Better, stronger, faster – How digital twins can accelerate commercialisation

Fleur Doidge speaks to technologists about how building digital replicas of a system can massively improve processes over time.

Digital twinning is diversifying beyond 3D design into advanced process improvements that accelerate commercialisation across the board – expanding into “everything manufacturers do on the shop floor”, according to Sushant Sharma, UK Vice President for Manufacturing at IT consultancy Tech Mahindra.

“When you have feasibility to create a complete simulation of what’s happening, that delivers end-to-end production visibility for quality parameters, testing assets and any predictive analytics on those supply lines,” says Sushant, “even before you start making any live production, you can tell the OEM that this model works or doesn’t, and make changes accordingly.”

Building a digital replica of a system enables aggregation of all the data from different siloed systems and processes to track quality parameters in real time, with one manufacturing customer reducing lead times and effort by 30% as well as ‘scrap and rework’ in production by about 15%.

For pharmaceutical customers this includes checking the levels of particle sizes, concentrations, flow properties and more – all of which can impact the final product’s complex biological processes. If fed into an artificial intelligence (AI) model, this reveals patterns and can improve control, according to Jacob Paul, UK Head of Healthcare and Life Sciences at Tech Mahindra (pictured below right).

                                                                       

“A lot of wastage typically occurs when making medicine,” Jacob says. “The whole industry has this dream of bringing in a ‘golden’ or perfect batch. The way to get there is through process.”

It can take 12 years and US$2bn to develop a vaccine, and rejections of “almost 30%” are commonplace. However, one pharmaceutical customer’s cutting-edge digital twin project of exactly this type has reduced wastage to 10%.

“The focus is to reduce batch rejection, heighten throughput and get real-time insights on how the processes are working. We know exactly what is changing and are able to take rectification measures to make sure final output becomes far better.”

Why make ’em – when you can fake ’em?

Alan Prior, Vice President for EuroNorth at Dassault Systèmes (pictured below right), confirms that in life sciences and consumer packaged goods (CPG) “you’re looking at volume, rather than complexity”.

                                     

“If making a deodorant rather than an aeroplane, many feel it’s quicker to build a little plastic injection plant and make prototypes – but if you’re going to make a billion of these things worldwide, you need to understand your manufacturing process pretty well,” Alan says.

Entering CAD details and physical parameters into a virtual-twin programme to create 10,000 or so iterations of a shampoo bottle falling on the floor from different angles can massively reduce wastage. Once the full costs of building physical prototypes are calculated, the argument for digital or virtual twinning strengthens.

“If a business selects the right solution for its requirements and can scale, then absolutely value can be created very fast,” says Alan.

Digital twinning can reduce time to market, improve process efficiency, minimise risk and drive sustainability. Not only can manufacturers slash the time taken to set up a production line or create prototypes, but they can adapt with agility to changing demands – quickly and efficiently re-engineering production methods to serve customers better or move into a new niche.

Jaguar Land Rover completely reinvented the way it does things with the Reimagine programme that it has just launched in the last year, using a huge amount of virtual-twin on Dassault Systèmes 3DExperience platform,” Alan says. “So, it doesn’t have to be a new start-up, it can be an existing OEM.”

Alan says some twinning tools “almost train themselves” to find the combinations of settings and variables that boost process efficiency, reduce errors or maximise throughput, whether for safety or sustainability.

Experimenting on a digital twin is often safer and cheaper – and can enable digital collaboration between different groups or datasets. But simulations should be carefully developed with attention paid to ‘people’ factors including education and skills. Inputs and calculations must be accurate and relevant.

“If you don’t get the virtual model right, everything you learn from it is useless,” Alan warns.


Header image courtesy of Shutterstock