Fracturing/pressure pumping

Digging up New Information on What Fractures Really Look Like

During a panel discussion on what fractures look like, one expert added a significant qualifier to the title: what they “may look like.”

jpt-2018-03-fracturing-digginghero.jpg
Diagram shows how a single slant well (SCW) was used to collect core samples from two production wells (AUWC, AMWC) at the Permian Hydraulic Fracturing Test Site.
Source: URTeC 2697483.

During a panel discussion on what fractures look like, one expert added a significant qualifier to the title: what they “may look like.”

That addition by Roberto Suarez-Rivera, scientific advisor for W.D. Von Gonten & Co., was in keeping with a discussion at the SPE Hydraulic Fracturing Technology Conference that offered ample evidence that when hydraulic fracturing is closely examined, results often vary from expectations.

He followed a presentation by ConocoPhillips that offered multiple examples of why experts need to be cautious about their pronouncements. The rare study gathered samples of fractured rock around a well, among many other tests. A conference technical paper (URTeC 2670034) described its findings as “very different from the simple view of the stimulated reservoir volume that are commonly modeled or predicted with current fracture models.”

jpt-2018-03-fracturing-digginggraph.jpg
Four observation wells (names beginning with S) were drilled by ConocoPhillips to monitor and sample fracturing in an Eagle Ford production well (P3). Source: URTeC 2670034.

 

For example, when the authors collected the core samples, they were surprised that the rough surfaces of the cracks held little of the sand injected in the nearby well to prop open fractures to ensure continued production.

“There is sparse evidence of abundant proppant beyond 75 ft,” said Kevin T. Raterman, director of reservoir engineering technology for ConocoPhillips, who was involved in the project.

That is significant because propped fractures are thought to be the source of much of what is produced. This study and others are raising questions about where the oil and gas is coming from and where all the proppant is going.

Even with five observation wells equipped with the latest monitoring equipment, ConocoPhillip’s team was left wanting more. When Raterman was asked where all that proppant went, his response was, “I do not think I can answer that. This volume (of reservoir) was drastically under-sampled,” he said.

The problem was the closest well was 30 ft above the well fractured and 75 ft beyond it. The company did not get samples from beneath the well, and Raterman would like to look at samples closer to it. Getting close to producing wells is risky, but he hoped that someday someone “would have the courage to drill a well that intersects another well.”

That brave soul would also need deep pockets. The cost is likely in the tens of millions of dollars and the high level of technical expertise is a barrier to these projects.  But, “Do you know anything without taking core?” asked Stephen Holditch, a professor at Texas A&M University who led the panel.

Since that 2014 test in the Eagle Ford, a lot has changed as the center of the unconventional universe has shifted to the Permian Basin and the industry used the downturn after the oil price collapse to revamp its completions methods.

In the past couple of years, a joint industry project with US government support built a fracturing test site in the Permian. The Permian Hydraulic Fracturing Test Site (HFTS) led by Laredo Petroleum, had a budget of $25 million with backing from 10 oil and service companies and the US Department of Energy’s National Energy Technology Laboratory.

That was likely a smaller budget than the ConocoPhillips’ test had, which drilled more wells at a time when service costs were higher. The Permian test used a slanted observation well to collect nearly 600 ft of core samples and data from two wells, one section near a producing well in the upper Wolfcamp, and the other near a lower Wolfcamp well, according to a technical paper delivered at another industry conference (URTeC 2697483).

The paper did not announce the group’s results—it plans to do so this year at the URTeC conference in Houston in July. Chances are it will not confirm the status quo.

“Whenever we directly measure created hydraulic fractures or the stimulated reservoir volume, we are always surprised and our mental images are altered and begin to change,” said Kent Perry, executive director of the Gas Technology Institute, who is involved in the HFTS project.

Growth Options

A theme in these rare studies is that the picture of “the number, location, dimensions, and characteristics of the created fractures, and altered natural fractures, is very complex,” Perry said.

The ConocoPhillips study observed that “fractures tend to swarm. It is not unusual to find 15 to 20 fractures in 10–15 ft,” Raterman said. And it concluded that much of the permeability created that led to production was due to hydraulically created fractures. Which is not to say that fracture formation was dictated by the fracturing design. Raterman said that “there is a weak correlation in the fracture swarms and the fracture spacing.”

At the hydraulic fracturing conference, Shell reported that fluids injected in one stage frequently traveled sideways outside and then inside the casing, ultimately entering the formation through an opening down the well.

High-pressure injection in one stage is thought to create stress shadows that can bend and stunt the growth of fractures in later stages. But the ConocoPhillips paper said, “The apparent side-by-side propagation of closely spaced, near parallel hydraulic fractures also differs from the output of currently accepted fracture models and may call into question the role of stress shadowing in hydraulic fracture propagation.”

Many Looks

An artist asked to create a realistic picture of a typical fracture would likely have been frustrated by the discussion about what fractures look like. The comments after all were aimed at engineers paid to generate production, not pictures.

These small cracks in the rock, which are small compared with their impact, can travel thousands of feet and their appearance changes based on multiple variables, not all of which are known.

Suarez-Rivera highlighted seven factors that can alter fracture development, from design decisions, like the number and spacing of fracture stages, to geologic factors, to how the natural stresses direct fracture growth. Some observations resist explanation. Propped fractures have been found far from wells in places sand should never reach. “It might not happen everywhere but it happens,” he said.

The path of fractures is affected by rock layering, weak intersections between layers, and the intersection of the many rock layers that can use step overs, where the path of a fracture takes a short sideways detour between the layers before returning to its original direction.

Injected fluid is likely to follow weak spots in the laminations between layers, making fractures more likely to grow out rather than up, said Suarez-Rivera. The ConocoPhillips study observed that fracture networks tended to be wider than taller.

Fracture growth is also controlled by local stresses. For example, if the minimum and maximum horizontal stress is nearly equal, fracture growth is expected to be more complex, increasing the appetite for more detailed maps of the stresses.

Mark Zoback, a professor at Stanford University who was the third speaker on the panel, is working on feeding that appetite with a new, more detailed stress map. The project for Zoback, whose name has long been associated with a widely used map showing geologic stress trends, has new data related to his other claim to fame: as an expert in induced seismicity—earthquakes linked to factors such as deep underground water disposal.

The surge in earthquake activity in Oklahoma and nearby oil producing states has led to stepped-up seismic monitoring producing the raw material for local stress maps, including a new one of the Permian from Zoback.

The Delaware basin in the western Permian “is one of the most interesting states of stress I have seen” with the axes of the minimum and maximum stress rotating from north to south, Zoback said. In contrast, the Midland basin is stable.

For those working in the field, or on models, there is a lot more data available now. The hard part is keeping up with it. “While much progress has been made, the complexity has continued to increase faster than modeling capability,” Perry said.