Taurus: a powerful mix of industry standard and lessons learned through experience

Automotive drivetrain engineers aim to perfect and refine electric drive lines to the point where they operate right at the edge of what is physically possible. This requires simulation models, to act as cost function in the design process or to train reduced order models. These latter models should incorporate all physical loss and performance mechanisms and should be computationally efficient at the same time. Often however, important contributions to efficiency or performance drops are hidden in empirical build factors. These factors can only be quantified too late in the design process, i.e. after testing of the first prototypes.

Based on our experience we developed Taurus, a simulation tool chain that uses industry standard tools, but adds the necessary embedded experience to capture detailed impact from component level, all the way up to the drivetrain level.

Use case: Taurus eliminates the use of build-factors for motor loss prediction

Detailed modelling in Taurus allows to calculate in a computationally efficient way the loss contributions in the motor induced by manufacturing effects and high-frequency operation. This includes degraded material properties at the cutting edges and mechanical pressure for the manufacturing effects. PWM switching induced losses in the magnets, copper, and iron are also calculated. Figure 1 illustrates the delta in amount of losses and their spatial distribution by including these effects. Prior to the calculations, additional material characterizations have been carried out.

Figure 1: ignoring PWM-induced losses and manufacturing effects leads to incorrect loss distribution data.

This spatial loss information is calculated for all operating points and thus allows further detailed analysis utilizing the data in time-based simulations to identify hot spots or to calculate cycle consumption for the full driveline.

Figure 2: Cycle loss breakdown: industry standard tool (build-factors) vs. Taurus (first principles).

The net effect of this increased fidelity on a drive cycle consumption is visible in Figure 2. It compares the loss predictions from an industry standard tool (using build factors) with the first-principles approach from Taurus and shows the correlation with measurements.

Conclusion

With limited additional characterization, a detailed and computationally efficient calculation allows to optimize cycle consumption and avoid local hot spots. By adhering to two main principles: leveraging industry standard tools and including loss mechanisms on a first principles basis, Taurus enables fast driveline R&D support from concept to troubleshooting.

Introducing DRV Solutions

DRV Solutions is an engineering partner for advanced electric drive lines, supporting from concept to validation. We combine motor & power electronics design, full driveline analysis, and system integration expertise in our Taurus Toolkit. This enables us to accelerate your design and troubleshoot performance issues.

How deeptech is already revolutionising EV powertrain engineering

Simon Shepherd, Head of eDrive and Chief Product Officer, Monumo

Deeptech is transforming EV powertrain engineering by introducing new levels of computational freedom, speed, and system integration, allowing companies to achieve levels of performance and cost reduction previously out of reach through existing methods.

Nature-Inspired System Design

Modern engineering increasingly draws inspiration from nature, where efficiency is achieved through adaptation and system-level harmony. EV developers are now using advanced computational tools to explore organic, highly optimised component shapes and system architectures; mirroring, for example, how a tree or a bird’s wing achieves a perfect balance of forces. Unlike natural evolution, which takes millennia, digital tools allow engineers to iterate and converge on superior solutions within weeks, critical for keeping pace with rapidly changing industry and regulatory demands.

Overcoming Conventional Barriers

Classic engineering methods, which adjust just a handful of design parameters, can no longer solve today’s complex powertrain optimisation challenges. Next-generation computational platforms let engineers evaluate vast combinations of component geometries and system parameters – tens of millions, in some cases -far beyond what’s possible with manual approaches or simple brute-force computation. Intelligent automated routines and novel search strategies deliver the breakthroughs needed when conventional software and human time simply cannot scale

Real-World Results at Scale

Breakthrough deeptech platforms – like Monumo’s Anser® Engine – are now enabling EV makers to rapidly generate and assess hundreds of thousands of valid designs for components such as motors and gearboxes.

• In recent projects, 250,000+ viable motor designs were created in just three days, trimming magnet use by more than 8 % and cutting costs by 4 %, saving €15 per vehicle. [1]
• Expanding system optimisation to include additional variables (e.g. stator dimensions, gears, and motor length) produced over 550,000 valid solutions in five days, achieving savings as high as 11 % (€43 per unit), or reducing losses by 12.5 % at no added cost. [2][1]

The Next Frontier

The emergence of “generative design” tools – able to propose promising designs directly – suggests future engineering cycles will be radically compressed, possibly delivering optimal solutions within minutes instead of days. Monumo’s technology roadmap forecasts continued gains as greater systemlevel freedom, parametric control, and digital intelligence are brought to bear:

• Motor-only optimisation: ~5 % cost reduction
• Whole-system integration: 10 %+
• Projected potential with full-system and freeform optimisation: 20 %+ cost reduction.

Deeptech is no longer a future prospect: it is already driving major advances in EV powertrain cost, weight, and performance, for the manufacturers prepared to fully adopt it.

Simon Shepherd is Head of eDrive and Chief Product Officer at Monumo, a deeptech startup based in Cambridge and Coventry, UK, specialising in AI-driven engineering solutions. Ready to explore how deep-tech can transform your powertrain development? Contact Simon.Shepherd@Monumo.com, see more at monumo.com listen to him speak about our latest developments at 11:15 on Wednesday 3rd December in Deep Drive Track L.

