This paper investigates filtering and compression of real-time production data by means of wavelets. We conduct a comprehensive analysis examining a series of possible wavelets for this type of data, and demonstrate that spline39 and spline4246 have excellent properties for cleansing of pressure and rate data, respectively. We extend the wavelet filtering technique for E&P data in several ways. We demonstrate that the methods frequently used for noise estimation give erroneously results, and that the noise estimation will depend on the wavelet utilized. New noise estimators are developed. We show that the new noise estimators combined with a median filter are suited for removal of outliers. We demonstrate that the SCAD shrinkage rule gives best results together with spline39 for pressure and spline4246 for rate. We show the choosing a higher filter level than those common used will be beneficial. A new empirical threshold estimator is proposed, giving excellent results compared to the other commonly used threshold estimators. A new compression routine for automatically compression of signals is proposed, finding the optimal compression ratio for piecewise constant signals like rate data. The filter approach proposed in this paper requires no user interaction, and can thus be performed automatically. In most cases, noise reduction at 85% or above is achieved.

You can access this article if you purchase or spend a download.