Late-Time Transient Data

Key Takeaways

• The sequence of flow regimes in a pressure transient test progresses from early time through middle time to late time, each identifiable on the log-log diagnostic plot (log of pressure change and log of pressure derivative plotted against log of elapsed time): the early-time wellbore storage period (where the compressibility of wellbore fluid dominates the pressure response) appears as a unit slope (45-degree line) on the log-log plot; the middle-time infinite-acting radial flow period (where the pressure disturbance propagates radially outward without reaching any boundary) appears as a flat derivative on the log-log plot (the "Bourdet derivative" stabilization) and a straight line on the semi-log Horner or MDH plot with slope m = 162.6 * q * mu * B / (k * h) that gives permeability-thickness product; the late-time boundary effects then deviate the pressure derivative from the flat stabilization in characteristic patterns depending on the boundary type -- a sealing fault causes the derivative to double (derivative rises from one stabilization level to twice that level, indicating that only half of the reservoir is infinite-acting in the direction of the fault), a closed boundary causes the derivative to rise above the radial flow level on a unit slope (pseudo-steady-state), and a constant-pressure boundary (aquifer, gas cap contact, or injector support) causes the derivative to drop below the radial flow stabilization.
• Sealing fault detection from late-time transient data is one of the most important applications of boundary analysis in pressure transient interpretation: when a single sealing fault lies at a distance L from the well, the pressure behavior transitions from infinite-acting radial flow to a "hemi-radial" or "channel" flow regime in which the effective drainage area is reflected across the fault plane, creating a doubled pressure derivative characteristic; the time at which the fault effect first appears on the log-log diagnostic plot (the onset of the derivative doubling) is approximately t = 1688 * phi * mu * ct * L^2 / k (in hours, field units), from which the fault distance L can be calculated if the reservoir properties are known; for wells near two parallel faults (a channel reservoir), the derivative first doubles as the closer fault is reached, then doubles again (to four times the radial flow value) when the farther fault is reached, producing a two-step derivative doubling signature characteristic of a channel bounded by two parallel sealing faults; the ability to detect and locate faults from late-time pressure transient data without drilling additional wells or acquiring additional seismic makes late-time analysis a powerful tool for understanding reservoir compartmentalization that might otherwise only be revealed by well interference tests or production history mismatch with simulation models.
• Closed reservoir boundary effects in late-time pressure buildup data are characterized by the pseudo-steady-state flow regime, in which the average reservoir pressure declines linearly with cumulative production as all boundaries have been reached; the transition to pseudo-steady-state appears on the log-log diagnostic plot as a unit slope derivative increase (indicating that the pressure drop per unit log time is proportional to the rate of depletion of the closed reservoir volume) and on the Cartesian pressure versus time plot as a straight line with slope m* = -0.2339 * q * B / (Vp * ct) (in field units) from which the drainage pore volume (Vp) can be calculated; the pore volume calculated from the late-time pseudo-steady-state slope provides an independent estimate of the reservoir volume that can be compared to the volumetric reserve estimate from geological and petrophysical data, with agreement between the dynamic (pressure-derived) and static (volumetric) estimates providing confidence in the reserve assessment and with significant disagreements suggesting that either the volumetric calculation or the pressure transient interpretation contains an error; well-to-well interference and multiple well interactions in a field context complicate the single-well pseudo-steady-state analysis by spreading the pressure transient signal across multiple drainage areas, requiring simulation-assisted history matching to decompose the individual well contributions.
• Constant-pressure boundary effects from aquifer support, gas cap expansion, or producer-injector well pairs appear in late-time pressure data as a derivative that declines below the radial flow stabilization level, converging toward zero derivative (pressure stabilization) at late time; constant-pressure boundaries maintain a fixed pressure at the boundary (approximately the initial reservoir pressure for an active aquifer, or the injection wellbore pressure for a connected injector) that limits the pressure decline at the producing well during drawdown and provides pressure support during buildup; the time at which the constant-pressure boundary effect begins to suppress the derivative provides an estimate of the boundary distance using the same relationship as for sealing faults (but with different flow geometry), and the shape of the derivative decline after boundary contact characterizes whether the boundary is a line source (injector well), a planar source (fault-limited aquifer), or a volumetric source (large connected aquifer system); identifying constant-pressure boundary support from late-time pressure data has important implications for reserves (a well supported by a large aquifer will recover a larger fraction of OOIP than a solution-gas-drive well with no aquifer support) and for enhanced recovery design (injector placement can be optimized based on the inferred boundary geometry).
• Test duration requirements for capturing late-time boundary effects are significantly longer than for the middle-time radial flow analysis, since the pressure disturbance must propagate from the wellbore to the boundary and return to the well as a reflected wave before the boundary effect appears in the pressure record; for a 10-millidarcy formation with porosity 0.15 and total compressibility 10 x 10^-6 psi^-1, a sealing fault at 500 meters from the well first affects the pressure data at approximately 24 hours, while a fault at 2,000 meters requires approximately 380 hours to appear; the implication is that late-time boundary analysis requires extended test durations (several days to weeks for low-permeability formations or distant boundaries) that may not be economically feasible for every well test but should be planned for key appraisal wells or well tests designed specifically to characterize the reservoir boundary geometry; drillstem tests (DSTs) with typical flow periods of 4 to 12 hours rarely capture late-time boundary effects in anything other than very small or very permeable reservoirs, while production well tests with flow periods measured in weeks or months routinely sample the late-time boundary response; the advent of high-resolution downhole pressure gauges with sub-psi resolution and digital recording has extended the interpretable late-time signal by allowing very small pressure changes from distant boundaries to be detected above the gauge noise floor.