Reference Data
Don’t start from a blank sheet — someone has usually modelled the adjacent process. Find them here.
Under review. Entries are catalogued, not yet individually assessed. Treat this as a starting point for your own search, not a vetted shortlist. Sector tags are assigned by a classifier from each source’s own labels — good enough to browse by, not to count with.
How it was built. Systematic search across labs, agencies, journals and code repositories, seeded and reviewed with TEA Commons partners and experts. Every link was fetched before publication; paywalled and by-request resources were excluded.
Know a model we’re missing? Send it to hello@tea-commons.org — a landing URL is enough.
Links only. Every entry points to a publicly accessible page on the publisher’s own site — nothing is rehosted here. We have not cleared linking rights entry by entry; if you are a publisher and want an entry changed or removed, tell us at hello@tea-commons.org and we will.
Cataloguing is not endorsement. Inclusion means the artifact exists, is reachable, and estimates cost. Check base year, currency, scope and financing assumptions before reusing anything — most of these disagree with each other. Escalate to a common base before comparing (cost indices).
Formula density measures liveness, not quality. formulas ÷ cells, measured by opening each workbook — it separates a live model from a published results table. A 60%-formula workbook can be a badly-parameterised mess; a 2% one can be a clean lookup over a solid dataset. It tells you whether inputs propagate, nothing more.
34 entries sit behind a free registration wall (BatPaC, H2FAST, RETScreen and similar) — flagged in the row.