Nobody knew where the oil and gas was trapped inside shale until a scanning electron microscope (SEM) showed the pores within the organic material inside.
Now there are a growing number of machines creating three-dimensional (3D) digital replicas of these rocks by combining SEM imaging with a focused ion beam (FIBSEM), which cuts off ultra-thin slices like a meat slicer, revealing multiple layers within.
These 3D digital images are incredibly detailed, but they are based on cubes whose longest side is 10–20 micrometers—millionths of a meter—making them small enough to fit into the spaces in porous sandstone. When it comes to building reservoir models, it leaves engineers wanting more.
“The real challenge is we are creating micron-sized images and dealing with square miles of resources,” said Carl Sondergeld, a professor at the Mewbourne School of Petroleum and Geological Engineering at the University of Oklahoma (OU). “If we can transcend those scales we can do a better job of characterizing reservoirs.”
In the pursuit of more, he and Chandra Rai, professor and director of the Mewbourne School, have built up one of the largest unconventional rock research laboratories, backed by 12 oil companies. The focus has been on understanding the workings of the nearly impermeable rocks, which has changed how some properties are measured.
The large lab displays Sondergeld’s constant hunt for improved tools for analyzing how rocks and reservoirs perform. The laboratory recently added the latest generation of scanning electron microscopes from one of the biggest makers of these instruments, FEI.
One thing that sets this FIBSEM machine apart is its ability to stitch together thousands of micron-scale images to create 2D pictures covering rock surfaces measured in millimeters rather than micrometers. These hyper-detailed pictures offer an overview of an area, with the option of zooming in for a closer look to see if low-resolution impressions are correct. They often are not.
Another plus is its ability to create images using extremely low power—as low as a 100 V compared with 30,000 V for older machines—creating images that may offer better ways of measuring critical details, such as the amount of kerogen—the organic material that is the source of the oil and gas.
And then there is the deal that brought that machine and two related devices to the lab—a technology development collaboration between the OU shale lab and FEI. For Sondergeld it is a way to gain access to state-of-the-art equipment that was not in his budget, and is the first of what he expects will be a series of such partnerships to help develop improved tools for
rock analysis.
For FEI the collaboration is seen as a way to make its machines faster and easier to operate. But more importantly it needs to make the machines more useful by finding ways to combine imaging with other rock-testing devices to generate meaningful numbers to help build better reservoir models.
For the industry, this collaboration is an example of how testing techniques are evolving to bridge the gap between the formation, whose scale is huge, and the defining details in these rocks, which are minute.
Making It Quantitative
Oil and gas exploration and production is FEI’s fastest growing market for FIBSEMs, but there is room for growth.
Weatherford Laboratories owns one and uses it to answer questions about the characteristics of unconventional rocks, such as pore distribution and morphology, said Mike Dixon, manager of geologic services in Houston at Weatherford Laboratories.
“Currently our use of the FIBSEM is more on our geological side, not as much on the engineering side,” Dixon said. “It is used to answer questions about what are the controls on reservoir quality and the porosity type.”
When it comes to measurements used to build reservoir models, such as permeability, Raymond Bruce, petrology group manager at Weatherford Laboratories, said that it relies on standard industry quantitative lab-analysis methods. Special sample-handling steps have been added to ensure repeatable results with shale.
While the company is steadily finding new ways to use its FIBSEM, serious questions remain about whether certain detailed measurements of little bits of rock are representative. “Engineers want data accuracy. They want to know if you say 12% it is 12%,” Bruce said. “Engineers think in black and white. Geologists not so much.”
The Big Picture
The rise of unconventional oil exploration has created a hunger for greater magnification. Prior to the shale boom FEI would sell a couple of electron microscopes a year to oil and gas companies, and not the most powerful ones, said Herman Lemmens, oil and gas market development manager at FEI. A top-of-the-line machine was not needed to see the details in porous, conventional reservoirs.
But in recent years, Lemmens said sales have soared for its most powerful FIBSEMs to oil operators and service companies around the world trying to better understand formations where they are investing billions of dollars with mixed results. The instrument maker is trying to replicate what it did in the semiconductor business, where its machines are part of the manufacturing process, Lemmens said.
The new machine that will be installed in the OU lab when building upgrades are completed creates images so large they are measured in gigabytes—1 billion bytes—showing details measured in nanometers—1 billionth of a meter.
An example of what it could do is an intricately detailed mosaic combining 12,800 images that it stitched together in the way some cameras can combine pictures. It is possible to zoom in and out of the 150-GB computer file covering an area the size of a thumbnail, like using Google Earth.
The presentation by Lemmens of that picture of a Marcellus shale starts with an overview, which looks like a granite countertop. The background is a range of grays that is spotted, streaked, and flecked with black spots where organic material is likely and the nearly white ones with minerals.
A much closer look at the dark spots reveal some looking like fractured lumps of coal that is not likely to hold oil or gas; and another, microns away, showing kerogen pockmarked with holes, which likely contain hydrocarbons.
For Sondergeld, having a machine designed to create detailed images of larger areas can make rock analysis more meaningful by revealing the complex fabric of the rock. Now, digital rock analysis is based on picking 20 or so locations for 3D imaging, based on measurements that do not offer nearly as much detail.
