A scenario report documents what came out of a run:
solve status and objective, the generation, capacity and new-capacity
mixes, a sub-annual dispatch profile, emissions and costs, and the
verify_solution() checks. It is the counterpart of
vignette("report-model"), which documents the assumptions
going in.
The one below is pypsa_eur_41 solved on a sampled
calendar — one day per month, 288 of the 8,760 hours — which keeps every
region and process while making the solve tractable:
library(energyRt)
library(reneuro)
cal <- calendars$d365_h24_1dpm
scen <- interpolate_model(pypsa_eur_41, cal, name = "eur41_m12")
scen <- write_script(scen, solver = solver_options$julia_highs_barrier)
scen <- read_solution(solve_scenario(scen, wait = TRUE))
report(scen)Two notes on running it.
The sample shortens the solve, not the interpolation. Parameters are interpolated and only then filtered to the declared timeslices, so interpolation costs much the same whichever calendar it is given – about 35 seconds here – while the solve drops to about 220 seconds from a calendar thirty times larger.
Use HiGHS. GLPK solves the five-node models in
seconds, but at 41 nodes it was still iterating after ten minutes on a
problem the barrier closes in under four. open_solvers()
reports what this machine has.
Beyond the defaults: run = reports a run other than the
active one, verify = FALSE omits the checks section,
levcost = TRUE adds an ex-post cost table, and
report(compare_scenarios(list(a = ..., b = ...))) puts
several runs side by side.