Reservoir characterization

What Do Fractures Look Like? A Picture Says A Lot, Even When It is Wrong

When it comes to fracturing, experts argue about many things, but they agree that fractures do not look like lightning bolts, tree roots, or shattered glass.

jpt-2015-05-fractureshero.jpg
The 300-ft tall fractured outcrop in upstate New York shows the lower portion of the gray Ithaca shale over the upper part of the black Geneseo gas shale.
Photo courtesy of Terry Engelder, Penn State University.

When it comes to fracturing, experts argue about many things, but they agree that fractures do not look like lightning bolts, tree roots, or shattered glass.

The sight of an image in a recent National Geographic story showing a hydraulic fracture that looked like a lightning bolt spurred Terry Engelder, a professor of geosciences at Penn State University to call the editor and complain.

The geologist, whose work has long focused on natural fracturing, was spurred to action by a long-standing complaint: “Invariably illustrations of fracturing look nothing like the real thing,” he said. Real hydraulic fractures are likely to look like natural fractures. Hydraulic fracturing typically opens existing cracks, which means the fractures will generally run along the planes of natural fractures.

So the picture of hydraulic fractures used to stimulate wells in the Marcellus Shale should be like the natural fractures crisscrossing the vast formation, which have been studied for more than a century.

Engineers have also noticed the problem with fracturing cartoons, and even SPE has gotten complaints.

“They do not look like tree roots. Things do not shatter like glass and run in all directions,” said Norm Warpinski, a Halliburton fellow for Pinnacle, Halliburton’s geophysical testing service. Fracture “complexity has to be consistent with pre-existing geologic features.”

jpt-2015-05-fracturesfig2.jpg
This illustration is typical of what is available online to show what hydraulic fractures look like. Experts say illustrations such as this are not representative of what fractures look like. Courtesy of FracFocus.

Some illustrations, like the one above are doubly bad because the design sends the message that drinking water aquifers are found near fractured zones, rather than thousands of feet away, creating a misleading impression that fracturing puts drinking water supplies at risk.

But these fracturing cartoons continue to appear, likely because they are easy to find online, and offer a quick fix for designers looking for a hydraulic fracturing image other than a photo of massed pumpers on a job site.

More life-like illustrations alternatives are rare. This story offers drawings that convey the blocky, angular reality of how fracturing follows natural fractures but they will not likely be the choice of an artist creating the cover for a technical conference program.

jpt-2015-05-fracturesfig3.jpg
This stream bed in the Utica formation shows the crossing planes created by natural fractures. Photo courtesy of Robert Jacobi, University of Buffalo.

Points of View

This story grew out of a speech by Engelder at the Unconventional Resources Technology Conference last August in Denver about what fractures look like. When others were asked that question, the comments and images collected offered many perspectives. They say things about what we know about fracturing, the ways it has changed, and how those views are affected by one’s professional point of view.

jpt-2015-05-fracturesfig4.jpg
A high-pressure stream of water injected from a well at the center of the illustration can spread out through a natural fracture system. Image courtesy of Maurice Dusseault, University of Waterloo.

 

Creating a simple picture of fracturing requires capturing the essence of something going on miles underground in rock that is complex, unpredictably variable, and only observable in limited ways.

When asked how he visualizes fracturing, Mike Smith, the founder and president of the fracturing consulting firm NSI Technologies, said after some thought, “My mental image varies by formation—by the geological environment. It depends on where you are.”

After decades of work on modeling fracturing and rock stress analysis, he is not one to generalize. One given is the path followed by a high-pressure stream of water flowing through a perforation into a reservoir will vary based on what nature offers in the form of interconnected fractures that can be reached from a wellbore. And the process is influenced by the alignment and relative strength of the stresses on the rock, both natural and manmade.

jpt-2015-05-fracturesfig5.jpg
These Barnett Shale samples collected in the Delaware Basin in west Texas by Julia Gale of the University of Texas show complex natural fracture networks.

“It goes to hit the first natural fracture and turns whichever way, then goes along until it hits another fracture, splits off again, and starts off across the country,” Smith said. His professional focus is on finding ways to stimulate rock to produce more hydrocarbons.

Engelder has a different time frame in mind. While engineers are likely to use the verb, fracturing, geologists are likely to describe it as a noun, fractures.

Engelder’s work on rock stresses helps explain why fracturing works and how natural fracturing altered huge formations over millions of years. Petroleum engineers, however, are focused on what happens over a period of hours while high-pressure streams of water and sand are injected into a well from many points along a wellbore.

