Protecting a single profit pillar is not a winning strategy

Global powertrain markets are diverging. BorgWarner CEO Joseph Fadool explains why “making speed the moat”, re-regionalization, and AI will shape the automotive propulsion industry – and why policy shouldn’t push technology

Joe, what does it mean for BorgWarner when, as is currently the case in global markets, propulsion concepts are diverging dramatically?

Maybe some context is helpful here. In the past, let’s say 30 to 40 years ago, much of the powertrain development was driven by emissions and fuel economy improvements, with Japan and Germany leading in part. Each region followed more or less the same path, adopting technology that led in one region and then flowed to others three to five years later. What’s changed now is that each market requires a fundamentally different mix due to a combination of local regulations and consumer behaviors. In China, over 50% of vehicles are hybrids or pure BEVs; Europe is approaching 18-19% EV share; the U.S. is stepping back, with government incentives withdrawn and EV penetration expected to remain around 7-8%. For BorgWarner, as a global company serving all major OEMs around the world, the good news is that we have a resilient portfolio capable of serving all markets, no matter the propulsion type. As an industry, we must return to a customer-first mentality rather than letting governments legislate what people will buy. I think that was a complete disaster. The OEMs know what consumers want to buy. You’re going to continue to see regionalization and differences between the markets. Long term, however, we still believe in electrification; it is the only way to truly decarbonize. But it will happen at different speeds, it is dependent on many factors like infrastructure and rare earth mineral availability, and it won’t be without disruption.

In your plenary speech at the CTI symposium in Novi, you used the term “moat”. What are the decisive factors in stabilizing or widening it, especially in competition with China?

Moat is a term Warren Buffett and others used – a reference to castle moats that slow down or stop the enemy. For our industry, it means a couple of things. First, innovation: more value at lower cost, more efficient powertrains – better fuel economy for combustion, smaller batteries for the same BEV range. Second, and this is what’s changed in the last five years: speed as a moat. China is teaching the rest of the world that you must constantly reinvent and bring better products to market faster. The companies that move at speed will win; the slower ones are stuck in the old paradigm. We see this reflected in the OEM landscape too: when I joined the industry 35 years ago, growth was driven by Ford, GM, Volkswagen, Toyota. Today, the only OEMs really growing are seven or eight Chinese companies, plus Hyundai. Hyundai is still growing and doing well. Everyone else – GM is out of Europe, Stellantis retreating from India, Ford down to 3% in Europe – is shrinking. Retreating to protect a single profit pillar is not a winning strategy, in my opinion.

How does regionalization align with the traditional concept of economies of scale through a “world product”? And what are the risks of speed over scale?

This is a great question, because for 20 to 25 years, as global vehicle volumes grew from 50 to 90 million, scale was the name of the game. We were investing heavily in technology, developing suppliers, and building factories that needed to run efficiently. What we now see is that beyond a certain point, other factors become more important than scale – specifically speed and local accountability. That doesn’t mean abandoning scale; it means finding a new balance. Housing and mechanical parts, even factory assembly, require far less scale than before because these are readily available products, and suppliers are more regionalized. Semiconductors, on the other hand, remain a scale game – chip companies pay more attention to Tier 1s and OEMs that give them high volume. The downside of scale is loss of local agility. Large global competence centers that push technology out to regions are less effective now – they’re expensive and slow. When a region has to route

decisions back to a distant center of competence, that’s time lost, and people far from the customer are making local market decisions without fully understanding the pressure on the ground. What we find more effective is giving regions greater authority and competence – a democratization of know-how, with regions learning from one another rather than relying on a central hub. BorgWarner has a decentralized operating model for this reason.

Would a concept of “similar but not identical” components across markets work for you as a supplier?

Definitely. Take turbochargers: they spin at 300,000 RPM and can be dangerous if they fail. Engineered for the German market – high speeds, high temperatures, autobahn use – they’re built to be virtually indestructible. In China, the use case is mainly stop-and-go traffic with 1.5-liter engines. The load profile is fundamentally different, so you don’t need the same robustness. We’ve reengineered our turbo line for China accordingly – 20% lower cost than the European or North American equivalents. Five or ten years ago, we would have carried over the European or North American product into China. And we found we were no longer competitive. We were sometimes over-engineering for markets that didn’t need it – and while we told ourselves we were gaining scale, the design simply wasn’t affordable in every market.

How can leadership culture help to handle the change?

