1

Topic: initmodule vs initialcondition

Hi all,
would like to know your opinion how on the following problem:
let assume I have a sequence of two different engineering problems, where the solution of the first one should be used as initial state for the second one. Basically, there are two options:
1) Solve the problem as staggered problem and map the variables. This is ok, the disadvantage is that the first problem occupies memory, when solving the second one, while no more needed.
2) Second solution would require to solve problems separately, save the state of the first problem into file (using custom export module) and using this saved state to initialize the second problem. Now there are two options:
  2a) use initModule to read and set initial state. Should conceptually work, however, many engineering models then override this by values from initial conditions. The initModule works only well for internal variables, the primary unknowns are overridden by values from initial conditions.
  2b) use custom initial condition to read primary unknowns from file and combine it with initModule.

Let assume that the approach 1 could not be used (memory restriction). It would be nice to have one generic way to initialize the state, either extend initial conditions to support internal variables as well and discard initialization modules or to keep initialization modules. The advantage of init modules is that they are more flexible at least in the current implementation, where they can initialize and set up virtually everything. If we keep the init modules, how do you see the co-existence of both options? How to solve potential renundance?

Borek

Re: initmodule vs initialcondition

I have been thinking about replacements for specifying temperature fields.
Right now, they are specified as body loads on elements, but there are numerous problems with this.
Primarily, they simply aren't actually loads, and specifying a constant temperature difference (quite common) requires entries on each element line.

Inside the code, we simply want a PrimaryField (of suitable derived type), and we want to be able to specify this as input.
So, should we simply add a line to the input file which describes fields?
This would probably just replace the InitialCondition line (as that would just be a simple "ConstantField"-type).

The code would simply initiate the unknowns from this external field not caring if it's

So, this is basically the custom initial condition, except I think we should just refer to it as a field in the input records, so that we could use it to e.g. control the temperature in structural problems.
In-code, i think we should end up with a Field (which is completely agnostic to how and where it is being used).