What is the difference between predictive and prescriptive maintenance?
Posted: August 25, 2020
Predictive Analytics is a form of advanced analytics which examines data or content to answer the question “What is going to happen?” or more precisely, “What is likely to happen?”, and is characterized by techniques such as regression analysis, forecasting, multivariate statistics, pattern matching, predictive modeling, and forecasting.
Prescriptive Analytics is a form of advanced analytics which examines data or content to answer the question “What should be done?” or “What can we do to make _______ happen?”, and is characterized by techniques such as graph analysis, simulation, complex event processing, neural networks, recommendation engines, heuristics, and machine learning.
Predictive versus prescriptive maintenance: Is it just industry hype, or a long-yearned-for development that will make a real impact on productivity? In my view, it’s the latter.
With automation becoming increasingly prevalent across industries, mechanical devices are being replaced by electronic components in manufacturing, industrial, and factory environments. This evolution means more sensors are being used to capture additional types of data. The more in-depth data these sensors can capture, the greater the visibility and insight for the Owner Operator (OO). At least – that’s how it should be.
Let’s take a look at the reality. An industry leader high up on the APM maturity ladder and fully embarked on the predictive analytics journey was receiving many alerts from their sensors. All the alerts indicated that there was an anomaly detected – but they didn’t provide a solution to the problem.
This leader was left wondering whether the mitigating actions they were considering to solve the anomaly were actually the best ones. How would they know the timeframe in which to execute the necessary maintenance tasks in time? Was there room to postpone the alert until the next shutdown? And what were the best practices for handling this specific maintenance task? The bottom line is: what good is anomaly detection if you don’t know the right way to mitigate the failure deriving from it?
When running a predictive analytics pilot, alerts are easy to understand and you have time to address each one. But when you move beyond the pilot phase and scale your sensors and alerts throughout the plant, you may not see the forest for the trees. If you can’t estimate the value of an alert and can’t manage it accordingly, it’s often disregarded, creating a real threat to your asset performance.
For predictive analytics to add true value, you need to be able to react to incidents. Prescriptive analytics will help you to better prepare your mitigating actions.
The maturity level for prescriptive analytics
I often hear that prescriptive analytics is only a relevant goal when you reach the top of the APM maturity ladder. I strongly disagree with that opinion. Prescriptive analytics is needed in any APM strategy, whether you are executing a reactive maintenance plan or a fully predictive approach. The best strategies are a mix of different approaches for different equipment throughout the plant. Nevertheless, it’s still of great value to minimize downtime for all equipment and assets.
Prescriptive analytics and its content
To truly add value with predictive analytics, you need to get away from “alert mania” and prepare your mitigating actions using using prescriptive analytics.
Imagine you have thousands of alerts. To manage them, you need three things:
- An understanding of the alert’s criticality translated into value (what is the impact of this alert?)
- The urgency of the specific alert (how much time do I have to repair?)
- The best mitigating action (What do I need to do, what skills do I need to do this, what tools do I need, and do I need a spare part for this?)
When you have full transparency on those three key points, you are able to manage your alerts.
So, is prescriptive maintenance a game changer? Yes, you need it to create an alert strategy. You will minimize downtime significantly and avoid costly delays. But most importantly, make sure that your data is of top quality to maximize results. At AVEVA, we offer libraries with quality data to help you get started.