Digital Twins: Models Everywhere

Keynote by Professor Øystein Haugen

Østfold University College, Halden, Norway

https://www.hiof.no/iio/itk/english/people/aca/oysteiha/index.html

EXTENDED ABSTRACT

Digital twins are catching on. People of different

trades talk about models and appreciate that models

can serve to improve our relationships with an

increasingly complex reality. This is very promising,

but as modelers we should make sure that we know

properly what a model is, and why we want the

models that we make or use.

Modeling is not new. Models have helped to

understand the actual world for a long time. In the

early days of computing, models were created and

realized in programs to simulate situations of the real

world. This was used to plan track layout of railroad

stations and strategies for vaccinating the population

for tuberculosis. The pioneers did not talk about

digital twins at that time, but they could have if they

had had our vocabulary.

While the pioneers had to wait a long time

between their digital results and the changes of the

represented reality, the time gap has decreased

drastically and now the feedback from the digital

twins may affect the represented reality in

milliseconds. But this is not new, either. We have

created real-time systems for many years, where the

digital components have been programmed to

represent and control some piece of reality in the right

direction.

So, what’s new with digital twins?

Technically, there is not much new technology

with digital twins, but it may still have a great impact.

Here are some aspects that make digital twins useful

and promising:

1. People of different competences find digital

twins fruitful as “living” representations of

complex systems that may serve as their

common base for understanding.

2. Modern systems often involve sensors and

actuators that are directly connected to the

actual systems, such as smart homes or

advanced production. This makes digital

validation necessary for the safety of using

the actual system.

3. Physical systems are not like computer

programs. They are affected by physical

wear and tear, and digital twins can serve as

their cleaner counterparts which are still

accurately updated by data from the actual

systems.

4. Modern systems are more dynamic than

before and therefore it becomes important to

have digital representations that behave in

ways corresponding to the dynamic world,

and that can be visualized in many different

ways depending on the purpose.

5. Modern systems should work 24/7. They

must evolve without having to stop and

reboot. They must handle the data in real

time. They cannot wait until later to do the

analysis because the data keeps coming.

We need concepts and notation for describing

digital twins and their evolution.

Digital twins seem to come in three categories

depending on their rough purpose:

1. Digital twins for monitoring. The digital

twin visualizes the actual twin and prepares

human decisions. The presentation may

include historical data.

2. Digital twins for simulation.The digital twin

executes with current data to project

behaviors of the future. This gives an even

better background for human decisions as

they can see what-if scenarios, but the model

must be more advanced.

3. Digital twins for control.The digital twin is

used to control the actual twin. This is

automation. There is no human in the loop

and the digital twin must be even more

accurate and reliable.

Digital twins may evolve from monitoring

through simulation to control as the trust in the digital

representation increases and the purposes are well

defined.

Digital twins are models for the future.

Haugen, Ø.

Digital Twins: Models Everywhere.

DOI: 10.5220/0013454700003896

In Proceedings of the 13th International Conference on Model-Based Software and Systems Engineering (MODELSWARD 2025), page 5

ISBN: 978-989-758-729-0; ISSN: 2184-4348