The energy industry isn’t born digital. Legacy portfolios and ways of working require a different approach to implement AI successfully.
The energy industry generates huge volumes of high value data on a daily basis - a modern offshore oil and gas platform generates 2 terabytes of data per day. SCADA systems and supporting instrumentation in offshore wind farms can produce up to 25 terabytes of data daily. 1 million consumer smart meters sample 96 times a day and produce 35Bn records in aggregate, equivalent to 2900 terabytes of data.
Safety, reliability and efficiency are critical for the energy industry. We are here to ensure that AI can deliver business value with no risk to people, equipment or the environment.
Rapid prototyping, Scrum and other agile frameworks work when technology and business cycles are moving very quickly, and mistakes can be amended with no operational impact. However, these approaches need to be adjusted for the energy industry where reliability, efficiency and HSE considerations demand more scrutiny.
Traditional ‘waterfall’ software development methodologies have been replaced by agile methods in many cases, and for good reasons. But there is a noticeable lack of proactive risk management in agile projects, especially for risks that are external to the project such as governance risks and operational risk. Companies in the energy industry rightly only want to deploy technologies that have been designed and tested with appropriate risk controls in place.
AI technology companies are often industry agnostic, and almost none focus exclusively on the unique challenges of the ienergy sector. They lack sector-specific knowledge, know-how and experience, making it frustrating for both sides to communicate, engage and deliver data-driven solutions that are immediately useful to the business.
Many businesses in the energy industry have not yet implemented corporate AI strategies, resulting in a disjointed and hobbyist approach to AI solution development. There are many good reasons to take a consultancy driven approach to each AI project and evaluating it on how it delivers strategic value to the business.
The energy industry has a strong view on the potential value of its data – resulting in restricted access to third parties. This is especially acute when there is a lack of a clearly defined business objective or when data safeguards are limited. Cloud computing has many advantages such as ease of access for storage and computing processing, but these benefits need to be weighed carefully against privacy, security and compliance requirements.
Storing massive amounts of data has become easy and relatively cheap. Preparing data for later use is time-consuming and many companies underestimate the effort required for data extraction, cleansing, normalising and wrangling. It is important to understand that machine learning still needs a lot of human ingenuity before useful results can be achieved.