Matthias, just a return question: As your assumption is the existence of different resistance branches, why don't you just take the resultant mass flow share (to be calculated strictly in advance from given/measured resistances) as boundary condition at the T junction exits? I guess you regard the junction itself as constant pressure system, so the actual distribution will form by itself, or rather by potentially intentional microscopic disturbance settings at the inflow.

If your desires are pointing towards a calculation of the branch resistances themselves I'd assume that you will have a very hard job, since this would require to model the complete branch in detail to calculate the effects of potentially very complex geometric structures on the resultant flow in that branch. Here an experiment would probably be the far more simpler approach.

Cheers,

Petra

Petra, thanks for your kind response. I totally agree on your suggested approach, which would have been both sufficient and simple. However, my mind is set on transient simulations and I would like to calculate the interactions in a hydronic heating system when associated radiator models are incorporated in space models. My intention is that thermostatic valves (TRV) should open and close as the temperature in the spaces drops/increases, and as one TRV closes, the remaining radiators will be afflicted with a higher flow rate, which in turn causes an increased temperature in the associated spaces.

Once again, thanks for your response.

Mattias,

I'd say that this is the absolutely typical case for modern hydronic heating systems. But the changes that are imposed on your T junctions of interest should be more or less quasi-static: compare the mass flow vs. adaption time of the individual radiator branches, that do not shut down the flow all of a sudden..

Accordingly, I'd suggest to batch process a two-dimensional series of static flow conditions, defined by the resistances for the two outflow settings of your T.

There is a slight chance of missing an evolutionary flow profile development that depends on former flow states and that only could be approached by setting up a computation with time-dependent boundary conditions (w/r/to  branch resistances or mass flows in the outflows). But I don't think that this is very likely.

My first answer assumed that you are interested in a very detailed flow field analysis at the T junction, but your response indicates that this might not be the case, and you are only looking for a 1-dimensional + time resolved development description. Even in this case I'd suggest to set up a mostly quasi-static system model that assumes fixed mass flow distributions within a given time interval.

Cheers,

Petra

Petra,

Yes, you are completely correct, and I apologies for the unclear details in my initial question, I am interested in a 1-d model to solve pressure-flow relations in time.

I will considered your suggestion, thanks a lot!

By the way, given your expertise, can you suggest any appropriate references in the field?

Mattias, to be honest: This is exactly my blank spot that I did not yet address in my work. :-) I know though that most colleagues use MatLab for such tasks. Just in case you wish to model run lengths (i.e. different branch volumes and different places where the temperature dependent throttling happens, resulting in differing time constants) it might be wise to directly use FEM modelling. A possible system is ElmerFEM (https://www.csc.fi/web/elmer). I took it for real 3d heat spread modelling only.

Cheers,

Petra