Seeing is Believing !
When executing a Digital Transformation strategy, the introduction of new technology and new ways of working can make many people uneasy. Everybody believes they should be going digital, but they don’t necessarily know how to do it. The benefits of implemented solutions may not be tangible at first sight, and the best approach to implement a project is not always clear. It becomes difficult then to justify an investment without proven results. Which is the right direction? What is the right first step?
The First Step
Asset Predictive Analytics is an important piece of any companies Digital Transformation strategy. More and more, it´s becoming the first step in a digital journey because it´s possible to ensure big results before acquisition, if you have the right technology and approach.
The terms “Artificial Intelligence (AI)”, “Machine Learning”, and “Pattern Recognition” are used everywhere, but what does it mean in real life operations? To truly understand that, you need a solution that is:
- Easy to use, so you don't need to be a computer scientist to navigate
- Gives you access to how the model is handling your data. No black box!
- Flexible that allows you to explore your information in different ways to get insights about the best way to operate your plant
Creating Effective Maintenance Strategies
The development of the Maintenance Strategy is part of an ongoing journey towards continuous maintenance improvement involving the collaboration of people, processes and assets through digital technology. It doesn’t happen all at once but instead gains speed and velocity over time as people, processes and assets are digitally fused together to eventually bridge the operations technology and information technology gap. Start small in your strategy and adoption, but start now to maintain or improve your competitive level and market position.
A comprehensive Maintenance Strategy combines enterprise data capture with asset management, prescriptive analytics and risk-based management. Work orders can be automatically generated to relieve maintenance issues. Analytics capabilities continue to evolve from predictive to prescriptive, i.e. from what will happen to what should be done.
Read the Chemical Processing article now: Maintenance Strategies in the Era of IIoT
A True Story: COVESTRO
To advance their predictive maintenance strategy, Covestro evaluated AVEVA’s Predictive Analytics solution, PRiSM, in the most effective way. They proved the benefits during the pilot phase by running offline models with information from real past cases.
In their evaluation, Covestro selected two incidents that caused significant impact in terms of production and/or overall results to the company’s bottom line. The first case was a large motor failure that caused the trip of a 3-stage compressor from a supplier. The trip stopped the supply of a critical feedstock that could not be replaced, leading Covestro plant to shut down. Thermal and mechanical models were developed using historical data, and they trained the models for a 1 year period. The mechanical model was rigorous enough to demonstrate that the failure could be predicted at least 8 months before it happened. If they had AVEVA Predictive Analytics application at the time, they would have enough time to identify the problem, plan the maintenance activities, and avoid the feedstock supply interruption.
In the second case, Covestro demonstrated how AVEVA’s Predictive Analytics solution could be used to predict failures based on very subtle indications; proving machine learning really works. One of the Covestro plants shut down for two weeks because of major corrosion damage. A heat exchanger operating with solvent and water had an internal leakage. Because this solvent reacts with water generating a corrosive compound, the corrosion spread to different equipment leading the plant to shutdown. The Covestro team was skeptical in this case because they didn´t have all the measurements they thought they would need to build a rigorous model. Based on the correlations made by AVEVA’s solution, it was possible to calculate the important variables to predict the leakage. The model results demonstrated that the leakage could be identified about 10 days before the plant shutdown, avoiding the major damage caused in many units and reducing the overall downtime.
Another large Industrial Gases customer has quantified their savings from the detection of equipment failures before they occur and the ability to prevent unscheduled system downtime at more than of $500+K.
Clear Results and Insights
In both cases, the AVEVA Predictive Analytics solution benefits were clear proving the losses they could avoid related to the described incidents were significant. As well, Covestro had access to how the models worked (opened calculations), so the technical team had insights about how to use machine learning to solve different problems.
Watch the webinar where Jane Arnold, Head of Global Process Control Technology at Covestro, shares the aforementioned case studies and how this is part of Covestro’s largerDigital Transformation strategy. For more insights about Maintenance Strategy in the Era of IIoT, read the Chemical Processing article now: Maintenance Strategies in the Era of IIoT
Fernanda Martins is the Global Marketing Manager for the Oil & Gas Downstream and Chemicals industries at AVEVA, holding a B.Sc. in Chemical Engineering from the Universidade de São Paulo. In her almost 20 years experience, she has helped companies to adopt and explore a variety of transformational technologies in areas like process engineering, optimization, workforce empowerment and enhanced operations and maintenance.