import logging
from dataclasses import InitVar, dataclass, field, replace
from enum import Enum
from typing import Dict, List, Literal, Optional, Tuple, Union
import message_ix
from genno import Quantity
from message_ix_models import Context, ScenarioInfo, Spec
from message_ix_models.project.navigate import T35_POLICY as NAVIGATE_SCENARIO
from message_ix_models.project.ssp import SSP_2024, ssp_field
from message_ix_models.project.transport_futures import SCENARIO as FUTURES_SCENARIO
from message_ix_models.report.util import as_quantity
from message_ix_models.util import identify_nodes, package_data_path
from message_ix_models.util.config import ConfigHelper
log = logging.getLogger(__name__)
[docs]@dataclass
class DataSourceConfig(ConfigHelper):
"""Sources for input data."""
#: Emissions: ID of a dump from a base scenario.
emissions: str = "1"
#: Light-duty passenger vehicle techno-economic data.
LDV: Union[None, Literal["US-TIMES MA3T"]] = "US-TIMES MA3T"
#: Non-passenger and non-light-duty vehicles.
non_LDV: str = "IKARUS"
[docs]def quantity_field(value):
"""Field with a mutable default value that is a :class:`.Quantity`."""
return field(default_factory=lambda: as_quantity(value))
[docs]@dataclass
class Config(ConfigHelper):
"""Configuration for MESSAGEix-Transport.
This dataclass stores and documents all configuration settings required and used by
:mod:`~message_ix_models.model.transport`. It also handles (via
:meth:`.from_context`) loading configuration and values from files like
:file:`config.yaml`, while respecting higher-level configuration, for instance
:attr:`.model.Config.regions`.
"""
#: Information about the base model.
base_model_info: ScenarioInfo = field(default_factory=ScenarioInfo)
#: Values for constraints.
#:
#: "LDV growth_activity_lo", "LDV growth_activity_up"
#: Allowable *annual* decrease or increase (respectively) in activity of each LDV
#: technology. For example, a value of 0.01 means the activity may increase by 1%
#: from one year to the next. For periods of length >1 year, MESSAGE compounds
#: the value. Defaults are multiples of 0.0192 = (1.1 ^ 0.2) - 1.0; or ±10% each
#: 5 years. See :func:`ldv.constraint_data`.
#: "non-LDV growth_new_capacity_up"
#: Allowable annual increase in new capacity (roughly, sales) of each technology
#: for transport modes *other than* LDV. See :func:`non_ldv.growth_new_capacity`.
#: "* initial_*_up"
#: Base value for growth constraints. These values are arbitrary.
constraint: Dict = field(
default_factory=lambda: {
"LDV growth_activity_lo": -0.0192,
"LDV growth_activity_up": 0.0192 * 3.0,
"non-LDV growth_activity_lo": -0.0192 * 1.0,
"non-LDV growth_activity_up": 0.0192 * 2.0,
"non-LDV growth_new_capacity_up": 0.0192 * 1.0,
"non-LDV initial_activity_up": 1.0,
"non-LDV initial_new_capacity_up": 1.0,
}
)
#: Scaling factors for costs.
#:
#: ``ldv nga``
#: Scaling factor to reduce the cost of NGA vehicles.
#:
#: .. note:: DLM: “applied to the original US-TIMES cost data. That original data
#: simply seems too high - much higher than conventional gasoline vehicles in
#: the base-year and in future, which is strange.
#:
#: ``bus inv``
#: Investment costs of bus technologies, relative to the cost of ``ICG_bus``.
#: Dictionary with 1 key per ``BUS`` technology.
#:
#: - Used in ikarus.py
#: - This is from the IKARUS data in GEAM_TRP_Technologies.xlsx; sheet
#: 'updateTRPdata', with the comment "Original data from Sei (PAO)."
#: - This probably refers to some source that gave relative costs of different
#: buses, in PAO, for this year; it is applied across all years.
cost: Dict = field(
default_factory=lambda: {
#
"ldv nga": 0.85,
"bus inv": {
"ICH_bus": 1.153, # ie. 150,000 / 130,000
"PHEV_bus": 1.153,
"FC_bus": 1.538, # ie. 200,000 / 130,000
"FCg_bus": 1.538,
"FCm_bus": 1.538,
},
}
)
#: Sources for input data.
data_source: DataSourceConfig = field(default_factory=DataSourceConfig)
#: Set of modes handled by demand projection. This list must correspond to groups
#: specified in the corresponding technology.yaml file.
