|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
The main goal of hydraulic fracture treatments is to achieve a target geometry and conductivity within given operational and budget constraints. Evaluating past frac designs can help one to understand completion performance by highlighting and summarizing relevant engineering considerations. For example, to simulate fracture growth, the model requires a pumping schedule as an input. This study describes an automated procedure that identifies the different steps of a pumping schedule and generates several pertinent statistics based on the hydraulic fracturing time-series data collected in the field. Essentially, it returns an actual or "as pumped" fracture treatment schedule, which may differ from the designed one.
The dataset analyzed in this study includes the slurry rates and proppant concentrations for 577 stages from all major North American basins. The algorithms were calibrated using 112 stages and tested on the remaining 465 stages. The procedure first identifies the start and end times of various sections of the pump schedule (pad, acid pad, slurry, sweeps, and flush) by smoothing, normalizing, and then "rounding" the proppant concentration signal. The procedure then isolates sustained intervals with positive proppant concentration to identify acid pads and proppant stages. The remaining time intervals are associated with pads, sweeps, and flushes. Various statistics such as volumes, durations, and averages are computed for each interval.
Each slurry interval (proppant ramp) is then further broken down into its proppant steps. This is accomplished using quantization ideas from digital signal processing. Quantizing the signal maps the observed proppant concentrations in each step (noisy signal) to a representative value for that step. If the stage is pumped as planned, then these representative values should be close to the designed concentrations. The authors have two techniques to accomplish this: the first is based on clustering, and the second uses a piecewise constant regression based on recursive partitioning (decision-tree regression). The representative values are used to identify the proppant steps, and once the steps are identified, the process generates statistics for each step.
Hydraulic fracturing time-series data is an ideal target for analysis using signal processing techniques. The process correctly recognizes the start and end times of the various pumping steps: 95% of the picks (target events) identified by the procedure are within three seconds of the manual picks. This automation significantly reduces the time required for picks, allows fast auditing of existing picks, and provides an efficient method for analyzing historic pumping schedule data that is available only as PDF files. This is the first paper that describes a computational method to automatically extract a pumping schedule from hydraulic fracturing time-series data. The method uses techniques from digital signal processing and is accurate, robust, transparent, and fast.
One of the challenges of unconventional resource development is the identifying and preventing casing failures caused by the hydraulic fracturing process. Multiple mechanisms may be responsible for casing deformation and/or failures, starting with the rock properties of the formation, the wellbore configuration, quality control of tubulars, and operational aspects during drilling and completion. This paper presents two case studies where casing issues were discovered during the drill out of frac plugs following multi-stage fracturing treatments. The objectives of these studies are (a) to determine the cause and nature of the casing failures, (b) to recommend changes to future completion programs to prevent similar operational issues, and (c) to develop a model that automatically identifies these failures.
The subject wells are located in two very different basins: the Eagle Ford trend in the Brazos Valley (BV) area of south Texas and the Powder River Basin (PRB) in Wyoming. In both studies, the casing issues could be directly correlated to Abnormal Pressure Behaviors (APBs) observed during fracturing. A total of 486 stages, completed in 12 different wells, were reviewed using a cloud-based application that allows stages to be examined individually, or as groups. Since then, five additional wells have been added to the data set. After problem stages were identified, the completion team worked with the drilling engineers and geologists to determine the mechanisms causing the casing damage.
Tight spots encountered during frac plug drill out in the BV wells directly correlated with stages completed in geological transition zones between the Eagle Ford and Woodbine formations. Once this was recognized, the team implemented operational contingencies to fracture designs for stages completed in BV transition zones. In the PRB wells, after reevaluating the post-mill inspection of the casing, the damage was found to be poor casing quality control. The location of casing deformations and/or failures directly correlated with stages that displayed evidence of frac plug failure. Moving forward, the PRB completion supervisors were made aware of potential issues, and alternative procedures were developed for both fracturing and drill out operations that utilized the questionable casing. As of this time, no additional casing issues have occurred.
In these studies, identification of the problem stages was initially performed manually (stage-by-stage) using a cloud-based analytics platform (CBAP). During the process, it was recognized that the two types of problem stages had their own characteristic pressure signature. A machine learning algorithm was developed that automatically identifies plug failure, which is indicated by a sudden unexplained pressure drop in the absence of rate changes. Transition stages could be easily identified through the use of stage variance plots (e.g., comparing maximum/average rates and pressures across multiple stages and wells) and also through machine learning algorithms that identified unexpected pressure increases followed by sharp pressure drops.
