Digital Twin: A phrase not a product

The digital twin might reduce development time, improve product quality, cut waste and ultimately, lower costs. However, Deputy Chief Engineer Jonathan Butt, Digital Engineering at the NCC, challenges the Digital Twin as a single solution, one size fits all for manufacturers, and talks about the digital transformation process. 

‘Before-and-after’ photo stories appear regularly on our social media feeds, from house renovations to exercise regimes: they are designed to impress and inspire and create a visual shorthand for transformation.  Yet, in one swipe of the screen, the stories leapfrog a huge number of steps.  In reality, transformation is all about the details in-between, the sum of multiple, potentially break-through steps and gains that together, deliver real change.

The Digital Engineering Technology & Innovation (DETI) programme drills into the detail of digital manufacturing transformation. As engineers, taking a forensic approach comes naturally: we research, test, alter and test again, until we find the right solution.  Digital is fundamentally changing every aspect of the product manufacturing lifecycle, and data is the new keystone.  As we begin to extract data from each manufacturing process, the challenge is transforming that data into manufacturing intelligence, so that we can feed in knowledge from each process to enable us to engineer-out problems early-on. It’s the detail that counts, being sure we’re asking the right questions, to understand the real need and problems, before applying the appropriate technology – but before we start this process, the initial challenge frequently comes down to semantics.

Take the ‘digital twin’. Broadly, we all understand this to be a virtual replication of the physical product and manufacturing process that should enable us to better understand and predict the performance of the ‘physical twin’.  The digital twin incorporates high-fidelity simulation and modelling, data analytics and machine learning in order for operators to see clearly the impact of variables, such as changes to the design, the manufacturing processes, environmental conditions and materials on the product.  The digital twin might reduce, even mitigate, the need for physical prototypes, as well as reduce development time, improve product quality and performance, cut waste and ultimately, lower costs.

On paper digital twin should answer any of the questions manufacturers have and yield positive results out of the box, however - a digital twin isn’t a single solution, there is no one-size-fits-all transformational product, so we need to be careful not to assign it without consideration.

Manufacturers are ready and willing to embrace digital – but we need to ensure we understand their real needs.  In our experience,  manufacturers want to “make it right first time, make it better and make it at lower cost”. They believe that having the right digital toolkit could unlock this innovation. We want to ensure this expectation is met.

In DETI, we do two key things to digitally transform a manufacturing process; using a systems engineering approach we capture a thorough set of requirements to understand the true end goal of the transformation, and we use toolkits developed in-house to build a deep understanding of the manufacturing process we plan to transform. It’s only at this point, that we can consider the digital technology application that will form the eventual solution.

One method that we use to break down a manufacturing process, in order to build our knowledge up, is called Process Variable Mapping (PVM) . PVM allows us to extract the network of complex variable relationships that exist within manufacturing processes, so we can understand where process influence exists, and suggest where it might exist. We then integrate this knowledge into our digital transformation design so that we can be confident that our solution has access to the critical variables we need to monitor; whether these be related  to raw materials, environmental variations such as temperature or complex, process-specific variables. 

Once we understand the desired output from a digital transformation and have built up the in-depth knowledge of a manufacturing process, we can start to design the solution. For manufacturers that would like to realise “right every time” manufacturing we’re looking at digital twin solutions that enable Closed Loop Manufacturing (CLM). We’ve been investigating and designing CLM approaches to improve systems that currently require manual adjustments and interventions to overcome quality issues, deploying machine mounted sensors that measure the critical variables and feed data into a high-fidelity process model.  This model, which depending on the process complexity, might incorporate machine learning algorithms, can be used to predict and generate machine commands in near real-time to overcome or adjust process variations in order to deliver right, every time manufacturing.

We’re trialling this approach for Liquid Resin Infusion (LRI), a composites manufacturing process.  LRI is designed to enable the use of dry fibre material types, which are lower cost when considered against their prepreg counter parts, with the resin being infused through the entire component before curing.  LRI offers a higher production rate and lower production cost but is a highly skilled technique. Its outcome is highly dependent upon variables that are, for the most part, beyond the control of the technicians who setup the infusion, these variables include raw material properties (such as fibre density) and preform bulk (how tightly packed fibres are in the component). In short, LRI is a one shot process, once started the infusion can’t be undone or stopped due to the chemical reactions taking place within the resin.  It is this chemical reaction which often means the process takes place in an oven, so operation interventions can only be made at the periphery of the process. Furthermore, the infusion takes place right at the end of the composite part production process,  so there is a lot of cost already invested in the component. The  LRI process is high risk and requires a huge amount of work to develop an infusion strategy that can overcome these variations and specialist setup for each infusion.

Our Closed Loop Manufacturing system for LRI is using the outputs gained from the process variable mapping to identify the critical parameters that influence the success of this process. We embed this knowledge into data collection, so that at the point where infusion starts we understand the status of our component, and we use our process model to take real-time data from the infusion process to determine if the current infusion strategy is producing a result that will deliver a successful product.  If, at any point, the process doesn’t fit the boundaries of the high-fidelity simulations embedded within the process model, the infusion strategy is automatically changed.

As industry grows its use of LRI to infuse more complex, high-performance structures from entire boat hulls and aircraft wings to super-sized wind turbine blades, companies will look to digital solutions to ensure that their products can be produced to the right standard every time.  With each successful demonstration of a digital solution for a manufacturing process, more will be deployed, in turn, the pool of data around this process will grow and with it the opportunity to make drastic changes in manufacturing process performance through the application of machine learning and artificial intelligence.

Applying digital at the individual process scale helps us to understand the ‘real world’ complexities and limitations of this capability, so we can honestly assess its benefits and applicability.  If the complexity is relatively low, the closed loop systems we’re developing could be implemented at a macro scale enabling companies to transition to the ‘digital machine twin’ step quickly. On production lines that have a high degree of variability in the product their making, the resultant complexity of the closed loop manufacturing solution might not currently meet the business case requirements, the value proposition might not be there yet – but this is why manufacturers are using DETI as their digital testbed.

We’re taking our DETI customers on the journey with us, so they can see the analytical processes we’ve developed and the challenges we faced along the way. This helps those customers understand how to assign the appropriate digital transformation and provides them with the confidence of producing a solution that yields the results they want. We want customers to take away the value proposition for digital transformation – building the business case foundation to invest in the right digital solution first time. A journey that is applicable across all manufacturing sectors – not just composites.

Digital twin might be the common shorthand for what we think we need to deliver to embrace digital, but it’s the detail of getting from the ‘before’ to ‘after’ that’s showing what we often need to focus on is a certain element of ‘right first time’.  Building the infrastructure, enabling engineers to understand how these systems should be put together – understanding what decisions need to be made – what drives them and how these decisions affect the end process - adapting analytical tools not creating new ones.

Sign up for DETI news and updates