Learning Geomodel by its QC
In my early journey of learning how to build a geomodel, there was a time when I just started to comprehend of a geomodeling process, and my hierarchy asked me to do a specific project: “to simplify the quality control of a geomodel”, in which the objective is to make a practical list of tasks that need to be checked, prior layers of quality review meetings.
Before many things, the quality control definition, according to many online resources, is a process to ensure that a product has a maintained or improved quality. So, in my understanding, the QC job should be an important key step in a geomodeling project, since it is to ensure geomodel is well constructed, accompanied by a quality that can be accounted, ‘likely’ done at the end of geomodeling steps.
The project was started by gathering knowledges and experiences from geologists who worked on a geomodeling construction projects, concerning general geomodeling process and their practical way to QC a geomodel.
In addition, the project is also based on company guides and manuals for QC’ing Geomodel. If the simplification was also referred to a company guide and manual written by a colleague, a senior geologist, that already quite comprehensive and well organized. Why it needs to be simplified?
Later, I realized that we live in a period when everything moves fast, the data grows, the process and geomodeling workflow evolves. Along with demands of an accountable geomodel and also for various reasons in a customer focus point of view.
Long story short, the simplification project was done, by publishing a single page of A4 size, covered most of geomodeling workflows, categorized by layers of quality review meetings. It was an interesting project, because it helps me to view and understand comprehensive geomodel and how to build it.
Learning Geomodel by QC a geomodel, is like learning backwards, since the product or the object or the geomodel itself already done, while QC process make us looking every detail of the geomodel including the way of geomodel is constructed.
As Geomodel QC requires the knowledge of previous work, i.e the objectives of geomodel, the geomodeling tools, up to the geological background, behind the geomodel. This process helps us to get a general framework of why and how geomodeling is constructed, since (for example) every geomodeling software has their black box or every geologist has their politics.
If an A4 page of QC list can be more simplified, geomodeling QC should (at least) considered 2 main subjects, the 2-D’s, Data-results Visualization and Descriptive Statistics.
Data-result Visualization
It may be quite straightforward way to QC, that is to say, extracting maps from created properties. It is expected that all the initial geological concepts and modeling objectives are in harmony. Then, it is also required that input data are well respected in the visualization.
I used to overlay properties maps (i.e net-sand, porosities, net-pay, hcpv), expecting to have a good relationship on each map, checking all deterministic interpretations (i.e channel’s fairway, fluid contacts limit, dynamic units) are well respected visually, and also checking simulated values are being coherent to well data. Furthermore, well-logs visualization on correlations, stratigraphic markers, until logs-input comparison to blocked-logs is also essential, especially when there is possibility to lost some of important heterogeneities.
At this point, as nothing is perfect, some missed or 'anomaly' result can be expected, as long as it could be explained and accounted. It can be as small as twisted grid or serious problems as mis-interpreted facies.
Descriptive Statistics Review
Since we are dealing with data points, and interpolation/simulation process, this review is needed. When data analysis was well cooked, before the geomodeling processes, the QC process on the descriptive statistics can be comfortably to work on, by comparing data-input and output, hoping that the frequency, general tendency, and variation is in coherency to each other. This oftenly done by visualizing histogram and statistics summary. Sometimes, the statistics can explain some of 'problems' that is seen on maps or results visualization.
As Hernaldo Turillo in 'Important Benefits of Quality Control', said that quality control can reduce your inspection costs because you will have a better idea of your processes and more confidence in your company's ability to turn out high-quality goods without error. Also, Scott Young Newsletter explain about the idea of learning backwards.
In my case, I strongly agree. It benefits me, to have an idea of 'why previous geologist do this and not that', and allow me to better understand the geomodel workflow. Along with helping me to continue learning, as I think that every field is unique.
Finally, geomodel QC itself isn't easy task, especially for those who originally built the geomodel. But, thankfully, it helped me in comprehending the geology of the brown field, including geomodeling workflow at a time. The experience might not be perfect, but at least it gave me 'map' of foundation of where to start, even though (it must be) by learning backwards.