Hydraulic pressure can change the physical forces affecting fracturing, such as the direction of stress, which has a major impact on fracture development. “It is irritating that there is so little work trying to measure stresses. That is very important,” Warpinski said. “One reason it is difficult to talk to geologists and geophysicists about what we are doing in a reservoir is they do not have a good perspective on how we are changing stresses during hydraulic fracturing.”

jpt-2015-05-fracturesfig6.jpg
Hydraulic fractures connect to natural fractures, reopening calcite-filled natural fractures, and following them as they turn. This illustration by Julia Gale, a research scientist at the Bureau of Economic Geology at the University of Texas, Austin, summarizes work done in the Barnett Shale. Image courtesy of AAPG Bulletin, v. 91, no. 4, p. 606.

A full picture of fracturing would cover a mind-boggling range of scales, from connected nano-size pores filled with kerogen, to micro fractures, and complex fracture networks in basins spread over multiple states. At each of those scales are experts, each with their own mental image of what affects production.

Combine all those points and it makes a simple picture for fracturing seem all the more unlikely.

One View

For a picture of fracturing, Engelder said the best illustration was by created by Julia Gale, a research scientist at the Bureau of Economic Geology at the University of Texas, Austin, which appeared in the AAPG Bulletin in April, 2007.

The image was based on her work on creating detailed descriptions of the birthplace of unconventional exploration and production, the Barnett Shale. What it shows is the natural fracturing in a brittle shale known for its complex fractures.

Natural fractures are offered as a template because fracturing is thought to use hydraulic force to take advantage of the path of least resistance, such as natural fractures that are either open or weakly cemented bedding planes. Maurice Dusseault, a professor of geological engineering at Waterloo University said the path is defined by the principle of minimization of work—the hydraulic fracture seeks maximum volume with minimum work.

The productive value of complex networks of fractures has given rise to the phrase, “creating complex fracture networks.”

That phrase can be a test of how a consultant explains fracturing. Hearing that phrase prompted Ibrahim Abou-Sayed, president of i-Stimulation Solutions to quickly respond: “You cannot create a fracture network! You can only activate the fracture network.”

His insistence is a reflection of the growing appreciation of the role natural fractures play in reservoirs where stimulation is also a given.

“There is a hotly contested debate about the extent that rocks in situ are filled with natural fractures. And if they are not filled with natural fractures to what extent are engineering fractures filling the volume,” Engelder said.

While Engelder sees natural fractures as the primary source of fracture networks, he said the odds are against a well providing conductive pathways to the natural fracture network without hydraulic fracturing, even if the wellbore contacts a modest fracture set. “If you drill a well and put in casing and perforate through it, the chance of that hitting on a natural fracture on the other side is not very good,” Engelder said. “At least initially I agree that as the fracture stimulation leaves the wellbore it has to make new fractures.”

Engelder and Gale’s work has been focused on early gas-producing shale formations, the Barnett in Texas and the Marcellus in Pennsylvania and adjoining states. Unconventional production has moved beyond those formations into reservoirs where the rock is not actually shale, though that label has stuck for unconventional plays.

Engelder said the work done on the physics of fracturing in black shale formations will apply in other unconventional formations where the geology is different, because “the physics that applies to one formation will apply to the next,” he said.

But based on experience and diagnostics, in the plays that followed early efforts in the Barnett and Fayetteville, in Arkansas, the rock looks different to those doing hydraulic fracturing.

“A lot of people took it for granted shale should be complicated. And that is true in Barnett and Fayettville,” Warpinski said. “In other areas we have to work on creating complexity.”

Sound Checks

The most direct way to image what is happening during fracturing is to listen to the popping sounds of rocks using microseismic testing as hydraulic force is applied. The sounds can be used to map the location of the events due to the force involved in the job—such as rock faces shearing—but cannot be used to draw a picture of the fracture network.

jpt-2015-05-fracturesfig7.jpg
Sounds recorded during microseismic testing (stars) can be caused by fractures near the wellbore (center) that are propped, fractures further out that are not propped, and some that are not connected to the well. Interpretations of actual microseismic data cannot reliably create such detailed pictures. Image courtesy of Maurice Dusseault, University of Waterloo.

“With simple planar fractures, microseismic can provide reliable information on fracture geometry. As the fractures get more complicated, as in many of the horizontal shale stimulations, it gets harder and harder to interpret the microseismicity,” Warpinski said. “Just connecting the dots to make a discrete fracture network is largely guessing about the fracturing results unless there is other supporting information.”

Warpinski has been a part of the evolution of microseismic from the days when he played a key role in a government research project at the Sandia National Laboratory to turn these observations into a useful fracturing observation tool leading to a competitive business with consultants trying to squeeze more meaning from the data.

Like ancient astronomers looking at connecting stars into constellations, some try to tie together that array of dots into a fracture network. They can be used to suggest geometric fractures or even tree roots depending on the point of view of the interpreter.

Some sounds may be the sound of rocks shearing as they adjust to the pressure change. Those may add an outer branch to the fracture network, which is unpropped and likely short-lived, or create fracture that is not part of the productive fracture network.

