Structured Dataiku Flow
Ingest your data coming from different sources of your downstream operations stored on AWS, prepare them, analyze and get insights
The goal of this solution, built in collaboration with AWS, is to show central energy managers how Dataiku and AWS can be used together to create the foundation necessary for customers to drive business critical workloads for market competitiveness, enabling high-value initiatives including process optimization, predictive equipment maintenance, process planning, and GHG emissions management.
Within downstream operating facilities, there is a complex network of equipment and infrastructure. Facing increasing regulations regarding product specifications and emissions, downstream operations need to focus on operational excellence. Reliability of heaters, pumps, compressors and other equipment in the facilities plays an instrumental role in guaranteeing sustainability and efficiency of downstream activities. The Equipment Health & Maintenance (EHM) solution developed by AWS and Dataiku provides predictive equipment analytics to monitor equipment health and performance and boost preventative maintenance strategies for all refining activities. EHM helps reduce equipment downtime, improve reliability and performance while enabling organizations to optimize operations and reduce maintenance costs.
Thanks to Dataiku’s native integration with AWS Services and products, refining organizations can quickly scale with AI. Dataiku’s visual, end-to-end collaborative AI platform empowers users in cleaning and enriching data from the AWS ecosystem, building advanced data pipelines and machine learning models in a visual interface and easily deploying production-ready AI projects with built-in automation and monitoring. The solution comes with several pre-built Dataiku projects to accelerate delivery of critical use-cases: predictive maintenance, anomalies detection, and monitoring and surveillance. From data to outputs, all elements are available to take a strong start: preparation steps, modeling, results as well as end visualization with ready-to-use dashboards. All projects can be quickly customized to adapt them to specific needs.
Ingest your data coming from different sources of your downstream operations stored on AWS, prepare them, analyze and get insights
Transform the data coming from your different systems into a unified view where you can get operational insights on all the critical equipments of your refineries
Use Dataiku advanced analytics capabilities to understand the root cause of past equipment failures and identify inefficiencies in your maintenance operations.
Equipment such as pumps, valves, usually represent 10% to 15% of the capital investment in the refinery processes. They are also the most important type of equipment for plant reliability (followed by blowers and compressors). Using historical data of your pump failures, train ML models to identify the mechanical pieces with the highest probability of having the same problem in the future.
Once your model is in production, you can know the probability of failure, and efficiently identify two groups of equipment pieces: the parts to be checked in the 2 next days because they were most likely to have issues, and the parts that could wait a few extra days to be checked. Use Dataiku MLOps capabilities to monitor your models and detect drift in your data to make sure you have reliable results through time.