Trace Id is missing
August 17, 2023

OriGen accelerates reservoir simulations by 1,000 times with Azure AI infrastructure

New York City–based software developer OriGen is revolutionizing the energy industry with proprietary AI models supported by Microsoft Azure AI infrastructure. OriGen’s AI solutions focus on three key areas—energy reservoir simulations, geological modeling, and operational optimization—to deliver extremely high speed and accuracy and empower better decision-making for energy industry customers. Using Azure AI infrastructure, OriGen has fast, easy access to the compute resources required to drive its NVIDIA GPU-based solutions and the means to deploy its powerful offering as a software as a service platform.

OriGen

Bringing AI to the energy industry

Like most businesses these days, energy companies are striving to do more with less. Historically, the energy industry has been slow to adopt new digital technologies, but the need to realize greater returns with lower outlay is driving the development of exciting new solutions. That’s where OriGen comes in. The company started in 2018 to focus the power of AI on solving energy industry challenges, such as accelerating reservoir simulations, using Azure AI infrastructure.

OriGen developed three primary AI technologies for its solutions. Its reservoir simulation solution uses available drilling and production data along with geological factors to simulate well and reservoir behavior. The company’s second major solution, geological model generation, rapidly generates realistic geological formations, allowing customers to quickly find geological models that match real-life data. With its third solution, OriGen conducts energy industry optimization by using AI simulation to help customers improve crucial energy management practices like factoring production yields, drilling, and managing wells.

The speed and accuracy of OriGen’s solutions is game-changing. In the case of reservoir simulation predictions, traditional CPU-based solutions consistently require thousands of simulations—with long processing times for each—to achieve an accurate prediction. OriGen managed to break that bottleneck. Ruben Rodriguez Torrado, Founder and Chief Executive Officer of OriGen, states, “OriGen’s models on Azure AI infrastructure reduce the time of a single simulation from days and hours down to milliseconds.” With that speed comes ever greater returns in accuracy as well because OriGen’s AI optimization improves through each simulation.

“OriGen’s models on Azure AI infrastructure reduce the time of a single simulation from days and hours down to milliseconds.”

Ruben Rodriguez Torrado, Founder and Chief Executive Officer, OriGen

AI assisted with Azure

OriGen developed and architected its solutions using Azure AI infrastructure that runs on Linux and is purpose-built for reservoir modeling and other AI workloads. The company takes advantage of the easy scalability and fast deployment of Azure resources based on fluctuating demand, along with additional Azure services used in its complex AI models. OriGen also values the ability to deploy its solutions as a software as a service (SaaS) platform. Torrado notes, “Our customers trust a lot in Microsoft as a cloud partner.”

The company runs its AI models using a combination of Azure Kubernetes Service (AKS), Azure Blob Storage, and Azure Virtual Machines. It chose the NVIDIA GPU-enabled virtual machines rather than CPU specifically because of the demanding need to run neural networks with billions of parameters. OriGen loads the data via Azure Blob Storage and sets up the configuration on its physics-informed neural network architecture on Azure, which uses AKS to automatically deploy the platform to customers. OriGen uses the newest high-performance computing resources, Azure high performance compute optimized virtual machines for multicore processing for several workflows, and a range of GPU-accelerated VMs (NCas T4 v3-series, NCv3-series, and NC A100 v4-series) powered by NVIDIA A100 Tensor Core GPUs, T4, and previous generation GPUs for accelerated computing essential to run its neural network models.

By using Azure Kubernetes Service, OriGen can provide stability and high reliability while eliminating the need for its customers to maintain their own infrastructure to run the AI models. Through its SaaS option, customers get sophisticated data management, elastic computing, and efficient resource utilization. OriGen considers it crucial to have cloud-based resources that it can scale effortlessly to meet its AI models’ demands. “With Azure, we can scale our workflows up and back quickly and easily, and we have access to all the latest hardware and software to run our simulation, modeling, and other AI workloads effectively,” Torrado says.

Making the impossible possible

With OriGen’s proprietary AI models running on Azure infrastructure, its customers get rapid simulation processing, more accurate geological models, and optimized workflows. According to one such customer, Ecopetrol, those add up to advantages in the marketplace.

