2  Repository

A database is constructed through the python repository GBuildings. It is built with:

• python >=3.9: major release of the Python programming language
• SQLAlchemy: Python library allowing to interact with SQL databases
• geopandas: Python library to manipulate geographic data (geospatial extension of pandas)

The project is composed of a model folder where the database is built and a data analysis folder where additional analysis on the data obtained from the model can be performed.

2.1 Model

The model has a 5 folders:

• Aggregated/Processed/Smoothed: Those three corresponds to the construction of the corresponding database schemes.
• Postprocessing: Contains scripts to perform modifications of the database a posteriori from its construction.
• Scripts: Groups the rest with run examples, useful functions, etc

2.1.1 Aggregated

run.py : main file that uses the following dependencies to work

Settings

Class to read and import database.ini into the class “settings”, for each project

Dictionary
CRS	projection
input_directory	location of the input layers
building	cadastre + 3D + roofs + facades
building_register	RegBL + link table
output_directory	location for export
task	list of the tasks to be done
output_dict	name of the columns for each table, corresponding to the ones of the db
meteo	Contains temperature and irradiance for each hours of the year

Meteonorm

Class to read meteofile as as function of “mask” location

attribute	type
hy	int64
m	int64
dm	int64
h	int64
G_Gh	int64
G_Dh	int64
G_Gk	int64
G_Dk	int64
G_Bn	int64
Ta	float64

Building

Class that import external layers on buildings characteristics (cadastre, building3d, roof, facade) and aggregated them into a ‘footprint’. That footprint is then used as the reference geometry to retrieve every data that falls into it.

buildings
Attribute	Type	Method	Formula	Reference	Comments
area_envelope_m2	float64	Computed	area_envelope_m2 = area_roof_m2 + area_facade_m2 + area_footprint_m2		Computed area_envelope_m2
area_facade_cadastre_m2	float64	Computed	area_facade_cadastre_m2 = length_footprint_m x height_sb3d_m		Computed area_facade_cadastre_m2
area_facade_m2	float64	Computed	if length_facade_m > length_footprint_m (area_facade_m2 = area_facade_cadastre_m2) else area_facade_m2 = area_facade_solar_m2 + (length_footprint_m - length_facade_m) * height_sb3d_m		Computed if length_facade_m > length_footprint_m (area_facade_m2), otherwise calculated differently
area_facade_solar_m2	float64	Exterior data		Sonnendach	Exterior data
area_footprint_cadastre_m2	float64	Exterior data		TLM_footprint	Exterior data
area_roof_m2	float64	Produced	area_roof_m2 = max(area_footprint_m2, area_roof_solar_m2)		Produced area_roof_m2
area_roof_solar_m2	float64	Exterior data		Sonnendach	Exterior data
coord_X	float64	Produced	representative_point()		Produced representative X coordinate
coord_Y	float64	Produced	representative_point()		Produced representative Y coordinate
coord_Z0	float64	Computed	min(z)		Computed minimum Z coordinate
coord_Z1	float64	Computed	mean(z >= z0 + (z2 - z0) / 2)		Computed mean Z coordinate
coord_Z2	float64	Computed	max(z)		Computed maximum Z coordinate
geometry	geometry	Produced	RegBL, SB3D, TLM_footprint	RegBL, SB3D, TLM_footprint	Produced Geometry object, polygons
height_overall_sb3d_m	float64	Produced	height_overall_sb3d_m = coord_Z2 - coord_Z0		Produced maximum height of the building
height_sb3d_m	float64	Computed	height_sb3d_m = coord_Z1 - coord_Z0		Computed mean height of the building
id_building	object	Produced		UUID from Qbuildings by database	Produced UUID from Qbuildings by database
id_building3D	object	Exterior data	SB3D	SB3D	Exterior data
length_facade_m	float64	Exterior data	Sonnendach	Sonnendach	Exterior data
length_footprint_cadastre_m	float64	Exterior data	TLM_footprint	TLM_footprint	Exterior data

Register (RegBL)

