







































E. Kurvinen et al. /Future Technology                                                                                 November 2022| Volume 01 | Issue 03 | Pages 
03-05 

3 

 

 

 

Perspective 

Physics-based dynamic simulation opportunities 

with digital twins 
Emil Kurvinen*, Amin Mahmoudzadeh Andwari, Juho Könnö 

Machine and Vehicle Design (MVD), Materials and Mechanical Engineering, University of Oulu, P.O. Box 4200, FI-

90014 Oulu, Finland 

A R T I C L E   I N F O 
 

Article history: 
Received 05 July 2022  
Received in revised form 
06 August 2022 
Accepted 11 August 2022 
 
Keywords: Physics-based simulation,  
digital twin, machine design, systematic design 
 
*Corresponding author 
Email address: emil.kurvinen@oulu.fi  
 
DOI: 10.55670/fpll.futech.1.3.2 

A B S T R A C T 
 

This paper aims to provide a viewpoint on the exploitation of physics-based 
dynamic simulation in product development and discrete manufacturing 
products. The dynamics models can be represented with computationally light 
models when the product and its dynamics are well known and thereby 
analyzing the performance e.g., with AI methods rapidly and accurately. The 
recent developments with methodologies, sensor development, measuring 
techniques and increased computing capacity are making the simulation world 
closer to reality and the ability for real-time operation simulations paralleled to 
the real system. This enables the exploitation of the digital twin paradigm at full 
capacity together with high-maturity digital twin models.   

 
 
 

 

 
1. Introduction 

1.1 Physics-based simulation  

In advanced high-technology countries, such as Finland, 

the utilization of physics-based simulation has long been 

employed. Machine manufacturers have the most information 

about the product and its behavior, thus the utilization of 

physics-based simulation enables them to predict the actual 

performance in the early design phase when the machine is 

built. This allows for tailoring the products more to customer 

needs and assessing the requirements that influence the 

machine performance. Thereby, the possibility to meet the 

customer requirements can be considered at a higher level 

than without the physics-based simulation. 

2. Digital twins 

With the recent advancement in the theme of digital 

twins, the interest in the utilization of existing simulation 

models with the actual product through its lifecycle has been 

of interest. Specifically, real-time capable simulation tools are 

attractive, since they can be utilized in cases where a human 

is operating the machine e.g., in mobile heavy machinery 

applications [1]. The physics-based simulation has been 

supporting machine design and other disciplines since the 

analytical equations were formulated. Currently, the digital 

transformation is rapidly making the demand for simulation 

technology even higher. Especially the digital twin paradigm 

and its development have been the driver for the clarification 

and definition of the physics-based simulation and its role 

and potential for business [1]. Figure 1 depicts an example of 

a high-speed electric machine rotor, where the dynamics are 

defined by the high-speed rotor and a conceptual example of 

the digital twin solution to it.  

While working in a computer environment and 

exchanging information in a digital format enables us to 

assess the information from various points. It thereby enables 

the development of systematic methods for analyzing and 

assessing the information, for example for decision-making 

purposes, with high accuracy. Current trends, such as 

sustainability and energy efficiency, are driving the 

development further and developing computationally 

efficient means of simulation. 

Dynamics simulation has been in a central role in 

product development in large machines, for example, off-

highway vehicles for decades. The dynamics can be assessed 

with computationally efficient models and can be even used 

in real-time applications for assessing the machine’s 

performance when a human is acting as an operator. Thus, the 

recent development in algorithm development especially in 

the computationally efficient dynamics calculation methods 

has progressed rapidly. For example, multibody system 

dynamics is used in many industrial applications as a basic 

methodology for conducting the virtual design for the 

dynamical behavior of the product. In most cases, it means e.g. 

avoiding the resonance frequencies while operating the 

application.  

 

 

Future Technology 

Open Access Journal 

https://doi.org/10.55670/fpll.futech.1.3.2 

 

 

November 2022| Volume 01 | Issue 03 | Pages 03-05 

Journal homepage: https://fupubco.com/futech 

 

ISSN 2832-0379 

mailto:emil.kurvinen@oulu.fi
https://fupubco.com/fuen
https://doi.org/10.55670/fpll.futech.1.3.2
https://fupubco.com/futech
https://fupubco.com/


E. Kurvinen et al. /Future Technology                                                                                 November 2022| Volume 01 | Issue 03 | Pages 03-05 

4 

 

 

 

 

The combination of physics-based simulation and 

measured data enables computationally efficient methods for 

creating neural networks e.g., for fault identification tasks and 

the transferability of the developed neural networks is of the 

essence i.e., not solely for single-purpose utilization but also 

beyond it to another similar type of products [2]. 

One approach to utilizing physics-based simulation is to 

utilize the validated computational models as a base for 

investigating the parameters and their sensitivity. It is worth 

noting that a deep understanding of the machine should be 

defined prior to exploring the parameters and their 

sensitivity to the dynamics. In these cases, the measured data 

from real machines is of importance, as that is used to validate 

and verify the simulation models. Figure 2 depicts a 

conceptual solution for creating the identification software 

with design information. 

Fundamentals of dynamics can be captured with simple 

models, see for example [3]. The 3 DOF model computational 

time is approximately 10 seconds solved in the frequency 

domain, the 27 DOF model solved in the time domain takes 8 

hours, and the full measurements approximately one week, 

and the resonance frequency should be avoided to ensure the 

safe operation. However, it should be well known what the 

application and expected behavior are to decide when for 

example a simplified model is justified to be utilized. 