Leadership has to start by accepting the uncertainty. A big part of the job is looking around the corner and anticipating the future – that has become much harder. It means thinking in scenarios rather than toward a single point, even while maintaining a true north. The second shift is toward flexibility: a resilient portfolio, a flexible supply chain, and a manufacturing footprint that can serve multiple customers on the same production line rather than running just one. Leading today is about providing clarity on facts and priorities, while helping people navigate the uncertainty. The disruption increasingly comes from outside the industry, and that’s accelerating. In China, many successful automotive players came from consumer electronics – Huawei being the obvious example. In the U.S., Apple, Google, Waymo, and Tesla have demonstrated that companies with no traditional automotive background can become highly relevant because of software competence and systems thinking. These players don’t focus on individual components – they think in terms of the experience they want to deliver to the customer.

BorgWarner has been shaped strongly by powertrain hardware. What does it mean when intelligence is shifting into software – and the SDV?

We’ve seen software grow in the powertrain space for a long time – ECU development, electrification driving demand for complete drive modules with integrated software. The software-defined vehicle is a parallel innovation aimed at reducing development costs and enabling over-the-air updates long after a vehicle has left the dealership. If you think of how a smartphone works, the SDV follows the same logic: new features come through software, not hardware replacement. Electrification and SDV actually accelerate each other. The hardest part of implementing SDV is doing it on a legacy platform with hundreds of distributed electronic modules. The goal is to consolidate to a zonal controller architecture. At BorgWarner, we see a future where the Powertrain controller integrates into the front-end zonal architecture. Some competitors have expanded from their ECU supplier role into zonal suppliers – we’re looking at the same path. Most OEMs currently keep Powertrain as a separate subsystem, but we expect that to change, and we want to drive it.

Where is AI most important for BorgWarner as a company?

We’ve used machine learning and AI in our factories for a long time. Generative AI with large language models adds value in three areas. First, product development: replacing routine engineering tasks with agents and tools – code generation, test automation. The highest benefit

comes from automating the repetitive tasks that consume significant engineering hours, freeing time and resources up to manage more strategic work. Second, factory automation: AMRs and robots are increasingly AI-enabled, and generative AI makes it significantly easier to train them. We expect a step-function increase in automation potential in our factories. Third, personal productivity and end-to-end process improvement: using tools like Copilot or Claude for daily work, and automating manual processes such as monthly financial closes and forecasting. Generative AI is one of the biggest changes we will witness. Personally, I find it liberating – you can delegate the monotonous groundwork, the research, the routine write-ups, and focus on the work that actually creates value.

How will increasingly software-defined vehicles change buyer expectations, particularly in North America, where combustion still dominates?

The two trends are closely linked. SDV is easier to implement on an EV platform. As EV batteries become more affordable, adoption will grow; as EV adoption grows, users will experience more features; that experience will pull more people toward EVs in the future. It’s a reinforcing cycle. We expect SDV to reaccelerate electrification as electric vehicles become cost-competitive. Europe will play this out first; North America will follow. Buyers will demand over-the-air updates, vehicles that have more features and deepen integration with their home and mobile environment. You know, these cars are awesome. So I think these two things are converging, and it will reaccelerate electrification to some degree.

Interview:

Gernot Goppelt

Innovation Powered by High-Fidelity Multiphysics Modeling

A Holistic System Optimization of BorgWarner’s Next Generation Integrated Drive Modules

Dr. Arnaud Leblay, System Engineer Technologies & Innovation
Eric Bourniche, Engineering Supervisor Technologies & Innovation
Dr. Pascal David, Engineering Manager Technologies & Innovation
Adrien Bossi, System Engineer Technologies & Innovation
Harsha Nanjundaswamy, Engineering Director Technologies & Innovation
all BorgWarner PDS

1. Introduction

The development of cost effective, energy efficient and reliable Integrated Drive Modules (iDM) for electric vehicles is closely tied to optimization challenges. Addressing these challenges requires Multiphysics holistic approaches, that span magnetic, electrical, thermal, mechanical and fluid domains. Such an engineering methodology has been adopted by BorgWarner.

Across the industry, numerical methods such as Finite Element Analysis (FEA) and Computational Fluid Dynamics (CFD) are widely used to evaluate systems involving one or more physical phenomena within multiphysics analyses. These Tools form the foundation for higher level modeling approaches such as Digital Twins, which enable full system vehicle simulations and the evaluation of electric machine control strategies under real driving conditions. Combined, these Methods provide critical metrics that support decision making and guide technology development roadmaps.

BorgWarner PowerDrive Systems Technologies & Innovation group has developed a high fidelity, multiphysics based Digital Twin to support the development and innovation of iDM. The model has been rigorously validated through comparison with 78 physical signals sourced from both CAN communication channels and external sensor data [1]. In addition, some numerical models of sub components have been developed and validated in partnership with universities.

