Oil Settings

settings reference for the dynamic drainage volume (ddv) model on oil wells every field on the liquid ddv settings input is documented here defaults are tuned to work across a wide range of wells — most wells never need any of these changed, but understanding them helps you diagnose an average reservoir pressure result that looks wrong what this model does the ddv model uses each well's daily rates and pressures to answer two questions how much reservoir volume is this well actually draining , and how has average reservoir pressure declined over time? it does this by matching the well's flowing behavior — fitting the relationship between production and pressure decline, then solving for the drainage volume and pressure history that best explain the observed data the outputs (connected volume, average reservoir pressure over time, remaining pressure support) feed the pi based forecast and downstream optimization the oil variant has two ways to identify the underlying trend, controlled by the first two settings below an automatic identification method that learns the trend shape from the data (the default), and a classical prescribed shape curve fit (the fallback) trend identification use sindy ( usesindy , default on) — with the default, the model identifies the integral rate normalized pressure (irnp) vs material balance time (mbt) flow regime trend directly from the data using an automatic identification method, rather than forcing the history into a prescribed decline shape this adapts better to wells whose behavior genuinely changes over their life turned off, the model falls back to the classical curve fit, and the pressure normalized rate fit settings below take effect example a short history well where the automatic method has too little data to identify a stable trend — switch it off and let the bounded classical fit carry the match use segmentation ( usesegmentation , default off) — when on, the automatic method breaks the history into pieces at detected behavior changes and identifies the trend piecewise useful on wells with distinct eras — a workover, an artificial lift conversion, arrival of offset interference — where one continuous trend visibly fails to describe the whole history left off, the model identifies a single trend across the training window piecewise sindy step ( piecewisesindystep , default 365 days) — the window length used when identifying the rta flow regime trend piecewise on the irnp vs mbt trend shorter windows react faster to behavior changes but amplify noise; longer windows smooth over noise but can blur a real flow regime flow regime changes are not expected to happen quickly, so the default is recommended global match error threshold ( globalmatcherrorthreshold , default 0 05) — the safety net on the piecewise approach if the piecewise match error exceeds this value, the model retries with a single global identification over the whole history and keeps the better answer lower it to fall back to the global approach sooner (stricter on piecewise quality); raise it to tolerate noisier piecewise matches before giving up on them reservoir pressure initial reservoir pressure tolerance ( initialreservoirpressuretolerance , default 0 05) — how far the model may move initial reservoir pressure away from your configured value while matching history, as a decimal fraction (0 05 = 5%) within the tolerance, the solver will trade a small pressure shift for a better overall match example your initial pressure comes from a recent buildup and you trust it — tighten to 0 01 or 0 02 so the match can't drift from it if initial pressure is a rough regional estimate, the default or wider lets the data speak reservoir pressure fit settings ( reservoirpressurefitsettings ) — starting guess and allowed range (min / base / max for magnitude, early decline, and curvature, plus a hyperbolic to exponential switch) for the fitted average reservoir pressure trend all fields default to 0, which means unset the model picks its own starting point from the data only set these when the automatic pressure trend converges somewhere unphysical relative to what you know from offsets or surveillance saturation bounds water saturation min / max ( watersaturationmin / watersaturationmax , defaults 0 1 / 0 9) — the range the solver may consider for water saturation when resolving the drainage volume the defaults are deliberately wide, excluding only the non physical extremes example petrophysics puts water saturation at 25–35% in this zone narrowing to 0 2–0 4 removes a family of wrong but well fitting solutions and typically makes the volume answer both faster and more repeatable well to well data quality minimum average drawdown ( minaveragedrawdown , default 100 psi) — the smallest average drawdown the solver requires before attempting a solution a well flowing with almost no drawdown carries almost no pressure signal, and a drainage volume resolved from that data would be noise dressed up as an answer; this guard makes the model decline to answer instead example a genuinely low drawdown well (strong aquifer support, heavily choked) that you still want a result for — lower the guard deliberately, and treat the resulting volume with matching skepticism fit tuning pressure normalized rate fit settings ( pnrfitsettings ) — the bounds and starting guesses for the classical curve fit relating the well's rate response to pressure only used when usesindy is off each parameter has a min / base (starting guess) / max field what it bounds yimin / yibase / yimax magnitude of the fitted trend at time zero logdimin / logdibase / logdimax how quickly the trend changes early in the history bmin / bbase / bmax curvature of the trend over time switchpercent decline level at which the fitted trend transitions from hyperbolic to exponential behavior (0 = no forced transition) the base values are where the solver starts; the min/max are hard walls it cannot cross if a fitted parameter lands exactly on a bound, that bound is binding — widen that one wall slightly and re run rather than moving several at once caution the defaults in this group have been finely tuned against a large well population treat them as last resort levers for an individual problem well, not as field wide knobs