Discussion Forum

Posted: 7 years ago Oct 6, 2017, 2:14 a.m. EDT

Hi Hopefully you have progressed since you published this request, but if not I have some comments that might help: To do MULTIphysics problesm one nedd to start gradually, one physics at the time, this I believe you have done as your HT and SPF physics, taken independently are correct. But you have now to add the couplings, COMSOL cannot invent these you must first identify which one you want and then help COMSOl to set up the correct couplings: i.e. Your static HT case must be made more general, as the SPF includews fluid transport: add a Fluid HT node for the air domain to tell COMSOL to consider the motion of the air that is to come. (even in static analysis you may ahave a stready state velocity, hence transport phenomena. Check the equation tab of your physics to see what COMSOL adds for each one. Then in HT as we talk about transport of air, you must also define the boundaries where thermal energy flows through, the inlet and outlet SPF boundaries introduces material enthalphy and removes heat energy, add the corresponding Outflow and Inflow Heatflux (linked to the inflow ambient temperature you select) boundary nodes. Still one more to come for HT see below. On the SPF side the main issue is to ensure SPF sees the true temperature distributions calculated from the from HT, this can either be done by adapting each T field in all SPF nodes, but COMOSL can do itfor you in ine go: Right click the Multiphysics node and add some multiphysics sub-nodes already preapared with relevant couplings. here you have three choices relevant for your case (and a few other for going further into other details). These three overlap: you can either add "Flow coupling FC " alone i.e. getting the transport velocities of SPF automatically added to the relevant inputs of the velocity fields in HT. And/or add "Themperature Coupling TC" to allow the HT calculated temperature T to be entered into the relevant fields of SPF. These two are "mono-directional" couplings. And last, the bidirectional "Nonisothermal Flow NITF" node that adds the two previous cases + some more optional physics. If you add all three with NITF as last one you will see that it is overriden by the two others hence will not be considered, therefore you need to select which one(s) of the three you want. The NITF is the most general one and the default settings should do as a good starter. Once you have adapted your model like this, be sure that your already defined solver nodes are really reset to reflect these new node settings, I prefer to delete themn hand have them fully reset, but thsi also removes all "Results node" which can represent quite come work to set up again, then a "Solver Configuration - rightclick- Reset Solver to Default" can do. If you do directly. A "Study - rightclick - ShowDefault Solver might generate a sedcond set of solver subnodes, handy for advanced users, but this will be confusing for new users so I do recommend to clear and start from scratch. Still once your mode is running you need to go through a VV&C process (Verification, Validation and Calibration - against measuremennts". My recommendations here are to check the energy balances what is going in, what is going out (through the boundaries), and how much is stored internally. And does this fit with your analytical calulations ? But, be aware that flux calculations with FEM tools (all not specifically COMSOL) are errror prone as they depend a lot on mesh density and related dependent variable gradients, hence should be looked at in a relative wway (expressed in %) I use the criteria < 10% unbalance is OK for a normal mesh, to get better results you need to work on your mesh, based on your first results, thi8s is an iterative process. Particularly in the time domain ... be aware. Hope this helps on the way, Multiphysics is fun, but can be tough, be sure you have sketed out, beforhand, the full set of phyics phenomena and their relations / intersections for all physics, and all corresponding nodes are there, often you need an additional model tree sub-node per physics and per interaction, minus the ones COMSOL proposes under the multiphysics model node Have fun COMSOLing Sincerely Ivar