As a major oil and gas provider in Colombia with an energy transport business as well, Ecopetrol partnered with OriGen in 2021 to see how it might take groundbreaking AI technologies into the energy industry to maximize efficiency and solve some serious challenges. Jose A. Villasmil, Reservoir Development General Manager at Ecopetrol, recalls, “We needed to improve and speed up our decision-making with the help of big data. OriGen was open to innovating and solving problems with us, and they presented a solution that really fit our needs.” Using physics-informed neural networks to process data far faster, OriGen and Ecopetrol increased efficiency across one of Ecopetrol’s larger Colombian reservoirs.

OriGen worked with Ecopetrol to conduct historical production matching on a complex reservoir consisting of thousands of wells and 40 years of historical data dating back to the ‘80s. By running thousands of simulations that factored data alongside geological plausibility, OriGen successfully created a model that matched 85 percent of the wells in the oilfield of study. “Everybody said that conducting a history match on more than 1,000 wells was impossible, but at OriGen, we make the impossible possible,” Torrado says. “We’re pioneers in using AI to conduct history matching with geological plausibility, and we delivered an 85 percent match by running our model on Azure AI infrastructure.” It also now runs at a speed 1,000 times faster than conventional simulations, and in many scenarios, it can run even faster depending on the complexity of the field, type of simulation, and hardware used.

With that enhanced accuracy and insight, Ecopetrol now makes more informed decisions about how to glean the highest yields and best manage its reservoir. This means more efficient work with higher yields more quickly attained. Villasmil states, “The biggest benefit of Azure AI infrastructure is faster time to market. We’re optimizing production in our fields using information from the OriGen simulations. It helps us make big decisions, like how much production can be extracted or how much fluid can be injected into our wells. Using AI simulations really shortens our decision-making time.”

Ecopetrol employees with specialized skills now make better use of their time, too. “Instead of having to spend time on a lot of back-end analysis tasks, our engineers get ready-made recommendations to maximize reservoir performance, which they can immediately start working with,” says Villasmil. Ecopetrol can deploy far more quickly and optimize efficiency in production. Villasmil continues, “Using OriGen’s AI simulation solution, we see things we simply could not have before. This will help us make our industry more efficient and effective each day. I think it’s going to be a big enabler.”

OriGen can focus on what it does best—developing and delivering revolutionary AI models to help transform the energy industry, like it’s doing with Ecopetrol—because it can depend on powerful, purpose-built Azure infrastructure to fully accommodate its AI workloads. Tyler Friesen, Chief Product Officer at OriGen, says, “We use Azure AI infrastructure because it gives us the ability to quickly move our customers’ data into our AI solutions so that we can deliver real value quickly and remove the bottlenecks.”

The company has big plans for its AI capabilities, like developing CO2 sequestration solutions, becoming a leader in hydrocarbon optimization, and exploring other renewable energy solution opportunities. It also aims to make the most of emerging AI technology, like adding ChatGPT functionality to its platform to enhance the user experience.

Having Azure AI infrastructure in place gives OriGen the freedom to take on complex problems within the energy industry. “Relying on Azure removes the extra overhead we would have to carry on our own,” says Friesen. “We can combine the trust our customers have in Microsoft with the value proposition of our technology to really move our solutions forward.”

Torrado adds, “With Azure, we have instant access to the speed and scale we need to test new AI simulation and optimization ideas, which was impossible in the past. And we can deliver our platform as SaaS, so, amazingly, our customers can run our complex AI models from an iPad.”

Find out more about OriGen on LinkedIn and about Ecopetrol on LinkedIn

Learn more about NVIDIA transforming subsurface workloads with AI and high-performance computing.

“We use Azure AI infrastructure because it gives us the ability to quickly move our customers’ data into our AI solutions so that we can deliver real value quickly and remove the bottlenecks.”

Tyler Friesen, Chief Product Officer, OriGen

Take the next step

Fuel innovation with Microsoft

Talk to an expert about custom solutions

Let us help you create customized solutions and achieve your unique business goals.

Drive results with proven solutions

Achieve more with the products and solutions that helped our customers reach their goals.

Follow Microsoft