One thing microseismic has shown is that classic images of hydraulic pressure causing symmetrical planar fractures was a bad model in ultratight unconventional rock.

In the early days of microseismic testing, Warpinski said they used it to monitor fracturing in the Barnett Shale, which was then still far from proven.

“We knew in all reservoirs where we create fractures that they propagate along the path of least resistance,” he said. But in unconventional reservoirs, it is so unpredictable that early tests of microseismic led to questions about whether it was working.

“It went every which way. It was pretty complicated. A lot of people didn’t like that and thought it was a mistake,” he said.

That proved to be an accurate view of a formation that was the first of a generation of reservoirs that required a new a way of looking at fracturing.

“When you did a hydraulic fracture treatment in the Barnett, four or five wells nearby would get bashed,” with fluid killing production, he said. “Hydraulic fracturing fluid was not just going in one direction. It was moving out laterally instead of creating a nice planar feature. There was a wide zone of microseismicity.”

Changing Pictures

Fracturing pictures offer markers of change. One step in the evolution of microseismic interpretation was the stimulated reservoir volume (SRV). It estimated the volume of rock likely fractured based on the area filled by the swarm of microseismic signals recorded.

jpt-2015-05-fracturesfig8.jpg
Stimulated reservoir volumes (SRV) from three wells are shown in a 2010 paper. The concept was developed as a way to show the area that was likely stimulated. Source SPE 119890.

Slides showing the SRV were a common sight at the SPE fracturing conference in 2012, but were not seen in 2015. Warpinski said SRV was created as a way to visualize the impact of fracturing, based on what was known at the time. “We started calling it SRV—the data on fracturing was so complicated we looked at the area stimulated,” he said.

It fell out of favor because it was not a useful measure of the likely output from an area.  “You cannot do engineering by putting shrink wrap around an area and call it the SRV. How are the fractures performing? That is why we focus more on discrete fracture networks,” he said.

This year, a slide showing up again and again at the fracturing conference showed how the stresses exerted on the rock when fractures are tightly spaced stunts the growth of later fractures. There were multiple mentions on how adding hydraulic pressure at one stage can hinder fracture development at the next one.

jpt-2015-05-fracturesfig9.jpg
Colored chalk displays a fan pattern on this exposed fracture face, marking the gradual progression of a natural hydraulic fracture. Photo courtesy of Terry Engelder, Penn State University.

 “It is real and it is bad,” Smith said. He offered an example of a well in which the first and fourth fractures are dominant, stunting the two stages in between and said “putting clusters 17 ft apart is madness.”

The consultant has spoken to engineers at an independent producer, which is not a client, who would like to see the company move to fracturing designs that reduce the number of locations fractured to one or two per stage, which would be a significant reduction. That would reduce the stress interference caused by closely spaced fracturing for more productive wells long term, he said. But they have been unable to convince those in charge of operations that less would mean more.

Visual images of what is happening below are based on data that can be interpreted in many ways. The same set of production data can be seen as the product of a short fracture, which is highly conductive, or a long one that is less so. And where one expert sees stress shadowing, another sees formation differences. “I think everyone, largely has their point of view on that. And when they do have a point of view, it is difficult to get them out of it,” Warpinski said.

At the fracturing conference Smith said those doing fracturing need to do more to understand hydraulic fracturing rather than sticking with “Let’s do what we did last week. That worked pretty good.”

While using a fracturing model is an abstract exercise compared to drawing a picture, both represent an effort to gain a fundamental understanding of what is going on. That thought process may lead to a better way of doing things.

“What choice do we have if we really wish to improve? If we want to improve we have to use models and work to ensure we have the real physics in the model,” Smith said. “Outside the US they use fracture models routinely. They are a very powerful tool. In North America it gets a question mark.”

More accurate images of fracturing will require multiple data sources, Warpinski said. These range from microseismic testing data with pressure analysis, fiber-optic data gathering, well logs, production data, and even tiltmeters.

No one has found a way to see the productive fractures, and which sections of them are propped open, in the way a doctor can X-ray a bone.

This inexact reality has a political dimension. Dusseault served on a hydraulic fracturing review panel for the Nova Scotia government in 2014 and is currently advising with three other provinces. He has met with opponents of fracturing who argue that if the industry cannot say exactly where fractures networks go, it cannot be sure fracturing does not damage water supplies.

“People who are against development of fossil fuels, they criticize us for trying to predict fractures. Can you tell us exactly how long fractures are?” he said. While exact dimensions and shapes are not possible, the available methods do offer a dependable measure of the stimulated region ensuring that the fracture is far from causing harm.

Fracturing technology development has been focused on controlling the area fractured, an increasingly important consideration as wells are spaced more tightly horizontally and vertically. And increased in-ground monitoring is offering more measures of what rock has been fractured. Much has changed, but the cartoons used to illustrate fractures do not reflect it.