2 Replies Latest reply on Jul 19, 2018 9:36 AM by Bill McEachern

    High Reynolds number modeling

    John Bankert

      Hello,

       

      I am doing validation modeling of drag in immersed flow of air over a streamlined body, however, I am interested in high Reynolds number flow (i.e. 20e6).  As I understand it, the turbulence model flow sim uses is validated either to 6e6 (Bill McEachern's comment in this thread https://forum.solidworks.com/thread/20131) or 8e6 (Bill McEachern's comment in this thread https://forum.solidworks.com/thread/57334).  Further, in the validation white paper, I only saw data up to a Reynolds number of 1e6 (https://www.solidworks.com/sw/docs/Flow_Validation_Methodology-Whitepaper.pdf ).

       

      My questions are the following:

       

      1) Is it completely useless to model drag for flows in the 20e6 Re realm?  Alternatively, if exact answers are not valid at that Re value, is there any validity in comparing two designs relative to each other, to indicate which design is "best."

      2) From a physics standpoint, for a streamlined body in air, would design decisions made on a body modeled at a lower Re like Re=8e6 (or whatever the limit is) give any useful information on the behavior of that body at higher Re values? I'm imagining that the boundary layer behavior is different at Re=8e6 compared to Re=20e6, and so the behavior at the lower Reynolds number would be non-representative at higher Re.  Am I better off modeling it at 20e6, or 8e6, if I decide I'm interested more in comparing designs than absolute results?

      3) The items I am modeling do show boundary layer separation and this appears to be important to the drag; would modeling in the time domain provide improved results at Re=20e6 vs a steady state model @ Re=20e6 (see comments in this thread https://my.solidworks.com/reader/forumthreads/187084/solidworks-flow-simulation-accuracy)? Or is this another case of "there's no use in modeling at this Reynolds number, no matter what you do?"  It also looks like in the validation white paper (re-linked: https://www.solidworks.com/sw/docs/Flow_Validation_Methodology-Whitepaper.pdf  ), Figure 17 "Drag over a cylinder" uses time-averaged results, implying time-domain study. 

       

      I am less familiar with fluid mechanics than I would like, but given what I've read about flow sim, I am worried about my results if I don't understand the numerical effects of modeling outside of the validated range.

       

      Thanks all!  

        • Re: High Reynolds number modeling
          Bill McEachern

          I would forego speculation and find 2 or more different cases at the target R'number and then run them in flow and see what you can make of it. Typically in smooth attached flows, in my experience at any rate, Flow Simulation will over predict drag. You are probably modeling something like and airship or a submarine and there maybe some public domain or historical data for an airship around or or perhaps some other body. My guess is that trends should be predictable but you  would need to confirm it. The limitation of Flow Sim for high R'number flows came up with an airship development back in and around the early 2000's or so as I recall.

          • Re: High Reynolds number modeling
            Bill McEachern

            on the time average thing: you can run a transient study and let the results settle out. or you could just look judiciously at the steady results. Instead of looking at the goal value at the end (last iteration) take a look at the averaged value which would be the average value of the goal (drag) over an interval which might be a better metric if significant unsteadiness is present.