This paper presents a systematic procedure for integrating 3-D seismic data and production data to develop a detailed reservoir description. The procedure involves three steps. Step one includes the inversion of the seismic data to construct impedance distribution with the well core and log data. Step two includes the simulation of porosity values consistent with the impedance data. The last step involves the estimation of permeability values consistent with its relationship with porosity as well as the dynamic information. All these three steps are discussed in detail. The procedure is further validated by applying it to both synthetic and the field data. The results obtained have been satisfactory illustrating the suitability of the method.
3D seismic data are often collected for field exploration. Compared to the well data, seismic data have better areal coverage, but poor vertical resolution. The seismic data mainly reflect the acoustic property of formation, such as the acoustic impedance which describes a static property of the reservoir. By integrating the seismic and well data, a better impedance distribution within the reservoir can be obtained. This requires a better inversion technique which can integrate these two types of information. A better integration of seismic and well data may reduce the non-uniqueness of the solution and enhance the resolution to describe the heterogeneities in the reservoir. Although improved, the integrated static description of the reservoir may not be sufficient to reproduce the dynamic behavior. For a better reservoir management, it is critical to predict the future performance of the reservoir. Before predicting the future performance, we first need to reproduce the prior performance. As a result, we need a reservoir description which is not only consistent with the static properties. but is also capable of reproducing the past performance.
In 1977, Lavergne and William presented a work regarding the conversion of seismic amplitude data into pseudo-velocity by deconvolution. In 1979, Lindseth also presented a similar method to obtain the pseudo-log. In this type of inversion, the wavelet was determined in the well location. The other seismic traces are inverted using this wavelet by applying the deconvolution method. Back to 1970, Backus and Gilbert proposed the generalized linear inversion (GLI) theory. Since then, many investigators applied this theory for seismic data processing and inversion. Cooke applied the technique to invert the CDP (Common Depth Point) seismic trace into impedance. GLI remains a popular inversion technique.
The conventional methods discussed above have some disadvantages:
the solution is non-unique;
they require knowledge of the boundary impedance, i.e., the starting impedance.
In 1991, Sen and Stoffa applied simulated annealing and genetic algorithm for the layer velocity determination for seismic data processing. These methods have the capability to integrate other information and may Overcome some of the difficulties encountered in conventional methods. Recently, Huang and Kelkar implemented these algorithms for 3D seismic inversion by integrating the well information and seismic data into the inversion process. In addition, a new efficient algorithm called modified stochastic hill climbing was also used for the inversion process.