Business plan plot

[TomFer97]

No description provided.

[TomFer97]

I've renamed the back but it still shows sankey_branch should I do git fetch?

[davide-f]

Have you tested the execution of the function?

[davide-f]

Nice implementation!

I've added some comments.
It could also be nice to add a test once finished.

may you attach a plot of the image you obtain here for discussing it?

Awesome PR!!!

[davide-f]

This PR is getting a very nice shape! :D

[davide-f]

You didn't touch this line as I guess it is not clear.

In the original split function all this values where 1D arrays by user that represent the total discounted values by entity.
Therefore, by making the appropriate differences we calculated the total discounted reward allocation by user.

Now with each variable being 2D that is a bit tricky to do.
I think we should first identify the total annualized reward allocation and then (assumption) uniformly distribute it across the years.
Given that all the structures are 2D there is for sure a smart way to do that; I propose an approach, but feel to propose an alternative.

Basically, what we may need to do is to create a proxy total discounted cost of all terms but NPV so to reproduce the old (CAPEX .+ OPEX .+ Ann_Replacement .- Ann_Recovery).
For example, something like:
(CAPEX .+ OPEX .+ Ann_Replacement .- Ann_Recovery).data * ann_factor (note that since they are matrix operation, there may be the need for some transpositions

Then, the resulting vector shall be a 1 column or 1 vector and we can create the equivalent 1D cost vector, so that we can do (total_discounted_reward = NPV .- new_vector).
Then, we can do total_discounted_reward ./(sum(act_factor) - 1) and this should be a 1D vector of the yearly reward allocation by user, that can be exploded into 2D by simply duplicating the entries.

[davide-f]

The year set may be obtained from the axis of any function of business_plan, what do you think?

[TomFer97]

I'll think about it

[TomFer97]

Unfortunately the last commit allow me to obtain something like this.

[davide-f]

I see.
May you please check and post the screen of the table calculated by business_plan?

[TomFer97]

This is the dataframe. For sure OEM, EN_SELL, REWARD and RV must be 0 the first year. And REP 0 except the year of substitution. EN_CONS and PEAK is always 0 except for last year

21×9 DataFrame
Row │ Year CAPEX OEM EN_SELL EN_CONS REP REWARD RV PEAK
│ Int64 Float64 Float64 Float64 Float64 Float64 Float64 Float64 Float64
───┼───────────────────────────────────────────────────────────────────────
1 │ 0 3.96488e5 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
2 │ 1 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
3 │ 2 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
4 │ 3 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
5 │ 4 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
6 │ 5 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
7 │ 6 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
8 │ 7 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
9 │ 8 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
10 │ 9 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
11 │ 10 0.0 5568.01 2337.41 1686.68 -1.07268e5 1666.37 52369.5 0.0
12 │ 11 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
13 │ 12 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
14 │ 13 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
15 │ 14 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
16 │ 15 0.0 5568.01 2337.41 3373.36 -1.07268e5 1666.37 52369.5 0.0
17 │ 16 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
18 │ 17 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
19 │ 18 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
20 │ 19 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 0.0
21 │ 20 0.0 5568.01 2337.41 0.0 -1.07268e5 1666.37 52369.5 36447.2

[TomFer97]

I forgot in which part of the code insert ECModel.results[:R_Reward_agg_tot] = 0 for non_cooperative in order to create a reward variable that is = 0 for non cooperative and >0 for cooperative and ANC. You showed it to me but I forgot and I didn't apply the changes

[davide-f]

Cool Tommaso :D
I think we are close to finalize :D
Could you add a sample plot as well?

[davide-f]

I forgot in which part of the code insert ECModel.results[:R_Reward_agg_tot] = 0 for non_cooperative in order to create a reward variable that is = 0 for non cooperative and >0 for cooperative and ANC. You showed it to me but I forgot and I didn't apply the changes

Here :)

EnergyCommunity.jl/src/non_cooperative.jl

Line 692 in fc3420f

# Nothing to do

You can get inspiration from here:

EnergyCommunity.jl/src/aggregated_non_cooperative.jl

Lines 203 to 265 in fc3420f

	# Power of the aggregator
	ECModel.results[:P_agg] = JuMP.Containers.DenseAxisArray(
	[sum(ECModel.results[:P_us][:, t]) for t in time_set],
	time_set
	)
	# Shared power: the minimum between the supply and demand for each time step
	ECModel.results[:P_shared_agg] = JuMP.Containers.DenseAxisArray(
	[
	min(
	sum(ECModel.results[:P_P_us][:, t]),
	sum(ECModel.results[:P_N_us][:, t])
	)
	for t in time_set],
	time_set
	)
	# Total reward awarded to the community at each time step
	ECModel.results[:R_Reward_agg] = JuMP.Containers.DenseAxisArray(
	[
	profile(ECModel.gen_data,"energy_weight")[t] * profile(ECModel.gen_data, "time_res")[t] *
	profile(ECModel.gen_data, "reward_price")[t] * ECModel.results[:P_shared_agg][t]
	for t in time_set],
	time_set
	)
	# Total reward awarded to the community in a year
	ECModel.results[:R_Reward_agg_tot] = sum(ECModel.results[:R_Reward_agg])
	# Total reward awarded to the aggregator in NPV terms
	ECModel.results[:R_Reward_agg_NPV] = (
	ECModel.results[:R_Reward_agg_tot] * sum(1 / ((1 + field(gen_data, "d_rate"))^y) for y in year_set)
	)
	# Total reward awarded to the aggregator in NPV terms
	ECModel.results[:NPV_agg] = ECModel.results[:R_Reward_agg_NPV]
	# Cash flow
	ECModel.results[:Cash_flow_agg] = JuMP.Containers.DenseAxisArray(
	[(y == 0) ? 0.0 : ECModel.results[:R_Reward_agg_tot] for y in year_set_0],
	year_set_0
	)
	# Cash flow total
	ECModel.results[:Cash_flow_tot] = JuMP.Containers.DenseAxisArray(
	[
	((y == 0) ? 0.0 :
	sum(ECModel.results[:Cash_flow_us][y, :]) + ECModel.results[:Cash_flow_agg][y])
	for y in year_set_0
	],
	year_set_0
	)
	# Social welfare of the users
	ECModel.results[:SW_us] = sum(ECModel.results[:NPV_us])
	# Social welfare of the entire aggregation
	ECModel.results[:SW] = ECModel.results[:SW_us] + ECModel.results[:NPV_agg]
	ECModel.results[:objective_value] = ECModel.results[:SW]

Basically all expressions mentioned there should be available in the NC configuration as well, with appropriate values (including many 0s)

Apologies for the late reply, I didn't see the message at first

[TomFer97]

I Think that It's almost completed but I still get this problem with the label of the plot. Probably collect(keys(plot_struct)) is making some problems. I tought that with a list it would have been resolved

[davide-f]

Maybe its the key: "labelS".
Test it locally

[TomFer97]

The key labelS=bar_labels, kwargs...) cancel completely the following keys. Maybe it's because it's a vector and maybe should require a list but I've tried with [keys(plot_struct)] and It gives me a strange error. Maybe list(keys(plot_struct))?

[davide-f]

It may be the type of object that is a matrix instead.

[1 2 3] is a 2D matrix 1×3
[1, 2, 3] is a 1D vector

hcat(collect(keys(..))...)
May work. Maybe the collect is not needed but I'm unsure

[davide-f]

@TomFer97 Merged :D