Case Study | Enagás: Anomaly Detection in Cryogenic Pumps
Enagás is an international standard bearer in the development and maintenance of gas infrastructures and the operation and management of gas networks. Throughout their 50-year history they have built the main infrastructures for the Spanish Gas System, transforming it into a benchmark for security and diversification of supply. It has around 12,000 Km of gas pipelines in Spain. Since 2011, they have also developed and operated infrastructures outside Spain, and they currently operate in eight countries.
Enagás develops national and international projects to contribute to the decarbonisation process and to improve air quality. Their initiatives are focused on energy efficiency, the use of natural gas in transport, specially maritime transport, or the development of renewable gases.
Evolution to a Predictive Maintenance System Based on Your Own Data.
The maintenance of the cryogenic pumps installed in the Liquefied Gas (LNG) tanks is one of the most important tasks associated with the processes carried out in the plants.
The pumps have very changing operating conditions, and are submerged inside the tanks at an approximate temperature of -163ºC, which makes their maintenance complex and makes it difficult to make a clear judgment about the possible malfunctioning of the pump. Enagás wanted to evolve from its current model of planned maintenance to one that would allow it to anticipate possible failures by observing abnormal behavior in the data. In this way, maintenance can be scheduled before a failure occurs that would affect the quality of the gas supply.
Although these anomaly detection methods represent a considerable advance on ‘routine maintenance’ and ‘repair after failure’, their effectiveness is limited by the quality and detail of the information that is gathered.
Another extra challenge in Enagas use case is they don’t have any abnormal data point labeled so the proposed solution must be unsupervised.
Solution on AWS
Traditionally, in industrial environments maintenance processes have so far been carried out based on rules or advice from the supplier of industrial units. More recently, a conversion to a data driven approach is underway in these ecosystems to take advantage of the enormous amount of data being generated.
In the case of predictive maintenance use there is not much data on industrial units that have suffered failures (thanks to these maintenance policies). For this reason, an unsupervised approach has been chosen, under the premise that in general, the normal behavior of a machine should not vary excessively, and any sufficiently large changes in the sensor data should indicate a deviation to evaluate.
The proposed model compares the different measurements with each other and then infer whether a measurement is an anomaly or not based on how far is this one from the rest of the data. This solution provides Enagás a model for every pump helping their technical team to evaluate the functioning of every pump individually.
Finally, we have developed an API so that the trainings, inferences or updates of the models can be performed easily by business users.
AWS services used were as follows
AWS S3 as our main repository.
AWS API Gateway to implement all client requirements with API requests.
AWS Lambda to be used as an API controller and to process all requests.
AWS Sagemaker to train the models and evaluate new data.
Amazon ECR to store our docker containers for the models.
Amazon DynamoDB as a database for the pumps.
Amazon Cloudwatch to alarm the users via email if an error has occurred.
Benefits for the client
Enhancing Operations
Enagás’ main goal is not just to anticipate pump failures, but to detect anomalies in the operation and maintenance of the equipment early. This solution allows them to complement their current predictive maintenance program, with a prognosis based on their own data that will help reduce costs by optimizing the operation and extending the life of the equipment.
Better Decisions
Due to this project, Enagás has now more information about both normal and abnormal behaviour of pumps. This information will help to consolidate its know-how and to objectify the decision making.
Faster Processes via API Integration
Despite the number of pumps, thanks to the API, users can perform multiple trainings/inferences faster, making this task more agile.
Efficient Data Storage
Using Amazon DynamoDB we can store information about the pumps, their models and the hyperparameters. This information can be also shown by the API.
Keepler is a full-stack analytics services company specialized in the design, construction, deployment and operation of advanced public cloud analytics custom-made solutions. We bring to the market the Data Product concept, which is a fully automated, public cloud services-based, tailored software that adds advanced analytics, data engineering, massive data processing, and monitoring features. In addition, we help our customers transition to using public cloud services securely and improve data governance to make the organization more data-centric.
Would you like to talk about your business?
We can help you leverage the power of data to enhance your operations.