Transfer of energy potentials

[espenfb]

At the moment there is only support to transfer quantities around the system. For more advanced technical modelling it can be benefitial to have resources that both have an energy quantity and an energy potential.

E.g. electricity has both a flow rate in current and an energy potential in voltage. Or for gases with flow rates and pressures.

Both of these types of properties exist in the same resource and is not seperatable. Such that it is not preffered to model them as seperate resources. At the moment both power and current can be transferred around the system as they are summarized in the interface/junction between nodes and links. Voltage on the other hand is transferred by equality on the same interface/junction.

Possible solution:

• The addition of a new variable to transfer the energy potential along with the flow_in/flow_out and link_in/link_out variables.
• The new variables only needs to be added for specific types of more advanced resources, such that it would not be added unncessesary and not result in breaking changes.

[JulStraus]

I do not really know whether we want to have here variations in Base or in Geography. Pressure in a local energy system is in general less of a crucial component (while I guess voltage is). I hence also includedin my answer below. Note that a similar problem arises with pooling problems so we should get in this situation a unified solution. Pinging @hellemo in for that.

Thoughts for implementation in Base

Given your current thought process, the variables must

1. be available for the connected links and nodes and
2. have their balance solved between links and nodes.

The corresponding function is

constraints_couple(m, 𝒩::Vector{<:Node}, ℒ::Vector{<:Link}, 𝒫, 𝒯, modeltype::EnergyModel)

In this function, if only to be created for a subtype of resources, we would need to either include a conditional statement or a subfunction for dispatch with an empty function for all other Resources.

I think the interesting question here is which index set we would like to have these variables for. Based on my feeling, we would need to create them for the specific nodes/links, time period, and resource. That is the same index sets as flow_in/flow_out and link_in/link_out. The variable creation would then be done in the functions

variables_flow(m, 𝒩::Vector{<:Node}, 𝒳ᵛᵉᶜ, 𝒯, modeltype::EnergyModel) # and
variables_flow(m, ℒ::Vector{<:Link}, 𝒳ᵛᵉᶜ, 𝒯, modeltype::EnergyModel)

again, either with a subfunction or through a conditional statement.

Thoughts for implementation in Geography

Should these potentials be a part of the TransmissionMode or do you see it more as an Area specific concept? Do we need to have access to them at every single point?

While I do not have sufficient knowledge regarding the electrical system, we can argue that the pressure of a Resource should be dependent on the region. Different pressure levels could then be achieved through a compressor Node (although I really do not have too much an idea how to incorporate that in the current version nicely without having to add the Area as well as field to a compressor to extract the pressure variable).

We could in this situation extend on the compute_trans_in and compute_trans_out functionality in EnergyModelsGeography.

[espenfb]

I was working on implementing it the way you suggest for the Geography package. And it can be implemented that way to allow power flows with small changes to EMB.

For power systems I think we want to make the voltages available for the nodes in the local system for more advanced modelling of the technologies.

• Power and voltage relationships of generators/storage for providing voltage support (most important)
• The effect of voltages on sink demand
• Possibly local power flows on links to model low-voltage side locally (not sure if this is the way to go)

The most important thing is that the voltage is available to the nodes within an area, not that it is differentiated within the local system (but that could open the door for even more applications).

As I see it the two best ways of trasferring the voltage from the area/availability to a generator within the same area is:

1. The solution that is most consisten with the current design:
   -- normally varables are trasferred trough links, the same can be done with potentials as suggested above (also see brach under "Development" in the right-hand-side menu).
2. Transferring the area variables to the nodes without using the links
   -- This can be done using 𝒳ᵛᵉᶜ from constraints_elements which then must be passed on by create_element and create_node:

function constraints_elements(m, 𝒳::Vector{<:AbstractElement}, 𝒳ᵛᵉᶜ, 𝒫, 𝒯, modeltype::EnergyModel)
    for x ∈ 𝒳
        create_element(m, x, 𝒯, 𝒫, modeltype)
    end
end

create_element(m, n::Node, 𝒯, 𝒫, modeltype::EnergyModel) =
    create_node(m, n, 𝒯, 𝒫, modeltype)