The future of manufacturing: Why you can’t have a digital twin without simulation
Author : Zohair Mehkri, Director of Digital Twin at Flex
12 January 2023
In this Q&A, Zohair Mehkri, Director of Digital Twin at Flex, explores the biggest improvements to simulation software technology over the past few years, the benefits of remote connectivity and networking, and what the future holds for simulation.
What can you tell us about Flex's simulation software technology?
Simulation is the ability to model a real-world environment in a digital one, so that you can more quickly and cost effectively test, configure and optimise until your desired result is achieved.
In a manufacturing setting, simulation technology can model every element that makes up a manufacturing process. This includes the buildings, machines, materials and people, as well as the characteristics and factors that may affect them, such as humidity, temperature, power consumption, weather, etc. The model also considers space and time, so the digital model closely reflects reality.
Our goal is to model as many factors as fully as we can. This allows us to mimic the way reality functions and provides us with the ability to have a digital representation of a physical manufacturing environment.
What is the most innovative/efficient simulation project with which you've been involved?
Our team is lucky in that we've done a lot of amazing things and worked with many great customers. It’s amazing to see an excited customer when they see their production lines in virtual reality.
While the majority of the work we do is to optimise manufacturing production, we have worked on some really innovative projects to optimise entire factories. In one example, we made a digital twin of an entire medical device production operation and far exceeded our customer’s goal and expectations.
Our sites can be very large, and one of the more unusual projects was to build a model to improve traffic flow by changing the traffic light sequence at one of our sites. We even showed the local council a simulation of how the public traffic light sequence near our factory could be improved.
Those are some of the innovative things that we have had a chance to work on and we're privileged to be able to do that.
There is sometimes confusion between simulation and digital twins, what’s the difference?
It’s very common for there to be some level of confusion between simulation and digital twins. Simulation is the foundation for digital twins. You can’t have a digital twin without simulation.
Simulation replicates the physical environment in the digital world, based on data sets that relate to every aspect of the situation. A model will require information on a variety of things, including the composition of the walls, sunlight levels, temperature, how energy levels of workers change over the length of a shift, the power consumption at each stage of production etc. It then creates a model based on that data.
The digital twin is built on simulation data but, importantly, is connected to real-time data sources in the physical environment. This connection means information can flow in either direction to create a cyber-physical system where a change in one affects the other.
The power of a digital twin is when the system receives information from the physical environment, does calculations, optimisations and improvements, and then sends them back to the physical environment. Each system becomes a twin of the other – the physical and digital talking to each other in a continuous loop to unlock even more precise and accurate modelling. Add artificial intelligence on top of that, and you have a supreme level of optimisation.
To set up an environment like that is highly specialist software required?
I would say that this is not as much about the software as it is about the expertise and know-how to create the models. You need to understand what data you need to build the most accurate model. For example, you will need information about the people who will operate the production line – but what information do you need?
You may want to know how many kilograms a person in a warehouse can move each day, and how productivity changes over the length of a shift. The data about the warehouse operator may contrast with that of a line operator, and therefore have implications for how much material can be processed on each shift. These are the sorts of inputs that can be important.
In terms of computing, simulation takes a lot of processing power. There are a lot of variables and a lot of data to be captured, analysed, optimised and returned. What’s more, each element is interdependent, connected and constantly changing.
In this multi-objective evolutionary environment, over the course of a few months, you'd be passing exabytes of data. That’s why simulation is such a good use case for quantum computing, and why our team is currently managing the technology strategy for Flex in this area.
How have remote connectivity and networking benefited from simulation software technology?
Connectivity is very important, and there is an increased requirement for these technologies to continue to evolve beyond 5G, so that we can share the results of models with customers, no matter where they are in the world. In just one example, we ran nine trillion combinations of optimisation for a customer over six hours. That’s too much data to share over conventional networks.
Digital twins will also drive the need for secure, high-speed connectivity, as more systems and physical assets are connected, and real-time information requirements become the norm. However, the factory of the future is very much a connected factory and simulation and digital twins are just one of the Industry 4.0 use cases that require high-powered, high-speed connectivity.
How do you think future innovations will impact simulation software technology, especially in manufacturing?
Data is at the core of simulation. More and better data means even better simulation models. So, I think improving the data will remain a focus in the short term. Success with simulation is really about having the expertise and skill to understand what goes into a good model. Improvements in the software are less important, but I’m sure that will happen too.
Longer term, as we connect more systems to access real-time data, we will see digital twins increasingly relied upon to deliver the insights manufacturers and customers need to make better operational decisions. Digital twins will be the key discussion point for customers in the relationship with manufacturers.
That leads to innovations in how people consume the simulation models that we're creating. We have seen the evolution in information consumption shift from paper to desktop PCs, to laptops and today, people want access to information on their phones and to interact with simulations on augmented reality headsets.
I also see a role for virtual reality to enable more collaboration between customer and manufacturer around simulated models in the sales process, and digital twin in the ongoing relationship.
When do you think the technology will advance to the point that engineers and IT professionals will not be required for installation and operation?
We’re already seeing some of that happening, with different aspects of simulation becoming a lot more user-friendly.
In the next three to five years, the ease of access to simulation software will improve. Right now, you can find ‘easier to use’ simulation software, and there are a number of start-ups looking to simplify this technology. As simulation technology advances, layers of simplicity will be added and things will start to become a lot more accessible.