DeployInst before simulation start

[meg-krieg]

What is this task about?

Consider the case of when user inputs a deployinst start year before the simulation start time.

Why is this task needed?

Facilities should start with full inventories even though the simulation perceives them as just being built. Their lifetimes should be shortened according to their existence/life before the simulation start.

How can this issue be closed?

Add a scenario to deployinst in which the user specifies this start year value. Do not replace/disable the other scenarios in which the start_year is triggered during the simulation time and after simulation end.

[gonuke]

I think it would be a good idea to have an additional optional variable that allows a user to specify whether "early" deployments should be allowed (shifting built times & lifetimes) or not (error).

There are instances where the accumulated inventory of multiple years of operation are important & cant be simulated by just deploying facilities at time 0.

[meg-krieg]

Are there specific facility types that this would be restricted to? Or, is it all facilities?

I have a small idea about the implementation for this using the "early" deployment decision variable that could be added.

Since the inventory of each facility is different based on prototype throughput/capacity specifications and the different outgoing streams, should the accumulated inventory for early deployments be calculated within the archetypes classes?

Each archetype could receive the notification to deploy early and do a back calculation (unique to the facility type) to anticipate how much material it accumulated and deploy on time step 0 of the simulation with that material ready.

[gonuke]

I think a first step is to simply have a flag that allows archetypes to be built at time 0 if they are otherwise indicated for early deployment, or not.

It is true that we could come up with various behaviors to make this more allowable in the future, but there may be some archetypes for which there is no way to do that.

The case that is at top of mind in our group right now is the deployment of CANDU reactors for which tritium production is a byproduct. These are typically modeled as sources of tritium right now, and they trade that tritium with a storage facility. Because of the relatively short half-life, the timing of when the tritium is produced is important and it might be challenging to get this right by backcasting the behavior of the facility.

For the time being, I'd prefer to just call it an error and require the user to change their simulation start year. Other solutions will require more careful consideration and could be added separate issues.