Summary Time-lapse seismic monitoring technology has recently been introduced to the Duri field reservoir-management process as a means of directly imaging changes in vertical and horizontal steam distribution over time. The technique consists of repeated acquisition of three-dimensional (3D) seismic surveys to measure steamflood-induced changes in the acoustic properties of hydrocarbon reservoirs. The resulting seismic images are processed, parameterized, and translated to relevant engineering parameters. In this case study, steam flow was monitored during the early stages of steam injection by mapping anomalous seismic-reflectivity response within the steam-flow layers. Seismic-reflectivity data and borehole information were combined by use of multivariate statistical analysis and image processing techniques to produce images of steamflood conformance at the flow-unit level. The images reveal important insights into steam-flow behavior and were used to design a conformance-improvement program for the case study focus area. Introduction Duri Steamflood. Duri field, operated by P.T. Caltex Pacific Indonesia under a production-sharing contract with Pertamina, is located in the eastern coastal plain of central Sumatra approximately 70 miles northwest of the city of Pekanbaru. Oil is produced from structurally trapped Miocene sandstones at depths between 200 to 900 ft. Duri field encompasses 30,000 acres and holds an estimated 5.4 billion bbl of original oil in place. The Duri steamflood (DSF) project began in 1985 and is currently the largest steamflood in the world. Today, approximately 50% of the field is undergoing steamflooding in seven areas of varying operational maturity. More than 900 injector wells inject approximately 1.25 million B/D steam [cold-water equivalent (CWE)] into the heavy-oil reservoirs. The field currently produces 300,000 BOPD from more than 2,700 producer wells. Steamflooding is expected to boost ultimate recovery by an additional 2.5 billion bbl of oil over primary recovery techniques. Reservoir Monitoring. DSF reservoir-management strategies and work processes are designed to maximize total oil recovery and minimize steamflood heat requirements. Efficient heat management focuses on equitable allocation of heat to reservoir flow units on the basis of displaceable pore volume (PV) and on taking measures to ensure uniform areal conformance within each flow unit. Uncertainty in reservoir architecture and the continuously changing fluid-flow dynamics of an active steamflood require constant monitoring and adjustment to achieve the desired efficiencies. Several conventional reservoir-monitoring tools are used in combination to obtain insight into steam-flow direction, rates, and sweep efficiency.1 Steam-injector profiling surveys monitor the volume fraction of fluid exiting the injector wellbore into each flow unit. At the producer wellhead, flowline temperatures and production rates, casing pressures, and casing-gas-effluent rates are indicative of steam and heat entering the production-fluid stream. Anecdotal evidence, such as damage of producer-well tubing and gravel-pack liner, is used to identify possible steam-breakthrough flow units. Observation-well wireline measurements, such as pulsed neutron capture and temperature surveys, yield direct measurements of reservoir fluid saturation and heat. These conventional methods describe, to varying degrees of resolution and certainty, the reservoir state in the region of the wellbore. However, because of reservoir heterogeneity, operational-induced pressure variations, and nonlinear steamflood fluid dynamics, a wellbore sampling of the reservoir state is often not sufficient to construct an accurate representation of complex interwell reservoir conditions and relationships. Time-lapse seismic-monitoring technology has recently been introduced to the Duri field reservoir-management process as a means of directly imaging steamflood evolution, subject to resolution and detection limitations of seismic measurements. The method employs the time-lapse acquisition of 3D seismic surveys over the same geographical area. By exciting rocks with seismic waves and recording spatial and temporal changes in seismic wavefield behavior, steamflood-induced alterations in reservoir properties can be deduced. A successful single-pattern pilot study was carried out between 1992 and 1995 and demonstrated both the technical and economic feasibility of using time-lapse seismic for improved reservoir-management of the steamflood.2 Since then, multipattern time-lapse seismic surveys have been acquired over several areas of the field. This paper reports on the first efforts to integrate time-lapse seismic data into the Duri field reservoir-management process. Acoustic Fundamentals. The underlying physical basis for seismic steamflood monitoring is that changes in reservoir temperature, pressure, and fluid saturation during steamflooding significantly alter sound-wave velocity. In a published account of the Duri seismic monitoring field trial results, Jenkins et al.2 report that compressional velocity, vp, of the unconsolidated Duri field reservoir sandstones is strongly dependent on fluid temperature and phase. Core velocity measurements indicate that, as the reservoir is heated from ambient temperature of 100 to 350°F while holding pressure at 430 psi, vp decreases linearly by approximately 10%. With further heating, as the liquid in the pore space undergoes the phase transition to vapor, an abrupt velocity decrease of about 30% occurs. These findings are consistent with those published by Domenico3 and Wang and Nur.4 Compressional velocity fundamentally determines the transmission and reflection behavior of sound waves. For Duri reservoirs, a decrease in vp resulting from steam injection is expressed as an increase in seismic-reflection-time and -strength measures. Seismic reflectivity is, to a first order approximation, a function of the contrast in acoustic impedance (the multiplicative product of vp and ?) at a reflecting-layer boundary.5 where subscripts L and U=upper and lower layer properties, respectively. The sign of R is dependent on the relative juxtaposition of high- and low-impedance layers, while the magnitude of R, reflection strength, is dependent on the size of the impedance change across a layer boundary. Thus, a large decrease in acoustic impedance resulting from the displacement of liquid pore fluid by steam vapor is manifested as a large negative reflection at the top of the steam layer and a large positive reflection at the base.