#:
#: .. todo:: Read directly from technology.yaml
demand_modes: List[str] = field(
default_factory=lambda: ["LDV", "2W", "AIR", "BUS", "RAIL"]
)
#: Include dummy ``demand`` data for testing and debugging.
dummy_demand: bool = False
#: Include dummy technologies supplying commodities required by transport, for
#: testing and debugging.
dummy_supply: bool = False
#: Various efficiency factors.
efficiency: Dict = field(
default_factory=lambda: {
"*": 0.2,
"hev": 0.2,
"phev": 0.2,
"fcev": 0.2,
# Similar to 'cost/bus inv' above, except for output efficiency.
"bus output": {
"ICH_bus": 1.424, # ie. 47.6 / 33.42
"PHEV_bus": 1.424,
"FC_bus": 1.563, # ie. 52.25 / 33.42
"FCg_bus": 1.563,
"FCm_bus": 1.563,
},
}
)
#: Generate relation entries for emissions.
emission_relations: bool = True
#: Various other factors.
factor: Dict = field(default_factory=dict)
#: If :obj:`True` (the default), do not record/preserve parameter data when removing
#: set elements from the base model.
fast: bool = True
#: Fixed future point for total passenger activity.
fixed_GDP: Quantity = quantity_field("1500 kUSD_2005 / passenger / year")
#: Fixed future point for total passenger activity.
#:
#: AJ: Assuming mean speed of the high-speed transport is 330 km/h leads to 132495
#: passenger km / capita / year (Schafer & Victor 2000).
#: Original comment (DLM): “Assume only half the speed (330 km/h) and not as steep a
#: curve.”
fixed_pdt: Quantity = quantity_field("132495 km / year")
#: Load factors for vehicles.
#:
#: ``freight``: similar to IEA “Future of Trucks” (2017) values; see
#: .transport.freight. Alternately use 5.0, similar to Roadmap 2017 values.
load_factor: Dict = field(
default_factory=lambda: {"freight": 10.0} # tonne km per vehicle km
)
#: Logit share exponents or cost distribution parameters [0]
lamda: float = -2.0
#: Period in which LDV costs match those of a reference region.
#: Dimensions: (node,).
ldv_cost_catch_up_year: Dict = field(default_factory=dict)
#: Lifetime of light duty vehicles.
#: Dimensions: (consumer type,). Units: [time].
ldv_lifetime: Quantity = quantity_field(
dict(modest=20, average=15, frequent=10, _dim="consumer type", _unit="year")
)
#: Tuples of (node, technology (transport mode), commodity) for which minimum
#: activity should be enforced. See :func:`.non_ldv.bound_activity_lo`.
minimum_activity: Dict[Tuple[str, Tuple[str, ...], str], float] = field(
default_factory=dict
)
#: Base year shares of activity by mode. This should be the stem of a CSV file in
#: the directory :file:`data/transport/{regions}/mode-share/`.
mode_share: str = "default"
#: List of modules containing model-building calculations.
modules: List[str] = field(
default_factory=lambda: (
"groups demand freight ikarus ldv disutility non_ldv plot data"
).split()
)
#: Used by :func:`.get_USTIMES_MA3T` to map MESSAGE regions to U.S. census divisions
#: appearing in MA³T.
node_to_census_division: Dict = field(default_factory=dict)
#: **Temporary** setting for the SSP 2024 project: indicates whether the base
#: scenario used is a policy (carbon pricing) scenario, or not. This currently does
#: not affect *any* behaviour of :mod:`~message_ix_models.model.transport` except
#: the selection of a base scenario via :func:`.base_scenario_url`.
policy: bool = False
#: Flags for distinct scenario features according to projects. In addition to
#: providing values directly, this can be set by passing :attr:`futures_scenario` or
#: :attr:`navigate_scenario` to the constructor, or by calling
#: :meth:`set_futures_scenario` or :meth:`set_navigate_scenario` on an existing
#: Config instance.
#:
#: :mod:`.transport.build` and :mod:`.transport.report` code will respond to these
#: settings in documented ways.
project: Dict[str, Enum] = field(
default_factory=lambda: dict(
futures=FUTURES_SCENARIO.BASE, navigate=NAVIGATE_SCENARIO.REF
)
)
#: Scaling factors for production function [0]
scaling: float = 1.0
#: Mapping from nodes to other nodes towards which share weights should converge.
share_weight_convergence: Dict = field(default_factory=dict)
#: Specification for the structure of MESSAGEix-Transport, processed from contents
#: of :file:`set.yaml` and :file:`technology.yaml`.
spec: Spec = field(default_factory=Spec)
#: Speeds of transport modes. The labels on the 't' dimension must match
#: :attr:`demand_modes`. Source: Schäefer et al. (2010)
#:
#: .. note:: Temporarily ignored for :pull:`551`; data are read instead from
#: :file:`speed.csv`.
speeds: Quantity = quantity_field(
{
"_dim": "t",
"_unit": "km / hour",
"LDV": 54.5, # = 31 + 78 / 2
"2W": 31,
"AIR": 270,
"BUS": 19,
"RAIL": 35,
}
)
#: Enum member indicating a Shared Socioeconomic Pathway, if any, to use for
#: exogenous data.
ssp: ssp_field = ssp_field(default=SSP_2024["2"])
#: :any:`True` if a base model or MESSAGEix-Transport scenario (possibly with
#: solution data) is available.
with_scenario: bool = False
#: :any:`True` if solution data is available.
with_solution: bool = False
#: Work hours per year, used to compute the value of time.
work_hours: Quantity = quantity_field("1600 hours / passenger / year")
#: Year for share convergence.
year_convergence: int = 2110
# Init-only variables
#: Extra entries for :attr:`modules`, supplied to the constructor. May be either a
#: space-delimited string (:py:`"module_a -module_b"`) or sequence of strings.
