Engineers and analysts spend 80% to 90% of their time cleaning up hybrid datasets for machine learning models and analytics. Listen in as Novi Technical Advisors Ted Cross and Kiran Sathaye talk about oil and gas industry challenges with an emphasis on managing data quality, cumbersome workflows and spacing calculation complexity. Novi Data Engine addresses […]
Machine learning in oil & gas blog
Read through our posts covering real world applications of machine learning in oil & gas using AI-driven software and data tools. Learn from seasoned, experienced industry veterans as they go into technical topics related to machine learning, artificial intelligence, risk, and digital oilfield.
We typically publish a post every two to three weeks sharing our machine learning models and use cases for clients. Topics range from ML basics to deal evaluations and groundbreaking completions studies. We also feature thought pieces that cover bias, uncertainty and decision making, like this post on the risks of developing big discoveries.
We welcome guest writers working in the oil and gas industry. They work with oil and gas machine learning models using AI software to solve complex problems and optimize capital. If you would like to contribute a post or have an idea for us, send us a note.
Technical Advisor Ted Cross and Data Scientist Kris Darnell talk about why PDP and PUD forecast uncertainty quantification is so critical to making smart business decisions and how easy it is to get it wrong. Listen in as they take a deep dive into our machine learning technology and the data science behind it. For […]
Machine learning models that forecast production for PDP or PUD oil and gas wells may increase accuracy, save engineering time, or replace deterministic models in comparison to in-house methods. These are good reasons to switch to machine learning models and, not coincidentally, these are often the focal points of machine learning sales pitches. However, when […]
It all started as an accident. We passed early time production to a “pre-drill” Novi model. When we saw the results, we were stunned — they were that accurate. Machine learning for PDP oil and gas forecasting was the next frontier. What we had discovered is that existing methods were leaving a whole lot of […]
If, like me, you haven’t seriously played chess since you were young, you may find the advancements in chess artificial intelligence (or chess “engines”) incredible. But can chess teach us about the future of artificial intelligence in oil and gas?
Chess has long been a focus of artificial intelligence research. The game offers both clear rules (to easily measure success) and sufficient complexity (with at least 10^120 possible moves, no computer will ever be able to “solve” the game, unlike checkers). We certainly have complexity in the oilfield – but we also have the ability to measure success in the form of barrels produced, feet of pay encountered, or time to process a seismic volume.
In this post, we’ll walk through the latest AI in oil and gas market developments in chess and how they might impact the oilfield. Read the full post here.
What drives unconventional oil production? Of course completions, parent-child, and well spacing all play a huge role — but it all starts with the rocks.
We walk through a powerful new way of visualizing machine learning insights — regional profiles.
Are you the House or the Player?
Industry execGarth Stotts continues his series on bias, probability, and risk in the Shale patch with his latest message: successful investments in shale require an understanding of the REAL probabilities of your opportunities, from inputs all the way to cashflow.
Only then will you be in control — and be able to win repeatedly.
After the Great Coronavirus Shut-In, Vista Oil & Gas turned back on their new Vaca Muerta pad. A pleasant surprise greeted them. Their oil and gas machine learning model showed the highest production the play had ever seen. Both the MDM-2063 and MDM-2061 wells produced over 2,000 bbl/d on average for the month, with peak daily rates breaching 3,000 boe/d.
Why did these wells produce so much? Was it a novel completion design, a geologic hot spot, or a heavy tailwind of flush production? Lets explore this oil and gas machine learning use case. We will put these wells under the microscope with our Vaca Muerta public-data model.
WHY CARE ABOUT PXRF DATA? Thanks to relatively short scanning times and low operating cost, portable X-ray Fluorescence (pXRF) scanning of geologic core samples has become a burgeoning big data in oil and gas industry, with the technique capable of generating high resolution datasets containing comprehensive major and trace element abundances present in the rock.
We will be using Scikit-learn, an open source machine learning library that supports supervised and unsupervised learning towards a variety of applications. Here we will be utilizing principal component analysis and k-means clustering algorithms to manipulate and segment the big data before visualization.
