|
3-D, well, core data, integrated in advanced reservoir imaging
Steve Sutherlin for Petroleum News
A revolution in machine learning and data analytics is rapidly expanding the capabilities of scientists to integrate multi-scale subsurface data and understand critical geologic features more quickly and in more detail than what was possible just a few years ago, according to Dr. Shuvajit Bhattacharya, assistant professor of geophysics and petrophysics at the University of Alaska Anchorage.
Bhattacharya employs advanced machine learning and geostatistical algorithms for facies, fracture and rock property classification, prediction and modeling. Before joining UAA, he worked in a few oil and gas companies and research organizations, including EOG Resources, Talisman Energy (now Repsol) and Battelle. He worked in various onshore and offshore locations in North America (e.g., Williston, Permian and Appalachian basins), Australia, South Africa and India.
Bhattacharya delivered a technical talk on “Integration of Tax-Credit 3-D Seismic, Wells, and Core Data for a Better Understanding of the Nanushuk-Torok Reservoirs,” May 31 at the technical breakout session at the state Geologic Materials Center in Anchorage. The session focused on the potential for new investigative technologies and machine learning systems to better assist geoscientists and resource companies to meet the challenges of interpreting Alaska geology.
“The Nanushuk formation is a thick fluvial-deltaic deposit, and the Torok formation is its basinward equivalent, mostly composed of sandstone and shale - and sometimes you’ll find coal in the Nanushuk,” Bhattacharya said. “The sandstone in the Nanushuk has been described by David Houseknecht (USGS) as coming from two different directions: it’s coming from the Siberian side from the Chukotsk Highlands ... and also we have the sand coming from the southern side of the Brooks Range which were deposited in different fashions.”
Bhattacharya said the Pikka discovery is in sands deposited along the shelf-edges; the basin-floor fan is the site of the discovery in Smith Bay by Caelus Energy.
Regular seismic data display not enough “The goal of my work is to identify these reservoirs using the seismic, well logs and core data,” he said. “We need a combination of all these three data sources to identify the lithology, porosity and oil and for that we have to use multiple different technologies because seismic and core alone cannot do everything that we want; we may get sandstone, but we may not find the hydrocarbons in it that we are looking for.”
Bhattacharya employed 3-D seismic data released from the Alaska Department of Natural Resources, which was made public due to having been funded in part by the state’s tax credit program.
“I’m going to show you the interpretations from different seismic surveys, and some of these surveys cover the discovery wells such as the Horseshoe 1 discovery well and western portion of the Smith Bay,” he said, adding, “I have some petrophysical data from a few wells, and I will show you micro-CT (“computed tomography”) scan data from the core samples today.
“I will talk about the seismic data first to understand the regional features of the Nanushuk and Torok. Then, I will talk about the petrophysical data to understand the reservoir complexity and identify the oil-saturated column in the rock. And then, I’ll show the micro-CT scan data to compute the effective porosity and fractures and use that for accurate reservoir estimates,” he said. “You cannot identify all the geologic features in the Nanushuk and the Torok formations using just one seismic survey. You have to use multiple seismic surveys to understand these formations at a regional scale and high-resolution, which may not be possible from some of the publicly available 2-D seismic lines, so I’m going to use three or four different 3-D seismic surveys and show you the results.”
Bhattacharya displayed a slide from the Horseshoe discovery in a cross-section from northwest to southeast, which featured the clinoforms as well as the Nanushuk and Torok intervals.
“The Nanushuk is expressed as the topset and the Torok and GRZ are the foreset and the bottomset,” he said.
“As a geophysicist, I’m more interested in the quantification of the reservoir, so I’ll show how we can use different types of seismic attributes to better understand the anomalies,” he said. “We computed a simple root mean square amplitude attribute ... we see all these anomalies on these clinoforms and we can map these anomalies in 3-D.
“We used some new attributes (such as coherent energy and similarity) that clearly show you the distribution of the high-energy sand bodies. You see two types of sand bars: one that’s along the shelf edges that continues for many miles, and also across the shelf-edge in the form of basin-floor fans with higher coherent energy anomalies.”
Bhattacharya said post-stack seismic amplitude anomalies, as regularly used in industry, must be calibrated and evaluated, “otherwise we might find water-saturated sandstone and that’s not what we are looking for - seismic amplitude anomalies have fooled us many times. We need to look at the pre-stack seismic data and calibrate that response with high-resolution well data.”
Branching out Seeking a regional scale, Bhattacharya took the study 50-60 miles or so northwest to the Smith Bay area, pulling up the northeast NPR-A 3-D seismic survey.
“It’s a really large seismic dataset,” he said. “We see both the low-stand and high-stand sand deposits, and we see the slumping of the sediment in the Torok.”
