News

Deep diving into data using specialised plots helps us to move from basic observation to resource insight.

​

Over the last few years, I've developed unique approaches to displaying borehole image data that improve interpretation and reduce the impact of bias on our results. In this post, I discuss geometric sample bias. In the next post, we will look at the importance of quantifying fracture width/aperture and using it when plotting data.

​

These lower hemisphere stereonets plot the poles to planes for data picked from two borehole image logs acquired at the Fish Lake Geothermal System. This figure is adapted from my 2023 paper (download here). 

​

I include contours of geometric sample bias on these stereonets. Fractures and beds that are perpendicular to your well have a high probability of being intersected. In contrast, those that are parallel or near-parallel to the well have an extremely low probability of being intersected and cause a data gap (the blind zone, highlighted yellow).

FL-1 is a vertical well. This means that the absence of high-angle fractures from this borehole image does not guarantee that high-angle fractures are absent from the reservoir. Instead, this absence may just be a function of the geometric sample bias.

​

Key takeaway: Understanding and visualising sample bias prevents us from over-interpreting data gaps.

​

The deviated well FL-3 intersected numerous high-angle fractures and some of these occur within the blind zone. Fractures with the orientation indicated by the red arrow are sufficiently numerous in the reservoir to balance the diminishingly small probability that they will be intersected. The grey squares indicated by the arrow are the top and bottom surface of a fault gouge. This is likely to be an important structural trend.

​

Key takeaway: Considering data frequency in the context of sample bias and geologic processes enables us to determine relative importance of fracture clusters.

​

The theory of geometric sample bias was first developed by Ruth Terzaghi ("Sources of Error in Joint Surveys". Geotechnique, 15(3), 287–304, 1965). Her proposed correction is built into most borehole image log software. That correction weights the available fracture data based on the angle between that fracture and the borehole trajectory. Because this correction weights the fractures observed in the borehole image, it cannot fully correct for the blind zone (c.f., the FL-1 example). This is why it's important to plot and consider the blind zone, even if a Terzaghi Correction is applied.

​

Key takeaway: When applying corrections to data, it is important to understand how they operate and their limitations.

​

If you would like to include contours of geometric sample bias on your stereonets, check out my open-source Python library fractoolbox.

​

[click on the image to expand]

I was recently invited to give a talk to the borehole image special interest group of the Society of Petrophysics and Wireline Log Analysists (SPWLA). I took the opportunity to challenge peers to find ways to better characterise fracture width and aperture. For those like me who work in fractured reservoirs, these data are critical for identifying the key structural features and determining which of them may influence fluid flux. Our current methods are flawed, especially the automated modelling of fracture aperture. We need more validation studies and research into width/aperture characterisation for all host rock types. My talk is available on the SPWLA YouTube channel. 

I've been developing a document that helps geoscientists and managers navigate the path from data to insight about what controls permeability in a geothermal resource. This new, updated version can be downloaded here. Free to use with acknowledgement.  

I've launched a new, expanded Cubic Earth website to highlight projects and services. Get in touch to let me know what you think.