The importance of data quality in both asset monitoring and employment of AI in manufacturing

Posted on 19 Jul 2024 by The Manufacturer
Partner Content

Data is all around us, everywhere, and businesses across a wide spectrum of sectors are benefitting from the insights that can be gleaned from it.

The industrial arena is one such industry achieving – and has huge potential to achieve – significant efficiencies to help negate production line downtime, create uptime and deliver significant bottom-line cost savings.

According to a McKinsey survey in 2023, AI leaders outperformed their industry peers by a factor of 3.4 and, globally, it is estimated AI has the potential to deliver additional total economic activity of around $13 trillion by 2030. Around $1tn of this value remains to be captured from the industrial sector, but AI adoption and technology in general, remains low here. This is despite the fact there is significant opportunity to extract value from existing infrastructure.

Wherever an organisation may be on their journey to using technology or AI to inform preventative or predictive maintenance programmes, they must ensure the data on which they are basing decisions is robust. This relies heavily on data collection device knowledge and selection. There are a variety of devices available for data gathering and considerations around how to communicate it. Knowing what and how to best measure data and what to do with that information can be fraught with complexity.

Assessing existing data sources and identifying gaps

The McKinsey survey highlighted most industry players don’t have robust programmes for managing the accuracy and reliability of critical process measurements. There are so many variables to consider, like whether the data is being sampled at the right frequency. A more frequent reading could add in unhelpful spurious data, while a less frequent reading may omit the granularity needed to identify a potential issue.

Similarly, if a reading is being gathered every hour, can it be guaranteed to be at exact hourly intervals, which may be crucial to measurement? And that’s only considering measurement of one single device. Finding a reliable way to track the many thousands of measurements critical for effective operations and maintaining quality of those readings, manually, is impossible.

But, even with the help of AI, the right measurement monitoring devices must be implemented in the correct places. A plethora of machines and devices gathering data are already in any industrial plant.

Therefore, identifying the low-hanging fruit in the already-existing data is a must. It can then help highlight where the gaps are and, accordingly, where additional data collection devices may be needed, to give a more holistic picture of what is going on within a specific piece of equipment.

This is the opposite of an increasingly common approach, which is to rush to measure anything and everything, often resulting in the potential to over-invest in an approach that will under deliver.

A step-by-step strategy whereby a specific problem is identified, the relevant data extracted using the right device and insights gained to help solve the issue, presents a scalable approach with low risk and investment. Working in this way can facilitate strategic prioritisation of the most critical assets.

Critical to this approach is a clear understanding of what is being measured and why. All too often, the high-level knowledge needed doesn’t exist in-house and, in such cases, industrial operators should look to industry-specific external experts and solutions providers to aid them.

Simplicity is hindered by the expanding measurement device options

There was a time when sensors and other monitoring equipment were more expensive and not as easy to integrate. Thankfully, both cost and complexity of integration have reduced, making the technology more accessible and retro-fitting a cinch.

However, simplicity is hindered as more choice is presented and so the risk of implementing a device that isn’t necessarily the best for the job is heightened.

There are many factors to consider: does the sensor need to be wireless or wired? If it is battery-powered, what is the battery life capacity and is it suitable for the job if frequent readings are required? How does the sensor communicate the data and is it compatible with the network, or AI solution, being used?

An organisation could spend £100 on a sensor or £1,000 plus. Knowing what to apply to which scenario not only makes for a more effective monitoring programme, but can significantly optimise spend. A sensor costing £100 will have many limitations, but may be more than adequate in some applications. While a sensor costing £1,000 or more may be perceived as an expensive option, but if the impact of failure of the equipment being monitored is great, it’s thousands well spent.

It’s clear to see  deep knowledge of the equipment being measured, an understanding of what needs to be gleaned and which device is best to achieve that, is vital, but not always available in-house.

This is where solutions engineers are invaluable to help identify critical assets, what needs to be monitored, how data can be communicated, analysed and actioned on.

Using solutions providers with a proven track record and experience in the sector will ensure existing valuable data within the plant is maximised, gaps are identified and the right solutions implemented to bridge them.

Those which also offer cloud solutions with AI capabilities will be able to advise on compatibility of equipment and networks to ensure successful implementation of a data platform. This can open up a whole new world of insights beyond maintenance, like energy consumption, all accessible in user-friendly dashboards and reports in real-time.

In an ever-evolving and increasingly competitive arena, with challenges like skills shortages alongside global economic factors facing industrial operators, not capitalising on the potential of data is much more than a missed trick.


About the author:

Richard Jeffers is service solutions and technical director for RS UK&I