Standalone implementation of lumped thermal model

[veera-visw]

I am using the BaseModel() class based on BasicDFN() model to solve an aqueous alkaline electrochemical system. As an extension, I wanted to include a thermal model and I chose to start with the lumped model.

The idea I had was to include the temperature equation as part of the model RHS and see the temperature evolution over time (since it is not a converged variable). Attached is a simplified code (based on half cell code, with a simple temperature equation included - see lines # 104-110). When I solve this case, the temperature drops from 300 K (initial temperature) to 20 K and then linearly grows with time (which is fine). I do not understand where this drop comes from and I was wondering if I am missing something basic here?

Note: I made this simple dummy code to make it easy to debug - the electrochemical part of the code is just to have a complete model for the solver to solve . My main question is related to how the temperature equation is coupled to an existing framework solving electrochemical equations.

Thanks!
Simple_model_with_thermal.zip

[veera-visw]

Any suggestions on this topic?
I looked at alternatives to get this to work, where I can solve temperature equation as an algebraic equation (writing the discretised equation). But I am not able to find a way to access the time step used by the solver. I see the "step" function in baseSolver class has a variable dt, but not sure how I can couple it to the main solver.

The main objective is to couple the lumped temeprature to properties that are varying as a function of temperature ( thermodynamic potentials, diffusivity etc).

[veera-visw]

I have checked the individual components of the equation:
temp rhs: is the RHS source term in the equation dT/dt: model.rhs[Tn_x]

temp init is the model. model.initial_conditions[Tn_x].

All these seem fine, its just the drop from initial temperature to 0 that I am not able to understand properly. Full source code attached in the first message FYI!

[veera-visw]

Coming back to this topic:

After playing around a bit, I realised that the model.rhs{} class works fine when the variable is assigned to one of the domains. For example, if I add ["Temperature solve [K]"] to positive domain, I get the right temperature profile. But with a variable that is not assigned to any domain, the "T_init" is neglected.

So I can ideally make my code work, if I take the result of model.rhs and add the initial temperature to this value. But, this seems like a hack and I hope I am not missing anything. Any suggestions?

Eventually, I will couple this temperature to properties so I want to make sure that it is working as supposed to.