Custom objective function for PV

[Anju-0]

Hello all,

Can I use PMD to minimize the reactive power injection by solar PV? I tried using the below (I am still new to PMD so it could be wrong),

function objective_reactive_min(pm)
    JuMP.@objective(pm.model, Min, sum(sum(sum(qg) for (i, qg) in PMD.var(pm, n, :qg)) for n in PMD.nw_ids(pm)))
end

But according to my understanding when I use (qg) does it minimize all the reactive power generation of vsource, generators and solar? Is it possible to isolate the reactive power of solar pv and consider only it as a variable?

Also I want the reactive power injection of solar to be bounded within 0<=qg(solar)<=qgsolar(max), I wrote the below constraint but it is not working.

function constraint_generator_reactive_power_dispatch(pm, i; nw=PMD.nw_id_default)
    for c in (PMD.ref(pm, nw, :gen, i, "connections")) #get the nodes each generator is connected to
        JuMP.@constraint(pm.model, PMD.var(pm, nw, :qg, i)[c] <=0.5)
    end
end

Thank you!

[MartaVanin]

Hi, there's multiple ways to do that, here are some suggestions:

1. you could define separate variables for the p and q of solar , but probably in this case it is not practical and error-prone, as you would have to re-write also all the constraints where these are used to make sure they are added where needed. Though it feels a bit overkill if the only distinction is whether some variables are in the objective function or not.
2. you add a key to the "gen" dictionary items (or the "load" dictionary, if you define solar as negative load) that "flags" whether that specific generator/load is solar or not. e.g.,

data["gen"]["1"] = Dict{String, Any}(......, "solar"=>true)
...
data["gen"]["14"] = Dict{String, Any}(......, "solar"=>false)

so in your objective you can go (you need to double-check the syntax, as I don't have a laptop with Julia here, but you get the idea):

function objective_reactive_min(pm)
    JuMP.@objective(pm.model, Min, sum(sum(sum(PMD.var(pm, n, :qg, i) for i in  PMD.ids(pm, n, :gen) if PMD.ref(pm, n, :gen,i)["solar"]) for n in PMD.nw_ids(pm)))
end

Regarding the bounds on the power, I am not sure why you want to rely on ``nodes each generator is connected to". You can add again some bounds on the "gen" dictionary

for (_, gen) in data["gen"]
    if gen["solar"]
        gen["pgmin"] = xxxx
        gen["pgmax"] = yyy
...

and so on for qg. An easy way to "fix" the injection value (e.g., for non-solar gens) is to set the lower bound = upper bound.

Hope this helps

[Anju-0]

Hello,

Thank you for the answer.

But what if I am parsing the data through an OpenDSS file?

[MartaVanin]

Shouldn't be an issue, you just save the data as a PMD dictionary, e.g., data = PMD.parse_file("path/to/Master.dss", ...) and apply the changes you need to that dictionary.

Note that PMD has an ENGINEERING and a MATHEMATICAL data model, see here

The underlying JuMP model that is eventually solved is built from a MATHEMATICAL model, so functions like PMD.ref "go look" in the MATHEMATICAL dictionary that you either created yourself, or that PMD has built implicitly before solving the optimization problem, if you gave it an ENGINEERING data dictionary instead.

My suggestion would be that you parse the data to a MATHEMATICAL dictionary and edit that dictionary so that it does what you expect, e.g., adding there the "solar"=>true mentioned above.

When you parse/convert to the MATHEMATICAL model, there are usually entries in the component dictionaries ("name" or something like that, if I recall correctly) that carry some "memory" of the original component name in the dss file, so you can probably use those to look up whether that gen or load is a solar one.

[Anju-0]

This worked for me. I had to refer to the manual again, but once I understood the basics, it became easy. Thank you for your explanation; it was really helpful.