What problem does it solve? Reservoir and process engineers often need a fluid model before any laboratory PVT study exists, or need to adapt one common reference fluid to many wells and fields. This Skill provides dependency-free Python helpers to calibrate a heavy-end split factor against measured data, generate representative fluid cases, blend well streams into a field composition, and build a documented best-guess fluid basis from analogues when no PVT report exists. ## Core Features & Use Cases - Split-factor calibration: match_split_factor runs a golden-section search over a caller-injected forward model to match measured saturation pressure, GOR, or stock-tank density. - Case generation and blending: generate_fluid_cases builds low/base/high fluids from a reference, and blend_compositions combines well streams by molar-rate allocation into one field composition. - No-PVT fluid basis: build_analogue_fluid_basis derives reservoir temperature from geothermal gradients, classifies fluid type from GOR/API bands, interpolates analogue depth trends, and produces an assumption register with provenance and confidence grading. - Use Case: For a North Sea discovery at 3590 m TVDSS with no samples, build a declared best-guess fluid basis with low/base/high GOR and gravity cases, then tune a NeqSim EOS characterization to the base case. ## Quick Start Ask the agent to build an analogue fluid basis for a discovery at a given depth and province, then calibrate a split factor against a measured saturation pressure using the reference_fluid helpers.
