Reservoir Engineering: Rock and Fluid Properties, Flow in Porous Media, Drive Mechanisms, Material Balance, Reserves, Geomechanics and Simulation

Porosity φ is pore volume over bulk volume; total porosity counts all pores, effective porosity only the interconnected ones that can deliver fluid. Saturation Sₒ, S_w, S_g is the fraction of pore volume each fluid fills, and they sum to one; the connate or irreducible water saturation S_wi is the water that cannot be displaced. Permeability k is defined by Darcy’s law for the laminar flow of a single fluid that fills the pores: in darcy units q = kAΔp/(μL), with q in cm³/s, k in darcy, A in cm², Δp in atm, μ in cP and L in cm; in oilfield units q = 0.001127kAΔp/(μL), with q in bbl/d, k in md, A in ft², Δp in psi and L in ft. Gas permeability measured at low pressure is higher than the liquid value because gas molecules slip at the pore walls (the Klinkenberg effect); extrapolating k_gas against 1/p̄ to infinite pressure gives the equivalent liquid permeability.

Layers in parallel (flow along the beds) average arithmetically by thickness, k̄ = Σkᵢhᵢ/Σhᵢ; beds in series (flow across them) average harmonically, k̄ = L/Σ(Lᵢ/kᵢ), so the tightest bed dominates. Wettability is the tendency of one fluid to spread on the rock in the presence of another, measured by the contact angle or by the Amott and USBM tests; most sandstones are water-wet and many carbonates oil-wet or mixed-wet. Capillary pressure is the pressure difference across a curved interface, Pc = p_non-wetting − p_wetting = 2σ cos θ/r for a capillary of radius r, and in a reservoir it equals Δρ g h above the free-water level, which is why saturation varies through the transition zone. Relative permeability k_r is the fraction of absolute permeability available to a phase when several flow together; each curve ends at its residual (immobile) saturation, and the curves cross at a water saturation above 0.5 in a water-wet rock.

Coring recovers rock for direct measurement: conventional cores cut with a core barrel and annular core bit give continuous full-diameter samples; sidewall cores (percussion or rotary) sample chosen depths after logging. Routine core analysis measures porosity (helium porosimeter by Boyle’s law, or summation of fluids), air permeability (corrected for Klinkenberg slip) and fluid saturations (Dean-Stark distillation or retort). Special core analysis (SCAL) measures capillary pressure (porous-plate, centrifuge, mercury injection), relative permeability (steady-state and unsteady-state displacements), wettability, rock compressibility and electrical properties for Archie’s constants. Rock compressibility c_f = (1/V_p)(dV_p/dp) and the fluid compressibilities combine into the total compressibility cₜ = Sₒcₒ + S_wc_w + S_gc_g + c_f that appears in every transient equation.

Oil is described at stock-tank conditions by its API gravity, °API = 141.5/SG − 131.5, where SG is the specific gravity at 60 °F; water is 10 °API and lighter oils have higher values. In the reservoir, oil carries dissolved gas: the solution gas-oil ratio Rs (scf/STB) rises with pressure up to the bubble-point pressure p_b and is constant above it. The oil formation volume factor Bo (reservoir bbl/STB) is the reservoir volume of oil plus its dissolved gas per stock-tank barrel; it rises slightly as pressure falls toward p_b (the liquid expands) and then falls below p_b as gas comes out of solution. The gas formation volume factor from the real-gas law is Bg = 0.02827zT/p ft³/scf with T in °R and p in psia (equivalently 0.00504zT/p bbl/scf), where z is the gas compressibility factor read from the Standing-Katz chart at the pseudo-reduced pressure and temperature. Oil viscosity falls with pressure down to p_b as gas dissolves and then rises below it.

The pressure-temperature phase envelope of a reservoir fluid is bounded by the bubble-point curve and the dew-point curve, which meet at the critical point. Its highest pressure is the cricondenbar and its highest temperature the cricondentherm. Where the reservoir temperature falls relative to the envelope classifies the fluid: to the left of the critical point it is oil (black oil far to the left, volatile oil close to it); between the critical temperature and the cricondentherm it is a retrograde gas condensate, in which liquid condenses in the reservoir as pressure falls below the upper dew point — the opposite of normal behaviour — and much of it stays trapped; beyond the cricondentherm it is a wet gas (liquids form only at the separator) or a dry gas (none form at all). PVT laboratories measure these properties by flash liberation (gas stays in contact with its oil, as in the separator) and differential liberation (gas removed as it forms, as in the reservoir below p_b).