Accordingly, the constant development of computationally 

efficient simulation models is ongoing, (refer to e.g., [4]).  The 

dynamics are a product of the product mass and stiffness 

properties.  When identifying changes in the systems the 

parameters which influence the mass or stiffness have the 

most effect on the system i.e., increased flexibility of 

structures is not a wanted phenomenon in the system. The 

system-level understanding is important to have in the virtual 

world as different configurations can be benchmarked prior 

to manufacturing the first prototype. 

 

 

 

In addition, current products include more software for 

control, and with the accurate virtual product, software 

development can be initiated prior to the first physical 

prototype being built. While neural networks have also been 

active in the focus of research, the need for labeled data is the 

main prerequisite for accurate and efficient neural network 

creation, especially in supervised learning. The physics-based 

simulation models can produce data for that purpose. 

Especially including non-idealities and faults in the datasets 

can be created with ease [5]. 

The research related to efficient and accurate simulation 

techniques is in progress, which is beneficial for the labeled 

data generation with a computer [4, 6]. Therefore it is 

expected that the simulation is capable of merging tighter 

with the real world more effectively. Accordingly, the benefits 

of both approaches can be used to generate an understanding 

of the applications. With the active development of the 

modeling techniques and requirements for measurements 

becoming clearer, the gap between simulation models and 

reality is getting smaller. For example, the recent 

development in the utilization of Kalman filters and 

connecting them with computationally efficient simulation 

methods [7] is a promising step to closing the gap between 

simulation and the real world. 

3. Conclusion 

The physics-based simulation has several possibilities to 

enhance the product lifecycle from early design to end-of-life. 

Simulation enables rapid design iterations to explore the 

behavior as it serves as the virtual object and simultaneously 

it helps to align and structure different stakeholders’ 

viewpoints and information in a clear and quantitive 

perspective. Currently, the exploitation strategies for 

different types of companies are under active research. 

 

Figure 1. Example case with a high-speed rotating rotor 



E. Kurvinen et al. /Future Technology                                                                                 November 2022| Volume 01 | Issue 03 | Pages 03-05 

5 

 

 

 

 

 

The virtual product and its exploitation are especially 

beneficial as the inertia related to tests is minimum when 

compared to physical prototyping. Simultaneously, the 

decision based on virtual products should be well validated to 

align the behavior with the real environment. 

Ethical issue 
The authors are aware of and comply with best practices 

in publication ethics, specifically with regard to authorship 
(avoidance of guest authorship), dual submission, 
manipulation of figures, competing interests, and compliance 
with policies on research ethics. The authors adhere to 
publication requirements that the submitted work is original 
and has not been published elsewhere. 

Data availability statement 
Data sharing is not applicable to this article as no datasets 

were generated or analyzed during the current study. 

Conflict of interest 

The author declares no potential conflict of interest. 

References 

[1] UKKO, J., SAUNILA, M., HEIKKINEN, J., SEMKEN, R. S., 

AND MIKKOLA, A. Real-Time Simulation for 

Sustainable Production: Enhancing User Experience 

and Creating Business Value. Routledge, 2021. 

http://dx.doi.org/10.4324/9781003054214  

[2] LEI, Y., YANG, B., JIANG, X., JIA, F., LI, N., AND NANDI, A. 

K. Applications of machine learning to machine fault 

diagnosis: A review and roadmap. Mechanical Systems 

and Signal Processing 138 (2020), 106587. 

https://doi.org/10.1016/j.ymssp.2019.106587 

 

 

 

 

[3] KURVINEN, E., VIITALA, R., CHOUDHURY, T., & 

SOPANEN, J. (2020, October). Simulation model to 

investigate effect of support stiffness on dynamic 

behaviour of a large rotor. In 12th International 

Conference on Vibrations in Rotating Machinery (pp. 

457-469). CRC Press. 

https://doi.org/10.1201/9781003132639-37 

[4] CHOUDHURY, T., KURVINEN, E., VIITALA, R., AND 

SOPANEN, J. Development and verification of 

frequency domain solution methods for rotor-bearing 

system responses caused by rolling element bearing 

waviness. Mechanical Systems and Signal Processing 

163 (2022), 108117. 

https://doi.org/10.1016/j.ymssp.2021.108117 

[5] BOBYLEV, D., CHOUDHURY, T., MIETTINEN, J. O., 

VIITALA, R., KURVINEN, E., AND SOPANEN, J. 

Simulation-based transfer learning for support 

stiffness identification. IEEE Access 9 (2021), 120652–

120664. 

https://doi.org/10.1109/ACCESS.2021.3108414  

[6] YU, X., ACEITUNO, J. F., KURVINEN, E., MATIKAINEN, M. 

K., KORKEALAAKSO, P., ROUVINEN, A., JIANG, D., 

ESCALONA, J. L., AND MIKKOLA, A. Comparison of 

numerical and computational aspects between two 

constraint-based contact methods in the description of 

wheel/rail contacts. Multibody System Dynamics 

(2022), 1–42. https://doi.org/10.1007/s11044-022-

09811-6  

 [7] KHADIM, Q., HAGH, Y. S., PYRHÖNEN, L., JAISWAL, S., 

ZHIDCHENKO, V., KURVINEN, E., ... & HANDROOS, H. 

(2022). State estimation in a hydraulically actuated log 

crane using unscented Kalman filter. IEEE Access. 

https://doi.org/10.1109/ACCESS.2022.3179591  

 

 

Figure 2. Utilization of design data for identification software building 

 This article is an open-access article 

distributed under the terms and conditions of the Creative 

Commons Attribution (CC BY) license 

(https://creativecommons.org/licenses/by/4.0/). 

https://doi.org/10.1016/j.ymssp.2019.106587
https://doi.org/10.1201/9781003132639-37
https://doi.org/10.1016/j.ymssp.2021.108117