Class to read different version of RegBL/RCB and map with a reference database (status, class, period, system, energy source) and integrate an association table (link points with building, district, grid ID). Compute era and add height to building layer.

register class attributes
Attribute	Type	Method	Formula	Reference	Comments
area_era_m2	float64	Mixed	area_era_m2 = 0.93 x area_footprint_m2 x count_floor	RegBL; KHOURY, Jad. Assessment of Geneva multi-family building stock: main characteristics and regression models for energy reference area determination. 2016	Either available in RegBL or estimated from RegBL footprint
area_footprint_m2	int64	Exterior data		RegBL
area_net_floor_m2	float64	Computed	area_net_floor_m2 = area_era_m2 / 1.245	KHOURY, Jad. Assessment of Geneva multi-family building stock: main characteristics and regression models for energy reference area determination. 2016
class	object	Exterior data		RegBL	RegBL building type classification
class_380/1	object	Exterior data		SIA 380/1	SIA building type classification
count_flat	float64	Exterior data		RegBL
count_floor	float64	Exterior data		RegBL
geometry	geometry	Exterior data		RegBL	Point
id_building	object	Produced			UUID from Qbuildings
ID_egid	object	Exterior data		RegBL
is_hotwater_system	bool	Computed		RegBL
period	object	Exterior data			construction period
period_renovation	object	Exterior data
source_heating	object	Exterior data		RegBL
source_hotwater	object	Exterior data		RegBL
standard	object	Produced
status	object	Exterior data		RegBL
system_heating	object	Exterior data		RegBL
system_hotwater	object	Exterior data		RegBL

SIA 380/1

SIA standard that describes the heating, electrical and hotwater needs per building type.

SIA 380/1 class attributes
Attribute	Unit	Type	Method	Formula	Reference	Comments
area_capita_380/1_m2/cap	m2/cap	float64	Exterior data	-	SIA 380/1	Estimated number of occupants for a given building type /m2 of ERA
area_envelope_m2	m2	float64	Computed	-	-	sum of the facades + roofs + footprint area
area_era_m2	m2	float64	Mixed	-	RegBL	-
area_form_factor_380/1	-	float64	Computed	area_form_factor_380/1  = area_envelope_m2 / area_era_m2	SIA 380/1	useful for heating needs computation
capita_cap	cap	float64	Computed	capita_cap = area_era_m2 / area_capita_380/1_m2/cap	SIA 380/1	Number of occupants
capita_presence_380/1_h/d	h/d	float64	Exterior data	-	SIA 380/1	Occupancy for given building type, useful to compute heat gains from people
class_380/1	-	object	Exterior data	-	SIA 380/1	Building type classification in SIA 380/1
electrical_factor_380/1	-	float64	Exterior data	-	SIA 380/1	Factor of heat gain from appliances
energy_el_380/1_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data	-	SIA 380/1	-
energy_el_380/1_kWh/y	kWh/y	float64	Computed	energy_el_380/1_kWh/y  = energy_el_380/1_kWh/(m2.y) * area_era_m2	-	-
energy_gain_cap_380/1_kWh/y	kWh/y	float64	Computed	energy_gain_cap_380/1_kWh/y = power_gain_cap_380/1_W x capita_presence_380/1_h/d *365/1000	-	-
energy_gain_el_380/1_kWh/y	kWh/y	float64	Computed	energy_gain_el_380/1_kWh/y = energy_el_380/1_kWh/y x electrical_factor_380/1	-	-
energy_heating_380/1_target_ratio	-	float64	Exterior data	-	SIA 380/1	Conversion to heating_limit to heating_target
energy_heating_base_380/1_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data	-	SIA 380/1	Base heating needs for a given building type
energy_heating_delta_380/1_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data	-	SIA 380/1	Variation around the base
energy_heating_limit_380/1_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data	-	SIA 380/1	Limit needs that the building should have
energy_heating_target_380/1_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data	-	SIA 380/1	-
energy_hotwater_380/1_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data	-	SIA 380/1	hotwater needed for  a given building type (not used)
flow_fresh_air_380/1_m3/(h.m2)	m3/(h.m2)	float64	Exterior data	-	SIA 380/1	-
id_building	-	object	Produced	-	-	-
ID_egid	-	object	Exterior data	-	RegBL	-
power_gain_cap_380/1_W	W	float64	Exterior data	-	SIA 380/1	-
power_gain_cap_380/1_W/cap	W/cap	float64	Computed	-	-	-
power_gain_cap_380/1_W/m2	W/m2	float64	Computed	-	-	-
standard	-	object	Produced	-	-	tells if the building is new or renovated
temperature_correction_380/1	-	float64	Computed	temperature_correction_380/1_C = 1 + ((9.4 - temperature_exterior_mean_C) * 0.06))	SIA 380/1	temperature used to computed the limit heating value
temperature_exterior_mean_C	C	float64	Exterior data	-	Meteonorm	-
temperature_interior_380/1_C	C	float64	Exterior data	-	SIA 380/1	temperature interior for the given building type

