The Wiggle that Makes a Difference: Using AI to get the Right Wavelet with Multi Well Input
H. S. N. Alkhanjari, M. M. Glukstad, S. H. Alqassabi, A. Alrubaiey, N. Ali, O. A. Albadi · 2023
Abstract A common problem when carrying out seismic interpretation in the Two-Way travel Time (TWT) domain is to choose the adequate wavelet when performing well to seismic tie. This step is also crucial in any Quantitative Seismic Interpretation (QI) project where the use of the right wavelet makes a big difference. Many of the fields in Oman have several wells with the appropriate data, i.e., Time-Depth (T-Z) pairs or Check shot data (VSP) and Sonic, but the quality of the log data and the short extend of the Sonic log requires calibration, making it very challenging to select the right well for the analysis. The interpreter often looks at several wells, in the same field, but with different data and quality. Although multi-well to seismic tie is available in most commercial software, the problem of dealing with variable data quality, having to select the right wavelet, that will make sense for all the wells, is still a problem that is very time consuming and often difficult to resolve. The interpreter has to make a decision but lacks any proper quantitative or qualitative approach on which to base the choice of wavelet. Recently an Artificial Intelligence (AI) tool was released, Automated Seismic to Well Tie (AWT) in Petrel version 2021, which uses the seismic volume and well data to obtain the best wavelet which provides the best tie when generating the synthetic seismogram, with minimum effort, in matters of few hours. The tool looks at all the possible wavelets and defines the best match and ranks the outcome for the different wells. Crucial part of this methodology is to provide the computer with as clean as possible data set. The team tested this tool in field XX and with technical support from the software owners, obtained spectacular results in a very short time saving weeks of work.