Hossam AbdelMeguid, Mohamed Kamel Elshaarawy, Mahmoud Bady, H. A. Dahab, M. A. Elazab
{"title":"Thermohydraulic Pareto optimization of a helical-flow shell-and-tube humidifier using surrogate models and applicability domain analysis","authors":"Hossam AbdelMeguid, Mohamed Kamel Elshaarawy, Mahmoud Bady, H. A. Dahab, M. A. Elazab","doi":"10.1007/s10973-026-16331-4","DOIUrl":null,"url":null,"abstract":"<div><p>Helical flow can increase air–water contact in a humidifier, but the longer and more restrictive flow path also raises hydraulic resistance. This study quantifies that trade-off for a shell-and-tube humidifier described by a one-dimensional steady heat, mass, and pressure balance model. A 5 × 10 × 10 full-factorial grid generated 500 geometries by varying helix pitch (<i>P</i>), humidifier length (<i>L</i>), and radius ratio (<i>RR</i>). Because the governing model assumes constant dry air density and constant transport properties, surrogate fitting was limited to the 456 cases satisfying Δ<i>P</i>/<i>P</i><sub>in</sub> ≤ 0.10; the 44 outside-envelope cases are reported separately. The analyzed cases span <i>RH</i> = 15.23–99.57% and Δ<i>P</i> = 2–10,088 Pa. Gaussian process regression (GPR), support vector regression, kernel ridge regression, and multilayer perceptron models were compared using a common 70/30 hold-out split, five-fold cross-validation, leave-one-level-out tests, and learning curves. GPR gave the lowest hold-out error for both responses: <i>R</i><sup>2</sup> = 0.9996, RMSE = 0.601 percentage points, and MAE = 0.216 percentage points for outlet relative humidity (<i>RH</i>); and <i>R</i><sup>2</sup> = 0.9997, RMSE = 22.8 Pa, and MAE = 5.64 Pa for pressure drop (Δ<i>P</i>). Five-fold GPR RMSE values were 0.659 percentage points for <i>RH</i> and 24.3 Pa for Δ<i>P</i>. Split conformal intervals gave empirical coverages of 97.1% and 98.5%, respectively. NSGA-II and MOPSO each produced 120 Pareto solutions inside the stated applicability domain. NSGA-II gave hypervolume = 0.8980, whereas MOPSO gave the lower IGD = 0.001609. The low-pressure knee lies near <i>P</i> ≈ 0.48 m, <i>L</i> ≈ 10 m, and <i>RR</i> ≈ 1.83–1.84; the nearest sampled case gives <i>RH</i> = 75.66% at Δ<i>P</i> = 35 Pa. An intermediate 90–95% <i>RH</i> region occurs around <i>L</i> ≈ 10 m, <i>RR</i> ≈ 1.46–1.56, and <i>P</i> ≈ 0.44–0.48 m, with Δ<i>P</i> of about 110–220 Pa. For near-saturation follow-up, <i>L</i> = 10 m, <i>RR</i> = 1.30, and <i>P</i> = 0.40–0.50 m give <i>RH</i> = 99.30–99.36% at Δ<i>P</i> = 874–953 Pa. These three regions define quantitatively supported design targets for subsequent high-fidelity CFD analysis and experimental assessment, providing a focused basis for evaluating thermohydraulic performance across low-pressure, intermediate, and near-saturation operating regimes.</p></div>","PeriodicalId":678,"journal":{"name":"Journal of Thermal Analysis and Calorimetry","volume":"151 22","pages":"19661 - 19682"},"PeriodicalIF":4.1000,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Thermal Analysis and Calorimetry","FirstCategoryId":"5","ListUrlMain":"https://link.springer.com/article/10.1007/s10973-026-16331-4","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"CHEMISTRY, ANALYTICAL","Score":null,"Total":0}
引用次数: 0
Abstract
Helical flow can increase air–water contact in a humidifier, but the longer and more restrictive flow path also raises hydraulic resistance. This study quantifies that trade-off for a shell-and-tube humidifier described by a one-dimensional steady heat, mass, and pressure balance model. A 5 × 10 × 10 full-factorial grid generated 500 geometries by varying helix pitch (P), humidifier length (L), and radius ratio (RR). Because the governing model assumes constant dry air density and constant transport properties, surrogate fitting was limited to the 456 cases satisfying ΔP/Pin ≤ 0.10; the 44 outside-envelope cases are reported separately. The analyzed cases span RH = 15.23–99.57% and ΔP = 2–10,088 Pa. Gaussian process regression (GPR), support vector regression, kernel ridge regression, and multilayer perceptron models were compared using a common 70/30 hold-out split, five-fold cross-validation, leave-one-level-out tests, and learning curves. GPR gave the lowest hold-out error for both responses: R2 = 0.9996, RMSE = 0.601 percentage points, and MAE = 0.216 percentage points for outlet relative humidity (RH); and R2 = 0.9997, RMSE = 22.8 Pa, and MAE = 5.64 Pa for pressure drop (ΔP). Five-fold GPR RMSE values were 0.659 percentage points for RH and 24.3 Pa for ΔP. Split conformal intervals gave empirical coverages of 97.1% and 98.5%, respectively. NSGA-II and MOPSO each produced 120 Pareto solutions inside the stated applicability domain. NSGA-II gave hypervolume = 0.8980, whereas MOPSO gave the lower IGD = 0.001609. The low-pressure knee lies near P ≈ 0.48 m, L ≈ 10 m, and RR ≈ 1.83–1.84; the nearest sampled case gives RH = 75.66% at ΔP = 35 Pa. An intermediate 90–95% RH region occurs around L ≈ 10 m, RR ≈ 1.46–1.56, and P ≈ 0.44–0.48 m, with ΔP of about 110–220 Pa. For near-saturation follow-up, L = 10 m, RR = 1.30, and P = 0.40–0.50 m give RH = 99.30–99.36% at ΔP = 874–953 Pa. These three regions define quantitatively supported design targets for subsequent high-fidelity CFD analysis and experimental assessment, providing a focused basis for evaluating thermohydraulic performance across low-pressure, intermediate, and near-saturation operating regimes.
期刊介绍:
Journal of Thermal Analysis and Calorimetry is a fully peer reviewed journal publishing high quality papers covering all aspects of thermal analysis, calorimetry, and experimental thermodynamics. The journal publishes regular and special issues in twelve issues every year. The following types of papers are published: Original Research Papers, Short Communications, Reviews, Modern Instruments, Events and Book reviews.
The subjects covered are: thermogravimetry, derivative thermogravimetry, differential thermal analysis, thermodilatometry, differential scanning calorimetry of all types, non-scanning calorimetry of all types, thermometry, evolved gas analysis, thermomechanical analysis, emanation thermal analysis, thermal conductivity, multiple techniques, and miscellaneous thermal methods (including the combination of the thermal method with various instrumental techniques), theory and instrumentation for thermal analysis and calorimetry.