For radial flow of oil to a well of radius r_w from an outer radius r_e, integrating Darcy’s law gives, in oilfield units, the steady-state rate q = 0.00708kh(p_e − p_wf)/[μBo ln(r_e/r_w)], with q in STB/d, k in md, h in ft, p in psi and μ in cP. A bounded reservoir that has reached pseudo-steady state (every point declining at the same rate) uses the average pressure p̄ and ln(r_e/r_w) − 0.75, and a skin s adds to the logarithm in both. Before the boundaries are felt the flow is transient and obeys the diffusivity equation, whose line-source solution is the basis of well testing. The productivity index J = q/(p̄ − p_wf) summarises a well’s inflow in STB/d/psi.

When water displaces oil, the fractional flow of water at a point, neglecting gravity and capillarity in horizontal flow, is f_w = 1/[1 + (k_ro/k_rw)(μ_w/μ_o)]. It depends on saturation through the relative permeabilities and on the viscosity ratio, so a viscous oil gives a high water cut early. The Buckley-Leverett frontal-advance theory moves each saturation at a velocity proportional to df_w/dS_w, which produces a shock front; the Welge tangent from the initial saturation to the f_w curve gives the front saturation and the average saturation behind it, hence the displacement efficiency at breakthrough.

Coning is the rise of an underlying water (or the fall of an overlying gas cap) toward the perforations of a producing well, because the pressure drawdown near the well exceeds the gravity head that holds the contact flat. Below a critical rate the cone is stable and the well produces clean oil; above it the cone breaks through, the water cut or gas-oil ratio jumps, and oil recovery is lost. Remedies: produce below the critical rate, perforate away from the contact, use horizontal wells (whose smaller drawdown per unit length cresting instead of coning), inject gel or cement barriers, or use downhole water separation. Reservoir pressure is measured with permanent downhole gauges, bottom-hole gauges run during tests, and wireline formation testers (RFT/MDT) that take a pressure at many depths in one trip; plotting those pressures against depth gives fluid gradients (about 0.43-0.45 psi/ft for water, lower for oil, much lower for gas) whose intersections locate the gas-oil and oil-water contacts.

Primary recovery is driven by the energy stored in the reservoir. In a solution-gas (depletion) drive the oil and its gas expand; above the bubble point only rock and liquid compressibility work and pressure falls fast, and below it gas comes out of solution, the gas-oil ratio climbs steeply and recovery is low. A gas-cap drive is pushed by an expanding cap of free gas, pressure falls more slowly and recovery is higher, especially with gravity segregation. A water drive is supported by influx from an aquifer, pressure is maintained, water cut rises steadily and recovery is the highest of the primary drives. Gravity drainage in thick, steeply dipping, permeable reservoirs can recover more still, slowly. Compaction drive comes from the collapse of pore volume in weak rock, with surface subsidence. Most reservoirs are a combination. Typical primary recovery factors are often quoted as roughly 5-30% of OOIP for solution-gas drive, 20-40% for gas-cap drive and 35-75% for water drive, with the actual value set by rock, fluid and rate.

The material balance equation treats the reservoir as a tank: the reservoir volume of fluids produced equals the expansion of what remains plus water influx. In the Havlena-Odeh form, F = N(Eₒ + mE_g + E_f,w) + W_e, where F = Nₚ[Bₒ + (Rₚ − R_s)B_g] + WₚB_w is the underground withdrawal, Eₒ = (Bₒ − Bₒᵢ) + (R_sᵢ − R_s)B_g the expansion of oil and its original gas, E_g = Bₒᵢ(B_g/B_gᵢ − 1) the expansion of a gas cap of relative size m, E_f,w = Bₒᵢ(1 + m)(c_wS_wi + c_f)Δp/(1 − S_wi) the expansion of connate water and shrinkage of pore volume, and W_e the water influx. It is straightened for interpretation: with no gas cap and no influx, F against Eₒ + E_f,w is a line through the origin of slope N. For an undersaturated oil with no influx the balance reduces to NₚBₒ = NBₒᵢcₑΔp, with effective compressibility cₑ = (cₒSₒ + c_wS_wi + c_f)/(1 − S_wi). For a volumetric gas reservoir it is the famous straight line p/z = (pᵢ/zᵢ)(1 − G_p/G), so the gas initially in place G is where the p/z line reaches zero.

