Machine learning-assisted design automation workflow of non-uniform flow distributors for multipass crossflow device

Aug 5, 2026

Multipass crossflow devices play a key role in process engineering across diverse applications, and additive manufacturing enables improved device efficiency through non-uniform flow distribution that can enhance heat transfer uniformity and reduce temperature gradients. This study tackles non-uniform distribution control by proposing a design that locally varies channel diameters at predefined positions to reach user-defined flow ratio targets. A machine learning-assisted design automation workflow integrates algorithmic modeling and CFD simulations to handle design complexity, with validation using experiments on an incompressible gas flow distributor supporting the reported findings.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Introduction

Multipass crossflow devices and the role of flow distribution

Consecutive manifolds (U-type and Z-type) and their limitations

Bifurcation geometries and additive manufacturing potential

Results in Engineering 27 (2025) 106931 Contents lists available at ScienceDirect Results in Engineering journal [homepage:](homepage: www.sciencedirect.com/journal/results-in-engineering)[ www.sciencedirect.com/journal/results-in-engineering](homepage: www.sciencedirect.com/journal/results-in-engineering) | Research paper
Machine learning-assisted design automation workﬂow of non-uniform ﬂow distributors for multipass crossﬂow device
Mertcan Kaya a, ,∗ , Julian Ferchow b, , Mirko Meboldt c, Christoph Klahn a,d,
| | | | | --- | --- | --- | --- | | a Karlsruhe Institute of Technology (KIT), Institute for Micro Process Engineering (IMVT), 76344 Eggenstein-Leopoldshafen, Germany b Inspire AG (ipd|z), 8092, Zürich, Switzerland
c ETH Zurich (pd|z), 8092, Zürich, Switzerland
d Karlsruhe Institute of Technology (KIT), Institute of Mechanical Process Engineering and Mechanics (MVM), 76131 Karlsruhe, Germany | | | | | A R T I C L E I N F O | | A B S T R A C T | | | Keywords:
ML driven design Design automation Chemical engineering Additive manufacturing Flow distribution Bifurcation design | | Multipass crossﬂow devices are essential in the process engineering industry due to their broad range of applications. The growing use of additive manufacturing techniques oﬀers opportunities to improve the eﬃciency of these devices by enabling non-uniform ﬂow distribution, which can enhance heat transfer uniformity to avoid temperature gradients across individual channels in chemical engineering applications such as heat exchangersand packed bed reactors. This paper addresses the challenge of controlling non-uniform ﬂow distribution by developing a novel design that locally varies channel diameters at predeﬁned positions within the distributor, to achieve user-deﬁned target ﬂow distribution ratios. The proposed design combines machine learning-assisted design automation workﬂow with algorithmic modeling and computational ﬂuid dynamics (CFD) simulations to manage the complexity of non-uniform ﬂow distribution eﬀectively. The results are further validated through an experimental study of an incompressible gas ﬂow distributor, which supports the ﬁndings presented in this work. | | 1. Introduction Multipass crossﬂow devices, such as heat exchangers, are widely employed in process engineering for heating and cooling applications. The selection of a suitable ﬂow arrangement depends on multiple factors, including thermal eﬃciency, ﬂuid paths, space constraints, temperature levels, and production costs [1–3]. Owing to their compact design, multipass crossﬂow heat exchangers oﬀer signiﬁcant advantages in industrial settings. However, their optimal performance depends on maintaining a uniform heat transfer coeﬃcient across all channels, which is strongly inﬂuenced by the inlet ﬂow distribution [4–7]. Even slight deviations in channel ﬂow can create local variations in heat transfer coeﬃcients, particularly in microchannel exchangers, thereby degrading thermal performance compromising the overall thermal performance of the system [5]. Notably, Guo et al. demonstrated that, under speciﬁc conditions, controlled non-uniform ﬂow can enhance heat transfer [4]. This strategy enhances overall heat transfer while promoting uniform heat transfer across all channels through a controlled, non-uniform ﬂow distribution [8]. Such enhancements are beneﬁcial not only for multipass crossﬂow heat exchangers but also for crossﬂow reactors used in exothermic processes. Isothermal conditions in these reactors en- * Corresponding author. [E-mail address:](E-mail address: mertcan.kaya@kit.edu)[ mertcan.kaya@kit.edu](E-mail address: mertcan.kaya@kit.edu) (M. Kaya). able higher conversion rates, emphasizing the importance of isothermal crossﬂow devices in these applications [9]. These ﬁndings highlight the broad applicability of controlled non-uniform ﬂow distribution in devices ranging from heat exchangers to chemical reactors and separa

FAQs

Why is controlling non-uniform flow distribution important for multipass crossflow devices?

Flow distribution strongly affects heat transfer coefficients across channels. Small maldistribution can create local thermal variations, degrading overall thermal performance.

How does the proposed design achieve user-defined target flow ratios?

It locally varies channel diameters at predefined positions inside the distributor to match specified flow distribution ratios.

What methods are used in the machine learning-assisted design automation workflow?

The workflow combines algorithmic modeling with computational fluid dynamics (CFD) simulations, and it is validated with experiments using an incompressible gas flow distributor.

Aug 5, 2026

Upload by Valentina