Experimental and Numerical Analysis of Tubular Heat Exchanger Sensitivity to Inlet Distortion

Paper in proceeding, 2026

This work evaluates the performance and sensitivity to inlet distortion in a radially flow tubular heat exchanger (HEX) operating as part of a waste heat recovery (WHR) system for aero-engine applications. This is part of the SWITCH project (Sustainable Water-Injecting Turbofan Comprising Hybrid-Electrics), within the EU Clean Aviation program related to future ultra-efficient short–to–medium–range aircraft for sustainable carbon-neutral solutions. A key metric in evaluating performance of gas turbines is the ideal thermal efficiency, which theoretically is above 60%, when operated at high pressure ratios, however all practical efficiencies remain below 40% due to thermal and mechanical losses. To further improve efficiency, new approaches are required, which e.g. utilize the otherwise wasted exhaust heat using a HEX located downstream the low-pressure turbine.  
 
In a former work (ASME Turbo Expo 2025, GT2025 153194) a modular 2.3m long, 30° sector WHR concentric tubular HEX was evaluated numerically and experimentally with a high degree of agreement, although the highly unstable and complex flow. A significant challenge was the variation of flow intothe heat exchanger. The HEX was sub-divided using internal baffles into 10 segments. At lower tip of the baffles, guide-vanes, scoops, were attached for shaving of a portion (ideally 10%) of the flow and turn the axially flowing exhaust gases into the HEX. It proved to be a significant cchallenge to disentangle these unstable large scale fluctuations from the guide-vanes from local diffusion and the ability of turning the flow into the HEX. Furthermore these fluctuations resulted in transient flow maldistribution with measurement challenges and issues to correctly assess and interpret heat transfer and pressure losses at specific locations.
 
To isolate these effects, two experimental rigs representing a single sector (a tenth of the earlier study) with the same topology, i.e. multiple staggered rows, were studied numerically and experimentally. The “idealized” rig provided undisturbed inflow to assess performance degradation of radial expanding vs orthogonal arranged HEX. The second rig, “turning”, re-used the HEX module, but - similar to the earlier study - also including upstream vanes, to shave off a portion of the flow and re-direct it into the HEX. The individual performance evaluation and direct comparison of these two variants enables both numerical verification and quantification of knock-down factors due to diffusion and turning in a controlled environment. In both rigs, heat transfer rate and total pressure loss were measured. Complementary CFD campaigns were conducted using various numerical settings, turbulence models, and computational meshes to assess prediction robustness and model sensitivity.
 
Results show that standard heat transfer and pressure loss correlations for orthogonally arranged tube matrices can also be used for radial tube matrices, provided minor modifications are applied to the correlations to account for flow deceleration. Numerical results matches the experimental data to a high degree with the k epsilon RNG turbulence model providing the most accurate predictions of heat transfer and pressure drop at ideal conditions. 
 
For the turning case, inlet distortions marginally penalizes overall heat transfer but significantly affected the thermal load distribution and pressure losses. As expected, accurate prediction necessitate detailed CFD of the vanes ability to turn the flow in order to capture the associated downstream non uniformities within the tube bank. 
 
This report provides the following outcome: i) benchmark case for radially expanding tubular HEX in ideal and highly disturbed inflow, ii) quantified knock down factors for both heat transfer and pressure loss due to altered inlet flow conditions; iii) a recommendation of k epsilon RNG as the preferred low fidelity CFD model for averaged heat transfer prediction; and iv) modified heat transfer and pressure loss correlations for non ideal, radially expanding tube matrix performance suitable for preliminary design and integration studies. These results decouple diffusion and turning contributions, remove ambiguity from flow maldistribution effects, and provide actionable guidance for sizing and integrating WHR HEX units in engines subject to non ideal inlet environments.