Transient pressure wave signal feature extraction for well leakage detection

Well leakage management is crucial for oilfield development, impacting well safety and production efficiency. To pinpoint leakage layers in the open hole section during drilling, a novel method leveraging the maximum kurtosis (MK) technique is introduced. By harnessing the transient pressure waveform's similarity to microseismic signals, the method accurately captures the characteristic waveform and determines the leakage location through signal time differences and wave speed analysis. To address noise in transient pressure signals, a singular value decomposition (SVD)–ensemble empirical mode decomposition (EEMD)–wavelet thresholding denoising (WTD) approach is proposed, enhancing signal clarity and preserving intrinsic features. The research shows that the SVD–EEMD–WTD denoising method effectively strips away noise, preserving signal characteristics and enhancing intrinsic modal component correlation by 79.5%; determining the time difference between the transient pressure wave arriving at the leakage layer from the wellhead and the wave velocity in the annulus according to the MK method can achieve the accurate positioning of the leakage layer location, and its positioning error is in the range of 0.12%–9.07%; under identical conditions, the MK method surpasses others in accuracy, with an error rate as low as 1.02%. The SVD–MK model can quickly identify the leakage location features and locate the leakage with high efficiency.