The study combined different types of seismic attributes together to better understand some of the geologic features.
“We see the nicely distributed sands along the shelf-edge and channels with high-amplitude anomalies, and we can see how this large canyon carved a deep area, transporting sand into a basin-floor fan,” he said.
“Sometimes you have the effect of the ice lakes and the permafrost on the seismic data that results in poor-quality subsurface imaging. … At UAA we are working on some new types of algorithms to remove the effect of the ice lakes and permafrost from seismic measurements,” he said. “It is going to be very helpful for many of the operators up here.”
The study then proceeded to the Puviaq 3-D seismic survey, southwest of the northeast NPR-A survey.
“Here we see the sinuous channels, and if we look at the amplitude attributes we don’t have the anomaly all along the channels,” Bhattacharya said. “What it’s indicating is that your sand to clay ratio is changing along the channels; that’s important for your well drilling in the future.
“As a quantitative geophysicist, I get excited when I look at these channels on the seismic, so I wanted to understand or infer the thickness of the channels because that’s important to estimate the net pay and map it.”
Accordingly, the study moved to spectral decomposition testing on the Puviaq 3-D survey.
“You decompose the seismic data in three different frequency elements in the zone of interest; in each, the frequency corresponds to thickness; high frequency indicates a thin channel,” Bhattacharya said, adding that channels of varying thickness are present in the Nanushuk.
Petrophysics; rock physics The study included three wells in the area of the Horseshoe 1 discovery well, including advanced petrophysical analysis, assisted by well logs and core.
Many operators divide the Nanushuk in different zones, as many as six or more, Bhattacharya said.
In the seismic and well data he found that inside those zones were multiple sub-zones, in which the ratio of oil to water can be mapped. The ratio can vary from the top to the bottom of each zone. Oil is present in a few sub-zones only. Identification and mapping of these sub-zones at the sub-seismic scale is important. Therefore, it is critical to integrate the results from the seismic with advanced petrophysical logs such as spectroscopy and Nuclear Magnetic Resonance, NMR, and core data.
“On the NMR log display, you see there’s a bimodal amplitude distribution that is indicative of hydrocarbons. When you put a magnetic field close to fluid, the protons in the fluid will align to the magnetic field,” he said. “Cut off that magnetic field, the protons will go back to the original position. If we have oil, gas or water, we’ve got different relaxation times related to the precession behavior of protons.”
In viewing the core data from the Horseshoe 1 well core under regular light, “you see the interbedded nature of the sandstone-shale sequence in the Nanushuk, of course, you see the oil-stained reservoir,” he said. “If you look at the same core under UV light you can directly identify the oil saturated zones.
“One of the interesting things we see that’s interbedded nature of these sand-shale sequences affects your well log responses, reservoir properties, and accurate reservoir estimates,” he said.
Bhattacharya also mentioned that rock physical measurement of the compressional wave (Vp) and shear wave (Vs) velocities (or Vp/Vs ratio) is critical to identify the oil-saturated sandstone reservoirs. This would be very useful to upscale core and log data to seismic scale and map the prospective sand bodies at the regional scale.
Pore visibility at the micron scale “We wanted to go beyond the regular workflow for reservoir characterization using 3-D seismic, well-log, and limited core data. We wanted to see the pore space in the rocks,” he said, “So we did micro-CT scans of some of the wells in the Umiat and other areas.”
“Micro-CT scanning (“computed tomography”) is an advanced non-destructive imaging technique that we can use to measure pore space in the rocks and fractures with the grain in 3-D ... for reservoir estimates and fluid flow simulations,” he said. “This is what we call digital rock physics; it’s one of the frontier areas of research now.”
The micro-CT scans of Umiat wells were displayed in video form.
The Umiat 1 well - a dry hole - had very few connected pores, compared to the Umiat 18 well. The size of the pores in the Umiat 1 well was too small to hold the hydrocarbon molecules, he said. In the Umiat 18 well, which had oil shows in the same reservoir, the technique indicated larger pore sizes of ~1 and 2 millimeters at places - good enough to hold hydrocarbon molecules, what’s more, the pores were connected to an extent, he said.
Bhattacharya moved on to an example of pore connectivity in the Torok formation, which exposed many fractures in multiple directions which help to understand the reservoir compartmentalization and mechanical issues that guide development planning.
“You will not be able to see any of these things if you just view the cores,” Bhattacharya said. “You have to do advanced imaging to see any of these things.”
What’s more important is the meaningful integration of multi-scale data for imaging and characterizing these complex reservoirs, he said. And, this has to be done quantitatively to make informed data-driven decisions.
- STEVE SUTHERLIN
| 