2. Methodologies Development for Deeper Insight

Methods form the foundation of engineering organizations. Sustaining innovation requires continuous improvement. Our methodology development strategy is grounded in in-house technical excellence, with experts actively contributing to Centers of Expertise to share best practices, lessons learned and identify improvement opportunities. This internal capability is further reinforced through strategic partnerships with academic institutions, start-ups, and technology providers. Our Analytical Simulation Design plan [1] is formed by simulation Cards whose inputs and outputs are bound to other cards to form a structured simulation framework. It enables pinpointing methodological gaps and areas where model predictions need refinement for greater accuracy and highlights phenomena that require deeper investigation through multiphysics analysis.

Those refinements are based on a dual approach combining advanced multiphysics modeling with high-quality experimental validation. The refinements can focus non-exhaustively on thermal, electric, magnetic, mechanical or fluid dynamics behaviors on overall iDM and its subcomponents.

One of these approaches concerns the methodology refinement applied to electric machine stator loss assessment. Stator loss can be split into iron loss and copper loss. The loss estimation enhancement is performed in collaboration with the Division of Industrial Electrical Engineering and Automation of Lund University in Sweden. The study was carried out in multiple steps to characterize the material properties, model the electromagnetic behavior, and finally perform transient thermal analysis.

The first step of the loss evaluation is to measure the lamination loss named iron loss and magnetic properties as magnetic hysteresis curves also known as BH curves. The experimental description can be found in [2].

The second step involves using the magnetic properties obtained in the first step and incorporating them into 3D Finite Element (FE) model of a complete wound stator under COMSOL Multiphysics, to assess the copper loss. The simulated impedances are compared to impedances measurements, and the relative error is not higher than 5% over a range of 100 Hz to 10 kHz.

The resulting current density with the conductor cross section can be observed on Fig. 1. The current density is not equally distributed resulting in the difference of losses within the conductors. A coupled electromagnetic thermal model was then employed to perform a transient analysis under adiabatic boundary conditions with the surrounding environment. The resulting temperature distribution after a 30 s transient event is presented in Fig. 2.

Figure 2: Temperature distribution after 30 s transient heat up in adiabatic conditions at 200 A(peak) and 1000 Hz

At system level, the inverter generates the current waveforms required to control the electric machine’s performance and control strategies are developed accordingly to enhance overall system behavior. Using the previously detailed multiphysics approach enables a precise assessment of how those current waveforms impact electrical machine losses. This capability permits us to optimize the influence of innovative control strategies at inverter level on vehicle efficiency as demonstrated in [1], as part of our model driven innovation and optimization capability.

3. Model Driven Innovation and Optimization

Innovation and optimization of the iDM are supported by several key pillars. One of them is Process and Design Development. Process and design tools are evolving in parallel, allowing engineers to explore innovative geometries and structures that were previously unattainable. Such optimization has been performed for a heatsink used in a power electronincs device, resulting in complex geometry as shown on Fig.3. A cooling liquid is turbulently flowing inside the heatsink to cool two heat sources. The topology optimization has been set to minimize the temperatures of the heat sources and their temperature differences, while being below a pressure drop threshold. It defines the inner geometry. The synergy between 3D metal printing and numerical multiphysics topology optimization has become a key driver for innovative solutions.

Figure 3: 41st iteration of 3D topology multiphysics optimization with a turbulent flow and two heat sources at the lowest surface

A second pillar is driven by AI breakthroughs in materials development. Historically, geometry/function and material selection have always been closely intricated in the development of components. Recent progress in artificial intelligence has significantly accelerated the growth of Computational physics, chemistry, and materials science. These approaches leverage quantum mechanics to predict the atomistic structure of materials and translate it into macroscopic physical properties. Machine‑learning‑based computational physics now enables tailoring a material’s atomic structure to meet specific performance requirements [3].

Figure 4: Integrated Drive Module iDM 180-HF, illustrating the inverter, its liquid cooled power module and its Viper power switches.

The third pillar is enabled by Digital Twin, which builds upon the first two approaches to optimize physical components and reinforce system‑level foundations. This pillar focuses on the development and evaluation of innovative control algorithms and new iDM control strategies [4]. Although the inherent complexity of the vehicle system impacts Simulation time, preliminary insights can often be obtained within only a few electrical cycles. While advanced control algorithms can be evaluated on a millisecond timescale, the full potential of Digital Twins is realized when combined with Reduced Order Models (ROMs), which significantly
accelerate simulation performance.

4. Use Case

The Digital Twin methodology makes it possible to extract high level, system wide performance indicators while still capturing the detailed behavior and interactions of individual components. Fig.4 illustrates this multiscale capability by zooming from the overall iDM architecture comprising the inverter, electric machine, and transmission, down to ist subcomponents. Within the inverter, for example, the power module is modeled at a granular level. It consists of a stacked assembly that typically includes heatsink, thermal interface material (such as a thermal pad), and the power semiconductor switch. This detailed representation ensures that local thermal electrical behavior including parasitic phenomena are accurately captured and propagated up to system level performance metrics.

