Truncation De-noising in Transient Pressure Tests

Truncation noise affects the interpretation of transient pressure tests when the total pressure change is very small compared to the total measurement scale of the tool. There are situations, mainly in high permeability reservoirs, where the maximum pressure variation is of the order of one thousandth of the total scale of the probe. Noise greater than one thousandth of the total pressure change hides pressure signals related to the characteristics of the reservoir; even when the accuracy of the tool could achieve one part by million of the total scale. This situation makes almost impossible the interpretation with the logarithmic derivative. It is demonstrated in this paper that the frequency spectrum of noise is in the same frequency band that frequencies associated to the reservoir signal. By using a fitting procedure and wavelets, the signal-noise ratio (SNR) of pressure tests with truncation noise is improved in most of the cases up to 8 dB. This situation is equivalent to measure the maximum pressure change of the reservoir with a tool of one bit of additional effective resolution. Greater improvements can be obtained when only Gaussian white noise is present in the pressure signal. This result offers the possibility to reuse old noisy derivative data logs, especially pressure tests from high permeability reservoirs. This contribution will permit to utilize derivative interpretation techniques in a more appropriate way. The efficiency of the proposed filtering technique is demonstrated with synthetic and field data pressure tests.