Predictive maintenance relies heavily on accurate data to maximize the return on investment.
In the age of IoT, more assets are being connected to digital platforms to maximize their performance and life span. One crucial way this is helping businesses save money is through enabling the implementation of predictive maintenance.
By deploying sensors connected to a digital platform—even for legacy assets—and using machine learning algorithms to monitor asset performance, companies can get ahead of potential machine breakdowns. Deploying those technologies could also enable businesses to automate their maintenance efforts.
Predictive maintenance differs from conventional preventative maintenance. Conventional preventative maintenance is based on a fixed schedule, where machines are pulled offline at regular intervals regardless of their state so an engineer or technician can examine them.
In contrast, predictive maintenance monitoring is condition-based and takes place in real time. This approach uses IoT-enabled sensors to collect and analyze machine performance data on a continual basis and identify, in real time, potential performance issues before the machines fail. This approach helps companies schedule maintenance when it’s most convenient and cost-effective, resulting in reduced downtime and lower cost. It also leads to shorter downtimes required for inspection and testing, a reduction in the number of repairs and replacement parts required and more time for maintenance technicians and engineers to dedicate to other tasks.
However, implementing a predictive maintenance program takes some prudent planning and preparation. It also requires a modest investment in the sensors and communication protocols required to gather the data and feed it into the predictive maintenance system.
When it comes to specifying sensors for a manufacturing environment, there are a lot of choice on the market. To be sure you make the right choice, there are five factors that need to be considered when determining the right sensor package for an operation’s predictive maintenance system.
Russ Freeman, product portfolio manager with RS Group, a global supplier of products and services for designers, builders and managers of industrial equipment and operations, says knowing exactly what you’re sensing is just the beginning. “If you know nothing about a sensor, but you get these five questions answered, it puts us in the best position to help you,” he said. By answering all five of these questions, you give an industrial products supplier the best chance pf making sure you have the right sensing options for your application.
First: determine what data you are trying to capture. The more precisely you can identify what you need to measure, the better the choice you’ll make in selecting a sensor. “How big is it? What color is it? What texture is it—wood, metal or plastic? Does it change textures? Does it change color? Does it change size?” he asked. While these may seem like obvious questions, it is still important to ask them.
Second: define the range from which the sensor will need to collect data. A machine being monitored from an inch away will require a different kind of sensor than one being tracked from several feet away. “Distance is important because if I need a sensor rated for three feet, I know I can’t use an inductive proximity sensor,” said Freeman. This will help narrow the range of sensors on the market to a list of suitable options.
Third: determine any constraints on power that you plan to use for the sensor. You must match the input power of the sensor to the power available at the site.
Fourth: establish the kind of output needed. A PNP (positive-negative-positive) sensor produces a positive output from the sensor’s input, while an NPN (negative-positive-negative) sensor produces a negative signal. The operator needs to know which input the application requires. Another factor to consider is whether a relay output is needed.
Fifth: what’s the operating environment? The configuration of machinery, available space, indoor or outdoor, or whether it is clean or dirty, will all influence the choice of sensor. A conveyor belt on a busy factory floor will need a different sensor than the same conveyor in a clean room or healthcare facility. “Different environments involve different considerations when selecting a sensor,” said Freeman, adding that you want to be sure the sensor you select will withstand the operating environment.
Automation Maintenance
While a solid predictive maintenance program can reap rewards for any manufacturing operation, highly automated factories may have the most to gain from moving to a predictive maintenance regime.
“Automation customers want to produce products at an increased rate, with higher quality and be more efficient doing it,” said Freeman. “No matter what control platform they choose to accomplish these goals, they are all dependent on repeatable, high-speed and accurate feedback from their sensors.”
There are two main data collection and analysis platforms for automation: cloud computing and edge computing. Cloud computing involves collecting data locally on the machine and then sending it to remote servers in the cloud to be processed and analyzed. The processed data is then sent back to the work location to aid in decision-making. With an edge computing system, the data is collected and processed at the machine, either in its control panel or on the machine itself. Rather than outputting the data to another location for analysis, the data being transmitted has already been processed. As a result, edge computing requires much less IT infrastructure than cloud computing, making it a more flexible option for smaller companies looking to retrofit existing automated equipment, and is more secure since it’s not transmitted to a remote location.
Predictive maintenance can deliver an excellent ROI for companies trying to generate better data and improve the efficiency of their operations. It’s essential for these companies to choose the sensors best suited for the job, especially if they aim to automate their predictive maintenance systems. Equipped with the right sensors, they will be able to generate the data they need to prevent costly breakdowns and maximize the benefits of Industry 4.0.