Heading for industrial autonomy

In 2030, two-thirds of the 500 industrial companies surveyed expect to be operating largely autonomously. This is evident from a recently published study by Yokogawa. In five years’ time, productivity will be significantly higher as more work and production processes operate more autonomously. The aim of industrial autonomy is not to replace jobs with robots. The goal is a robust industry that can compete with the world. Before industrial autonomy is achieved, however, there are still quite a few hurdles to be taken.

The first question that immediately comes to mind is how 'autonomous' should be defined. Marcel Kelder, director at Advanced Solutions and OT cybersecurity consultant at Yokogawa explains: "Autonomous operations certainly does not mean unmanned plants where only robots work. Humans will play an important role in almost every plant in the future, but it can be done with fewer people than today making maximum use of data and automation."


Let us return for a moment to the fear that robots will take over people's jobs. How big is this chance? "In the future, certain functions, for example where people are at risk, will be taken over by robots," Kelder believes. "Another example is functions - often repetitive functions such as inspection - that we would rather not fulfil, can possibly also be filled by robots. Will this lead to more unemployment in the process industry? No, simply because we already have a shortage of personnel in the industry, which will not change overnight. Industrial autonomy will also create many new jobs, some of which do not exist today."

In control

The current way of working in the industry is based on the fact that one mostly reacts to what has already happened. "Many solutions currently in use in the industry collect historical data from the process and use this as the basis for a visualisation of what has happened. Then certain knobs can be turned. With autonomous operation, there will be so much more information available from the process that the automation can predict which control is needed. So that's another dimension of automation. For many companies, data will therefore be much more decisive in the future for the autonomous operation of a plant".

Too little data for industrial autonomy

The most concrete and obvious example is maintenance. There is already a trend going on there that people want to move more and more from preventive maintenance to predictable maintenance. Kelder: "In maintenance, but also in operations, you see, however, that too little data is available to really predict the behaviour of the asset or the process. If you do want to predict behaviour with this limited amount of data, you currently need complex software. Often, the full potential of this type of solution is only used to a limited extent due to a combination of complexity and a lack of knowledge. If the number of sensors in the field increases significantly using for example IIoT sensors, the applications for analysis will become more user-friendly. Yokogawa has already delivered some IIoT projects for maintenance and the response from users is very positive, with a return of investment being less than six months."

Cheap storage

The emergence of IIoT and the availability of cheap storage for data in the Cloud can significantly accelerate predictable work, both in operations and in maintenance, according to Kelder: "The big bottleneck so far is not so much technology, but the budget," he says. "The collection and storage of data currently requires expensive licences for Data Historians. I expect this to change when data is stored in the Cloud or edge servers, where the costs will be significantly lower."
In addition to cheap storage, IIoT will provide a breakthrough. "Ask any maintenance worker what they would like to change in their work and the answer will often be more sensors in the field. Now, in many cases, that is not possible because a traditional transmitter costing a few hundred euros, with wiring, cross-wiring, connection to the control system, sending data to the Historian, updating the drawings, and so on, ends up costing several thousand euros. There is simply no budget for that."

Critical vs non-critical

Traditional (4-20 mA) sensors are indeed necessary for critical applications in the OT domain, but not specifically for non-critical applications. Kelder: "The critical assets in a plant usually only cover about twenty percent of the total plant. That means that eighty percent of the assets are not critical - but are important! - and on those assets you can use cheaper IIoT sensors. Currently, many of the non-critical assets are not equipped with sensors".
"On the critical OT assets, we continue to work with traditional sensors", continues Kelder, "but what if you could equip all non-critical assets with IIoT sensors, whose data goes directly to the Cloud? Then you end up with a fraction of the price of a critical OT sensor. This means that with the same budget, you can install considerably more sensors. Then we're really talking about a game changer." According to Kelder, the developments mentioned will play an important role in the transformation from industrial automation to industrial autonomy, also known as IA2IA.

Digital twin

In addition to data, integration plays a crucial role in autonomy. There is a certain need to make work in the process industry more predictable. "The outflow of personnel in western Europe is greater than the inflow of personnel," says Kelder. "While the processes are often becoming more complex. That means that you must be able to do more specialised work with fewer people, while processes mostly operate autonomously in the background. Then it becomes necessary to do more with data, for example in combination with a digital twin."’


In a plant, there are virtually no separate, stand-alone processes anymore. Kelder: "If you draw up a diagram of all the individual (sub)processes, you conclude that they are mutually connected in one way or another. So, you can't just automate part of your process and assume that other parts have nothing to do with it."
To achieve effective industrial autonomy, integration is a necessity, both horizontally and vertically. "Horizontal integration is about the integration between the different work processes such as supply chain, operations and maintenance. Vertical integration involves the entire chain, from sensor to Cloud. In practice, you see that many companies have tried to integrate, but often only vertically and also in a limited way: from ERP to truck loading, for example, and then also isolated on the horizontal scale. The challenge now is to make both horizontal and vertical integration part of the transformation process towards autonomy."

Starting point to transform

Companies that want to transform first need a starting point, Kelder knows. "In practice, we see that many plants do not know exactly where they stand and therefore do not know where to start. We have developed a methodology to determine the basic level and to link a benchmark to it. The basic level says how far a company is on the scale of smart manufacturing or Industry 4.0. If you know where you stand and what the end goal is then you link them to a master plan and a roadmap. The difference between starting point and goals along with the master plan ultimately determines the budget."

Priority for industrial autonomy?

The Yokogawa study shows that the corona crisis acts as a catalyst in the pursuit of industrial autonomy. Kelder: "We are only now seeing what the consequences are for a plant if a limited number of people can only work on the site. Companies are often much more vulnerable than people thought. Now it's COVID, tomorrow there will be another major disruptor, the fact is that plants want and need to be better prepared for situations like this. This used to be a lower priority, but we see that companies are changing. Even though many plants recognise the need for a transformation, this does not appear to be easy in practice due to the phenomenon of 'lean and mean' production. Due to the competition, many companies have had to economise and at the same time increase efficiency. As a result, both space and resources are limited for transformation. In the mid-term, this will become a problem. Companies therefore need to create the space or find a partner to start the digital transformation". Asian countries are less affected by the lean and mean bottleneck and are therefore able to take big steps. The survey also shows this: 71% of respondents in Asia-Pacific expect to be working completely autonomously in ten years' time. In North America, this is only 58% and in Western Europe 56%.

The transformation that the industry now faces is of a different kind than we have seen in recent years. "Where transformations were often very business driven before, this new transformation is technology driven. In this case, Yokogawa stands out for the fact that it has understood OT for more than 100 years and has now helped many companies plan and implement digital transformation.

Want to read more?

On yokogawa.com/en/DXebook you can download the ebook 'Digital Transformation in the process industry'.

Source (interview and image): processcontrol Magazine