AI-Powered Engineering Software for Electric Drives: OPED

Boosting powertrain development with agility, fast time-to-market and optimal product-market fit

Dr. Martin Hofstetter, Head of E-Mobility and Alternative Drivetrains Research Group, Graz University of Technology
Dr. Dominik Lechleitner, Senior Researcher, Graz University of Technology

Designing electric powertrains is challenging: engineers must quickly find competitive designs and optimize the system for multiple key performance indicators (KPIs) at once, e.g., efficiency, cost, and package. The industry-approved engineering software OPED (Optimization of Electric Drives) can do this automatically by combining parametric system models with an AI-based optimization algorithm and exploring hundreds of thousands of design variants within 24 hours.

Development of Electric Drives

The development of electric drives (e-drives) is a highly complex and interdisciplinary process. Engineers must simultaneously design numerous electrical and mechanical subsystems (see Figure 1) that must optimally work together while meeting ambitious system targets for performance, efficiency, cost, and packaging. These objectives are often conflicting – improving one typically worsens another. Moreover, this highly challenging task must be solved under strong time pressure as it is critical for ambitious time-to-market goals. Therefore, engineering of electric drives demands digital tools capable of handling multi-criteria optimization and cross-domain interactions in an integrated way to quickly provide solid answers to complex questions.

Revolutionizing the Development Process with OPED

The software OPED fundamentally changes how e-drives are developed. Instead of relying on sequential design steps and manual iterations, OPED uses parametric system models combined with an AI-powered evolutionary optimization algorithm to explore the full design space automatically. The outline of OPED is shown in Figure 2: Based on specified e-drive system requirements, the design problem is encoded as a multi-objective optimization problem. The intelligent design algorithms then generate different e-drive designs, which are evaluated by system analysis models. Based on the calculated design properties, the optimization algorithm rates the generated designs and aims at improving them based on the best found designs so far. This closed loop of design analysis and synthesis continues until no more improvements are observable and converging behavior is present. Furthermore, self-learning artificial neural networks boost the optimization performance by guiding the optimization algorithm and directing its search towards promising design regions. Within 24 hours of computation time, around 50 design parameters are varied, hundreds of thousands of possible e-drive designs are evaluated, and the most promising ones are identified based on multiple concurrent objectives such as

• performance,
• efficiency,
• cost,
• package integration,
• carbon footprint,

and any other design objective – everything that can be calculated, can be optimized. The result is a Pareto front of optimal solutions, providing engineers and decision makers with a clear overview of achievable trade-offs and design potentials. By merging simulation, optimization, and system understanding in one integrated workflow, OPED enables rapid development cycles. Agility is key: every request for quote (RFQ) or sudden project change request (PCR) is replied with fast and solid answer, tailor-fit to the specific requirements. Taking informed decisions early ultimately leads to an optimal product-market fit for the e-drive system and a fast time-to-market.

Figure 2: From e-drive system requirements to optimal e-drive solutions with OPED

Master Challenging Targets of Package, Efficiency and Cost within 24 Hours

OPED’s strength lies in its versatility. The software can be applied to a wide range of design questions – from component sizing and material selection to system-level trade-offs and product family development. Moreover, as system requirements are often vague and uncertain in the early development stages, OPED can be utilized for requirements engineering. Another powerful capability lies in the full 3D package investigation: OPED not only finds designs that comply with a given 3D target installation space, it also provides possible packaging options within the available space. An example is shown in Figure 3, which depicts two possible design solutions for an e-drive: One with the smallest possible length and one with the smallest possible height (e.g., providing additional trunk space for applications at the rear vehicle axle).

Besides package feasibility, both energy efficiency and cost are always critical and conflicting KPIs. To select the most suitable solution, OPED provides a Pareto front of e-drive designs (Figure 4), where each point represents one optimal design solution. Accordingly, engineers and decision makers are provided with a solid foundation for selecting the most suitable system solution respecting the specific goals of each vehicle application. As each design solution from OPED contains detailed technical information – including a 3D CAD model – a seamless and smooth transition from OPED results to the A-sample development is ensured. This makes OPED a powerful enabler for fast and digital electric powertrain development.

Figure 3: Example of e-drive packaging options; available installation space is shown in blue

Concluding, OPED enables engineers and decision makers to

• respond quickly to requests for quote (RFQs) and project change requests (PCRs),
• solve conflicting KPIs & do requirements engineering,
• develop competitive solutions with product-market fit,
• design optimal product families, utilizing commonality and carry-over-parts.

OPED is established in practical use at a leading global automotive tier 1 supplier – with high potential forscaling across other suppliers and OEMs.

Figure 4: Select the sweet spot of energy efficiency vs. production cost (vehicle model from [1])

Sources:
[1] Holiday, D. (2021). Jaguar I-pace Concept. https://sketchfab.com/3d-models/jaguar-i-pace-concept-3ea106994ec9442eb4b72906026fa215. CC Attribution: https://creative-
commons.org/licenses/by/4.0/, modified. [Online; accessed 13 February 2024