Aller au contenu

Data quality and confidence

A TESSA study draws on several layers of data — automated building extracts, statistical demand models, user-supplied overrides — and each layer carries a different level of certainty. Understanding where each value came from helps you identify which buildings need manual review and how much to trust the aggregate results.

Origin fields

For the three most important building quantities — Energy Reference Area (ERA), space-heat demand, and hot-water demand — TESSA records a text tag that says how the value was derived. These origin fields are visible in the buildings table export and in the building inspector.

Field Covers
era_origin Where the floor area used in the heat demand calculation came from
qh_origin How the space-heat demand was determined
qww_origin How the domestic hot-water demand was determined
qc_origin How the cooling demand was determined

ERA origin values

ERA (Energy Reference Area) is the heated floor area — the single most important input to the demand calculation. The value TESSA uses is determined by a cascade: the first usable source wins.

Origin tag Meaning
input ERA ERA was provided directly in the input data and was accepted as-is
total_surface * a_s Total building surface was multiplied by the SIA-based heated-fraction coefficient a_s for this building class
residential_surface * a_app Main-residence area was multiplied by the apparent-area coefficient a_app
total_surface Total surface area was used directly (no coefficient available for this building)
total_surface (from factory) ERA was derived from the footprint and number of floors in the source dataset

The further down the cascade, the less certain the ERA. A building tagged total_surface has had no heated-fraction correction, which can overestimate ERA for buildings with large unheated areas (basements, car parks, industrial sheds).

Demand origin values

Origin tag Meaning
AVG Value per SIA and Age Demand was estimated from Swiss SIA 380/1 norms averaged by building class and construction period
SIA Norm DHW demand was set from the SIA norm for this building type
Default Value per Archetype Demand was set from the TABULA archetype default for this building class and construction period

Values derived from archetypes and norms are statistical averages — accurate at the portfolio level but potentially off by ±50% for any individual building.

What the origin fields tell you in practice

flowchart LR MEASURED["Measured / provided<br/>──────────────<br/>era_origin = input ERA<br/>qh_origin = measured data"] MODEL["Statistically modelled<br/>──────────────<br/>era_origin = total_surface * a_s<br/>qh_origin = AVG Value per SIA"] FALLBACK["Fallback / uncertain<br/>──────────────<br/>era_origin = total_surface<br/>qh_origin = Default Value per Archetype"] MEASURED --> |"higher confidence"| MODEL --> |"lower confidence"| FALLBACK classDef high fill:#c8e6c9,stroke:#2e7d32,stroke-width:1.5px,color:#1b5e20 classDef med fill:#fff3e0,stroke:#ED6000,stroke-width:1.5px,color:#bf360c classDef low fill:#ffccbc,stroke:#bf360c,stroke-width:1.5px,color:#bf360c class MEASURED high class MODEL med class FALLBACK low

When reviewing a district:

  • Buildings with era_origin = input ERA and measured demand values have the highest confidence. No review needed unless the values look obviously wrong.
  • Buildings with archetype or norm-derived demand are worth cross-checking against any available metered data for the area, especially for non-residential buildings where the archetypes are less detailed.
  • Buildings where ERA fell back to total_surface should be inspected — their demand may be significantly over- or under-estimated. Consider editing the ERA directly in the building editor or supplying a corrected input file.

Confidence at the portfolio level vs individual buildings

Statistical demand models are calibrated against large datasets. At the district level (hundreds of buildings), individual errors tend to average out, and the total demand is usually accurate within ±10–20%. At the individual building level, errors can be much larger — this is expected and is not a defect in the model.

For network sizing, peak power is more sensitive to errors than annual energy. A building with an overestimated ERA will produce a higher peak demand estimate, leading to a slightly over-sized pipe. In practice, TESSA sizes from the district aggregate peak, so isolated outliers have a limited effect unless they represent a significant fraction of total load.

Improving data quality

To improve confidence for specific buildings:

  1. Edit buildings on the map — use the building editor to correct ERA, construction period, or building type for misclassified buildings.
  2. Upload a buildings table — provide your own ERA and demand values from energy audits or metered data. See Building data template.
  3. Use OSM data — OpenStreetMap can provide footprints and floor counts where the registry data is missing. See the user guide on using OSM data.

After editing, re-run "Generate building demands" to update the demand values and origin tags.

Where to go next