LVDH wrote:I have a feeling that something in how it is handled is wrong.
I only quoted a part of Mantium's comment but I agree with most of his thought.
I am convinced that there is no a real "solver" issue (even if a finer mesh over the inlets would help) but a "physical model issue" (I am only talking about the cooling strage behaviour) related to the bc.
In my tests I realized that the imposed flow model works quite well when the local total pressure is positive. On the contrary, low pressure (local, it can happen to have a positive integral with local low regions) sometimes generates odd effects.
To avoid this problems I used a purely empirical approach: I submitted a car with the highest pressure differential I could.
I can propose two possible solutions:
1)
Short term (emergency solution): I would suppress the inlet boundary condition (imposed flow),
estimating the cooling efficiency only by computing the contribution of normal forces (measured in the same way of the forces on the "*wings" files). A minimum normal force (inlet resultant - outlet resultant, free total area with very minimum requirements, around 25000mm2 for each inlet/outlet surfaces, with the same internal template used now) should be decided and included into the rulebook. Metrologically speaking a force (N) is homogeneous to the integral of pressure (Pa*m2). This solution is much more stable (even if less "realistic" in the volume very near to the inlet).
2)
Long term: implementing the porous media or an equivalent simulation workaround (es. Explicit geometry bottleneck) in order to avoid bc on inlets.
Last point: it is more probable that cars with low pressure differential (less than 10-20 mPa) will have instability issues if compared to cars with higher pressure on the inlets.
