Reactor geometry as a high-dimensional design variable: Representation, evaluation, and closed-loop design
Year: 2026
Type: article
Abstract: Reactor performance emerges from the coupled effects of kinetics, catalyst properties, operating conditions, transport phenomena, and reactor geometry. Among these factors, geometry is the central spatial design variable because the arrangement of flow paths, heat transfer surfaces, and catalytic interfaces governs how these effects are expressed across the reactor domain. However, design methodologies have long fixed geometry early and optimised it over a narrow parametric range. This contribution argues that an important limitation of current reactor design practice is structural: when geometry is confined to a small set of parameters, evaluated sequentially, and disconnected from manufacturing, the accessible design space is small by construction. The problem is therefore reformulated by treating reactor geometry as a high-dimensional design variable whose representation, evaluation, and physical realisation form three interdependent axes of a geometry-centric framework. Within this formulation, approaches usually treated separately emerge as instances of a common geometry-performance mapping, while additive manufacturing enters directly as a constraint that defines which geometries are physically realisable. The framework is operationalised through explicit integration criteria and applied diagnostically to published reactor design workflows. The analysis shows that the components required for a fully closed design loop are individually operational but have not yet been coupled within a single iterative system in which measurements from fabricated geometries update the geometry generation model itself. The remaining challenges (high-dimensional representation, evaluation cost, data-efficient learning, realised geometry characterisation, and manufacturing loop closure) define a research agenda toward autonomous reactor design capable of exploring architectures beyond conventional parametric families.
Source: Chemical Engineering Journal
Authors: Bogdan Dorneanu, Harvey Arellano-Garcia
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Keywords: reactor geometry, geometric representation
Open Access status: hybrid
Estimated APC paid: $4,830
OpenAlex ID: W7235029919
