Data Sources & Licensing
Every value in the cost database traces to one of eleven registered sources. The registry itself is queryable (GET /api/tea/sources) and license discipline is enforced by tests: no proprietary cost feed or licensed index series can be stored.
The licensing rule
Cost engineering data is a licensing minefield: the standard textbooks are copyrighted, CEPCI is a paid Chemical Engineering feature, and commercial price feeds (Richardson, ICIS, IHS) cannot be redistributed. Jasper's rule:
Coefficients enter the database only via open-source implementations or public government datasets, and every source's license is recorded.
This is why some source names look indirect — the provenance chain is the point.
All eleven sources
| Source | What it provides | License |
|---|---|---|
turton | Turton et al. purchased-cost coefficients, bare-module factors, material factors | published coefficients via an MIT-licensed port (see below) |
idaes-sslw | Seider/SSLW correlations from IDAES-PSE (SSLW.py) | BSD-3 (US DOE labs) |
bls-ppi-ces | The Jasper Cost Index inputs: metals PPI, machinery PPI, construction wages | US Gov public domain |
eia-public | Utility prices: electricity, natural gas; two-factor steam | US Gov public domain |
literature-prices | Representative chemical prices from published TEAs | flagged representative |
eurostat-ecb | EUR/USD reference rate, EU labor cost index | Eurostat / ECB public |
us-irs-statutory | US federal corporate tax rate (21%) | US Gov public domain |
turton-grassroots-contingency | Contingency by AACE class | concept attribution |
jasper-base | US-GC base-region definition (all drivers ≡ 1.0) | internal, definitional |
location-us-mw | US Midwest location drivers | flagged representative |
jasper-mapping | Unit-type → correlation mapping (30 rows) | internal, documented |
About turton
Turton et al., Analysis, Synthesis and Design of Chemical Processes (4th ed., Appendix A) is the de-facto standard source for module-costing coefficients — but the book is copyrighted, so Jasper does not transcribe it. The coefficients were ingested from an MIT-licensed open-source implementation that republishes them; the registry records the citation:
Turton et al., Analysis, Synthesis and Design of Chemical Processes, 4th ed., App. A.
The same pattern applies to idaes-sslw: Seider et al.'s correlations arrive via the BSD-licensed IDAES-PSE codebase, not the textbook.
What is deliberately absent
- CEPCI values — never stored. The Jasper Cost Index replaces it from public BLS data and is validated against it only in tests.
- Richardson / ICIS / IHS or any commercial feed — a test fails if such a source is ever registered.
- Invented numbers — equipment without an open-licensed correlation is gap-logged rather than given a made-up cost.
Representative vs. measured
Two source classes are honest approximations and flagged as such everywhere they're used:
literature-prices— chemical prices are representative published values, not live quotes. Estimates that use them say so, and per-project overrides replace them with your contract prices.location-us-mw— driver values are representative differentials vs. US-GC; the model (composed drivers) is real, the magnitudes are study-grade.