Oil & Gas Production Analytics with OmniSci
Download OmniSci Free, a full-featured version available for use at no cost.
Predictive analytics, along with big data, is the nirvana of data analytics today. It ties into all companies managing technology resources whether it is an IoT sensor, automobile location or a well producing oil and gas. But even after going through costly digital transformation exercises, many organizations still end up with an unusable mass of opaque big data holdings.
All industries have challenges for viewing and analyzing their big data collections. Many times this means looking backward at historical information to understand costs and outcomes from projects or campaigns. While many solutions seek to solve this historic problem, the rubber truly meets the road when data assets are used to uncover future predictions to avoid failures or to inform capital expenditure decisions. This is particularly the case for the oil and gas industry as asset performance is the name of the game.
As the leading revenue generating companies globally, the oil and gas market collects massive amounts of data, especially around production. Finding, acquiring and building new wells, for example, is a particularly important task, but it requires pulling together many different datasets and looking at only a few scenarios at a time.
To demonstrate the power of having both historic data and predictive analytics all in one platform, we've built the publicly accessible OminSci oil and gas upstream production dashboard demo. While this is just a slice of the kinds of analytics needed in the industry, it is an easy-to-understand example of the power of visual analytics on massive production datasets.
Let's take a brief walk through some of the components and then take a look at how we can tie in predictive analytics for Acquisition & Divestiture (A&D) use cases.
Upstream Oil & Gas Production Demo
The production data is from OmniSci partner UpstreamDB and provides decades of historic data, including 256 million data points spread across more than 1 million US-based wells.
Historically, it has been challenging to view more than just a single basin of operations, forcing analysts to continually narrow their search trends and opportunities. Standard Business Intelligence (BI) tools and desktop GIS were simply not designed for this volume of data.
In this case, OmniSci is able to view data across multiple regions, or it can focus on all the data for a region of particular interest. For example, by interactively clicking on entries in the chart, users can easily select Delaware (left) and Midland (right) basins and display them in a single map to show the relative value of wells between them. Meanwhile, the charts show the well counts and a breakdown by vertical depth and number of barrels. It is possible to see trends and details with just a few graphics and a few clicks.
Understanding the drop off of production yield curves is another common requirement. In this case, there is a chart showing both oil and gas production declines over time, for both selected basins. To drill in, select one of the basins in the horizontal bar chart and the production charts immediately update. Here is a side-by-side comparison of the decline curves for the Midland (left) and Delaware (right) basins - what does it tell you about the relative production difference between the two basins?
Comparing the performance of these basins shows how diverse the production rates of different elements can be. For example, the oil versus gas ratio can vary greatly depending on basin and this is shown in the other chart on the dashboard. Both basins combined show the oil (green) and gas (red) total production—notice how they both rise together after 2014.
If you look only at Midland basin you will see that the trend is quite different, with gas dropping off while oil accelerated.
None of these insights in this region or with this data is particularly new for those in the field, however, the ability to interactively explore any dimension of the dataset, at the speed of thought, is groundbreaking. Powerful charting and mapping capabilities help draw out insight, in real time, that were otherwise impossible at this scale of data.
Analyzing a Portfolio of Wells
Looking at the historic data is only half the battle. While many platforms and tools choke on vast amounts of historical data volume, the real business value lies in predicting the future. CPU and GPU accelerated platforms must tie into common data science tooling that can leverage the accelerated capabilities.
OmniSci connects directly with the JupyterLab platform—opening an array of standard data science tooling that can take advantage of OmniSci acceleration. Easily pass a dataset from the OmniSci dashboard into a Jupyter notebook dataframe for further accelerated analytics, using GPU acceleration when available. Resulting predictive datasets can then be saved back into OmniSci for dashboard interaction.
When looking to estimate production for a new acquisition, users never have to leave the OminSci platform to get the needed insight. Using built-in tools, a very large training set can be applied to a particular subset of wells to provide future intelligence. The following example shows how predictions for each well are made possible with very little additional work.
More Analytics—More Data Science
Stay tuned for more on using Jupyter notebooks for this workflow for enabling this with your custom data. In a future demo you will see how we generated production decline curves for a set of specific wells that were offered as part of an Acquisitions & Divestitures (A&D) package.
This last stage ties into statistical analysis that can greatly accelerate how much data can be analyzed for investment purposes. No longer are users limited to analyzing small portfolios on their own.
Learn how the OmniSci platform is revolutionizing data analytics for large, complex oil and gas datasets. Watch for an upcoming webinar to see how, for the first time, companies can interactively inspect their data, gather insights and share results across the organization.