Multi-resolution Analysis of Long-Term Pressure Transient Data Using Wavelet Methods
Jitendra Kikani, Meiqing He · SPE Annual Technical Conference and Exhibition · 1998
Abstract Large bandwidth transmission systems have made it easier and convenient to routinely collect large data from permanent monitoring systems. To effectively manage, filter, reduce and interpret these large data sets requires sophistication and methodologies that have sound bases. The standard data reduction procedures are fraught with difficulties and have biased modes which give rise to aliasing effects. In fact, a two step process is recommended wherein noise filtering and behavioral filtering are separated and allowance is made for visualizing the data at different dyadic scales of resolution. In this paper, the characteristics of simulated pressure transient data are investigated in the frequency domain. The power spectrum properties of both simulated data (with random noise added) and real data sets are studied. Wavelet transform is introduced to accomplish the time-frequency analysis of long- term pressure transient data. The characteristics of different family of bases of wavelets are studied, and a recommendation is made for use with pressure data. Application of wavelet transform in data denoising has been experimented with and the performance of different thresholds established. Two types of denoising methods are recommended for slowly changing pressure transient data. Multi-resolution approximation of the long-term pressure transient data using wavelet transform is also presented. Approaches for data culling are addressed. P. 117