[Maciej Majchrzak]

I have a problem with the heat exchanger calculations. The cool side of the exchanger has herringbone (zigzag shaped) fins - very narrow slots, 1 by 4mm, separated by even thinner sheet metal.
Despite a very accurate grid with correct wall layers (30 layers, y+ less than 1), I get pressure drop values several times higher than those measured in the experiment - (2500 Pa vs 500 Pa). The turbulence model used is kOmegaSST, the flow is velocity driven. The discrepancy is similar over a wide range of velocities.
I tried:
-various solver settings, from the most stable to the most accurate
-tuning of turbulence model coefficients
-transition to rhoPimpleFoam
-rhoPimpleFoam with LES turbulence model
-the transonic option
-using wall functions at a grid with y+ about 50
-different methods of determining wall distance,
obtaining only small changes in pressure drop.

Even more perplexing is that on the hot side of the exchanger the same settings give results almost identical to the measurements from the experiment - the difference being the completely different geometry, the short, straight ribs located further apart. Likewise, thermal analysis of the exchanger using chtMultiRegionSimpleFoam with a porous medium yielded results that almost perfectly coincided with the measurements.
The pressure increase is uniform along the length of the channel, the calculations are stable and convergent. Is there any reason why OpenFOAM could overestimate the pressure in a narrow channel? The surprising convergence of the result obtained by different methods and by two engineers even led me and my client to suspect measurement errors in the experiment.

I am attaching zip files with boundary conditions and solver settings.
I would appreciate any insights,
MM