PrimaVera project: the challenging path to Predictive Maintenance
Machines are becoming increasingly complex. They must be highly reliable, because of the pursuit of just-in-time productivity. With the rise of sensoring, machine learning and big data, Predictive Maintenance has become an important topic. A large consortium from knowledge institutions and companies initiated the Dutch PrimaVera project for research on Predictive Maintenance. THUAS lecturer and researcher Alieh Alipour gives us the latest information and a glimpse in the future of Prescriptive Maintenance.
The relatively young profession of asset manager has all to do with managing the capital goodsthat are of critical value to an organization. For instance, in a beer brewery the asset manager has to deal with a series of machines in a row that are all essential for the just-in-time production flow of cans or bottles of beer. A little malfunction can stop that flow, a risk that must be avoided. So, Predictive Maintenance is an important topic for the asset manager.
Importance of machine learning
In small production companies however it is still quite common that maintenance of machinery takes place based on the findings of the maintenance expert who makes his rounds through the factory and uses his intuition when planning maintenance. In this setting it is also normal to replace a part of the machine after x numbers of hours. Whether it needs to be replaced or not.
Machine learning can change everything. Based on machine data, collected for several years, an asset manager can get a prediction on when a machine will need maintenance and which parts will need to be replaced. The model becomes even better when you can create algorithms based on the data from a cluster of machines. In this way the nationwide PrimaVera project is working towards smart and predictive maintenance. PrimaVera stands for Predictive maintenance for Very effective asset management. The participants investigate how they can use big data to create algorithms that predict when capital goods need maintenance.
More certainty for factories
Alieh Alipour is a project leader of PrimaVera and lecturer Asset Management at The Hague University of Applied Sciences. “Predictive Maintenance will give us more certainty in a world with a lot of uncertainties. It helps us to gain more control over the inspection plan, maintenance plan, cost and logistic that we want to have for our machinery. More control over our resources: what spare parts should we buy in advance and when should we buy them? More control also over how many technicians we need for the inspections and maintenance.”
Collecting machine data is very important. Alieh: “We don’t get enough data yet from the participating factories. So, we will generate data. Students of the minor Asset Management and Maintenance are manufacturing a test set-up in the practical room. This test set-up will test bearings for short and long periods under different operating conditions. These tests are performed to gain a better insight into the behavior of bearings with the aim of generating large amounts of data so that algorithms for predictive maintenance models can be validated. More students, lecturers and local SMEs are invited to participate, to develop and manufacture all kind of facilities for the PrimaVera research.”
In the PrimaVera project researchers are for the first time combining big data in an optimized method. Alieh: “In this research we focus on three sectors: Maritime, Infrastructure and High-tech Production. In researching maintenance, you have to do with several work packages: collecting data, data prognosis, data diagnosis etc. All those work packages influence each other. But existing old-fashioned maintenance techniques do not take this mutual influence into account. Even existing predictive maintenance techniques only work for small-scale systems. They are difficult to scale up. That affects the quality of the predictions. Choices made in one place in the chain have an important influence on other processes in the chain. The choice of a certain type of sensors and measurements influences the type of predictions that can be made.
The PrimaVera consortium is looking for an innovative cross level optimization methodology. A methodology that for the first time does take into account that mutual influence.”
Share data with good quality
Companies are eager to know about the results of the PrimaVera project. Alieh: “Many of them have started with asset management and Predictive Maintenance. They are all curious about what we discover in this development phase. Any result of any methodology that can help them would be interesting for them.”
She tells about the big challenge for the factories. “To get Predictive Maintenance in their asset management they have to share data of good quality. The data should have the right volume. The history of the data should be registered very well. Factories can always call us for the right way to collect data.”
Predictive Maintenance or Condition-based Maintenance
In the development of smart maintenance Alieh Alipour distinguishes condition-based maintenance and predictive maintenance. “Both share the same goals: reducing equipment downtime and optimizing resources by performing maintenance work only when needed. Predictive Maintenance combines sensor measurements with precise algorithmic formulas to predict the exact moment when maintenance actions should be taken. Condition-based Maintenance can only provide information when given perimeters reach an unacceptable level, and action needs to be taken in the present moment to avoid equipment failure.”
On the horizon of the researcher’s imagination is Prescriptive Maintenance. Alieh: “That’s the future. In Prescriptive Maintenance the machine knows when a failure will happen and what he should automatically do to fix it. This seems a kind of science fiction. But we are in transition. We will never stop in maintenance development.”
Join the project or the lunch
Are you as a student interested in Predictive or even Prescriptive Maintenance? Join the minor Asset Management and Maintenance. Or ask for the possibilities to do an internship in the PrimaVera project.
More information? See https://primavera-project.com/ or sign up for the lunch lecture by Alieh Alipour on January 18, 2021.