Digital Twins — same, same but different…
How similar do digital twins have to be? Hyperreal or hyper-performant?
Digital Twins (DT) is one of the key enabling technologies that will be fundamental to the success of the industrial metaverse — especially for the manufacturing industry 4.0. In its current state, of low-resolution 3D representation of machines in factories, it is light-years away from the ‘physics-based, hyperreal replicas of real-world assets’ needed to simulate products and processes in the industrial metaverse, — before investing in any real world implementation of the same.
But, how real do they really have to be?
The industrial metaverse is banking heavily on the hope that digital twin technology will deliver on its promise — a market currently estimated at 10.1 billion (in 2023) and expected to grow rapidly at CAGR of 61.3% to reach 110.1 billion USD by 2028 [1], see image 1 below:
Note: There are other figures floating around in the market depending on the research group, such as, “the industrial digital twin market will grow from US$3.5 billion in 2021 to US$33.9 billion in 2030 at a 29% CAGR”, according to ABI Research. [3]
Convergence of emerging tech
Lyu and Fridenfalk (2023) [4] investigated key technologies of the industrial metaverse such as digital twins, cloud rendering, virtual-real integration, big-data visualisation and IoT. And their research found that digital twin technology that allows “the creation of virtual replicas of physical assets that can be monitored and analyzed in real-time”, was crucial to the development of the industrial metaverse meant for improving and optimizing manufacturing processes.
But it is not just digital twin technology, the industrial metaverse needs the help of a number of other emerging and fast evolving technologies as well. Such as (but not limited to), IoT (internet of things), AI/ML (artificial intelligence and machine learning), AR (augmented reality), VR (virtual reality), MR (mixed reality), XR (extended reality), 5G/6G connectivity, cloud and edge computing, etc., just to name a few. Forgot to mention… plain simple computing power.
Needless to say (but I’ll say it anyway), that these other emerging technologies are yet to reach maturity individually.
Back to the basics — the first digital twin from 1949
Perhaps, it is best to start at the very beginning with what can be considered the first digital twin ever made in 1949, by the economist Bill Phillips who used mathematical equations and water to model the British economy (see image above). [5] His two-meter tall machine called the ‘Moniac’ was used extensively in education, and most notably — to train economists at LSE in the ‘mathematical approach to economics’. [6]
Watch this < 2 minute video simulation of Bill Phillips ‘Moniac’ machine [7]:
Highly recommend you read the whole article here on Science Museum!
But, is this really the first digital twin from 1949?
Well, if Nick Szabo — who proposed the idea and coined the term ‘smart contracts’ in 1997 [8], can attribute the first smart contract back to a vending machine… then perhaps, we can accept that the ‘Moniac’ is indeed the first digital twin. Why not!
However, problems have since been pointed out with such mathematic models, which are complex yet simplistic at the same time. Mathematics now includes probability, i.e. the possibility that things might not always go as planned… to include the human element, irrationality in human behaviour, error in human judgement, or simply a mishap… an accident!
And then there is another claim to ‘the first digital twin’ from the 1960s — NASA’s Apollo 13. Read this incredible article for yourself on Siemens blog. Siemens’s own contribution to the mission isn’t worth missing either, read till the end of that article to find out!
Now back to the present day needs of the industrial metaverse, and the lamentation over the lagging DT technology.
Current implementations of DT vary in use cases and degrees of maturity, and could look something like this < 2 minute video simulation of a milling machine from the manufacturing sector [9]:
Or like this image below of a digital twin team collaborating in the industrial metaverse:
Or like this < 5 minute video presentation of BMW’s digital twin factory created in NVIDIA’s Omniverse (a replica of their factory in Bavaria):
Well, so how real do digital twins have to be — you ask?
There seem to be two schools of thought — those who advocate hyperealism, and then those who advocate hyper-performance.
For hyperreal — Starly et al. (2023) [11] in their study paper propose ‘ten technologies necessary for ‘Unreal’ factories,’ (such as in the milling video simulation above) going from factory illumination, shading and reflections on machines, to authentic audio and smell generation of the factories in the industrial metaverse.
For hyper-performance — Khan et al. (2022) [12] in their study put forward the functional element of the digital twin and highlight that its purpose is to provide a cost-effective way to test the physical system it is representing. They also point out risks related to high-fidelity DT (faithful depiction of the real-world asset) that can be monitored remotely, saying that it also doubles the attack surface — by allowing malicious actors (hackers) access to previously difficult to reach information (example: mains electricity); in addition to the ease of carrying out an attack remotely too.
Or could it simply depend on the use case? Going by the three distinct use cases (mentioned above)— with the milling machine, the DT team collaboration, and the BMW factory DT? And if we were to take into account the the OG (original) digital twins — the ‘Moniac’ and NASA’s Apollo 13, it could all very well just depend on the purpose the DT is meant to serve — optimization of a product, a process or a service.
Perhaps, it is best to wrap up here with this parting thought — an apt quote by Nobel-Prize-winning economist Peter Diamond, who famously said, “taking a model literally is not taking a model seriously”.
— — —
References:
[1] Digital Twin Market by Enterprise, Application (Predictive Maintenance, Business optimization), Industry (Aerospace, Automotive & Transportation, Healthcare, Infrastructure, Energy & Utilities) and Geography. https://www.marketsandmarkets.com/Market-Reports/digital-twin-market-225269522.html — Global Forecast to 2027, Digital Twin Market, June 2022
[2] What is a digital twin? IBM. https://www.ibm.com/topics/what-is-a-digital-twin
[3] Industrial Digital Twins Market Quickly Maturing to Reach US$34 Billion in 2030. https://www.abiresearch.com/press/industrial-digital-twins-market-quickly-maturing-reach-us34-billion-2030/
[4] Lyu, Z., Fridenfalk, M. (2023). Digital twins for building industrial metaverse, Journal of Advanced Research, https://doi.org/10.1016/j. jare.2023.11.019
[5] Willige, A. (2022). Digital twins: What are they and why do they matter? WEF. https://www.weforum.org/agenda/2022/05/digital-twin-technology-virtual-model-tech-for-good/
[6] How does the economy work? (2018). Science Museum. https://www.sciencemuseum.org.uk/objects-and-stories/how-does-economy-work
[7] Video simulation of Bill Phillips ‘Moniac’ machine on YouTube. 170519 Phillips Passive 1080 https://www.youtube.com/watch?v=iUrhENgn_m0
[8] Szabo, N. (1997). Formalizing and Securing Relationships on Public Networks. First Monday, 2(9). https://doi.org/10.5210/fm.v2i9.548
[9] Video simulation of a milling machine from the manufacturing sector on YouTube. Real-time HAAS UMC 500 milling visualization in Unreal Engine 5. https://www.youtube.com/watch?v=dzpxGhwO4jI
[10] On the path to the industrial metaverse. Capgemini. https://prod.ucwe.capgemini.com/wp-content/uploads/2023/01/Industrial-Metaverse-F5.pdf
[11] Starly, B., Koprov, P., Bharadwaj, A., Batchelder, T., Breitenbach, B. (2023). “Unreal” factories: Next generation of digital twins of machines and factories in the Industrial Metaverse. Manufacturing Letters. pp. 50–52, https://doi.org/10.1016/j.mfglet.2023.07.021.
[12] Khan, L. U., Saad, W., Niyato, D., Han, Z., & Hong, C. S. (2022). Digital-twin-enabled 6G: Vision, architectural trends, and future directions. IEEE Communications Magazine, 60(1), 74–80. https://arxiv.org/pdf/2102.12169.pdf