SIA 2024

SIA standard by rooms. Used in GBuildings for day of use and simultaneity but could be used to double check the values.

SIA 2024 class attributes
Attribute	Unit	Type	Method	Formula	Reference	Comments
annual_simultaneity		float64	Exterior data		SIA2024	Use for hotwater
area_capita_2024_m2/cap	m2/cap	float64	Exterior data		SIA2024
area_glass_fraction_%	%	float64	Exterior data		SIA2024	Glass fraction according to the net floor area
area_net_floor_m2	m2	float64	Exterior data		RegBL	reference area for SIA 2024
capita_2024_cap	cap	float64	Computed	capita_2024_cap = area_net_floor_m2 / area_capita_2024_m2/cap	SIA2024
class_380/1	-	object	Exterior data			Use to convert into class 2024
day_of_use_2024_d/y	d/y	float64	Exterior data		SIA2024	Number of day the hotwater is used
energy_el_2024_kWh/(m2.y)	kWh/(m2.y)	float64	Computed			Somme de toutes les demandes électriques par type de bâtiments
energy_el_2024_kWh/y	kWh/y	float64	Computed	energy_el_2024_kWh/y = energy_el_2024_kWh/(y.m2) x area_era_m2
energy_el_ap_2024_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data		SIA2024
energy_el_ap_2024_kWh/y	kWh/y	float64	Exterior data
energy_el_lt_2024_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data
energy_el_lt_2024_kWh/y	kWh/y	float64	Exterior data
energy_el_vt_2024_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data
energy_el_vt_2024_kWh/y	kWh/y	float64	Exterior data
energy_heating_2024_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data		SIA2024	Heating needs
energy_heating_2024_kWh/y	kWh/y	float64	Computed	energy_heating_2024_kWh/y = energy_heating_2024_kWh/(y.m2) x area_era_m2
energy_hotwater_2024_kWh/(m2.y)	kWh/(m2.y)	float64	Exterior data		SIA2024	Hotwater needs
energy_hotwater_2024_kWh/y	kWh/y	float64	Computed	energy_hotwater_2024_kWh/y = energy_hotwater_2024_kWh/(y.m2) x area_era_m2
flow_fresh_air_2024_m3/(h.m2)	m3/(h.m2)	float64	Exterior data		SIA2024
flow_hotwater_2024_l/(cap.d)	l/(cap.d)	float64	Exterior data		SIA2024
flow_infiltration_air_2024_m3/(h.m2)	m3/(h.m2)	float64	Exterior data		SIA2024
flow_infiltration_air_2024_m3/h	m3/h	float64	Exterior data		SIA2024
flow_water_2024_l/(cap.d)	l/(cap.d)	float64	Exterior data		SIA2024
id_building	-	object	Produced
ID_class_2024		object	Exterior data		SIA2024	Id of rooms in the building according to SIA 2024 classification
ID_egid	-	object	Exterior data
power_cooling_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_el_2024_W	W	float64	Computed
power_el_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_el_ap_2024_W	W	float64	Computed
power_el_ap_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_el_lt_2024_W	W	float64	Computed
power_el_lt_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_el_vt_2024_W	W	float64	Computed
power_el_vt_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_gain_ap_2024_W	W	float64	Computed
power_gain_ap_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_gain_cap_2024_W	W	float64	Computed
power_gain_cap_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_gain_intern_2024_W	W	float64	Computed
power_gain_intern_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_gain_lt_2024_W	W	float64	Computed
power_gain_lt_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_guain_extern_2024_W	W	float64	Computed
power_guain_extern_2024_W/m2	W/m2	float64	Exterior data		SIA2024
power_heating_2024_W	W	float64	Computed
power_heating_2024_W/m2	W/m2	float64	Exterior data		SIA2024
solar_gain_factor_2024_-/m2		float64	Computed	solar_gain_factor_2024_-/m2 = area_facade_m2 x area_glass_fraction_fraction_2024_%/100 /area_net_floor_m2 x power_gain_glass_transmission_facotr_2024 x power_gain_solar_reduction_factor_2024	SIA2024	perc of facades wrt to floor area times the glass fraction for this kind of room times the transmission of the glazing (g-value)
standard	-	object	Exterior data
standard_sia2024	-	object	Exterior data

