Overview: How and why a digital twin was created and tested for an automotive production line
CNH Industrial is a global leader in capital goods. It is financially controlled by the Italian investment company Exor and is comprised of 12 brands, including Case, New Holland, and Iveco. Through its brands, CNHI designs, produces, and sells a wide range of agricultural, industrial, and commercial vehicles and powertrains. It employs more than 63,000 people in 66 manufacturing plants and 53 research and development centers in 180 countries. The company is listed on the New York Stock Exchange and is a constituent of the Italian stock market index.
Fair Dynamics operates primarily out of Milan and provides a wide range of consulting services in a variety of industries, including banking, manufacturing, and public services. The company has recently been acquired by Engineering Ingegneria Informatica S.p.A., a provider of software and IT services, both in Italy and internationally. In 2017, the consolidated revenue was more than €1bn.
Fair Dynamics applies innovative technologies to solve industrial problems and improve efficiency. Their key approach is modeling and simulation, for which, the company has been using the AnyLogic Platform since 2010 and is also its Italian distributor.
Problem
Manufacturing processes are becoming increasingly digital. It is considered that now we are entering a fourth industrial era (Industry 4.0) and the transition towards smart factories has begun. Within smart factories, cyber-physical systems monitor physical processes, create a virtual copy of the physical world, and make decentralized decisions. Digital twins are core to the operation of these systems.
With digitalization already underway, many companies are trying out new technologies like artificial intelligence and cloud computing with the aim of gradually shifting to a smart factory and benefitting from the new phenomena.
CNH Industrial identified maintenance processes as a promising area to start applying new Industry 4.0 technologies. In the automotive and related industries, downtime costs can be large. For global companies like CNHI the cost of a single minute of downtime could be more than $160k and these figures increase year by year. As such, improving maintenance in order to reduce downtime can deliver significant success. By identifying the most critical areas, even a very small percentage improvement could save a lot of money.
Thus, CNH Industrial wanted to test a digital tool for evaluating and selecting different maintenance policies and agreed with Fair Dynamics on a pilot project. They decided to focus on a single manufacturing line dealing with Iveco Daily van chassis welding (the Mascherone line of the Suzzara plant, Italy). A digital twin, a representation of the line in a virtual environment, was to be created. The simulation would enable CNHI management to see the benefits of possible maintenance policies in various scenarios and make informed maintenance decisions.
The choice of the Suzzara plant for a digital twin was not random. CNH Industrial applies the principles of World Class Manufacturing (WCM), an innovative program for continuous improvement. At that time, CHNi had only one WCM Gold Level award and the Iveco plant in Suzzara was very close to a second one. CNHI wanted to see how the new technology could help attain it.
Solution

The digital twin project focused on a specific manufacturing line, Iveco van chassis welding. This line can be described as a conveyer which runs through a number of stations. Fair Dynamics were asked to focus their attention on the automatic welding stations (orange blocks in the picture). When a van stops at one of these stations, the robots work in unison to complete the welding.
The welding guns have an Achilles heel – the Lamellar pack (an electrical conductor which must flex during operation). The movement gradually leads to the damage of a pack’s copper layers. When the damage becomes critical, and sufficiently changes the conductivity, it can result in the melting of the Lamellar pack itself. While normally this component can be replaced in few minutes, it can take hours if the Lamellar pack has been damaged. A digital twin that monitors and forecasts the health of this component could provide significant downtime reduction.
Fair Dynamics built an agent-based digital twin with the following agents:
- Vans — There are different types of van agents in accordance with the types of vans to be produced. Each type requires different handling (different operations, stations, and robots could be involved) and this affects component degradation.
- Stations — Each station agent is characterized by the number of robots it contains and is regulated by particular rules.
- Robots — Each robot is fitted with a sensor which sends a signal about the robot’s actual condition to the simulation model. Each robot agent, in turn, is provided with a specific PHM (Prognostic & Health Management) model predicting the robot degradation in accordance with the signals received.
By building the digital twin this way, Fair Dynamics could introduce three basic maintenance policies for testing and use:
- Scheduled maintenance (components are replaced according to a schedule).
- Condition-based maintenance (components are replaced according to warning signals).
- Predictive maintenance (components are replaced on a schedule based on information from their state and use).
Within the project, AnyLogic software proved useful for digital twin creation. Apart from enabling agent-based modeling, it enabled the customization that made it possible for Fair Dynamics to include the prognostic ELF (machine learning) model. The integration of modeling and machine learning techniques has great potential in such systems.

Through the use of AnyLogic, the digital twin could connect to external data sources. The production sequence, welding point per van type, robot life cycle curve, and other data were imported from external sources and automatically read by an agent at runtime. Moreover, the system could be exported and delivered as a standalone application to multiple machines, easing data constraints and the demands on the IT department.
Outcome
With the help of the digital twin, CNHI management and specialists can get detailed and demonstrative information about the economic and production consequences for different maintenance policy configurations. This is done by running various what-if scenarios where the user can vary different core parameters (e.g. maintenance policy, production plan, working schedule, etc.). It is also possible to change the characteristics of the line or a robot, if needed.
The system can handle both the near and far future and, moreover, using the digital twin for simulation provides an easy-to-use tool to analyze and compare scenarios — enabling a quick understanding of how changes could impact maintenance cost. The digital twin provides a wide variety of data, including total production, maintenance time, total cost for spare parts, and the work cost of maintenance. In short, the digital twin is a detailed and comprehensive tool for establishing efficient production line operations.
The AnyLogic white paper, An Introduction to Digital Twin Development, contains further case studies and outlines digital twin benefits and development — download.
Project presentation by Luigi Manca, Project Delivery COO, Fair Dynamics Consulting Unit:
