data.datasets.sanity_checks.sanitycheck_emobility_mit()

Complexity

Conditions

24

Size

Total Lines	556
Code Lines	359

Duplication

Lines	0
Ratio	0 %

Importance

Changes

0

"""
    input_geo_thermal = db.select_dataframe(
        """SELECT carrier,
         SUM(capacity::numeric) as Urban_central_geo_thermal_MW
            FROM supply.egon_scenario_capacities
            WHERE carrier= 'urban_central_geo_thermal'
            AND scenario_name IN ('eGon2035')
            GROUP BY (carrier);
        """,
        warning=False,
    )["urban_central_geo_thermal_mw"].values[0]

    output_geo_thermal = db.select_dataframe(
        """SELECT carrier, SUM(p_nom::numeric) as geo_thermal_MW
            FROM grid.egon_etrago_generator
            WHERE carrier= 'geo_thermal'
            AND scn_name IN ('eGon2035')
            GROUP BY (carrier);
    """,
        warning=False,
    )["geo_thermal_mw"].values[0]

    e_geo_thermal = (
        round((output_geo_thermal - input_geo_thermal) / input_geo_thermal, 2)
        * 100
    )
    logger.info(f"'geothermal': {e_geo_thermal} %")


def residential_electricity_annual_sum(rtol=1e-5):
    """Sanity check for dataset electricity_demand_timeseries :
    Demand_Building_Assignment

    Aggregate the annual demand of all census cells at NUTS3 to compare
    with initial scaling parameters from DemandRegio.
    """

    df_nuts3_annual_sum = db.select_dataframe(
        sql="""
        SELECT dr.nuts3, dr.scenario, dr.demand_regio_sum, profiles.profile_sum
        FROM (
            SELECT scenario, SUM(demand) AS profile_sum, vg250_nuts3
            FROM demand.egon_demandregio_zensus_electricity AS egon,
             boundaries.egon_map_zensus_vg250 AS boundaries
            Where egon.zensus_population_id = boundaries.zensus_population_id
            AND sector = 'residential'
            GROUP BY vg250_nuts3, scenario
            ) AS profiles
        JOIN (
            SELECT nuts3, scenario, sum(demand) AS demand_regio_sum
            FROM demand.egon_demandregio_hh
            GROUP BY year, scenario, nuts3
              ) AS dr
        ON profiles.vg250_nuts3 = dr.nuts3 and profiles.scenario  = dr.scenario
        """
    )

    np.testing.assert_allclose(
        actual=df_nuts3_annual_sum["profile_sum"],
        desired=df_nuts3_annual_sum["demand_regio_sum"],
        rtol=rtol,
        verbose=False,
    )

    logger.info(
        "Aggregated annual residential electricity demand"
        " matches with DemandRegio at NUTS-3."
    )


def residential_electricity_hh_refinement(rtol=1e-5):
    """Sanity check for dataset electricity_demand_timeseries :
    Household Demands

    Check sum of aggregated household types after refinement method
    was applied and compare it to the original census values."""

    df_refinement = db.select_dataframe(
        sql="""
        SELECT refined.nuts3, refined.characteristics_code,
                refined.sum_refined::int, census.sum_census::int
        FROM(
            SELECT nuts3, characteristics_code, SUM(hh_10types) as sum_refined
            FROM society.egon_destatis_zensus_household_per_ha_refined
            GROUP BY nuts3, characteristics_code)
            AS refined
        JOIN(
            SELECT t.nuts3, t.characteristics_code, sum(orig) as sum_census
            FROM(
                SELECT nuts3, cell_id, characteristics_code,
                        sum(DISTINCT(hh_5types))as orig
                FROM society.egon_destatis_zensus_household_per_ha_refined
                GROUP BY cell_id, characteristics_code, nuts3) AS t
            GROUP BY t.nuts3, t.characteristics_code    ) AS census
        ON refined.nuts3 = census.nuts3
        AND refined.characteristics_code = census.characteristics_code
    """
    )

    np.testing.assert_allclose(
        actual=df_refinement["sum_refined"],
        desired=df_refinement["sum_census"],
        rtol=rtol,
        verbose=False,
    )

    logger.info("All Aggregated household types match at NUTS-3.")


def cts_electricity_demand_share(rtol=1e-5):
    """Sanity check for dataset electricity_demand_timeseries :
    CtsBuildings

    Check sum of aggregated cts electricity demand share which equals to one
    for every substation as the substation profile is linearly disaggregated
    to all buildings."""

