This patch release improves the correctness, transparency, and validation of consequential marginal-market modelling.
Highlights
-
lead time=False now explicitly uses the production-weighted market-average lead time.
-
lead time=True now applies technology-specific observation intervals under myopic modelling.
-
Perfect-foresight intervals remain anchored to the demand year.
-
Out-of-range observation years are mapped individually to the nearest available IAM year, with a warning.
-
Invalid combinations of range time, duration, foresight, lead-time mode, measurement method, and weighted-slope parameters now raise clear errors.
-
The area-based capital-replacement calculation and legacy zero-duration interval were corrected.
-
Constrained suppliers are now matched by exact name, preventing diesel and generic liquefied petroleum gas from being excluded accidentally.
Documentation and validation
The consequential documentation now explains:
- short and long demand-change intervals;
- average versus technology-specific lead times;
- myopic versus perfect-foresight modelling;
- measurement methods 0–5;
- legacy fallback behaviour and argument validation;
- effective intervals reported during database generation.
A reproducible validation compares Premise against an independent calculation for 11 IMAGE SSP1-M/WEU electricity configurations. All six measurement methods and both expanding and declining market branches matched exactly.
Important for consequential users
Results may change if you previously used:
- lead time=True;
- area-based measurement method 2;
- marginal diesel or LPG markets;
- observation intervals approaching the IAM time-series boundaries.
We recommend rebuilding consequential databases generated with earlier releases.
Installation
pip install --upgrade premise==2.4.9.1Full details are available in the changelog.