Reserves are estimated by four techniques in rough order of the data they need. Volumetric: in oilfield units the oil initially in place is N = 7758Ahφ(1 − S_wi)/Bₒᵢ STB, with A in acres, h in ft, 7758 bbl per acre-ft, and gas in place G = 43 560Ahφ(1 − S_wi)/B_gᵢ scf with B_gᵢ in ft³/scf; multiplied by a recovery factor they give reserves. Material balance needs production and pressure history. Decline-curve analysis extrapolates a producing well’s rate. Simulation combines everything. Arps’ decline family is q = qᵢ/(1 + bDᵢt)^(1/b): exponential decline (b = 0) is q = qᵢe^(−Dt), a straight line on a semilog plot of rate against time, with cumulative Nₚ = (qᵢ − q)/D; hyperbolic decline has 0 < b < 1; harmonic decline (b = 1) is q = qᵢ/(1 + Dᵢt) with Nₚ = (qᵢ/Dᵢ)ln(qᵢ/q). Exponential decline is the most conservative of the three.

Field development turns a discovery into a producing plan: the number, type and spacing of wells, the drive to rely on and whether to support it by water or gas injection, the surface facilities and export route, and the production profile — build-up, plateau and decline — that the economics are run against. Reservoir management is the continuing loop of data acquisition (pressures, logs, tests, production), model updating and decisions (infill drilling, workovers, injection changes, EOR) that aims to maximise economic recovery over the field’s life rather than rate in any one year.

Three principal stresses act at depth: the vertical (overburden) stress σ_v, the weight of the rock above — often near 1 psi/ft — and the maximum and minimum horizontal stresses σ_H and σ_h. Rock responds to effective stress, σ′ = σ − αp, where p is pore pressure and α the Biot coefficient (close to 1 for weak, porous rock). Under the common assumption of lateral constraint (uniaxial strain), σ_h = [ν/(1 − ν)](σ_v − αp) + αp, with ν Poisson’s ratio. A hydraulic fracture opens against the least principal stress and so propagates perpendicular to σ_h: vertical fractures in most reservoirs, whose azimuth follows σ_H. The fracture gradient is essentially σ_h per foot of depth. Production lowers p and raises effective stress, compacting the reservoir, closing natural fractures and, in weak rock, producing sand; injection does the opposite and can reopen faults. Wellbore stability is the same analysis around a hole: too little mud weight collapses it (shear failure), too much fractures it (tensile failure).

A reservoir simulator solves the equations of mass conservation for each component, Darcy’s law for each phase, and PVT relations, on a grid. The mathematical model is a set of non-linear partial differential equations; finite-difference discretisation replaces derivatives by differences between grid blocks (the second-derivative formula of the Taylor series), in space and in time. Explicit time stepping is simple but stable only for small steps; implicit stepping is unconditionally stable and needs a linear solve every step; IMPES (implicit pressure, explicit saturation) is the classic compromise. Cartesian grids of rectangular blocks model fields; radial (r-θ-z) grids, with block radii spaced logarithmically outward from the well, model near-well behaviour such as coning. The black-oil model has three phases (oil, gas, water) and three pseudo-components (stock-tank oil, surface gas, water), with gas allowed to dissolve in oil through R_s and all properties functions of pressure — the workhorse for most oil fields; compositional models track individual hydrocarbon components where composition changes matter. History matching adjusts the uncertain inputs (permeability, aquifer size, relative permeability) until the model reproduces the observed pressures, rates and water cuts; only then is it trusted to forecast. A history match is not unique, which is why forecasts are reported with ranges.