Francesco Casella
Francesco Casella
More specifically, I checked again the three test cases that are reported as exemplary of the 52 Buildings 7.0.1 models failing in the backend: - [Buildings.Fluid.HeatExchangers.ActiveBeams.Examples.CoolingAndHeating](https://libraries.openmodelica.org/branches/master/Buildings_7.0.0/files/Buildings_7.0.0_Buildings.Fluid.HeatExchangers.ActiveBeams.Examples.CoolingAndHeating.err), ticket [#6207](https://trac.openmodelica.org/OpenModelica/ticket/6207) - [Buildings.Experimental.DistrictHeatingCooling.Validation.HeatingCoolingHotWaterSmall](https://libraries.openmodelica.org/branches/master/Buildings_7.0.0/files/Buildings_7.0.0_Buildings.Experimental.DistrictHeatingCooling.Validation.HeatingCoolingHotWaterSmall.err),...
Further analysis of [#6207](https://trac.openmodelica.org/OpenModelica/ticket/6207) reveals that we have the following code: ``` algorithm beaCooHea.conHea.mod.temDif_mod.i := 1; for j in 1:2 loop if beaCooHea.conHea.mod.temDif_mod.u > beaCooHea.conHea.mod.temDif_mod.xd[j] then beaCooHea.conHea.mod.temDif_mod.i := j; end...
The other two test cases only have initial algorithms, but they only contain asserts, so the root cause must be different.
@phannebohm, the case reported by @rfranke makes this small improvement really high priority. It could actually solve a lot of issues in the testsuite. Can you please take care of...
> Shall we introduce an annotation (similar to tearingSelect), which allows for explicitly setting the attributes (min/max/nominal/etc.) for derivatives. In example: Real x annotation(derivativeAttributes(nominal=1e5, min=0, max=Inf)); That is one possibility...
Yeah, good idea! It is always best if we can avoid changing the language, for two reasons: - it normally takes _a lot_ of discussion and time to get anything...
@phannebohm, I checked the [regression report](https://libraries.openmodelica.org/branches/history/master/2021-08-31%2023:16:46..2021-09-01%2022:54:52.html). There are several improvement, but I understand they are all due to @perost's fixes of previous NF commits that broke some stuff, see the...
@phannebohm, I still have some trouble with this issue. After your commit, derivatives get a nominal value which is equal to the corresponding state's nominal value, divided by the reporting...
According to my understanding, zero-crossing functions shouldn't be discontinuous, just in case the solver uses something more sophisticated than bisection to locate the event time. For a relation such as...
If I understand it correctly, the code fragments you posted generate step ZC functions. In that case, it is relatively straightforward to implement what you propose. I'm not sure it...