This year oil and gas companies have been hit hard by a huge drop in oil prices, COVID-19 and (geo-) politics. On top of that, the industry has committed to Net-Zero programs due to environmental pressure. Companies are struggling with how to stay profitable and build a sustainable future under these pressures.
With reducing incomes, necessary investments, and changes in business models, data is crucial to come to new solutions. Digitalisation is happening in each area and department within the O&G sector. In some areas digitalisation is just starting, while in others data has been gathered for years. Operations has collected sensor-generated time-series data for years. For major critical equipment multidisciplinary analytics projects, with help of data scientists may have been a success helping organisations predict when maintenance is needed or reliability could be increased.
With the success of the data modelling projects, the data scientists became overloaded, while being scarce in the market. A higher pressure was put on the process engineers to analyse the data themselves, but the tools they had available did not lead to the same positive results as seen in the data modelling projects. Limited successes were achieved by Excel wizards. This resulted in deception and a reduced belief in the usefulness of data analytics by subject matter experts like process engineers.
Modern technologies and applications make it possible for new tools to work with the large volumes of data captured during production in siloed business applications. By giving tools to the people who can interpret the data best, the expectation is that the largest number of improvements to operational performance can be gained. This actually is proven in many use cases that process and asset experts were able to perform with such a modern tool like self-service industrial analytics. Self-service industrial analytics lets the users tap directly into the enormous amounts of time-series data, contextualise it with data from other business applications and use pattern recognition and machine learning to understand the data. This is what we call the democratisation of data, where the power of analytics is put in the hands of the many operational experts at the production facilities.
Figure 1. TrendMiner's Production Cockpit allows optimal surveillance and monitoring of wells and major exquiment. Image courtesy of TrendMiner.
Also in the oil and gas industry self-service industrial analytics is proving its value, resulting in rapid adoption and an increasing array of use cases. Some interesting upstream cases engineers have worked on are, among others:
- Well surveillance and optimisation
Starting up or beaning up too fast will cause a well trip sooner than expected, but the relationship between the bean-up time and well performance have only been a hypothesis but have not been quantified. After analysis using pattern recognition and creating a fingerprint from multiple good bean-up executions a quantified link between bean-up time and trip probability was created. Guidelines and exception-based surveillance was established, resulting in well integrity improvements, overall production gain and a reduction in trips to flare.
- Offshore compressor monitoring
At an offshore platform there has been a rework of the electrical system, which led to an increased number of compressor trips. In only a month there were four trips.
By using pattern recognition of a trip using the gas flow and overlay similar trips, the trips were statistically compared with normal operating periods. This helped verify the hypothesis that a valve had abnormal behaviour. Further investigation shows erratic or drifting behaviour of the identified valves several hours before the compressor trip. Based on the analysis, a monitor was created and used in a dashboard, functioning as a production cockpit for the operators and engineers. The real time monitoring plus trip prediction led to 1-2 hours per month of avoided downtime considerably reducing production deferment.
- Unexpected flare flow issues
An oil rig operator in the North Sea experienced an unexpected increase in flare flow for about 15 minutes. Although not a critical event on its own, leaving the root cause unchecked could lead to more severe incidents in the future. Furthermore, more frequent occurrences of similar events could add up to significant losses. The engineers found multiple similar incidents had occurred, which was more frequent than expected. The patterns were overlaid to graphically see the similarity in process behaviour. Further analysis showed the severity and the maximum volume of the flare.
Figure 2. Pattern recognition and machine learning are the basis of self-service root cause analysis. Image courtesy of TrendMiner.As the software uses machine learning, it can recommend possible influence factors for the issue. This helped the team to find different behaviour before the time of the flare and at the time of the flare. Through the software, they could quickly find the root cause of the unexpected flare: every time before the flare, a valve closed too quickly. They easily eliminated this by changing the control logic.
By eliminating the source of the flaring, the engineers were able to reduce emissions and related process safety risks. On the other hand they were also able to improve the reliability of the critical valves. And on the human side, the engineers now were able to remove the recurring source of distraction for the operators and engineers.
Concluding thoughts
Besides the described practical use cases, showing how engineers can use the data to improve operational performance, there are many more areas self-service analytics is bringing great value. Analysing the data can be done in low production periods and even in the current COVID-19 situation where engineers are working from home to search for improvement areas. The use cases are related to Net-Zero initiatives, asset reliability, safe operations, and efficiency improvements all leading to a reduction of production deferment. Empowering operational experts with advanced analytics is the key to leverage the ongoing data science projects or existing models. The combination of both initiatives can help oil and gas companies to move to next levels of operational excellence and face volatile market pressures.