    with db.session_scope() as session:
        cells_query = session.query(EgonCtsElectricityDemandBuildingShare)

    df_demand_share = pd.read_sql(
        cells_query.statement, cells_query.session.bind, index_col=None
    )

    np.testing.assert_allclose(
        actual=df_demand_share.groupby(["bus_id", "scenario"])[
            "profile_share"
        ].sum(),
        desired=1,
        rtol=rtol,
        verbose=False,
    )

    logger.info("The aggregated demand shares equal to one!.")


def cts_heat_demand_share(rtol=1e-5):
    """Sanity check for dataset electricity_demand_timeseries
    : CtsBuildings

    Check sum of aggregated cts heat demand share which equals to one
    for every substation as the substation profile is linearly disaggregated
    to all buildings."""

    with db.session_scope() as session:
        cells_query = session.query(EgonCtsHeatDemandBuildingShare)

    df_demand_share = pd.read_sql(
        cells_query.statement, cells_query.session.bind, index_col=None
    )

    np.testing.assert_allclose(
        actual=df_demand_share.groupby(["bus_id", "scenario"])[
            "profile_share"
        ].sum(),
        desired=1,
        rtol=rtol,
        verbose=False,
    )

    logger.info("The aggregated demand shares equal to one!.")


def sanitycheck_emobility_mit():
    """Execute sanity checks for eMobility: motorized individual travel

    Checks data integrity for eGon2035, eGon2035_lowflex and eGon100RE scenario
    using assertions:
      1. Allocated EV numbers and EVs allocated to grid districts
      2. Trip data (original inout data from simBEV)
      3. Model data in eTraGo PF tables (grid.egon_etrago_*)

    Parameters
    ----------
    None

    Returns
    -------
    None
    """

    def check_ev_allocation():
        # Get target number for scenario
        ev_count_target = scenario_variation_parameters["ev_count"]
        print(f"  Target count: {str(ev_count_target)}")

        # Get allocated numbers
        ev_counts_dict = {}
        with db.session_scope() as session:
            for table, level in zip(
                [
                    EgonEvCountMvGridDistrict,
                    EgonEvCountMunicipality,
                    EgonEvCountRegistrationDistrict,
                ],
                ["Grid District", "Municipality", "Registration District"],
            ):
                query = session.query(
                    func.sum(
                        table.bev_mini
                        + table.bev_medium
                        + table.bev_luxury
                        + table.phev_mini
                        + table.phev_medium
                        + table.phev_luxury
                    ).label("ev_count")
                ).filter(
                    table.scenario == scenario_name,
                    table.scenario_variation == scenario_var_name,
                )

                ev_counts = pd.read_sql(
                    query.statement, query.session.bind, index_col=None
                )
                ev_counts_dict[level] = ev_counts.iloc[0].ev_count
                print(
                    f"    Count table: Total count for level {level} "
                    f"(table: {table.__table__}): "
                    f"{str(ev_counts_dict[level])}"
                )

        # Compare with scenario target (only if not in testmode)
        if TESTMODE_OFF:
            for level, count in ev_counts_dict.items():
                np.testing.assert_allclose(
                    count,
                    ev_count_target,
                    rtol=0.0001,
                    err_msg=f"EV numbers in {level} seems to be flawed.",
                )
        else:
            print("    Testmode is on, skipping sanity check...")

        # Get allocated EVs in grid districts
        with db.session_scope() as session:
            query = session.query(
                func.count(EgonEvMvGridDistrict.egon_ev_pool_ev_id).label(
                    "ev_count"
                ),
            ).filter(
                EgonEvMvGridDistrict.scenario == scenario_name,
                EgonEvMvGridDistrict.scenario_variation == scenario_var_name,
            )
        ev_count_alloc = (
            pd.read_sql(query.statement, query.session.bind, index_col=None)
            .iloc[0]
            .ev_count
        )
        print(
            f"    EVs allocated to Grid Districts "
            f"(table: {EgonEvMvGridDistrict.__table__}) total count: "
            f"{str(ev_count_alloc)}"
        )

        # Compare with scenario target (only if not in testmode)
        if TESTMODE_OFF:
            np.testing.assert_allclose(
                ev_count_alloc,
                ev_count_target,
                rtol=0.0001,
                err_msg=(
                    "EV numbers allocated to Grid Districts seems to be flawed."
                ),
            )
        else:
            print("    Testmode is on, skipping sanity check...")

        return ev_count_alloc

    def check_trip_data():
        # Check if trips start at timestep 0 and have a max. of 35040 steps
        # (8760h in 15min steps)
        print("  Checking timeranges...")
        with db.session_scope() as session:
            query = session.query(
                func.count(EgonEvTrip.event_id).label("cnt")
            ).filter(
                or_(
                    and_(
                        EgonEvTrip.park_start > 0,
                        EgonEvTrip.simbev_event_id == 0,
                    ),
                    EgonEvTrip.park_end
                    > (60 / int(meta_run_config.stepsize)) * 8760,
                ),
                EgonEvTrip.scenario == scenario_name,
            )
        invalid_trips = pd.read_sql(
            query.statement, query.session.bind, index_col=None
        )
        np.testing.assert_equal(
            invalid_trips.iloc[0].cnt,
            0,
            err_msg=(
                f"{str(invalid_trips.iloc[0].cnt)} trips in table "
                f"{EgonEvTrip.__table__} have invalid timesteps."
            ),
        )

