Decline curve analysis software

What decline curve analysis is

Decline curve analysis (DCA) is the oldest and most widely used production forecasting method in upstream oil and gas. It treats the production rate of a well as a curve that declines with time in a characteristic way, fits that curve to the observed history, and reads the future rate off the extrapolation.

Its strength is that it needs nothing but production history — no reservoir model, no simulation grid, no petrophysics. Its weakness follows directly: it is empirical. A decline fit describes what a well has been doing under the operating conditions it experienced. It does not know that a workover is planned, that a compressor is being installed, or that a field is about to be shut in.

That is why the fit has to remain inspectable. The parameters are the engineer's handle on an otherwise blind extrapolation.

The Arps family of curves

All three classical Arps curves are governed by a single exponent, the b-factor, which controls how fast the decline rate itself decays.

Exponential (b = 0)

The decline rate stays constant. The most conservative of the three, and the usual choice for the late-life tail of a well.

Hyperbolic (0 < b < 1)

The decline rate decreases over time, so the curve flattens as the well ages. This is the general case that most producing wells are fitted with.

Harmonic (b = 1)

The limiting case of hyperbolic decline, with the slowest flattening and therefore the largest extrapolated volumes.

Why automated DCA needs a terminal decline

Unconstrained hyperbolic decline flattens forever, and integrating it to infinite time produces an unbounded cumulative volume. That is not a modelling nicety — it is how a batch job silently generates reserves that cannot exist.

The Modified Arps model solves it by imposing a terminal decline: once the hyperbolic decline rate falls to a defined minimum, the forecast switches to exponential decline at that rate for the remainder of the well's life. PronoHorizon uses this model, and the terminal decline is one of the three parameters exposed on every well.

What changes when DCA runs at portfolio scale

Fitting one well is an afternoon. Fitting several hundred introduces problems that single-well workflows never have to solve.

• Consistency: every well is fitted with the same declared method, so differences between wells are reservoir behaviour rather than analyst habit.
• Triage: each well carries a review status, so attention goes to the fits that need judgment instead of being spread evenly across the portfolio.
• Measurement: error is scored on a holdout window across the portfolio, which exposes systematic optimism that any single well would hide.
• Data reality: missing production months, allocated or estimated volumes and days-on counts stay marked rather than being interpolated into a clean-looking curve.

When to override the automatic fit

An automatic fit is a proposal, not a verdict. Overriding is the correct response when the history contains information the curve cannot represent: a rate change caused by a choke adjustment or well service rather than reservoir depletion, a period of allocated volumes after a facility restart, a shut-in that the fit reads as a steepening decline, or planned future work that changes the operating basis entirely.

In PronoHorizon the override is part of the record. The assumptions and engineer edits stay attached to the reviewed case, so the next person can see not only that a curve was changed but on what basis.