Solar

Class to read the data from the Sonnendach and Sonnenfassade (different versions)

Roofs

Roofs class attributes
Attribute	Unit	Type	Method	Formula	Reference	Comments
area_roof_solar_m2	m2	float64	Exterior data	NA	Sonnendach	Estimated area available
azimuth	degree	float64	Exterior data	NA	Sonnendach	orientation of the roof
egid		int	Exterior data	NA	RegBL	EGID of the building to which the roof belongs
heating_need_kWh_y	kWh/y	float64	Exterior data	NA	Sonnendach	Estimated heating need by EGID, done by the Sonnendach project
hotwater_need_kWh_y	kWh/y	float64	Exterior data	NA	Sonnendach	Estimated hotwater need by EGID, done by the Sonnendach project
id_roof		int	Exterior data	NA	Sonnendach	UUID from the project
lenght_roof_m	m2	float64	Exterior data	NA	Sonnendach	Length from geometry
mean_annual_irr_kWh_m2_y	kWh/(m2.y)	float64	Exterior data	NA	Sonnendach
roof_annual_irr_kWh_y	kWh/y	float64	Exterior data	NA	Sonnendach	Total irradiation received on the available solar area
roof_type		object	Exterior data	NA	Sonnendach
tilt	degree	float64	Exterior data	NA	Sonnendach	slope of the roof
uuid_swissbuildings3d		object	Exterior data	NA	Sonnendach	UUID from SB3D to which the roof belongs

Facades

facades class attributes
Attribute	Unit	Type	Method	Formula	Reference	Comments
area_facade_solar_m2	m2	float64	Exterior data	-	Sonnenfassade	Estimated area available
azimuth	degree	float64	Exterior data	-	Sonnenfassade	orientation of the facade
egid	-	int	Exterior data	-	RegBL	EGID of the building to which the facade belongs
facade_annual_irr_kWh_y	kWh/y	float64	Exterior data	-	Sonnenfassade	Total irradiation received on the available solar area
facade_type	-	object	Exterior data	-	Sonnenfassade	-
heating_need_kWh_y	kWh/y	float64	Exterior data	-	Sonnenfassade	Estimated heating need by EGID, done by the Sonnendach project
hotwater_need_kWh_y	kWh/y	float64	Exterior data	-	Sonnenfassade	Estimated hotwater need by EGID, done by the Sonnendach project
id_facade	-	int	Exterior data	-	Sonnenfassade	UUID from the project
lenght_facade_m	m2	float64	Exterior data	-	Sonnenfassade	Length from geometry
mean_annual_irr_kWh_m2_y	kWh/(m2.y)	float64	Exterior data	-	Sonnenfassade	-
uuid_swissbuildings3d	-	object	Exterior data	-	Sonnenfassade	UUID from SB3D to which the facade belongs

2.1.2 Processed

Settings

Similar to the one from Aggregated but this time, data are retrieved from the database and not external gpks.

