Homotopy for Modelica.Fluid---Wall friction in pipe models

[modelica-trac-importer]

Reported by msielemann on 15 Dec 2011 14:13 UTC In order to implement homotopy in Modelica Fluid, the pipe models shall be modified on the branch

https://svn.modelica.org/projects/Modelica/branches/development/Homotopy3.3-dev

This shall be done along the ideas presented in

https://www.modelica.org/events/modelica2011/Proceedings/pages/papers/04_2_ID_131_a_fv.pdf

Please make the modifications and implement two or more test models in ModelicaTest.Fluid

---

Migrated-From: https://trac.modelica.org/Modelica/ticket/665

[modelica-trac-importer]

Comment by msielemann on 15 Dec 2011 14:16 UTC This ticket should cover both wall friction and heat transfer

[modelica-trac-importer]

Comment by kilian_link on 7 Mar 2012 09:56 UTC I checked in the following changes:

pressure loss:

• in Fluid.Pipes.BaseClasses.FlowModels.PartialGenericPipeFlow: used homotopy 4 times to compute the mass flow / pressure loss not with the corresponding exact function, but with given nominal values, e.g. homotopy(..., m_flow_nominal/dp_nominal*(dps_fg - {g*dheights[i]*rho_nominal for i in 1:n-1}))
• in Fluid.Pipes.BaseClasses.FlowModels.NominalLaminarFlow used homotopy 2 times to compute the static head pressure loss with the nominal density instead of the actual density, e.g. homotopy(..., {g*dheights[i]*rho_nominal for i in 1:n-1})

heattransfer:

• in Fluid.Pipes.BaseClasses.HeatTransfer.PartialPipeFlowHeatTransfer used homotopy to compute the Q_flows with a constant alpha value (alpha0) instead of the exact one, e.g. homotopy(alphas[i], alpha0)

I also added 2 testcases under Fluid.Examples.PipeHomotopy, ran them with and without homotopy (set the flags Advanced.OnlyUseHomotopyMethod and Advanced.SupportHomotopyFunction to true or false accordingly) and both of them only succeed when using homotopy.

[modelica-trac-importer]

Comment by kilian_link on 17 Apr 2012 11:58 UTC Since there were some problems with the two testcases (they could not be initialized at all), we figured out that it is very important that the flag Evaluate = true is set in order to succeed.

Additionally the second example (PipesWithHeattransfer) has been updated and now should work as stated in the comment above (initialization only possible if homotopy is used).

[modelica-trac-importer]

Comment by pjordan on 24 Apr 2012 09:41 UTC The second test case PipesWithHeatTransfer is still sensitive to increased heat flow rates at the boundaries. For example, it will fail if heat1.Q_flow = 2.5 * 200e3. Convergence is better if the nominal values for pressure loss and mass flow rate in the FlowModel are set to the correct order of magnitude (e.g. m_flow_nominal=10, dp_nominal=0.05e5).

[modelica-trac-importer]

Comment by fcasella on 14 Dec 2015 23:00 UTC Accepted tickets

[modelica-trac-importer]

Comment by fcasella on 14 Dec 2015 23:08 UTC Not enough time to handle this for MSL 3.2.2 and test properly. Rescheduled to next minor revision

[modelica-trac-importer]

Comment by beutlich on 20 Dec 2016 19:50 UTC Unhide in agreement with msielemann and otter.

[casella]

Adding homotopy may require some new parameters, e.g. for nominal mass flow rate, which were not originally present in the model. We could give a start value to them, with a warning like "This parameter was introduced in v. 4.0.0, please put an appropriate value to it to ensure proper operation of the model". In most cases the model will work anyway, and in all cases there's proper diagnostics to fix the issue.

The same applies to #666, #667, #668, #669