A new comprehensive filtering model for pump shut-in water hammer pressure wave signals during hydraulic fracturing

Abstract

In the process of fracturing, the position of the fractures can be identified by analyzing the spectrum of the water hammer pressure wave signals after pump shut-in. However, the collected water hammer pressure data contain a lot of noises, which brings difficulty to the spectrum analysis and affect the accuracy of the final results. Removing multiple kinds of noises while protecting useful fracture response is the challenge. However, the filtering model for the water hammer pressure wave signals is absent. This paper proposes a new comprehensive filtering model for pump shut-in water hammer pressure wave signals. The filtering model contains three parts. First, we designed a two-step filtering method, including Hampel filter and FIR low-pass filter, to filter signals in the main channel. Second, multi-frequency-adaptive notch and median filter were combined to construct the signals in the reference channel, which can be used to protect the useful frequency of the water hammer wave signals and pressure response of the fractures from being eliminated. Third, the filtered signal is obtained by adaptive cancellation between the signals in the main channel and the reference channel. After that, the filtering effects of signals with and without fractures were studied by constructing simulated signals and collecting the signals in the laboratory experiment. The signal-to-noise ratio (SNR) gain and the mean square error (MSE) reduction were selected as indicators for the model validation, and the cepstrum was also used as the indicator to verify the preservation of fracture response characteristics. Besides, the noise adaptive range of the model was evaluated by increasing the noise intensity to the constructing signal. This research can provide a new and useful filtering method for water hammer pressure wave signals in the field. • A new comprehensive filter model suitable for water hammer signals was proposed. • For the water hammer pressure wave signal, the SNR gain is about 6.8 dB, and the MSE can be reduced by about 0.05. • The noise adaptive range of the model was evaluated by changing the noise intensity to the constructing signal.