Signature model

Use the data from the table “Aggregated”.buildings to compute the energy signature of the building.

signature class attributes
Attribute	Unit	Type	Method	Formula	Reference	Comments
class		object	Produced		Girardin	Building type corresponding to Luc Girardin’s typification, translated from RegBL class
duration_heating_signature_h/y	h/y	float64	Computed	duration_heating_signature_h/y = sum(T_ext -<= temperature_threshold_heating_C)	Count the mb of hours where you have to heat
energy_cooling_signature_kWh/(m2.y)	kWh/(m2.y)	float64	Computed	energy_cooling_signature_kWh/(m2.y) = sum(thermal_k1_cooling_W/(m2.K) x(T_ext - temperature_threshold_cooling_C)	Girardin
energy_gain_intern_2024_kWh/(m2.y)	kWh/(m2.y)	float64	Computed	energy_gain_intern_2024_kWh/(m2.y) = power_gain_intern_W_m2/1000 x duration_heating_signature_h/y
energy_heat_loss_signature_kWh/(m2.y)	kWh/(m2.y)	float64	Computed	energy_heat_loss_signature_kWh/(m2.y) = sum(thermal_k1_heating_W/(m2.y) * (T_ext - temperature_interior_C)
energy_heating_signature_kWh/(m2.y)	kWh/(m2.y)	float64	Computed	energy_heating_signature_kWh/(m2.y) = sum(thermal_k1_heating_W/(m2.K) x(T_ext - temperature_threshold_heating_C)	Girardin	why not temperature_interior_C instead of threshold? Do we assume it is the same?
energy_hotwater_signature_kWh/y	kWh/y	float64	Computed	energy_hotwater_signature_kWh/y = flow_hotwater_l_cap_d x capita_cap x cp x rho x DT x day_of_use x annual_simultaneity	SIA 385/2	Hotwater signature mixing data from SIA 2024, SIA 385/2 and the hottwater temperatures
energy_loss_air_renewal_380/1_m3/(h.m2)	m3/(h.m2)	float64	Computed	energy_loss_air_renewal_380/1_m3/(h.m2) = temperature_interior_C x flow_thermally_active_air_380/1_m3/(h.m2) x 364 x rhocp_a x 24/100	SIA 380/1	Ventilation loss according to formula from SIA 380/1, counting the air renewal
energy_solar_irradiation_signature_kWh/m2	kWh/m2	float64	Computed	energy_solar_irradiation_signature_kWh/m2 = sum(G_gh) when T_ext <= threshold_heating_C	Sum of irradiations when windows are closed
period		object	Produced		Girardin	Building period of construction, translated for Girardin’s typification
power_cooling_signature_W/m2	W/m2	float64	Computed	power_cooling_signature_W/m2 = thermal_k1_cooling_W/(m2.K) x temperature_cooling_dim_C + thermal_k2_cooling_W/m2	Sizing cooling power
power_heating_signature_W/m2	W/m2	float64	Computed	power_heating_signature_W/m2 = thermal_k1_heating_W/(m2.K) x temperature_heating_dim_C + thermal_k2_heating_W/m2	Sizing heating power
power_hotwater_signature_W	kW	float64	Computed	power_hotwater_signature_W = energy_hotwater_signature_kWh/y x 0.0003196456 x 1000	Polysun	Hotwater sizing power, from polysun profiles. The coeff is the sum of the power divided by the max power
rhocp_a	Wh/(m3.K)	float64	Computed	rhocp_a = (1220 - 0.14 x z) / 3600	SIA 380/1	Volumetric thermal capacity
solar_gain_factor_signature_-/m2		float64	Computed	solar_gain_factor_signature_-/m2 = -(energy_heating_signature_kWh/(m2.y) - energy_heat_loss_signature_kWh/(m2.y)) / energy_solar_irradiation_signature_kWh/m2 + energy_gain_intern_2024_kWh/(m2.y)	Part of the irradiation that covers for the heating needs
temperature_cooling_dim_C	C	float64	Mixed	max(T_ext)	Meteonorm	Sizing temperature for cooling technology
temperature_cooling_return_C	C	float64	Produced		Girardin
temperature_cooling_supply_C	C	float64	Produced		Girardin
temperature_heating_dim_C	C	float64	Mixed	min(T_ext)	Meteonorm	Sizing temperature for heating technology
temperature_heating_supply_C	C	float64	Produced		Girardin
temperature_hotwater_return_C	C	float64	Produced		Girardin
temperature_hotwater_supply_C	C	float64	Produced		Girardin
temperature_threshold_cooling_K	C	float64	Produced		Girardin
temperature_threshold_heating_K	C	float64	Produced		Girardin
thermal_k1_cooling_W/(m2.y)	W/(m2.y)	float64	Produced		Girardin	Heat transfer coefficient, for cooling
thermal_k1_heating_W/(m2.y)	W/(m2.y)	float64	Produced		Girardin	Heat transfer coefficient, for heating
thermal_k2_cooling_W/m2	W/m2	float64	Computed	thermal_k2_cooling_W/m2 = -thermal_k1_cooling_W/(m2.K) x temperature_threshold_cooling_C
thermal_k2_heating_W/m2	W/m2	float64	Computed	thermal_k2_heating_W/m2 = -thermal_k1_heating_W/(m2.K) x temperature_threshold_heating_C
waste_kg/(cap.y)	kg/(cap.y)	float64	Produced		Girardin	Solid waste production, only for residential buildings
waste_kg/y	kg/y	float64	Computed	waste_kg/y = waste_kg/(cap.y) x capita_cap
waste_kW	kW	float64	Computed	waste_kW/y = waste_kg/(cap.y) x capita_cap x 16710.7 / 1000 /8760	Girardin	Average power available over the year, from solid waste

