Maintenance priorities and modelling triggers

This page records implementation priorities for maintainers. The caveats for users are in limitations and open questions. New physics or a new site needs a separate specification.

The deferred work has explicit triggers:

Deferred item	Trigger for bringing it back
Programme schema and full cache invalidation, P-01/P-02	Repeated runs cannot be traced with the release JSON, or an external user requests it
Common event timeline, H-01	A4 or A5 changes the decision, or a lever is within 0.02 SG or 5 °C of a threshold
Coolant advection in the string, H-02	A3 straddles tool survival at a site whose verdict depends on it
Fluid displacement, W-03	A6 straddles the site verdict
Full flow and stress envelopes, D-03	A lever reopens a site and an operating range must be claimed
Tied bound labels: report every tied case in lo_by and hi_by, keeping case as the first, and have describe_bound() print "all cases" when every case ties	The next release that regenerates the evidence
Near-bit thermal model, growth law and thermo-poroelasticity	Specialist interest, usable observations or a clearly discriminating hypothesis justify a separately scoped study; the breakout formation-delay experiment becoming decision-relevant is another trigger
Report the frequency and extent of water-property lookups outside the cached grid	An evaluation approaches the grid limits, or a user needs evidence of property-domain coverage; requires separate instrumentation work

H-01 trigger met; not yet implemented. Soultz L2's 0.00137 SG trip width and its L2+L4 pair's 0.00949 SG drilling deficit meet the proximity trigger. The evaluated programmes still fail other gates, and no operating range is claimed. Revisit before claiming a near-threshold programme is feasible, or when new data make the timing approximation decision-relevant. Coolant delivery (H-02) and fluid displacement (W-03) retain their distinct triggers above.

Near-bit pre-cooling and the FORGE image-log follow-up can be justified by specialist interest, usable observations or a discriminating hypothesis.

FORGE image-log follow-up

The 16B image-log inventory identifies repeated coverage suitable for investigating breakout-width changes. The proposed follow-up is to obtain matched image intervals, establish logging direction and interpretation provenance, and compare widths at the same depths while accounting for the different circulation histories. Much of the overlap is in deviated hole, so predicting widths would require the corresponding stress calculation. The inventory is complete; the paired-width analysis has not been performed. This work requires a separate specification.