        # Check if charging demand can be covered by available charging energy
        # while parking
        print("  Compare charging demand with available power...")
        with db.session_scope() as session:
            query = session.query(
                func.count(EgonEvTrip.event_id).label("cnt")
            ).filter(
                func.round(
                    cast(
                        (EgonEvTrip.park_end - EgonEvTrip.park_start + 1)
                        * EgonEvTrip.charging_capacity_nominal
                        * (int(meta_run_config.stepsize) / 60),
                        Numeric,
                    ),
                    3,
                )
                < cast(EgonEvTrip.charging_demand, Numeric),
                EgonEvTrip.scenario == scenario_name,
            )
        invalid_trips = pd.read_sql(
            query.statement, query.session.bind, index_col=None
        )
        np.testing.assert_equal(
            invalid_trips.iloc[0].cnt,
            0,
            err_msg=(
                f"In {str(invalid_trips.iloc[0].cnt)} trips (table: "
                f"{EgonEvTrip.__table__}) the charging demand cannot be "
                f"covered by available charging power."
            ),
        )

    def check_model_data():
        # Check if model components were fully created
        print("  Check if all model components were created...")
        # Get MVGDs which got EV allocated
        with db.session_scope() as session:
            query = (
                session.query(
                    EgonEvMvGridDistrict.bus_id,
                )
                .filter(
                    EgonEvMvGridDistrict.scenario == scenario_name,
                    EgonEvMvGridDistrict.scenario_variation
                    == scenario_var_name,
                )
                .group_by(EgonEvMvGridDistrict.bus_id)
            )
        mvgds_with_ev = (
            pd.read_sql(query.statement, query.session.bind, index_col=None)
            .bus_id.sort_values()
            .to_list()
        )

        # Load model components
        with db.session_scope() as session:
            query = (
                session.query(
                    EgonPfHvLink.bus0.label("mvgd_bus_id"),
                    EgonPfHvLoad.bus.label("emob_bus_id"),
                    EgonPfHvLoad.load_id.label("load_id"),
                    EgonPfHvStore.store_id.label("store_id"),
                )
                .select_from(EgonPfHvLoad, EgonPfHvStore)
                .join(
                    EgonPfHvLoadTimeseries,
                    EgonPfHvLoadTimeseries.load_id == EgonPfHvLoad.load_id,
                )
                .join(
                    EgonPfHvStoreTimeseries,
                    EgonPfHvStoreTimeseries.store_id == EgonPfHvStore.store_id,
                )
                .filter(
                    EgonPfHvLoad.carrier == "land transport EV",
                    EgonPfHvLoad.scn_name == scenario_name,
                    EgonPfHvLoadTimeseries.scn_name == scenario_name,
                    EgonPfHvStore.carrier == "battery storage",
                    EgonPfHvStore.scn_name == scenario_name,
                    EgonPfHvStoreTimeseries.scn_name == scenario_name,
                    EgonPfHvLink.scn_name == scenario_name,
                    EgonPfHvLink.bus1 == EgonPfHvLoad.bus,
                    EgonPfHvLink.bus1 == EgonPfHvStore.bus,
                )
            )
        model_components = pd.read_sql(
            query.statement, query.session.bind, index_col=None
        )

        # Check number of buses with model components connected
        mvgd_buses_with_ev = model_components.loc[
            model_components.mvgd_bus_id.isin(mvgds_with_ev)
        ]
        np.testing.assert_equal(
            len(mvgds_with_ev),
            len(mvgd_buses_with_ev),
            err_msg=(
                f"Number of Grid Districts with connected model components "
                f"({str(len(mvgd_buses_with_ev))} in tables egon_etrago_*) "
                f"differ from number of Grid Districts that got EVs "
                f"allocated ({len(mvgds_with_ev)} in table "
                f"{EgonEvMvGridDistrict.__table__})."
            ),
        )

        # Check if all required components exist (if no id is NaN)
        np.testing.assert_equal(
            model_components.drop_duplicates().isna().any().any(),
            False,
            err_msg=(
                f"Some components are missing (see True values): "
                f"{model_components.drop_duplicates().isna().any()}"
            ),
        )