The process outlined in this blog post develops labeled datasets that can be tied to the core, ahead of the implementation of a neural network. Read the full post here.
Can a machine learning model learn where the play sweet spots are, just from raw well logs? The answer is YES. Using logs (or any input geo variables), plus a principal components analysis, gives the model everything it needs to learn what drives production.
This is the subject of our most requested URTeC 2020 paper — geoSHAP: A Novel Method of Deriving Rock Quality Index from Machine Learning Models and Principal Components Analysis. Click through to learn more!
How often does an engineer dash off a simple produced water analysis, doing something like applying a flat WOR to their oil prediction? It’s easy to ignore water, but unexpectedly high production, leading to more produced water, can damage well economics. In the worst cases, it can force shut-in if disposal capacity is full.
Fortunately, advanced machine learning methods developed for oil can be applied to water. These technique help disentangle the complex interactions of completions, geology, and spacing.
This is the subject of our URTeC 2020 (and JPT-featured) paper. Read the summary here.
Using a single scaling factor instead of a time series can ruin your well economics. An upsized completion might increase your production 20%, but knowing whether that applies to peak rate or EUR can have a huge impact on your well economics.
Machine learning models can predict a time series of production. This means you can evaluate the impact of completions design over the life of a well. Read the URTeC paper summary here.
Machine learning can help you build unbiased benchmarks for operator performance! Operators, financial services, and investors will all appreciate this one.
We know that many current students have had their summer plans upended with internships canceled or delayed. To give those students a chance to still differentiate themselves, we are announcing the Novi Labs Machine Learning in Oil & Gas Blog Student Contest. We will pay $500 each to any posts selected for publication to our […]
Why do operators stay in oil and gas plays long after it’s clear that they are uneconomic?
There’s a big difference between finding Big Tuna and getting it in the boat. You must understand uncertainty and feedback loops to make good decisions. Data-driven models offer one potential solution.
Whether exploring for oil offshore Brazil or scaling type curves in Lea County, engineers and geologists rely upon the power of analogy to estimate the productivity of a given area or engineering design choice. Tree-based machine learning models can quantify similarity based on what matters: how it contributes to production.
Capital allocation decisions made by engineers that work at E&P companies are completely rational and based on unbiased P50 oil and gas forecasts.
If this is your belief, and it must be a belief, don’t read this post!!
When Whiting declared bankruptcy, we put their acreage under the machine learning microscope.
What do production expectations and breakevens look like across their Williston portfolio? Read the answers in this post.
When Parsley Energy announced their acquisition of Jagged Peak in October 2019, WTI was trading at $55/bbl. Though the all-stock nature of the transaction protected their downside, Parsley still faced the question: “how can we economically develop the Jagged Peak acreage?”
In this post, we study completions and spacing options to maximize acreage value. Click through to read our analysis.
Optimizing DUC inventory is something that many operators unfortunately have to wrestle with after pauses in completions activity.
Machine learning models can help you design the right completions for each well — and Lead-in sentence then understand the risk and uncertainty to prioritize your capital allocation.
In this post, we analyze Noble’s Delaware DUC inventory during the great coronavirus shut-in. How can smart operators optimize their activity to survive low prices?
In this post, we analyze QEP’s Midland Basin inventory, studying the impact of a range of completions designs.
DIGITAL OILFIELD TECHNOLOGY OVERVIEW : Let’s face it, the last two to three weeks have been a complete train smash for shale operators and investors. We are price takers at the end of the day, so we as an industry are all trying to do what we always do – pick ourselves up by the bootstraps and find a way forward. I have some ideas utilizing digital oilfield applications that we can use as a foundation to stand on.
In this post, I’m going to discuss where things are at, Level set on what response vectors the shale industry has taken thus far and much more. Read the full report here.
With the current oil price downturn, many have started to wonder whether activity in the Bakken will ever return to its previous heights. Using our machine learning models, we estimate that <10% of the remaining inventory is Tier 1, with the real number perhaps <5% due to surface constraints. With lower-tier locations predicted to produce […]
How much of the oil & gas industry’s recent underperformance comes from misapplication of large completions designs with tight spacing configurations?