Augmenting insights into heat transfer performance of direct-contact evaporator: An interpretable data-driven hybrid strategy
The volumetric heat transfer coefficient (VHTC) of direct-contact evaporator (DCE) in an organic Rankine cycle (ORC) system was one of the key indicators to reflect the heat transfer efficiency and energy utilization. This research proposed an interpretable data-driven hybrid strategy for evaluating...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-10-01
|
| Series: | Case Studies in Thermal Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2214157X25011402 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The volumetric heat transfer coefficient (VHTC) of direct-contact evaporator (DCE) in an organic Rankine cycle (ORC) system was one of the key indicators to reflect the heat transfer efficiency and energy utilization. This research proposed an interpretable data-driven hybrid strategy for evaluating heat transfer enhancement performance in the low-temperature waste heat recovery systems. Firstly, the collected experimental operating conditions were preprocessed to obtain a series of highly reliable datasets. Subsequently, the augmented Dickey-Fuller test was used to reveal that whether the steam flow data has non-linear and non-stationary characteristics or not. Then, the steam flow rate was decomposed into a series of intrinsic mode functions by intrinsic computing expressive empirical mode decomposition with adaptive noise (ICEEMDAN) and the intrinsic mode functions were divided into high-complexity and low-complexity components by fuzzy entropy (FE). Next, the reconstructed values of the low-complexity steam flow subsequence were substituted into the least squares support vector machine (LSSVM) for training and testing, which obtains the forecasted value of VHTC. Compared to eleven comparative model, ICEEMDAN-FE-LSSVM-VHTC owns excellent accuracy since average absolute error decreases by 21.4 %–70.9 %, mean square error decreases by 62.5 %–94.5 %, root mean square error decreases by 16.7 %–67.9 %, and coefficient of determination increases by 1.9 %–58.0 %. In addition, the Shapley additive explanations values offer both local and global explanations of the proposed hybrid forecasting model. Overall, this study presents an in-depth analysis of complex heat transfer datasets to provide valuable insights into evaporator design to ensure efficient and consistent performance in real-world operation. Hence, the research on the forecasting of VHTC inside DCE of ORC can help promote the efficient use of energy and the cause of environmental protection. |
|---|---|
| ISSN: | 2214-157X |