        # Get all model timeseries
        print("  Loading model timeseries...")
        # Get all model timeseries
        model_ts_dict = {
            "Load": {
                "carrier": "land transport EV",
                "table": EgonPfHvLoad,
                "table_ts": EgonPfHvLoadTimeseries,
                "column_id": "load_id",
                "columns_ts": ["p_set"],
                "ts": None,
            },
            "Link": {
                "carrier": "BEV charger",
                "table": EgonPfHvLink,
                "table_ts": EgonPfHvLinkTimeseries,
                "column_id": "link_id",
                "columns_ts": ["p_max_pu"],
                "ts": None,
            },
            "Store": {
                "carrier": "battery storage",
                "table": EgonPfHvStore,
                "table_ts": EgonPfHvStoreTimeseries,
                "column_id": "store_id",
                "columns_ts": ["e_min_pu", "e_max_pu"],
                "ts": None,
            },
        }

        with db.session_scope() as session:
            for node, attrs in model_ts_dict.items():
                print(f"    Loading {node} timeseries...")
                subquery = (
                    session.query(
                        getattr(attrs["table"], attrs["column_id"])
                    )
                    .filter(attrs["table"].carrier == attrs["carrier"])
                    .filter(attrs["table"].scn_name == scenario_name)
                    .subquery()
                )

                cols = [
                    getattr(attrs["table_ts"], c) for c in attrs["columns_ts"]
                ]
                query = session.query(
                    getattr(attrs["table_ts"], attrs["column_id"]), *cols
                ).filter(
                    getattr(attrs["table_ts"], attrs["column_id"]).in_(
                        subquery
                    ),
                    attrs["table_ts"].scn_name == scenario_name,
                )
                attrs["ts"] = pd.read_sql(
                    query.statement,
                    query.session.bind,
                    index_col=attrs["column_id"],
                )

        # Check if all timeseries have 8760 steps
        print("    Checking timeranges...")
        for node, attrs in model_ts_dict.items():
            for col in attrs["columns_ts"]:
                ts = attrs["ts"]
                invalid_ts = ts.loc[ts[col].apply(lambda _: len(_)) != 8760][
                    col
                ].apply(len)
                np.testing.assert_equal(
                    len(invalid_ts),
                    0,
                    err_msg=(
                        f"{str(len(invalid_ts))} rows in timeseries do not "
                        f"have 8760 timesteps. Table: "
                        f"{attrs['table_ts'].__table__}, Column: {col}, IDs: "
                        f"{str(list(invalid_ts.index))}"
                    ),
                )

        # Compare total energy demand in model with some approximate values
        # (per EV: 14,000 km/a, 0.17 kWh/km)
        print("  Checking energy demand in model...")
        total_energy_model = (
            model_ts_dict["Load"]["ts"].p_set.apply(lambda _: sum(_)).sum()
            / 1e6
        )
        print(f"    Total energy amount in model: {total_energy_model} TWh")
        total_energy_scenario_approx = ev_count_alloc * 14000 * 0.17 / 1e9
        print(
            f"    Total approximated energy amount in scenario: "
            f"{total_energy_scenario_approx} TWh"
        )
        np.testing.assert_allclose(
            total_energy_model,
            total_energy_scenario_approx,
            rtol=0.1,
            err_msg=(
                "The total energy amount in the model deviates heavily "
                "from the approximated value for current scenario."
            ),
        )

        # Compare total storage capacity
        print("  Checking storage capacity...")
        # Load storage capacities from model
        with db.session_scope() as session:
            query = session.query(
                func.sum(EgonPfHvStore.e_nom).label("e_nom")
            ).filter(
                EgonPfHvStore.scn_name == scenario_name,
                EgonPfHvStore.carrier == "battery storage",
            )
        storage_capacity_model = (
            pd.read_sql(
                query.statement, query.session.bind, index_col=None
            ).e_nom.sum()
            / 1e3
        )
        print(
            f"    Total storage capacity ({EgonPfHvStore.__table__}): "
            f"{round(storage_capacity_model, 1)} GWh"
        )

        # Load occurences of each EV
        with db.session_scope() as session:
            query = (
                session.query(
                    EgonEvMvGridDistrict.bus_id,
                    EgonEvPool.type,
                    func.count(EgonEvMvGridDistrict.egon_ev_pool_ev_id).label(
                        "count"
                    ),
                )
                .join(
                    EgonEvPool,
                    EgonEvPool.ev_id
                    == EgonEvMvGridDistrict.egon_ev_pool_ev_id,
                )
                .filter(
                    EgonEvMvGridDistrict.scenario == scenario_name,
                    EgonEvMvGridDistrict.scenario_variation
                    == scenario_var_name,
                    EgonEvPool.scenario == scenario_name,
                )
                .group_by(EgonEvMvGridDistrict.bus_id, EgonEvPool.type)
            )