Decline-Curve Analysis (Arps)

Once a well is on production its rate falls, and decline-curve analysis fits that historical trend and extrapolates it to forecast future rate and estimated ultimate recovery. Arps' three classic forms - exponential, hyperbolic, and harmonic - describe how the decline behaves.

Petroleum EngineeringReserves & EconomicsFree preview
⏱️ About 18 min
Decline-Curve Analysis (Arps) — illustration
Illustrative image (AI-generated).

Once a well is producing and its rate is falling, decline-curve analysis fits the historical trend and extrapolates it to forecast future rate and ultimate recovery.

💡
The big idea: Decline-curve analysis (DCA) fits a well's historical rate-versus-time data and extrapolates the trend to forecast future rate and estimated ultimate recovery (EUR); Arps' three forms - exponential (b = 0, constant fractional decline, a straight line on semi-log axes), hyperbolic (b between 0 and 1), and harmonic (b = 1) - differ in how the fractional decline behaves, and DCA is empirical, assuming past operating conditions continue.
🎯 By the end, you'll be able to
  • Define decline-curve analysis (DCA) as fitting a well's historical rate-versus-time data and extrapolating the trend to forecast future rate and estimated ultimate recovery (EUR)
  • Name Arps' three decline forms - exponential (b = 0, constant fractional decline), hyperbolic (b between 0 and 1), and harmonic (b = 1) - and how they differ
  • Explain why exponential decline plots as a straight line on a semi-log rate-versus-time graph and why hyperbolic and harmonic forms forecast larger ultimate recoveries
  • State that DCA is empirical and assumes past operating conditions continue, so a change in drive mechanism, completion, or operating practice breaks the forecast
📎 Helpful to know first
  • Flow Assurance: Hydrates, Wax & Scale

What Decline-Curve Analysis Is

Once a well is on production its rate almost always falls over time as reservoir pressure depletes and the easiest oil is produced first. Decline-curve analysis (DCA) is the engineer's simplest forecasting tool: it fits a smooth curve to the well's historical rate-versus-time data and extrapolates that trend to predict future production and the estimated ultimate recovery (EUR) - the total barrels the well will produce over its life. The method needs only production history, no reservoir properties, which is why it is the first forecast an engineer reaches for. Its power is its simplicity; its weakness is that it assumes the past operating conditions continue - change the choke, start a waterflood, or stimulate the well, and the old decline trend no longer applies.

Arps' Three Decline Forms

J. J. Arps organised production decline into three classic forms, distinguished by how the fractional (percentage) decline rate behaves as the rate itself falls. The key parameter is the decline exponent $b$. In exponential decline ($b = 0$) the fractional decline is constant, so the rate falls by the same percentage each year and the rate-time curve is a straight line on a semi-log plot: $q(t) = q_i\,e^{-D t}$, where $q_i$ is the initial rate and $D$ is the nominal decline. In harmonic decline ($b = 1$) the fractional decline shrinks as the rate falls, so the curve flattens gradually. In hyperbolic decline (with $b$ between 0 and 1) the behaviour sits between the two - the decline slows over time but not as slowly as harmonic. Because hyperbolic and harmonic declines flatten out, they forecast larger ultimate recoveries than the steeper exponential form fitted to the same early data.

✨ DCA Is Empirical - Mind the Assumptions

DCA is an empirical curve-fit, not a reservoir model built from the flow equations - it does not ask why the rate is falling, it only assumes the fitted trend will carry forward. That makes it fast and transparent, but it demands judgement. The forecast only holds while the drive mechanism, completion, and operating practice stay the same: a well on natural depletion declines one way, the same well under a waterflood declines another, and a frac job reshapes the curve entirely. Extrapolating too far beyond the data, or across a change in mechanism, is the classic way a decline forecast goes wrong. Engineers use DCA as a quick, repeatable first estimate, then cross-check it against material balance and reservoir simulation when the stakes are higher.

📝 Worked example: A well begins production at an initial rate $q_i = 1{,}000$ STB/d and declines exponentially with a nominal decline $D = 0.10$ per year, so $q(t) = 1000\,e^{-0.10\,t}$. (a) Find the rate after one year. (b) Find the time to reach an economic-limit rate of $q_a = 100$ STB/d (the rate below which the well no longer pays to operate).
  1. (a) After one year, $t = 1$: $q = 1000\,e^{-0.10} = 1000 \times 0.905 = 905$ STB/d.
  2. The effective annual decline is therefore $1 - 0.905 = 0.095$, i.e. about $9.5\%$ per year.
  3. (b) Set $q(t) = q_a$ and solve for $t$: $100 = 1000\,e^{-0.10\,t}$, so $e^{-0.10\,t} = 0.1$.
  4. Take natural logs: $-0.10\,t = \ln(0.1) = -\ln 10$, so $t = \dfrac{\ln(q_i / q_a)}{D} = \dfrac{\ln(1000/100)}{0.10} = \dfrac{\ln 10}{0.10}$.
  5. With $\ln 10 = 2.303$: $t = \dfrac{2.303}{0.10} \approx 23$ years to reach the economic limit.
✓ (a) The rate after one year is about $905$ STB/d (an effective decline of roughly $9.5\%$ in the first year). (b) The well reaches the $100$ STB/d economic limit after about $t \approx 23$ years of exponential decline.

Check your understanding

1. Using exponential decline with an initial rate of 1,000 STB/d and a nominal decline of 0.10 per year, what is the production rate after one year?
After one year, $q = 1000\,e^{-0.10} = 1000 \times 0.905 \approx 905$ STB/d.
2. About how long does it take this well to reach an economic-limit rate of 100 STB/d?
Setting $100 = 1000\,e^{-0.10\,t}$ gives $t = \dfrac{\ln(q_i/q_a)}{D} = \dfrac{\ln 10}{0.10} = \dfrac{2.303}{0.10} \approx 23$ years.
✅ Key takeaways
  • Decline-curve analysis (DCA) fits a well's historical rate-versus-time trend and extrapolates it to forecast future rate and estimated ultimate recovery (EUR).
  • Arps' three forms are exponential (b = 0, constant fractional decline, a straight line on semi-log axes), harmonic (b = 1), and hyperbolic (b between 0 and 1); hyperbolic and harmonic flatten and forecast larger recoveries.
  • DCA is empirical - it assumes past operating conditions continue, so it breaks across changes in drive mechanism, completion, or operating practice.
➡️ Fitting the decline forecasts how much the well will produce - but how are those volumes described and graded for certainty? The next lesson surveys how the industry classifies reserves and resources.