Chalk has a very low permeability. Only in fractures, can oil flow more easily. But where are these fractures, how many are there, and what is their orientation? New fracture models that simulate the growth of the fracture network through geological time may give an answer. If successful, this technique could have significant commercial potential, not just in the Danish North Sea but worldwide.
Chalk is an important hydrocarbon reservoir rock in the Danish North Sea. It is made up of the skeletons of tiny plankton, crushed together very tightly. Because these grains are so small and so close together, chalk has a very low permeability, making it difficult for oil and gas to flow to through the rock to the wells.
However, chalk often contains cracks or fractures, formed when the rock has been subjected to stress over millions of years. The oil can flow much more easily through these fractures, so they play an important role when trying to control the movement of fluid through the rock. In order to plan how best to manage the chalk oilfields – where to drill wells, what pressures to maintain, etc – it is important to know where these fractures are, how many there are, and what is their orientation.
Unfortunately, you cannot see these fractures directly, except in a very few places where you have wells – and even then, you do not know how long they are or where they lead to. Standard industry practice is to build stochastic fracture models, where fractures are placed in random locations and given arbitrary sizes.
3D model of a North Sea oilfield, showing the fracture patterns predicted using the new technique. The geological layer (yellow) is cut and offset by a number of large faults (purple). Structures this large can typically be observed and mapped from seismic data. However, the fractures that control fluid flow in the field, shown in green, are too small to see on seismic data. The fracture pattern shown here was therefore generated by simulating the fracture growth, based on the large-scale geology and geomechanics. The close-up shows the level of detail possible to simulate.
These models are designed to be consistent with the fracture densities observed in the wells, and possibly with geophysical data or with fracture geometries in similar chalk exposed in onshore outcrops. However, there is still a huge amount of inaccuracy and uncertainty in these models, and they may not be geologically consistent. Therefore, they are poor at predicting fluid flow through the chalk.
The idea was to see if you could build much more realistic and accurate fracture models by simulating the growth of the fracture network through geological time, based on fundamental geomechanical principles. The hope is that this will give a fracture model which is consistent with the local and regional geology, and which can be built quickly and easily. This approach would also allow you to test the range of possible fracture geometries that would arise in different rocks under different geological histories, and therefore assess the amount of uncertainty in your predictions.
The model is built as a plug-in to existing software, used to create 3D models of the geology of oil and gas fields. This allows you to apply the model, not just to oil and gas fields, but also to fractured rock outcrops onshore, where you can check that the model predictions match the observed fracture patterns.
In order to plan how best to manage the chalk oilfields – where to drill wells, what pressures to maintain, etc – it is important to know where these fractures are, how many there are, and what is their orientation.
So far, the model has been tested on several onshore outcrops and in all cases, there is a good match between observed and predicted fracture geometry in all cases. Currently, the model is being tested on chalk fields in Denmark’s North Sea together with Total E&P Denmark.
If successful, this technique could have significant commercial potential, not just in the Danish North Sea but worldwide, including other industries such as geothermal energy and CO2 storage. Patent protection for the technique has been submitted, while deciding on the best way to take this forward for commercial application.
Michael Welch and Mikael Lüthje are both Senior Researchers at DHRTC
Simulating the growth of a fracture network through time: in this simplified model, fractures nucleate first in a small high stiffness zone, and then slowly propagate outwards through the rest of the rock layer.
Comparison of predicted (top right) and observed (top left) fractures at Robin Hood’s Bay, NE England. The model is able to predict the observed variations in both fracture orientation and geometry around the outcrop. Notice that in the first location the fractures are highly anisotropic, with a primary set of very long fractures connected by a secondary set of short fractures, while in the second location the fractures are much more isotropic, with long and short fractures in both orientations. Outcrop mapping is from Rawnsley et al. (1992).