A reduced order model for spatially resolved gas transfer prediction in full scale oxygenators

Journal of Membrane Science, vol. 742, pp. 125096

Abstract

Severe lung diseases can be clinically treated with extracorporeal blood oxygenation (ECMO), where blood oxygenators exchange gases across thousands of hollow fiber membranes. However, their large surface area increases blood trauma and the risk of device failure. Understanding and improving gas transfer efficiency through computational models can help to reduce the required membrane surface area for the design of smaller and safer oxygenators. However, existing models either oversimplify local flow and gas exchange or rely on experimental calibration. To address this gap, we developed a reduced order model (ROM) that predicts oxygen transfer in fiber bundles modeled as porous media, without requiring a priori experiments. The model provides locally resolved oxygen source terms that account for fiber configuration, Reynolds number, and oxygen concentration, and were derived from and fitted to data from in-vitro validated fiber-scale computational fluid dynamics (CFD) simulations (RMSE < 0.01). The ROM predicted oxygen transfer of the validated simulations with high accuracy (R-squared = 0.995) and an overestimation of 6 %. When applied to a lab-scale oxygenator, the ROM accurately predicted total oxygen transfer with a relative error of 2.3 % compared to experimental results and provided spatially resolved oxygen concentration distributions within the fiber bundle. We provide source term functions and coefficients for the four most common fiber configurations and a CFD implementation in a full-scale oxygenator for further use. • First time gas transfer simulation of full fiber bundle • Numerical results predict blood experiments • Oxygen source terms depend on fiber configuration, Reynolds number, concentration difference • Numerical model is ready to be applied on commercial oxygenators without a priori experiments

Authors 6

1. Jannis Focke Aachen Cardiovascular Engineering

   RWTH Aachen University

   Affiliation as printed

   Cardiovascular Engineering, Applied Medical Engineering, RWTH Aachen University, Aachen, Germany

2. Kai P. Barbian Aachen Cardiovascular Engineering

   RWTH Aachen University

   Affiliation as printed

   Cardiovascular Engineering, Applied Medical Engineering, RWTH Aachen University, Aachen, Germany

3. Paul-Luca Bonke Aachen Cardiovascular Engineering

   RWTH Aachen University

   Affiliation as printed

   Cardiovascular Engineering, Applied Medical Engineering, RWTH Aachen University, Aachen, Germany

4. Ulrich Steinseifer Aachen Cardiovascular Engineering

   RWTH Aachen University

   Affiliation as printed

   Cardiovascular Engineering, Applied Medical Engineering, RWTH Aachen University, Aachen, Germany

5. Jutta Arens

   University of Twente

   Affiliation as printed

   Engineering Organ Support Technologies Group, Department of Biomechanical Engineering, University of Twente, Enschede, the Netherlands

6. Michael Neidlin corresponding Aachen Cardiovascular Engineering

   RWTH Aachen University

   Affiliation as printed

   Cardiovascular Engineering, Applied Medical Engineering, RWTH Aachen University, Aachen, Germany