IoT and Predictive Maintenance (PdM). Are you Getting the Right Data?

The Internet of Things (IoT) capability has revolutionized a lot of areas in businesses. One such area is predictive maintenance (PdM). The success of PdM has and is largely dependent on incorporating sensors in the machinery to collect data that is at the centre of the prediction.

One industry that specifically reaps huge benefits from using IoT for PdM is the manufacturing sector.  All along the industrial revolutions, manufacturing has played a huge role in defining the progress of the revolutions.

IoT

During Industry 3.0, there was a big shift from analogue processes to automation using electronics. Slowly, data started to be harnessed. But it is Industry 4.0 that has made data the king of the industry. I am confident that data is not about to relinquish the kingship any time soon because we virtually need data to make decisions almost everywhere. And IoT is doing the bidding because of its efficiency in big data.

The infrastructure is important but as time has taught us, the software keeps on changing as does the hardware. Better algorithms are being churned out daily and with time the hardware, and sensors, will become affordable all in a bid to collect more and more data from whence innovations will emanate. Data seems the only thing that simply remains that, data.

So, across the industry, organizations should be investing in the right data. For IoT to amaze the data, it needs sensors, networks, intelligent actions and analysis. Sensors alone are quite expensive and incorporating them into already existing assets poses challenges. There is a demand for proper platforms to host the hardware and software, there is also a need for precise connectivity and the tools for controlling, collecting and analyzing the data. All these don’t come cheap.

It thus becomes apparent that it is not just about getting IoT tools and ending up collecting lots of data that may end up being not useful. The huge amounts of data that get generated need to be stored as well and that too needs expensive infrastructure.

PdM Maintenance

In the industry, maintenance is very challenging. Whilst there is preventive maintenance which is often scheduled and has to take place regularly albeit after certain periods, there are other maintenances that crop up unannounced such as an unexpected part breakdown.

These breakdowns can happen during times of high-volume production when timelines are tight and pressure to meet demand is high. These are the dreaded moments which any business manager or executive doesn’t wish for as they lead to loss of time and thus reduce turnover rates. But such corrective maintenance must be taken in such a situation for production to proceed.

It is akin to your car breaking up in the middle of the motorway when it has over 5000 miles to go before the next service and you are 100 miles away from home. The breakup will need to be sorted out as a matter of necessity which is what can be equivalent to corrective maintenance.

Every manufacturer would want their machines to run smoothly throughout the time of production to meet the demand and create value for the customer and that scheduled maintenance should take care of that smooth running. IoT has been heavily used in PdM in this regard by helping with predicting failure and thus have it scheduled for checking appropriately instead of deferring it were it not monitored and addressed.

The success of IoT and PdM is built on continuous monitoring of the various parts and simultaneously gathering and saving the data. The data is then sent to a prediction model which analyses the data and predicts when the part can be replaced. Previously collected data becomes integral in the creation of the prediction model.

Techniques such as those that can detect discrepancies in the vibrations or even ones that use infrared monitoring are used. Sensors have evolved as well and nowadays we have sensors that can monitor voltage, rotations, speed, current, pressure, thermal imaging, oil analysis or even noise-acoustics and much more.

The right data

It means then that with the current sensor technology and IoT capabilities, one can collect as much data as they want. But the focus now narrows to the data being collected. Is it important towards the intended objective?

Before organizations embark on collecting data for PdM, they need to set clear objectives about what they want to achieve. They should have the end in mind. Are they interested in the whole machine so that they put sensors everywhere or just the critical components? Perhaps by utilizing previous experience and data from the machine maintenance data sheets, they can decide on the critical parts or areas that need proper and closer monitoring and have sensors put there.

 And even then, the data needed should be narrowed down. Is it vibration monitoring, current, voltage or acoustic for example? Meanwhile, there should be a model in place which can aid in the prediction otherwise all the investment can be made and end up with data which cannot be made sense of.

With the continuous improvement we are seeing all around, we can only look forward to greater capabilities in machine learning, IoT, human- to computer -interaction (HCI), big data, and data analytics which hopefully will make data collection and analysis better and thus heighten the accuracy of PdM models which means reduced down times.