Cutting edge AI technologies such as Deep Learning, Machine Learning, and Reinforcement Learning are transforming the way that geoscientists work. Welcome to an interview with Naser Tamimi, NeuDax. NeuDax participated in AAPG’s U-Pitch at URTeC, which promotes and helps commercialize innovations and new technologies.
What is your name and your background?
My Name is Naser Tamimi and I have PhD in Geophysics from Colorado School of Mines. During the last 8 years in the oil and gas industry, I worked on more than 50 different oil and gas projects. From conventional to unconventional, from middle east onshore projects to offshore Brazil, and from engineering projects to advanced seismic processing. Since 2017, I started working with a few Colorado School of Mines alumni on a very exciting startup called NeuDax. NeuDax promise is to bring the modern AI (not the classic machine learning) to the upstream oil and gas industry.
What is your product and/or your process?
NeuDax main product is an AI Platform (SaaS) called FracDax. FracDax helps unconventional oil and gas decision makers to plan for well drilling and completion using cutting-edge AI technologies such as Deep Learning, Machine Learning and Reinforcement Learning. Our users are oil and gas operators or investors who are planning for a new drilling program in unconventional O&G fields. FracDax helps them to analyze hundred thousand of different well and completion design combinations in a few hours using advanced AI algorithms (it takes years for regular reservoir simulations to do it) and using optimization algorithms find the best well and completion designs in addition to estimated costs and economics evaluations. Then through interaction between the user and the AI platform, FracDax tries to understand the user’s utility function and rank the best designs. In a few words, FracDax is offering three levels of AI: Predictive, Optimization and Cognitive.
How is your product or process innovative, even potentially game-changing?
First, to our knowledge, it is the first effort in the upstream oil and gas industry to take advantage of Deep Learning and Reinforcement Learning technologies for well and completion design. In addition, the predictive algorithm behind this platform is one of the fastest and most accurate predictive models. Last but not the least, the cognitive unit that tries to understand the utility function of the user is a new technology not in the oil and gas industry, but in the other industries too. We believe companies and users have different multi-objective goals in their drilling programs. What might be a successful drilling project for Company/User A is not necessarily the best scenario for Company/User B (imagine a large E&P company compared to small investment company). Understanding the best outcome for each company and the corresponding well and completion design is the job of our unique Cognitive technology.
What made you recognize that there was a need for the product or process?
Two types of companies took our attention to this problem. First, we noticed many O&G operators make decisions after analyzing a few scenarios. We noticed that that they are losing up to 15% of their well potential production just because of not considering more drilling and completion options (without significant additional costs). The second group was the O&G investment companies who should decide about buying an asset and they need very fast valuation. For them they were two ways: months of valuations using reservoir simulations or fast decision without enough knowledge. We felt the market needs a product that predict and optimize the well and completion designs in a very fast way and increase the production of wells with considering all the scenarios.
Please describe an early example of implementation or product development.
We tested this approach in DJ Basin and Powder River Basin. Our initial results show that the production of many recent wells in DJ Basin can be increased up to 15%. Also, it showed that just a few wells in DJ Basins were drilled close enough to our AI recommendations and interestingly they are being considered as very successful wells in this basin. We are talking to a few investment companies to test this approach for asset valuation too.
Please describe a more recent example of implementation or product development.
The same as the previous answer (for us).
What were some of the lessons learned?
The oil and gas industry have not collected the decision-making data efficiently. When we ask companies why they made this decision (for example in well design) and not the other alternative, we usually notice that the companies and business units don’t remember the reasons and they don’t have good documentation of their decision-making processes. It is believed that just recording decision making data using an AI decision support system can increase the efficiency dramatically.
What are your next steps and short-term goals? What do you want to do next with the product or process?
The short-term plan is to finish the product and bring it to the market in 2019 for everyday users. We also need to expand our platform to basins like Permian, Eagle Ford, Bakken and Marcellus.
What are your long-term goals?
Until now, we have focused on AI applications before drilling. Our future platforms will cover the post-drilling challenges.
Please recommend a few books that you found inspiring.
To understand the importance of AI in decision making, I highly recommend “Prediction Machines” by Agrawal, Gans and Goldfarb. Also, “Unscaled” by Taneja is giving you a good understanding of AI revolution in different industries.
For those who like to learn the technical aspects, I have several recommendations:
- Data Science for Business by Provost and Fawcett
- Deep Learning by Python written by Chollet
- Decision Making Under Uncertainty by Kochenderfer
- Shale Analytics by Mohaghegh
What did you find valuable about U-Pitch?
For me, critics were useful. As developers of an early product, we don’t have too much chance to interact with future customers like those who have their product in market. Therefore, usually we miss this useful feedback. But at U-Pitch, I could hear the critics, as well as the excitements, and I am happy that I participated.