But it took 5 days to create that image. Methods have improved since then, reducing the time to 3 days. Either way, Dixon said it would be impractical for testing labs to spend that sort of time on imaging one sample.
Based on user feedback, Lemmens said, “They do not want to take three days. They want an image that takes one hour.”
Multiple Sources
Shales vary in unpredictable ways. Anyone sampling these heterogeneous rocks must have a measure of the degree of variablity to determine how many samples are required, and whether any single small sample is representative.
That degree, and an understanding of the features to be found in the reservoir area, will determine the number and location of the spots needed to use small samples to make useful inferences about a larger area.
Those doing rock analysis generally start with lower-resolution wide-area scans to get a general understanding of a sample, and then target specific areas for further detailed examination. When analyzing a core sample, Weatherford and some other testing labs start by doing a computed tomography (CT) scan which can detect changes in the makeup of the rock (lithology) and its density—for example, organic material is less dense than minerals—and fractures.
CT use, which goes back 15 years at Weatherford, has grown with the shale boom, leading to the purchase of a faster machine capable of scanning sections up to 200 ft in a day or 2.
The industry keeps searching for better, quicker, more detailed views of these difficult rocks. “Most recently we have added some new tools for shale analysis,” Dixon said. The challenges include characterization of very small pore systems as well as variability on a larger scale in rocks known for heterogeneity. “One advantage we have is not having to use one analytic tool. We look at a lot of different qualities.”
At Baker Hughes multiple SEM machines are used in conjunction with a growing array of reservoir analysis tools using various approaches— atomic force microscopy (AFM), nano-infrared (nano-IR) spectroscopy and Raman microscopy, said Gaurav Agrawal, director, enterprise research at Baker Hughes.
One of the devices lent to OU is FEI’s latest Qemscan, which does detailed mineral analysis. A smaller, more rugged version of that device is leased to oilwell service companies, which have adapted it for wellsite mineral analysis of cuttings. Weatherford’s Wellside Geosciences unit is a competitor in that service.
The goal of this hunt is to combine data from multiple tools is “to extract more from each image and measurement,” Sondergeld said.
Lemmens hopes to see a future wellsite device that will be able to measure the full range of rock properties, including permeability and porosity, using small core samples or cuttings, which are cheap and plentiful. This could be particularly valuable in countries without local labs, where getting rock to a laboratory for testing can mean long delays, said K.S. Chan, global director for oil and gas for FEI.
Sondergeld is looking for ways to build a wider range of testing capabilities into the FEI machines. One possibility is adding a nano-indenter, which tests properties such as hardness and brittleness by applying an exact force on a tiny area.
“What is happening in the instrument world is integration of all of these instruments,” said Sondergeld. “Adding multiple measurement systems can maximize the information you can extract from the same sample.”
His interest in the methods used for rock analysis in commercial labs goes back to one of the first problems posed by companies supporting the OU shale consortium: Different commercial labs often provided different measurements of properties, such as permeability, and the operators wanted to know who was doing it correctly. Sondergeld said that prompted more testing labs to disclose their measurement procedures, but greater openness is needed.
Weatherford said it has openly discussed its methods for determining shale rock properties such as porosity and permeability.
Now Sondergeld is seeking to change the tools used to measure unconventional rocks. “I view this as a different direction for university and industry cooperation,” Sondergeld said, adding “I think this will happen more and more frequently in the future.”
An Element of Surprise
The near-term goal for the OU-FEI collaboration is well defined: create easier-to-use machines that are able to consistently produce better images and quicker analysis.
FEI will be sending an employee to observe how things are done at OU—where Lemmens said the quality is consistently good—with an eye toward incorporating those techniques into algorithms that automate the process of creating multiple images, perhaps prompting the FIBSEM operator to adjust the settings for improved results in later shots. FEI will also be offering regular training programs for users on how to better use the devices.
In the longer term, the expectations are not as defined. They will be looking for practical uses for its new machine’s unique features, particularly its ability to operate at extremely low voltage levels. “I do not know what we will find. It is like being near-sighted and not realizing it until someone gives you a pair of glasses,” Sondergeld said.
The Mewbourne lab is being asked to combine what is observed with all rock analysis techniques it has available to determine what microstructural details govern production and create systems to efficiently and effectively test for those properties.
Even before the new FIBSEM machine was up and running, he was excited by images he had seen that were created using power levels approaching 100 V. The organic material appeared to be outlined by a bright, irregularly shaped halo that he had not seen in images created using higher power levels.
Early work in this area came from a project between FEI and Christopher Laughrey, who now works for Weatherford Laboratories, using a sample from his former employer, the Pennsylvania Geological Survey (AAPG 90122).
This phenomenon may offer a more efficient, objective way to measure the organic content, Sondergeld said. Currently the components and their quantities are determined by segmenting (thresholding) the various shades of gray in the image.
But work is needed to determine if low-voltage images can be used for measurement, or even why these look so different. “It may be a few years before we can use all the information in these low-voltage images,” Sondergeld said. “But the immediate benefits are obvious.”