#: Values prefixed with a hyphen (:py:`"-module_b"`) are *removed* from
#: :attr:`.modules`.
extra_modules: InitVar[Union[str, List[str]]] = None
#: Identifier of a Transport Futures scenario, used to update :attr:`project` via
#: :meth:`.ScenarioFlags.parse_futures`.
futures_scenario: InitVar[str] = None
#: Identifiers of NAVIGATE T3.5 demand-side scenarios, used to update
#: :attr:`project` via :meth:`.ScenarioFlags.parse_navigate`.
navigate_scenario: InitVar[str] = None
def __post_init__(self, extra_modules, futures_scenario, navigate_scenario):
# Handle extra_modules
em = extra_modules or []
for m in em.split() if isinstance(em, str) else em:
if m.startswith("-"):
try:
idx = self.modules.index(m[1:])
except ValueError:
pass
else:
self.modules.pop(idx)
else:
self.modules.append(m)
# Handle values for :attr:`futures_scenario` and :attr:`navigate_scenario`
self.set_futures_scenario(futures_scenario)
self.set_navigate_scenario(navigate_scenario)
[docs] @classmethod
def from_context(
cls,
context: Context,
scenario: Optional[message_ix.Scenario] = None,
options: Optional[Dict] = None,
) -> "Config":
"""Configure `context` for building MESSAGEix-Transport.
1. If `scenario` is given, ``context.model.regions`` is updated to match. See
:func:`.identify_nodes`.
2. ``context.transport`` is set to an instance of :class:`Config`.
Configuration files and metadata are read and override the class defaults.
The files listed in :data:`.METADATA` are stored in the respective
attributes, e.g. :attr:`set` corresponding to
:file:`data/transport/set.yaml`.
If a subdirectory of :file:`data/transport/` exists corresponding to
``context.model.regions`` then the files are loaded from that subdirectory,
for instance :file:`data/transport/ISR/set.yaml` is preferred to
:file:`data/transport/set.yaml`.
"""
from .structure import make_spec
# Handle arguments
options = options or dict()
try:
# Identify the node codelist used in `scenario`
regions = identify_nodes(scenario) if scenario else context.model.regions
except (AttributeError, ValueError):
pass
else:
if context.model.regions != regions:
log.info(
f"Override Context.model.regions={context.model.regions!r} with "
f"{regions!r} from scenario contents"
)
context.model.regions = regions
# Default configuration
config = cls()
try:
# Update with region-specific configuration
config.read_file(
package_data_path("transport", context.model.regions, "config.yaml")
)
except FileNotFoundError as e:
log.warning(e)
# Data structure that cannot be stored in YAML
if isinstance(config.minimum_activity, list):
config.minimum_activity = {
tuple(row[:-1]): row[-1] for row in config.minimum_activity
}
# Separate data source options
ds_options = options.pop("data source", {})
# Update values, store on context
result = context["transport"] = replace(
config, **options, data_source=config.data_source.replace(**ds_options)
)
# Create the structural spec
result.spec = make_spec(context.model.regions)
return result
[docs] def check(self):
"""Check consistency of :attr:`project`."""
s1 = self.project["futures"]
s2 = self.project["navigate"]
if all(map(lambda s: s.value > 0, [s1, s2])):
raise ValueError(f"Scenario settings {s1} and {s2} are not compatible")
[docs] def set_futures_scenario(self, value: Optional[str]) -> None:
"""Update :attr:`project` from a string indicating a Transport Futures scenario.
See :meth:`ScenarioFlags.parse_futures`. This method alters :attr:`mode_share`
and :attr:`fixed_demand` according to the `value` (if any).
"""
if value is None:
return
s = FUTURES_SCENARIO.parse(value)
self.project.update(futures=s)
self.check()
self.mode_share = s.id()
if self.mode_share == "A---":
log.info(f"Set fixed demand for TF scenario {value!r}")
self.fixed_demand = as_quantity("275000 km / year")
[docs] def set_navigate_scenario(self, value: Optional[str]) -> None:
"""Update :attr:`project` from a string representing a NAVIGATE scenario.
See :meth:`ScenarioFlags.parse_navigate`.
"""
if value is None:
return
s = NAVIGATE_SCENARIO.parse(value)
self.project.update(navigate=s)
self.check()