Digitalisation: it's time to walk the walk

When looking at areas for process improvement through digitalisation, the monitoring of rotating equipment stands out as having the greatest potential ROI.

IIoT, Big Data, Industry 4.0 and many other terms have been discussed for almost 10 years, but with little in the way of actual outcomes. Extravagant vendor marketing has dazzled us with possibilities yet failed to provide anything tangible. What has happened is disillusionment among many end users who have been convinced to part with hard-fought budget based upon promises that have failed to materialise. Not only have traditional vendors been involved, but the rise of IoT and its potential as a new market has attracted many new vendors and consultants to the process industries, many with an understanding of industrial processes that is questionable.

Rotating machinery is the most widely used form of mechanical equipment in process industries yet can also be the most problematic. Operational faults can not only lead to machine and process downtime, but also excess energy consumption, loss of production, safety incidents and the potential for catastrophic damage. It’s clear that when reviewing areas for process improvement, rotating equipment stands out above many others: that is, if you know what to look for.

Research has shown that the majority of equipment damage in this area is caused during installation, start-up or shutdown. Other longer-term factors can slowly erode performance, such as bearing wear, part degradation, lubrication issues and rising imbalance. Degradation will reveal itself in many ways, the challenge being to identify the signs and take corrective action early. In doing so you are gifted the potential to improve reliability and performance, increase overall life expectancy and maximise return on investment from your machinery.

With degradation occurring very slowly and over long periods it can, as mentioned, be difficult to identify. Rotating machinery in a slow state of decline will show irregular or increased vibration, elevated surface temperature and an altered acoustic profile. In the early stages these indicators are very subtle and challenging to pick up using normal diagnostic methods. Yet the benefits in doing so are huge, allowing early stage corrective action in the most efficient and least disruptive manner.

Each application has its nuances, but in general there are a multitude of factors that can impact the performance and life of plant process equipment.

• Environmental degradation: environmental conditions such as corrosive gas and dust, climate conditions such as desert and snow.
• Thermal degradation: conduction and radiation from heat sources.
• Mechanical deterioration: wear and tear.
• Location: foundation degradation, movement, shock.
   

Understanding the process and its environment — and how they impact performance — is the first step in planning and optimising maintenance strategies. Research suggests that approximately 40% of expenditure on rotating machinery is wasted in ineffective maintenance. Conventional preventive techniques performed during operator rounds are time-intensive and not always possible to perform effectively, and as a result degradation is often detected too late to be useful.

Utilising sensors designed for condition monitoring can provide precise readings at more frequent intervals and allow for more accurate decision-making without the maintenance overhead. Sensing vibration and surface temperature can equip operations personnel with the data required to make more informed decisions on an ongoing basis — that is if they are designed for the environment in which they are to operate, if location is carefully considered, they have the means to communicate effectively in confined environments, and there is a clear understanding of what you want to measure and why.

Let’s be clear that in pursuing such a strategy you will be opening up a huge well of data and it’s easy to become disillusioned and intimidated when faced with such volumes. Yet careful, early-stage understanding of how to deal with such a rich dataset along with expected outcomes can smooth the process.

Machine learning technology can perform advanced analytics and pattern recognition on sensor data, detecting anomalies faster than any human ever could, all while generating alerts and automated performance reports. In doing so the creation of large databases of asset health progressively improves the effectiveness of the software, thus building a broader understanding of how the components of your process impact overall performance. Creation of equipment profiles, comparing like-for-like performance of assets, guiding anomaly detection rapidly and with greater accuracy, all contributes to the protection of process integrity. This is where the real value lies and where many proposed solutions have fallen down in recent years.

You don’t need high-end process equipment or millisecond data points to measure effectively with this type of strategy. Samples should be taken at much greater intervals and over longer periods, so patience is a virtue. It’s all about the collective data profile and how the machine behaves compared to how it should behave.

Devices designed for this purpose need to be easy to install compared to fixed precision sensors, keeping overhead to a minimum. Ideally, they would require no wiring, utilise wireless technology and be battery powered. Easy-to-deploy and maintenance-free sensors not only gain favour with maintenance teams for these reasons alone, but do so for the resulting output.

Operations teams also understand how the same conditions that contribute to machine degradation also affect the control and monitoring equipment around them. The last thing they need is to add hardware that is not fit for purpose, so any sensors used need to be rugged and have the ability to withstand the same harsh exposure. Most end users understand that commercial products are not necessarily the best fit for harsh industrial applications. The trouble is that cheap ‘flick-and-stick’ sensors seem appealing from a cost perspective, only to later reveal significant consequential costs due to performance and longevity issues.

Beyond the sensors, selecting complementary technology for existing plant architecture is important. LoRaWAN is an example of an IoT communications protocol that won’t interfere with existing industrial wireless networks, has long-range transmission capability and reduces network hardware required to cover large plant areas.

With sensor and connectivity issues addressed you need to be mindful of where your data will be held and analysed. There’s a lot of conjecture around this topic, with many solutions being sold that are ‘all or nothing’: hugely complex cloud-based analytic solutions for full site monitoring from day one, when many operations personnel want to dip a toe in the water first before being teased into the ocean. Although this may be a long-term goal, you need the initial flexibility and reassurance that you can start small and scale up as and when desired.

In our experience, machine learning should have the option to be performed in a private cloud or on-premises if and where possible. This provides scalability with minimal worry about cost, security or sovereignty while keeping your data close and your industrial network secure. Local solutions also allow for easy integration with other control systems without the hassle of third-party APIs.

Outcomes matter, so consider all aspects of what you want to achieve with condition monitoring and keep compromise to a minimum. If it’s not fit for purpose, there is a lack of understanding of your needs or it’s an ‘all or nothing’ offering, then it just might not be for you.

Yokogawa created the Sushi Sensor range to address all the above, and more. The company has been walking the walk for over 30 years in Australia and New Zealand ensuring the right technology, application and outcomes to support manufacturing in the region.

Image: ©stock.adobe.com/au/imantsu