This article covers:
- Big data and its impact on the manufacturing industry
- An insight into predictive maintenance
- The main differences between predictive and preventive maintenance
- Advising factories about integrating predictive maintenance programs into their processes
In this article, we will review what the term predictive maintenance refers to. Machine manufacturers, component manufacturers and system integrators can utilize this article as a guide to gain further understanding of what predictive maintenance is all about. We will discuss the differences between predictive maintenance and preventive maintenance. Additionally, machine manufacturers will learn more about how to advise factory owners to integrate coordinated, cost-effective, predictive maintenance programs into current processes.
Big data and its impact on the manufacturing industry
The modern-day factory floor consists of sensors, actuators, machinery, PLCs, embedded HMIs, delivery vehicles and SCADA systems, which not only produce and deliver the main factory outputs but also generate a large amount of data. This data can, in turn, be used to optimize factory processes in a variety of ways.
A manufacturing company can use big data for example, to reduce the risk of delays in the delivery of products to customers. By analyzing the data of the entire supply chain transport route the delivery trucks are using (fitted with sensors and antennae), a manufacturer can accurately identify the points of historical and real-time delays. ,Consequently, the manufacturer can look at using an alternate route that avoids those areas in the future or in real-time. ,Thus, not only do customers get their products on time, but there is also an added benefit of a reduction in fuel consumption.
This is just one example of the way, big data is being used in the smart factory environment and in order to optimize standard plant processes. Data-driven plant optimization is becoming the norm in many factory and manufacturing environments, with many factory owners integrating at least one big data solution into their production and assembly lines.
An insight into predictive maintenance
The failure of just one machine that is a critical part of a mass production assembly line, can result in costly downtime and incur other expenses, which is something that machine manufacturers, component manufacturers and system integrators have to keep in mind. For example, in the case of a typical automobile assembly line, the workstation responsible for the installation of the steering wheel components experiencing downtime will result in a significant loss of profits, due to a delay in the production of the cars themselves.
Furthermore, even if the machinery at that workstation has not completely broken down, but is defective, the cars produced from that assembly line may have to be recalled once they are already completely assembled. In a worst-case scenario, these cars may end up being purchased before they are recalled and could be safety hazards on the road.
Machine maintenance and monitoring are vital in other words, in order to avoid downtime and costly product recalls, as well as extend the lifespan of the actual machine itself. Therefore, maintenance is an important part of the services that machine manufacturers, component manufacturers and system integrators need to consider offering to factory owners.
Predictive maintenance relies on big data. It involves evaluating data generated by a certain machine or machines. It predicts the chances of machine failure before the actual failure takes place, and can schedule the maintenance before the failure, to deal with the issues. This is all possible by the analysis of the data.
In standard predictive maintenance techniques, a sensor or sensors from the machine of interest collect performance related data. These sensors can monitor and collect data in real-time about the specific machine’s temperature, condition, pressure, vibration, and outputs. The sensors send the data about the machine to a processing unit, and any deviations from pre-determined reference values or machine learning models are noted. These reference values or machine learning models are based on the values observed before a fault in the machine occurs. Consequently, maintenance is scheduled before the actual failure in the machine occurs, if deviations are noted. If a factory owner does not have a cloud-based dashboard, with predictive maintenance KPIs and alerts, then the machine manufacturer could offer a “white label” dashboard gained from technology provider.
Other types of predictive maintenance techniques analyze the products produced by the machine, comparing historical product quality data to current product quality data, to check for defects or changes in quality over time, and therefore make inferences about the condition of the machine.
The main differences between predictive and preventive maintenance
Preventive maintenance involves scheduling maintenance on machinery, at regular intervals in order to decrease the chances of equipment failure, similar to the way owners of cars take their cars for service at certain times of the year.
Predictive maintenance involves using big data, machine learning models and statistics in order to predict the chances of machine failure and then schedule maintenance accordingly. Predictive maintenance is regarded as a proactive approach to maintenance issues. The actual maintenance would be conducted by the machine manufacturer, system integrator, or component manufacturer.
The main issue with preventive maintenance is determining the correct schedule for a specific machine to be serviced. This is often based on the manufacturer of the machine’s recommendations and not on the machine’s actual performance or status in the factory. Another issue is downtime incurred due to unnecessary maintenance, which may not be needed at that point in time. It is a reactive, rather than a proactive approach. It is reactive because if a machine starts having issues with performance, these issues will only be dealt with on the scheduled maintenance date, rather than in real-time resulting in wear and tear on the machine, a greater chance of failure and a greater chance of non-conformance in product output.
Advising factories about integrating predictive maintenance programs into their processes
An overview of the steps involved in the development and integration of a simple predictive maintenance program or predictive maintenance as a service program is discussed below. Machine manufacturers could offer this service to factory owners in order to assist them with implementing predictive maintenance programs in their factories:
- It is advisable that the factory starts with a small pilot program first and does not just implement a predictive maintenance solution for all machines at once.
- The correct software or SaaS tools for the aggregation and statistical analysis of the data should be purchased from the machine manufacturer first. The machine manufacturer should have consultants and a sales team to assist with this process.
- Each machine in the pilot program should be fitted with a sensor(s) that can record values and indicators relevant to maintenance. Usually, modern equipment comes with these sensors but older machinery may require the purchase and installation of these sensors.
- The development of machine learning models is needed, which will be based on the data obtained from the sensors. These models will use algorithms for pattern detection and real-time data evaluation.
- The models generated will need to be integrated into the existing ERP and MES of the factory and triggers developed that can alert staff that performance issues have been detected and that maintenance is needed. The machine manufacturer should provide an HMI with a dashboard that can alert the factory owner about predictive maintenance.
- Once staff receive those alerts they can schedule maintenance accordingly for that specific machine.
Predictive maintenance is just one of the modern day factory processes that utilize big data for plant optimization. There are many advantages to predictive maintenance and it is recommended that factory owners invest in predictive maintenance programs, in order to reduce downtime and increase overall machine efficiency.