Background data selection with ecoinvent 3.12: activity, geography and electricity mix

Most of your LCA result comes from processes you never measured: the production of your raw material, the electricity grid, the truck's diesel chain. This background data comes from ecoinvent 3.12 — over 25,000 activities across hundreds of geographies. The richness is also the risk: picking the wrong activity or geography can move a GWP-fossil result by 30-50%. This post covers the three critical selection decisions.

Decision 1: market or transforming activity?

ecoinvent has two activity types per product. The transforming activity is the production process itself ("portland cement production"); the market activity is the regional supply mix plus average transport ("market for portland cement"). The rule is simple: if you are not modelling your supplier and its transport explicitly, use the market activity — while avoiding double-counted transport. When you enter supplier-specific data (e.g. from the supplier's own EPD), you drop to the transforming activity and add transport yourself.

Decision 2: the geography hierarchy

Each activity exists in specific geographies: country codes (TR, DE), regions (RER — Europe), RoW (Rest-of-World) and GLO (global). The selection rule is the geography closest to your production site; if your country is missing, step up the hierarchy. A caution: RoW is "the rest of the world" and is often more representative than GLO for non-European production; using RER for Turkish production produces a systematic error because of the electricity mix difference.

• Production country first (TR, DE, CN...) — always preferred when available
• Then region (RER, RAS...) — if the production site is inside it
• If the region is not representative, RoW — consider before GLO for non-European production
• GLO — last resort; a global average that can mislead for electricity-intensive products

Decision 3: the electricity mix

Electricity is the single largest background contribution in most manufacturing processes. clca ships electricity country mixes: pick your plant's country and the grid mix binds to the correct activity in the background. If you hold a green power contract (PPA, I-REC) you can define a custom mix — but EN 15804 verification requires contractual evidence; a bare "we buy green tariff" claim is not enough.

System model: cut-off

ecoinvent publishes three system models; EN 15804 practice uses the cut-off model: recycled content arrives burden-free, and the benefit of material sent to recycling is reported separately in module D. Results computed with APOS or the consequential model are not comparable with an EN 15804 EPD — always confirm which system model your dataset comes from.