comparison demonstrated that the ANN model produced higher R² values and better agreement with field observations, indicating a superior capability to capture the coupled hydraulic and thermal behavior of the geothermal brine vaporizer system under fouling conditions. A sensitivity analysis of the ANN’s input parameters shows that flow rate has the greatest impact on power generation compared with pressure drop across the vaporizer and time.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Controlling mineral precipitation in geothermal power plant brine vaporizer through screening control and artificial neural network

  • 期刊论文

  • 发表日期：

    2026年6月24日

查看全文
  • Hakkı Aydın

摘要

Mineral precipitation is a challenging subject in geothermal heat exchangers, directly impacting the heat transfer efficiency of vaporizers and consequently, the overall performance of geothermal power plants. This study investigates the characterization of a flocculation problem within a geothermal brine vaporizer utilized in a combined geothermal power plant located in Alaşehir, Turkey. The investigation identified fine particle migration originating from high production rates of geothermal wells and silicate mineralization as the primary contributors to flocculation within the shell-and-tube heat exchanger. To mitigate this issue, a two-pronged approach was implemented. Firstly, screen filters were employed to eliminate fine particles from the brine. Secondly, an Artificial Neural Network (ANN) was utilized to optimize flow rate through the vaporizer. The ANN model was trained on data encompassing flow rate, pressure drop, power generation and time series data across the vaporizer. The results demonstrated the effectiveness of ANN in optimizing heat transfer from the brine vaporizer, enabling estimation of power production from the heat exchanger, and facilitating the development of a predictive maintenance program for cleaning processes. The hydraulic resistance calculations were based on the relationship between pressure drop and the square of the flow rate (ΔP∼Q²). The resulting fouling/clogging indices exhibited logistic-growth band behavior due to operational variations and changing formation-particle transport conditions during geothermal production. A techno-economic analysis was performed by considering ANN predictions, maintenance downtime during cleaning operations, and operational costs associated with vaporizer performance degradation. The results indicated that an operating flow rate of approximately 1600 m³/h was optimal for maximizing net power generation while maintaining acceptable cleaning frequency and operational sustainability under manpower and maintenance limitations. The developed ANN framework was additionally compared with benchmark machine-learning algorithms commonly used for nonlinear industrial prediction problems, including XGBoost and Random Forest. The comparison demonstrated that the ANN model produced higher R² values and better agreement with field observations, indicating a superior capability to capture the coupled hydraulic and thermal behavior of the geothermal brine vaporizer system under fouling conditions. A sensitivity analysis of the ANN’s input parameters shows that flow rate has the greatest impact on power generation compared with pressure drop across the vaporizer and time.