The dataset ships ONE model: twelve provinces, hourly, 51 objects. This cuts smaller cases from it – one province, one zone, a sample across zones – in a single call, resolving as it goes the things that are tedious and error-prone to do by hand.
Usage
build_topia(
regions = NULL,
preset = NULL,
layout = "honeycomb",
level = "province",
calendar = "d365_h24",
horizon = "BY2025_50_by10",
techs = NULL,
modules = NULL,
clusters = TRUE,
demand = NULL,
weather = NULL,
levers = TRUE,
as = c("kit", "model", "repository"),
boundary_prices = NULL,
quiet = FALSE
)Arguments
- regions
province codes, zone codes, the nation code, or a count for the first n in canonical order.
NULLkeeps all twelve.- preset
one of
"full","island"(R01, no wire and no domestic fuel) or"zone"(Z3, carrying the nuclear station and the reservoir). A preset expands to the arguments below and reports what it chose.- layout
"squares","honeycomb"(the default, and the one the shipped model carries),"island"or"continent".- level
the geoframe the model's own regions are. Only
"province"is built so far.- calendar
a calendar name or object. It must be a sample of the model's own – its timeslices a subset of
d365_h24's – whichd365_h24_1dpm(301) andd365_h24_1dps(101) are. Anything else is refused, naming the slices that do not match.- horizon
"BY2025_50_by10","BY2025", or ahorizonobject.- techs
which technologies to keep, or to drop with a leading
-(c("-EBIO", "-H2ELC")). Two keywords are derived from the objects rather than listed:"vre"is a technology declaring a weather series,"thermal"one that burns its input. Keeps and drops cannot be mixed.- modules
which of the article's nine sub-repositories to keep, or to drop with a leading
-:commodities,supply,weather,generators,storage,hydrogen,transmission,fuel_trade,demand.modules = "-hydrogen"strips in one word the chain the article declares across three sections.- clusters
not built yet.
- weather, demand
the name of a shipped dataset (
"sample","simulated") or adata.frameshaped like the matching element oftopia_profiles()– so the output of one is the input of the other. The frames are written into the shipped objects, which keeps each series' name, units and timeframe. A weather frame must cover every(cluster, region)pair the object's roster declares; a demand frame carryingloadis a relative shape and is scaled to the annual energy the shipped object holds, one carryingdemandis taken as given.- levers
TRUEfor all four,FALSEfor none, or names.- as
"kit"(the named list the dataset itself uses, so the two are interchangeable),"model"(a readynewModel()with the geoscale set), or"repository".- boundary_prices
passed to
subset_model_regions()to price the corridors a subset cuts.- quiet
suppress the one-line report of what was built.
Details
build_topia() authors nothing. Every object it returns was declared in
TOPIA I: building the model, which generates the shipped data files; this
selects from them. A technology TOPIA does not declare cannot be asked for,
and new processes are added to the article rather than here.
With every argument at its default the result is topia$modules$electricity
unchanged, which is what tests/testthat/test-topia-build.R pins.
regions takes codes at any geoframe. A zone code expands to the provinces
beneath it, so a request never cuts a zone in half by accident – which
matters because COA and GAS balance at the zone, and a subset that
splits one is refused by subset_model_regions() rather than averaged.
Selecting processes pulls a dependency closure behind it: drop every
coal burner and COA, SUP_COA, IMP_COA and both coal corridors go too,
because nothing is left to use them. A storage and a corridor are not
counted as consumers or producers while deciding this – one time-shifts
and the other moves, so counting them lets a commodity keep itself alive
after its last real user is gone. CO2 is reachable only as an emission
declared on COA and GAS, and is kept on that basis. A lever naming a
process that is gone goes with it, since it would otherwise reference what
the model does not declare.
A cheaper calendar re-keys the series and the demand curve onto it, and
the annual energy is held fixed: a sample is a cheaper view of the same
year, not a smaller year, and two cases are only comparable if the demand
they must meet is the same. Only a sample of the model calendar is
accepted. The measured wind and solar grades exist on d365_h24 alone, so
a calendar with timeslice names of its own – s4_h24, whose series are a
simulated stand-in with no grades at all – would leave every clustered
technology with no weather row and running unbounded.
The two levers that carry a spatial shape are narrowed with the model:
GREEN_H2, whose for.each names every province, and CO2_CAP, whose cap
is a per-province allowance times the province count and so is rescaled by
the count kept. CT_CO2 and EARLY_RET have no province dimension.
layout chooses which of the four pictures the model is drawn on. The
corridor network is the SAME on all four – fourteen corridors frozen in
data-raw/topia_links.R, not derived from any one layout – so a layout
changes the map and nothing the model solves.
See also
topia for the dataset, subset_model_regions() and
aggregate_model_regions() for the reduction verbs this sits beside.
Examples
if (FALSE) { # \dontrun{
# the shipped model, unchanged
identical(build_topia(), topia$modules$electricity)
# one zone, ready to interpolate
mod <- build_topia("Z3", as = "model")
# the island, by name
kit <- build_topia(preset = "island")
} # }
