This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 209030, “Complete Analysis of a Complex Sand-Screen Completion in a Single Run in Hole is Enabled by Combination of Novel Passive Acoustic Tools With Conventional PLT and Caliper Techniques,” by Duncan Troup, SPE, Archer. The paper has not been peer reviewed.


Wells with sand-control measures may still exhibit the onset of sanding during production, often because of isolated damage to individual screen sections. Positive identification of sand-production location allows targeted mitigation while retaining as much hydrocarbon flow as possible. The complete paper discusses novel acoustic techniques used to identify productive zones and areas of sand production in a well with a sanding event.

Digital Signal Processing

Most modern tools will perform one or two fast Fourier transforms per second, but the tool design presented in the paper has a very fast sampling rate and completes 220 transforms per second. This allows the output frequency information to be further sampled in statistically meaningful ways to provide three additional noise parameters. The time-filtered noise level is the general, steady-state noise amplitude with outliers removed and generally is used for leak and flow detection. The mean noise level is composed of all data samples and is an indication of the total acoustic energy present. The third parameter, the transient noise level, is an indication of time variability of the received signal and is sensitive to short-lived transient acoustic signals. It is this parameter that is of the most interest in the detection of sand impact, because, by its nature, a sand strike is a transient event.

The data generated can be displayed as a variable density log of frequency response against time or logging depth, with the amplitude of each frequency band represented by a color map, and a total noise energy curve may be derived over any frequency range desired.

To test the suitability of the signal processing for viable sand discrimination, the background flow noise was increased every 2 minutes and sand particles were drizzled onto the sensor plate for 1 minute of each flow period. When no background flow noise existed, the sand-impact noise shows up well. When flow noise is introduced, the baseline of the noise energy curve is shifted higher but good contrast with the sand signal remains.

A further test placed a tool inside a water-filled 5-in. liner section and a stream of water was directed against it (Fig. 1). The rate of water flow was controlled, and sand was introduced at known mass rates up to 1.17 gal/min. The results were plotted as noise energy against fluid velocity for each of the sand mass flow rates. Good correlation exists between the rates of both liquid and sand rates and total energy, and good correlation exists for liquid and sand flow rates. At low liquid flow rate, better discrimination between the rates of sand in the flow is observed, indicating that the full bandwidth transient parameter will provide better resolution when sand and fluid flow rates are low.

This content is only available via PDF.
You can access this article if you purchase or spend a download.