The automotive sector is undergoing a period of rapid and disruptive transformation from both a business and technical standpoint. The industry is increasingly using simulation and modelling to help both experienced OEMs and new players get their latest products to market as quickly as possible.
Alan Prior, senior director of technical sales, EuroNorth at Dassault Systèmes, said: ‘The last three to four years have brought more change than the last three decades. The key challenge is that all parties need to be able to work faster – and this is the challenge simulation and modelling can address for the automotive industry.’
This intrinsic need to work faster comes about as a result of a range of interlinking factors in the automotive sector. ‘There are three major and current trends pushing change: the emergence of new players in the business; new technologies such as electrification and automation; and the shift to a systems-based approach,’ Prior explained.
MDGo is one such startup in the automotive space. The company has designed a system that automatically alerts first responders and hospitals when there’s an incident on the road and provides them with information based on measurements taken from the vehicle’s sensors during a collision. Research shows that up to 44 per cent of people who died in car crashes could have been saved if first responders and hospitals had real-time, detailed information about the victim’s injuries.
Eli Zerah, co-founder and VP of R&D at MDGo, explained: ‘Our algorithm requires a massive amount of diverse data in order to learn how to predict the forces applied to passengers, given the vehicle sensor measurements. Unfortunately, most crash tests are limited to a few specific scenarios dictated by regulation. Relying on these, solely, results in small amounts of data, with very limited scenarios lacking the representation of a lot of real-life cases.’
The company used Altair Radioss to conduct its crash simulations. By combining these results with actual sensor data from car accidents, the system could predict potential occupant injuries. Zerah explained: ‘Simulation and modelling gave us the much-needed flexibility to gain results of any real-life scenario we like. We were able to grow from five specific [crash] scenarios to dozens of different cases differing in angle, barrier type, velocity, and so on. It also gave us the opportunity to increase the total amount of cases that we use for our algorithm training.
‘Simulations are a key enabler of our R&D process, therefore, we plan to continue using this tool in order to diversify our data in all verticals, scenarios, vehicle models and human models,’ he added.
This highlights another important application for simulation and modelling in the changing automotive industry. Namely, that any new technology entering the market requires thorough testing to ensure it meets the industry’s strict safety standards.
But new ways of working need to be developed, because of the new technologies entering the market. Gilles Gallée, business development director for autonomous driving at ANSYS, explained: ‘Autonomous vehicle and embedded software safety will make the expenditure for validation increase by a factor of 106 to 107. Because traditional statistical validation is not suitable anymore, autonomous systems require completely new release strategies.
‘New strategies include using simulation to achieve the complete digital safety of the car [and] driving millions of miles virtually, in order to produce 99 per cent of the validation through simulation, while completing the remaining testing with physical on-road driving,’ he added.
ANSYS is focused on five ‘major disruptions’ in the automotive industry, according to Gallée, each of which brings further challenges to the sector. These include autonomy (where development and testing require more than eight billion miles of road tests, equivalent to 1,000+ years of drive time); electrification (where battery costs must be reduced three-fold); safety (where new qualifications require ten times the validation effort); additive manufacturing (for a vast range of new topologies); and smart connectivity (where vehicles now rely on millions of lines of code), Gallée explained.
Altair is also working ‘aggressively’ in three main areas, according to Uwe Schramm, CTO of HyperWorks core development at the company. These areas include the simulation and modelling of lightweight designs, the e-powertrain (found in electric vehicles to replace standard internal combustion engines) and advanced driver assistance systems (ADAS).
ADAS affect the design of every part of the vehicle, according to Schramm, who said: ‘We’ve applied Altair Feko to the design of front bumpers/fascias for the effective mounting and operation of adaptive cruise control (ACC) radar systems. In addition, we’ve assisted with the design of rear fascias to ensure that they are highly visible to ACC.’
Electrification brings further change to the world of simulation and modelling. Schramm said: ‘Electric vehicles enable and demand changes to traditional vehicle architectures. Combined with a change in attribute priorities – especially if the crash regulations for autonomous vehicles diverge from current passenger cars – means past experience is not a good guide to future designs.’
Elaphe Propulsion Technologies develops and manufactures off-the-shelf and custom in-wheel electric powertrain technology for the automotive industry. The company uses a range of FEM simulation and modelling tools across its design and development lifecycle. The company’s CTO, Gorazd Gotovac, explained some of the challenges the team faces: ‘Since we design the most torque-dense e-motors in the world, we often need to take into account a lot of details already during the optimisation steps, or else we risk choosing a non-optimal design, which leads to late changes in the design process and costly overruns.’
‘Some of these details, especially the ones that depend on geometric nuances, are obtained much more accurately with finite element analysis than by analytical equation – however, optimisation algorithms require a large number of calculations and transient multi-physics calculations tend to be too slow to allow algorithm convergence,’ he added.
As a result, the simulation needs to perform the minimum required numerical calculations and obtain results from the underlying analytical model. ‘The linking of the two worlds – numerical and analytical – is challenging but very rewarding in terms of result accuracy and speed balance,’ Gotovac explained.
This need for speed was apparent when Volkswagen Motorsport recently beat the time record at the Pikes Peak International Hill Climb. Its first fully-electric race car was developed with the help of ANSYS’ simulation solutions and driven by racing professional Romain Dumas.