2.1.3 Smoothed

This layer is made up to filter the buildings containing incoherent data. There are many ways to define and treat outliers. Here, it is decided to reject first some buildings using ratio between values from different set of data. Then, columns are normalised by the ERA and are considered outliers values that are outside 3 std from the mean column value.

2.2 Data Analysis

For now, only one data analysis is built on GBuildings model, the district_archetypes.

From Terrier, Loustau, and Maréchal (2023)

2.2.1 District archetypes

This project’s goal is to propose a typification of districts, to help obtaining results from REHO at a larger scale. It uses 2 clustering approaches (Kmedoids and GaussianMixture) and gives some tools to compare their results.

The folder is composed of 5 parts:

• clustering_factory: the clustering model.
• data: personal additional data that can be used for the clustering, from some pre-treatments (e.g. the geothermal capacities for each districts in Geneva).
• plotting: library for plots of clustering results.
• postprocessing: methods to evaluates the clustering quality once it has been computed.
• scripts: scripts to run the model or the postprocessing part.

2.2.1.1 Clustering factory

Data recovery

The clustering retrieves the data from GBuildings.

cluster_class = ClusterDistrict(db='Suisse', db_schema="Processed", ...)
cluster_class.connect_database()

One of the tables defines what is a district. By default, it is the transformers table - that is one district corresponds to one low-voltage transformer as estimated by Gupta, Sossan, and Paolone (2020) - but you can add another table. For instance, if one wants to study the archetypes under administrative boundaries in the canton of Geneva, add the geo_girec table, link each building to a geo_girec districts and ask as reference table for the geo girec.

cluster_class.get_database_data(ref_table='geo_girec')

If not all the districts available in the database should be taken in account for the clustering, the user can filter them using any fields of the reference table. For instance, the transformers can be filtered by canton, cities, High-Voltage transformer to which they are connected, etc.

cluster_class.get_database_data(criteria={'id_canton': 2})

Finally, the user can have additional data out of the database.

Data treatment

Data that are not part of the reference districts table can be added. In this case, they are group by districts. Therefore, they should added through a dictionary that precises the name of the field and the method used to agglomerate them.

crit = {'egid': "count", "roof_annual_irr_kWh_y": "sum", "area_roof_solar_m2": "sum"}

Moreover, several methods are available to treat the data before the clustering, detailed in the table hereafter.

