The Oilgas AI ecosystem applies artificial intelligence and machine learning. (Source: petrovietnam.petrotimes.vn)Hanoi (VNA) – The Vietnam Petroleum Institute (VPI), a member of theVietnam Oil and Gas Group (PetroVietnam), has built an Oilgas AI ecosystemusing artificial intelligence and machine learning. The system is used tocollect and analyse in-depth data in the oil and gas sector, with products suchas crude oil, petroleum, liquefied petroleum gas and natural gas.
As a partner of Microsoft, VPI is taking measures to optimise the analysis, useand management of data, thus helping firms improve work efficiency. The OilgasAI was built on the back of Microsoft’s Power Pages and incorporated with thelatest processing, analysis, connection and management tools such as Power Apps(app development), Power Automate (task automation), Power BI (interactive datavisualisation) and Power Virtual Agents (smart virtual assistants).
For each product, Oilgas AI performs an analysis of market data, priceforecasting, supply and demand, infrastructure, and inventory. It also help users easily find, tap and interact withvivid data through charts, reports, dashboards, and more. The data will beconstantly updated, making it easier for users to monitor real-time data.
In particular, petroleum suppliers, importers and sellers can monitor the retailpetrol price forecast through petroleum products at the Oilgas AI ecosystem.Its petrol price forecast model uses machine learning algorithms to model inputdata sources. It also continuously update new factors to provide the mostaccurate prediction of provisions and expenditures using the Petroleum PriceStabilisation Fund. It also inputs base prices of domestic and importedpetroleum products.
VPI Data Director Le Ngoc Anh said Oilgas AI targets becoming a network ofexperts, scientists, organisations, businesses and consumers at home and abroadon the back of using the latest technological advances like cloud computing, AIand machine learning.
This ecosystem creates acompetitive advantage through the optimisation of analysis, mining and use ofdata in a safe and convenient manner. It also employs the application and useof models and algorithms that have been developed by VPI experts and otherleading specialists in the oil and gas sector who are cooperating with VPI.Energy market products and services will provide solutions to help businessesmake faster and more effective decisions in daily production and businessactivities, as well as devise long-term plans and strategies.
VPI has started providing a free data mining service for users to experiencethe trial version of the Oilgas AI ecosystem at https://oilgasai.powerappsportals.com. Following the trial period, VPI willprovide optional service packages to meet the diverse needs of customers.
Previously, VPI also applied AI and machine learning to quickly identify thepresence of the fractured granite basement. The system has an accuracy ratingof over 80%, helping save exploration time and costs.
The discovery and successful exploitation of fractured granite basementreservoirs at Bach Ho (White Tiger) and other oil fields in Vietnam havechanged the traditional oil and gas exploration viewpoint and contributed tothe formation of a new one in the region and the world.
However, the identification of fractured granite basement by using traditionalmethods needs specialised equipment that may require suspension of oil welldrilling. This lengthens the rig rental duration (the current rental is about65,000 - 68,000 USD per day), and raise expenses.
To optimise this process, the VPI has worked out a fracture prediction solutionthat applies artificial intelligence and machine learning, helping to accuratelydetect the presence of fracture systems basing on real-time data.
Relevant data will be used as inputs into machine learning algorithms, and thenmodels will be classified, ranked, and assessed to find out the best forfracture prediction.
The VPI has tested the accuracy of this prediction model on drilling data of 12wells in some fields with similar geological structures, showing a predictionaccuracy rate of a whopping 80%./.