Volkswagen Motorsport engineers carried out complete virtual drive tests of the entire race to help optimise the battery cooling system with minimal weight and aerodynamic drag loss. The engineers also replicated the course’s extreme driving conditions using this multiphysics solution.
François-Xavier Demaison, technical director at Volkswagen Motorsport and I.D. R Pikes Peak project, said: ‘Behind the wheel of the 680hp sports car prototype, Dumas mastered the track and the battery cooling system performed precisely as simulations predicted. ANSYS provided us [with] the competitive edge to outperform the high altitude and challenging turns and set a new world record.’
Moving to multiphysics
Multiphysics simulation and modelling tools help optimise the many interacting systems now found in our vehicles. Schramm explained: ‘HyperWorks provides a broad solver and workflow portfolio that enables the automotive industry to use more designs with optimisation and multiphysics simulations of interacting structural, mechanical, thermal, electromagnetic and fluid behaviour for all manufacturing methods.’
That’s a lot of systems and subsequent factors to optimise - but such optimisation is vital for automotive companies to gain a competitive advantage. For example, sound quality ‘will play a critical role in the differentiation of new vehicles,’ according to Schramm, who added: ‘The Statistical Energy Analysis (SEA) embedded into the SEAM software, Altair’s most recent acquisition, allows engineers and designers to identify and solve noise and vibration problems early in the design cycle, saving critical time and money, shortening the product development cycle and improving the user experience.’
As a result, simulation and modelling tools are increasingly required across every stage of the design and manufacturing lifecycle for the industry. Schramm said: ‘Massive deployment of Altair’s structural optimisation technology at the concept stage (including topology, topography and free-size) through to the detailed design (size, shape, and free-shape) and into manufacturing (for gauge, composites, additive manufacturing) is becoming more common.’
To truly get the industry up to speed with the world of simulation and modelling, improvements in user education must be realised. Prior explained: ‘The speed of deployment in the automotive industry is in line with the speed of adoption [of simulation and modelling tools]. We need to develop the workforce of today and the future to adopt this software and use it in a productive manner.’
While machine learning and AI-guided user workflows can help, according to Prior, educational initiatives such as La Fondation Dassault Systèmes can help businesses harness the power of simulation and modelling and understand its impact across all industries, workflows and research. ‘If we do not have a user community, it’s owning a racing car when you have only just passed your test,’ he added.
The University of Toronto Formula Society of Automotive Engineers (SAE) Racing Team is facing similar educational challenges. This student-run club designs, builds and competes with an open‐wheeled race car at the global Formula SAE (FSAE) Collegiate Design Series of student competitions.
Jonathan Lee, head of chassis at the University of Toronto Formula SAE Racing Team, said: ‘The most difficult challenge for the team, as a whole, is knowledge transfer. Due to the naturally high turnover rate of a student design team, it is difficult to properly educate the new members, while juggling all the other responsibilities entangled with being on a student design team and university classes.
‘Add to the fact that participation in the club is voluntary, it becomes difficult to determine who is willing to put the necessary hours in early on, which shortens the amount of time available to bring valuable recruits up to speed. The most recent attempts at fixing this revolved around better recruitment campaigns, dedicated recruitment leaders, as well as focused mentorships to bring the talented new members up to speed.’
The team used the ACTnowHPC on-demand cloud solution from Advanced Clustering Technologies (ACT) and Simulia’s Star-CCM+ to run computational fluid dynamics simulations in order to develop numerous iterations of the vehicle’s front-wing endplates.
These endplates have a significant effect on the performance of the wing and aerodynamics of the car. Lee said: ‘With regulations restricting the maximum height of the front wing, and a minimum ground clearance set by some basic calculations, the next least complicated part of the front wing to find maximum benefits from was the front wing endplates.
‘In recent years, owing partially to the advent of cloud computing, teams have started utilising CFD software to take advantage of aerodynamic forces, just like in motorsports. By properly utilising and designing a complete aerodynamic package, teams can drastically improve the handling performance of their cars, and the use of aerodynamic forces is now basically a requirement to compete at the bleeding edge of FSAE,’ he added.
The University of Toronto SAE team started using CFD three years ago and has been ‘slowly building the knowledge required to use an inherently very complicated tool,’ Lee said. ‘With the knowledge built up, the team is now surveying the possibility of utilising CFD to create a complete aerodynamic package, one that takes into account more nuanced aspects of aerodynamic stability, like pitch and yaw sensitivity,’ he added.
‘ACT’s suite and expertise in high-performance cluster workloads allowed us to use more accurate simulation tools and models, in order to better understand the car the team develops and manufactures. Specialised knowledge with such workloads allow for efficient use of computationally-heavy simulations, which can drastically improve vehicle performance when used correctly,’ Lee explained.
The automotive industry is increasingly embracing simulation and modelling to gain a competitive advantage and keep pace in this vastly changing sector. As more tools are developed, the industry stands to benefit greatly in the years ahead.
Schramm concluded: ‘Whether defining more effective design engineering workflows, using fleet-wide data acquisition to improve operational efficiency, or enabling the predictive maintenance of each vehicle, the true potential of combining advanced physics simulations and data science solutions throughout the vehicle lifecycle is just being realised.’