Method	Function
filter_3_std	Filters out the districts where value of any numeric attributes > |μ +/-3σ| of the dataset
add_id_class	Adds the building’s class as a criteria for the clustering
normalise_era	Normalises the numeric attributes by the total ERA in the district
split_dataset	Divides the dataset randomly by smaller ones in case the original one is to big
geothermal	Considers the geothermal capacity as an additional criteria (only if previously added in data)

The methods to be used should be passed to the class via a dictionary with the method’s name as the key and a boolean as the value.

methods = {"filter_3_std": True,  "geothermal": False,
           "add_id_class": False, "normalise_era": True,
           "split_dataset": False}

crit = {'egid': "count", "roof_annual_irr_kWh_y": "sum", "area_roof_solar_m2": "sum"}

cluster_class = ClusterDistrict(methods, crit, db='Suisse', db_schema="Processed", ...)

Clustering algorithm

With the data prepared, the clustering can be performed. The user can choose which algorithm should be used (GaussianMixture or Kmedoids) and the number of iterations.

Some methods are also available on how the run should be performed, detailed on the table hereafter.

Method	Function
cmethod	Algorithm used
find_cluster	Does the clustering with asked algorithm and number of iterations and keeps the best one (i.e the most stable)
find_cluster_n	Finds the optimal number of clusters that explains the dataset provided
plotting	Plots the results of the find_cluster_n research
with_export	Saves the plots and the .csv of scores and occurrences
save_clustering	Exports the districts with an additional column cluster

The final call to the function should now be ready.

methods = {"filter_3_std": True,  "geothermal": False,
           "add_id_class": False, "normalise_era": True,
           "split_dataset": False,
           "find_cluster": False, "find_n_cluster": True,
           "plotting": True,      "save_clustering": True,
           "with_export": True,   "cmethod": "kmedoids"}

crit = {'egid': "count", "roof_annual_irr_kWh_y": "sum", "area_roof_solar_m2": "sum"}
cluster_number = 8
iter = 500

cluster_class = ClusterDistrict(methods, crit, iteration=iter, key_clustering_parameter=cluster_number, db='Suisse', db_schema="Processed")
cluster_class.connect_database()
cluster_class.get_database_data(geo_girec=False)
result = cluster_class.run_clustering()

The search for the optimal number of clusters should be conducted with great care as the quality off the clustering depends a lot on it. The plotting is made to help that decision. The method is detailed in Loustau, Terrier, and Maréchal (2023), along with a case study conducted on Geneva.

2.2.1.2 Postprocessing

This part should contain everything that helps analysing the clustering result.

Currently the only script consists in gathering the results from REHO and compare the difference between the mean obtained with every districts and the result from the representative district. An example of this comparison is given in Figure 2.1.

Figure 2.1: Comparison of installed capacity for the energy technologies obtained with the two approaches. All districts stands for the results obtained by optimising every district in the canton of Geneva, while Representative districts is obtained by running the representative districts and scaling their solution to the districts of their cluster. The figures in red indicate the representative difference between the two approaches. Boiler=Natural Gas Boiler, EH=Electrical Heater, HP=Heat Pump, PV=Photovoltaïc Panel, WT=Water Tank

The results should have been computed previously in REHO. This is not a stable script and the paths are yet to be defined properly.

2.2.1.3 Plotting

This folder groups some plotting functions that works with the clustering.

Gupta, Rahul, Fabrizio Sossan, and Mario Paolone, eds. 2020. “Countrywide PV Hosting Capacity and Energy Storage Requirements for Distribution Networks: The Case of Switzerland.” Applied Energy. https://doi.org/10.1016/j.apenergy.2020.116010.

Loustau, Joseph, Cédric Terrier, and François Maréchal. 2023. “Clustering and Typification of Urban Districts for Energy System Modelling.” proceedings.com. 2023. https://www.proceedings.com/069564-0288.html.

Terrier, Cédric, Joseph Loustau, and François Maréchal. 2023. “From Local Energy Communities Towards National Energy System: A Grid-Aware Techno-Economic Analysis.” proceedings.com. 2023. https://www.proceedings.com/069564-0305.html.