Time-lapse seismic reservoir monitoring

Time-lapse seismic reservoir monitoring has advanced rapidly over the past decade. There are currently about 75 active projects worldwide, and more than 100 cumulative projects in the past decade or so. The present total annual expenditures on 4-D seismic projects are on the order of $50-100 million US. This currently represents a much smaller market than 3-D seismic, but the use of 4-D seismic has grown exponentially over the past decade and is expected to continue to do so. The major opportunity provided by 4-D seismic data is its ability to image fluid flow in the volumetric region not sampled by wells. Fluid flow is thus directly imaged by 4-D seismic data, father than solely predicted by flow simulation. In contrast to 3-D seismic, which is an exploration and development tool, 4-D seismic is quickly becoming a vital engineering reservoir management tool. Time-lapse seismic images can identify bypassed oil to be targeted for infill drilling, and add major reserves to production to extend a field's economic life. 3-D seismic can monitor the progress of costly injected fluid fronts (water, gas, steam, CO2, etc.) that can save hundreds of millions of dollars in optimizing injection programs. 4-D seismic can map reservoir compartmentalization and the fluid-flow properties of faults (sealing versus leaking), which can be extremely useful for optimal design of production facilities and well paths in complex reservoir flow systems. Rock physics measurements made in the mid 1980 Stanford University predicted that thermal enhanced oil recovery (EOR) processes, especially steam injection, should be visible in repeated surface seismic surveys. The first field tests of the concept were conducted in the mid-late 1980s and early 1990s in Canada, the US, and Indonesia at several steam injection sites and one fireflood site. These early projects showed conclusively that large anomalies related to steam and heated gas were indeed strikingly visible in time-lapse seismic data. This was followed by projects to monitor isothermal gas-fluid movement, particularly by early work in the North Sea and Paris basin. These gas monitoring experiments were also successful, but it became clearer that the interaction of the reservoir rock, fluid, and gas components could enhance or degrade the time-lapse seismic signal depending on specific reservoir conditions. The past five years has seen an increased focus on monitoring of oil-water systems. Here, the 4-D seismic technique works well in unconsolidated high-porosity sands with high gas-to-oil ratio (GOR) oil and salty brine, works moderately well in somewhat consolidated porous rocks with live oil, and is technically challenging in applications involving dead oil or stiff rocks like carbonates or cemented sandstones, independent of the rock's fluid content. It is now recognized that successful monitoring of fluid flow depends on critical reservoir rock and fluid properties pressure and temperature values, and high-fidelity seismic acquisition, processing, and interpretation. Much work has been done on feasibility risk analysis, and on modeling time-lapse seismic data from 3-D heterogeneous reservoir models, rock physics data, and flow simulation results. Acquisition and processing service companies continue to enhance repeatability and accuracy of their field hardware and processing algorithms. Multicube quantitative interpretation and inversion techniques for time-lapse seismic data are evolving, especially in ways that integrate 3-D seismic data with other reservoir data types such as well logs, pressure, temperature and saturation data, core measurements, flow simulations, and production data. Current research and the road ahead involves ocean-bottom and borehole seismic, permanent sensor installations, real-time reservoir monitoring instrumentation, sear waves, nonseismic techniques (such as electromagnetics, gravity and radar), inversion for fluid-flow properties with uncertainty analysis, reservoir data integration, seismic history matching and reservoir model updating.