Abstract A breakthrough patent-pending pressure diagnostic technique using offset sealed wellbores as monitoring sources was introduced at the 2020 Hydraulic Fracturing Technology Conference. This technique quantifies various hydraulic fracture parameters using only a surface gauge mounted on the sealed wellbore(s). The initial concept, operational processes, and analysis techniques were developed and deployed by Devon Energy. By scaling and automating the process, Sealed Wellbore Pressure Monitoring (SWPM) is now available to the industry as a repeatable workflow that greatly reduces analysis time and improves visualizations to aid data interpretations. The authors successfully automated the SWPM analysis procedure using a cloud-based software platform designed to ingest, process, and analyze high-frequency hydraulic fracturing data. The minimum data for the analysis consists of the standard frac treatment data combined with the high-resolution pressure gauge data for each sealed wellbore. The team developed machine learning algorithms to identify the key events required by a sealed wellbore pressure analysis: the start, end, and magnitude of each pressure response detected in the sealed wellbore(s) while actively fracturing offset wells. The result is a rapid, repeatable SWPM analysis that minimizes individual interpretation biases. The primary deliverables from SWPM analyses are the Volumes to First Response (VFR) on a per stage basis. In many projects, multiple pressure responses within a single stage have been observed, which provides valuable insight into fracture network complexity and cluster/stage efficiency. Various methods are used to visualize and statistically analyze the data. A scalable process facilitates creating a statistical database for comparing completion designs that can be segmented by play, formation, or other geological variations. Completion designs can then be optimized based upon the observed well responses. With enough observations and based on certain spacings, probabilities of when to expect fracture interactions could be assigned for different plays.
The snarled red lines on the chart look more like a plate of spaghetti than a source of fracturing insights. It looks like a meaningless mess, which is generally how the ups and downs of difficult stages are viewed. To Adam Hoffman, a completion engineer for Chesapeake Energy, those 47-stages-worth of data look like a valuable opportunity. “We see so many stages with so many odd spikes and drops or chatter. We chop it off and say that was an odd stage. In my mind when we are looking at all those stages, we should wonder, ‘what was that pressure spike telling us,’” he said. That curiosity became a research project after Chesapeake encountered a spate of blockages in recently fractured Eagle Ford wells. The investigation into the cause of the casing damage led to a collaboration with Well Data Labs to look for connections between pressure changes and what is happening in the wells. Based on hundreds of stages of data from 19 wells fractured in the Eagle Ford, and later in the Powder River Basin, they reported finding a distinctive pressure signature that provides a reliable, but not foolproof, guide to when casing damage is likely. Well Data Labs has automated the search for those signatures as it looks for the meaning of the terabytes of fracturing data in this overwhelming number of seemingly random, squiggly lines. The oilfield data and software company is working on ways to monitor changes in the fracturing-fluid chemistry, the proppant intake into perforations, and an explanation for the pressure spikes seen before the pressure falls, said Jessica Iriarte, research manager at Well Data Labs. The troubleshooting and pressure analysis were covered in paper SPE 201484 presented at the 2020 SPE Annual Technical Conference and Exhibition (ATCE). It described how engineering trouble-shooting revealed that geological stresses were the likely source of problems in one case, and faulty pipe in the other. It followed up with data analysis, which used machine learning to identify distinctive patterns that provide an early warning of what is happening in the well faster and more objectively than a completion engineer studying the chart. Based on the troubleshooting, Chesapeake made changes that largely eliminated those costly problems. But it was also a costly learning process. In the Eagle Ford, they identified the underlying problem by investigating why multiple coiled-tubing runs were blocked while they were trying to drill out plugs after fracturing. When that happens, Hoffman said, “it can mean a week lost working past it.” Failure to drill out a plug can block access to the productive rock further down the lateral. A reliable automated treating-pressure analysis in the daily report could alert the completion team to problems while fracturing is in progress. They could then make adjustments on later stages and create a plan to limit the time lost when drilling out plugs on stages where they are likely to encounter tight sections.
It looks like a meaningless mess, which is generally how the ups and downs of difficult stages are viewed. To Adam Hoffman, a completion engineer for Chesapeake Energy, those 47-stages-worth of data look like a valuable opportunity. "We see so many stages with so many odd spikes and drops or chatter. We chop it off and say that was an odd stage. In my mind when we are looking at all those stages, we should wonder, 'what was that pressure spike telling us,'" he said. That curiosity became a research project after Chesapeake encountered a spate of blockages in recently fractured Eagle Ford wells.