This blog post has been co-authored by Kapil Raval, Principal Program Manager, Microsoft.
Bluware, which develops cloud-native solutions to help oil and gas operators to increase exploration and production workflow productivity through deep learning by enabling geoscientists to deliver faster and smarter decisions about the subsurface and today announced its collaboration with Microsoft for its next-generation automated interpretation solution, InteractivAI™, which is built on the Azure implementation of the OSDU™ Data Platform.
The two companies are working together to provide comprehensive solutions combining Microsoft Cloud implementation of OSDU™ Data Platform with Bluware’s subsurface knowledge. As the world’s energy companies retool for the future, they are juggling priorities between new forms of energy, carbon emissions, and maintaining the growing demand for fossil fuels. Innovative solutions such as cloud computing and machine learning are playing an important role in this transition.
InteractivAI™ is utilized by the organization’s exploration and reservoir development teams to accelerate seismic interpretations and improve results by assisting geoscientists in identifying geological and geophysical features that may have been previously missed, incorrectly interpreted, or simply too time-consuming to interpret.
Using a data-centric approach, the application is unique in its ability, allowing users to train and infer simultaneously. Imagine running deep learning in real-time where the interpreter is providing feedback that the operator can actually see as the network suggests on-the-fly interpretations. This even includes results on data that is either not readily visible to the human eye or very difficult to see. This interactive workflow delivers more precise and comprehensive results in hours compared to months resulting in higher quality exploration and reservoir development.
The interactive deep learning approach
Bluware is pioneering the concept of ‘interactive deep learning’, wherein the scientist remains in the figurative ‘driver’s seat’ and steers the network as it learns and adapts based on the interpreter’s teachings. The adjustment and optimization of training the data set provides immediate feedback to the network, which in turn adjusts weights and biases accordingly in real-time.
Bluware differs from other deep learning approaches which use a neural network that has been pre-trained on multiple data sets. Users must rely on a network that was trained on data they have not seen, created with a set of unknown biases, and therefore something they have no control over.
The basic parameterization exposed to scientists in these traditional approaches gives the illusion of network control without really ceding any significant control to the user. Processing times can be days or weeks, and scientists can only supply feedback to the network once the training is complete, at which point training will need to run again from scratch.
The interactive deep learning approach is a data-specific approach that focuses on creating the best learning and training model for the geology the user is working with. Unlike traditional deep learning approaches, the idea is to start with a blank, untrained network and train it while labeling to identify any feature of interest. This approach is not limited to salt or faults, but can also be used to capture shallow hazards, injectites, channels, bright spots, and more. This flexibility allows the expert to explore the myriad of possibilities and alternative interpretations within the area of interest.
The energy company initially conducted a two-month evaluation with multiple experts across their global asset teams. The results were remarkable, and the organization is continually adding users. Additionally, Bluware has provided a blueprint for the company’s IT team for an Azure Kubernetes Service (AKS) implementation which will accelerate and expand this Azure-based solution.
A seismic data format designed for the cloud
As companies continue to wrestle with enormous, complex data streams such as petabytes of seismic data, the pressure to invest in digital technology intensifies. Bluware has adapted to this imperative, delivering a cloud-based format for storing seismic data called Volume Data Store™ (VDS). Microsoft and Bluware have worked together to natively enable VDS as part of the Microsoft Cloud implementation of OSDU™ Data Platform, where developers and customers can connect to the seismic data stored and provide interactive AI-driven seismic interpretation workflows by using the InteractivAI™ SaaS from the Azure Appsource.
Bluware and Microsoft are collaborating in parallel to support major energy customers through their seismic shift initiatives including moving petabytes of data to Azure Blob storage in a cloud-native VDS environment.
Revolutionizing the way energy companies store and use seismic data
Bluware designed InteractivAI™ not only with seismic workflows in mind but also with an eye on the trends shaping the future of the energy sector. Creating a cloud-native data format makes it scalable for energy companies to do more with their data while lowering costs and speeding up workflows, allowing them to arrive at more accurate decisions faster leveraging the power of Azure.
In 2018, a group of energy-focused software companies, namely Bluware, Headwave, Hue, and Kalkulo AS merged to become Bluware Corp. to empower change, growth, and a sustainable future for the energy sector.
As companies pivot from fossil fuels to cleaner energy sources, the combination of new industry standards, cloud computing, and AI will be critical for companies to adapt quickly, work smarter, and continue to be profitable. Companies that adapt faster, will have a significant advantage over their competition. For more information, visit Bluware’s website.
Source: Azure Blog Feed