Write back solve results to model for outer loop

Q&A
1

hi all,

i have a simple oemof model and i want to use it in an outer loop.
since v3.0 there is a native way to solve a single time step, but how can i feedback the solve results to my model and start the next time step?
i found this file git hub and line 84 should
# write back results from optimization object to energysystem optimization_model.results()
but this only returns a dict or did i misunderstood something?

thx, niels…

2

Do you want to access parameters on the optimization model and change them?

1 Like
3

yes - after the model is solved in the first time step, i want to use the new SOC (found by the solver) of the storage for the next time step. furthermore, i have to change the demand of some sinks. these values are given by an outer loop.
the next time step is to be optimized with these new parameters.

4

You then need to set the parameters and upper bounds directly on the optimization model (which is a pyomo model).

An example can be found here: https://github.com/oemof/oemof-examples/blob/master/oemof_examples/oemof.solph/v0.3.x/flexible_modelling/add_constraints.py

Unfortunately, the access differs depending on the actual object you are changing. But generally it should work like this whereas om is your optimization model:

# Flow: Set upper bound
om.flow[(source, target), timestep].setub(my_value)

# Flow: Set value
om.flow[(source, target), timestep].value = my_value

# GenericStorage: Set upper bound (capacity)
om.GenericStorageBlock[(my_storage,), timestep].setub(my_value)

# GenericStorage: Set initial/end state of charge
om.GenericStorageBlock[(my_storage,),om.TIMESTEPS[-1]] = my_value * 0.5
om.GenericStorageBlock[(my_storage,),om.TIMESTEPS[-1]].fix()

# Other components: Set upper bound
om.MyComponent[my_component_instance, timestep].setub = my_value
# Other components: Set value
om.MyComponent[my_component_instance, timestep].value = my_value

Depending on the component you want to change, you might also have a look into the modules here which hold the different model blocks (blocks, components, custom): https://github.com/oemof/oemof/tree/dev/oemof/solph

General information on how to work with pyomo models can be found here: https://pyomo.readthedocs.io/en/latest/working_models.html

Good luck!

Cord

1 Like
5

unfortunately this is not the 1-line-solution i expected, but it leads me in the right direction!
thx
:+1: