CLH: how to predict demand, maintenance and leak detection with AI
The proofs of concept developed demonstrate that through the use of artificial intelligence we can predict monthly and daily fuel demand to achieve more accurate forecasts, which improve efficiency and timing in supply logistics.
By applying predictive maintenance techniques, we are able to forecast anomalies more effectively and thus anticipate potential transmission pump failures. We can also use computer vision for the automatic detection of leaks with continuous checking in real time, reliably and avoiding false alarms.
CLH, leader in fuel transportation and with more than 53 storage facilities worldwide, trusted Plain Concepts to develop three innovation projects in Industry 4.0 using Artificial Intelligence and Internet of Things (IoT), high-value technological services with great potential for cost savings.
Existing fuel demand forecasts were to be improved using techniques such as data analysis, data engineering and deep learning and integrated with the company’s existing tools.
In addition, plan a fully automated data and task flow, where predictions of potential failures would be updated on a daily basis. And finally, to develop a deep learning model capable of recognising plant leaks and adapting to changing environmental conditions.
Results
- Specific models were obtained for each station, on a daily and monthly basis. With very precise estimates, which would be scalable and adaptable in the future, to facilitate decision-making in logistics and personnel.
- Possibility of planning maintenance effectively, in terms of time and costs; avoiding the negative impact of a possible unforeseen failure.
- Real-time leak detection and alarming allows security personnel to react quickly and manage incidents in the most appropriate way.
We are ready for new challenges
