Digital twinning in the plant

NASA developed the first digital twinning project in the early days of space exploration to operate and maintain systems that were out of physical proximity. Digital twinning has advanced exponentially since then, with factories now able to duplicate their machines for better monitoring.

The global market for digital twins is expected to grow 38 per cent every year to reach $16 billion by 2023, spurred on by growing number of companies are investing in the technology. Sectors such as oil and gas, aerospace, automotive and even utilities like water and energy are all benefiting from the development of digital twins. But what is digital twinning and why has it grown so much in the last decade?

Digital twins are digital representations of physical assets. Based on historical data from sensors they intend to mirror the physical asset’s current state and behaviour. For example, a remote engineer could look at the digital twin of a motor to see how it is performing and decide whether maintenance is needed or not.

Some digital twins also use historical data and predictive analytics. These allow the system to look for trends and predict where a failure is likely to occur in the future, notifying the plant manager so they can take action. 

Digital twins were previously restricted to industries working with large budgets and maintaining expensive machinery, however an increasing number of companies are now able to take advantage of digital twins.  According to Deloitte, “significantly lower costs and improved power and capabilities have lead to exponential changes that can enable leaders to combine information technology (IT) and operations technology (OT) to enable the creation and use of a digital twin.”

The increase in connected technology and the advancement in machine learning has made it possible to create intelligent digital twins that can contribute to the management of products.

Learn, model and predict

Danny McMahon, head of metrology and digital manufacturing at the University of Strathclyde’s Advanced Forming Research Centre, argues that digital twins are important for the entirety of the product’s lifecycle: “A digital twin will enhance the understanding of a product’s full lifecycle, so that future designs can be smarter, more efficient and effective.”

The digital twin achieves this by collecting data from the physical asset, gathering this historical data and learn its behaviour. It’s worth noting that the capability to capture and process machine data in digital twinning software directly affects the effectiveness of the twin itself. This will vary between software providers.

To make it easier to explore the full range of benefits of digital twinning, we’ll consider it in relation to a digital twin built on GE Digital’s Predix platform. This is because the platform is capable of handling the vast sums of data generated during industrial processes, and also includes machine learning algorithms that can analyse the data automatically.

Choosing a platform such as GE Digital’s Predix ensures that the twin can cope with the high volume of data that it collects from physical assets and convert it into something useful for the plant manager. This is the first stage of the digital twin process of learn, model and predict. Once the machine has gathered the data, it can see where trends are happening, such as increased wear on a certain part of a device.

The digital twin can then predict the wear that will take place over time if the current situation is allowed to continue. This identifies the root cause of the problem before it can develop any further and create unplanned downtime.

The next stage of the digital twinning process is for the system to monitor real time behaviour. The machine learning system looks at the current signals and based on historical behaviour tries to predict failures or underperforming situations.

The Predix platform allows the digital twin to run thousands of simulations to decide what the top resolution for the problem would be. These options are all based on what is the right option for the specific plant based on risk and the digital twin’s confidence that it will deliver the right outcome.

The next stage is for the digital twin to inform the operator what should happen to change the undesirable situation, perhaps to minimise wear. If the operator has chosen to perform the operation manually, the digital twin will have simulated the exact measurements required to reduce the wear. For example, it could prescribe the load rate, the ramp rate and the steam temperature, for the operator to input.

Alternatively, a digital twin system like the Predix platform can suggest an app that can be used to automatically control these inputs. This means that they can be monitored and adjusted precisely to ensure the optimum performance for the hardware.

Using historical data

One of Novotek’s customers in the Netherlands, a waste disposal company, uses digital twins to reduce downtime in the company due to failing equipment. A particular problem was with a steam turbine, where due to excessive wear, one of the blades has broken. This resulted in a cost of millions of euros, as well as a long period of downtime.

To minimise the risk of this happening again, the company gathered process data from a variety of different sensors over a long period of time. Using the Predix Asset Performance Management tool, the data was analysed to decipher which signals were occurring that led to the breakdown and where they were occurring. This information was then used to monitor the behaviour in real time and meant that plant managers could foresee something going wrong.

For plant managers struggling with downtime or a lack of efficiency in their plant, a digital twin can be an effective way to prohibit these kind of serious issues. By having a virtual model that can simulate thousands of different scenarios to devise the right way to mitigate the problems that could occur, this means that the machines are running at optimum efficiency and reduces unplanned downtime.

While some machines may not be as expensive as NASA’s space rovers, by using digital twins, plant managers can ensure that the investment they make in their machines is not wasted.