In this study, a vehicle Digital Twin embedding all multiphysics complexity into one single environment is employed. Electric machine, inverter and transmission models are interconnected and brought into a full vehicle representation. This includes detailed, non-exhaustively, thermal management, intelligent control strategies, thermal models, cooling system, lubrication.

Based on the vehicle Digital Twin simulation over a WLTP cycle, the results presented in Fig. 5 demonstrate the benefit of combining detailed component-level analysis with a system-level approach to assess global performance indicators.

The power switch temperatures can be accurately estimated at both granular and helicopter views. The evolution of the maximum power switch temperature over a single WLTC for three different thermal conductivities of the pad can be evaluated. The temperature profile exhibits alternating peak and cool‑down phases, driven by several factors such as instantaneous current demand, DC-link voltage, switching frequency, and the applied control strategies. In addition, the absolute power losses are strongly influenced by the semiconductor die Technology itself, which affects both switching and conduction losses. While many studies assume a constant coolant temperature at the heatsink inlet, vehicle thermal management has a major impact on component thermal behavior. Radiator valve actuation, front fan operation, and the overall thermal Management architecture and control strategy directly influence the inlet coolant temperature evolution. Consequently, these effects lead to different maximum temperature levels for the power electronic components, highlighting the importance of a fully coupled electro‑thermal‑system simulation approach.

5. Conclusion

The development of new numerical methodologies, validated through experimental investigations, enables a deeper understanding of coupled field effects, such as fluid, mechanical, thermal, electric, magnetic, and parasitic phenomena, and their impact on system behavior. By combining detailed component level multiphysics models with vehicle level driving cycles, BorgWarner’s Digital Twin provides system wide performance indicators while preserving physical fidelity. This capability transforms simulation from an analysis tool into a decision making and optimization platform for architecture selection, material choices, and control strategies. The ability to assess component-level impacts at granular view while simultaneously maintaining a helicopter view at system level is a key enabler for rapid innovation and robust holistic optimization.

Figure 5: Overview of Digital Twin capabilities from granular to helicopter view providing global performance indicators at vehicle system level.

Sources:
[1] Bossi, A., Bourniche, E., Leblay, A., David, P. et al., „Digital Twin, A Multiphysics Numerical Tool Chain for Next Generation Electric Drive Design,“ SAE Technical Paper 2025-01-8624, 2025.
[2] Colombo, L., Reinap, A., Fyhr, P., Alaküla, M., „Enhancing Core Loss Tracking Accuracy in Stator Cores: A Comparative Assessment of Static and Dynamic Jiles-Atherton Model Formulations,“ IEEE Transactions on Magnetics, vol. 61, no. 8, pp. 1-12, Aug. 2025, Art no. 7300512
[3] Boziki, A. “HPC in Physics: Enabling Simulations and Accelerating Data Processing,” Presentation at SCynergy, April 2025.[3] [4] Nanjundaswamy, H., Deussen, J., Mayer, A. et al., “A Step Beyond Two, “Next Generation Multi-Level Traction Inverter with Clean Wave Technology”, 46th International Vienna Motor Symposium 2025, ISBN: 978-3-9504969-4-9

No-Compromise Propulsion for the Heavy-Duty Electric Pickup

How a 2-speed integrated e-Beam axle removes the final barrier to practical heavy-duty electrification

Dan Ouwenga, Director – Product Management, Dauch

The promise of electric propulsion has always been straightforward: instant torque, smooth drivability, reduced NVH, and lower lifetime operating costs. For light-duty passenger vehicles, that promise has largely been delivered. But for the heavy-duty pickup truck – the Class 2b and Class 3 work vehicles that North American customers depend on to tow, haul, and perform under sustained load – first-generation electric architectures have fallen meaningfully short.

The technical challenge is the absence of a propulsion solution that can simultaneously deliver the launch torque, continuous power, and thermal robustness that HD truck customers require. A single-speed electric drive unit, however well-engineered, faces an inherent constraint: optimizing for peak launch torque forces a tradeoff against motor speed, mass, and efficiency at cruising conditions. The result is a system that can feel impressive off the line but struggles under sustained towing load which is precisely the use case that defines this segment.

Dauch’s HD 2-speed electric rear e-Beam was developed specifically to address that tradeoff. The architecture integrates a nested 2-speed planetary differential within the axle housing that enables low-range torque multiplication without placing additional torque loads on the upstream components. In low range, the system delivers between 16,000 and 25,000 Nm of wheel torque against a gear ratio exceeding 35:1 – numbers that comfortably exceed the demands of the most aggressive HD towing and gradeability cycles. In high range, the system transitions seamlessly to a configuration optimized for efficiency and highway operation, with the motor running in its peak efficiency zone.

The 2-speed design also enables the use of a smaller, lighter, highspeed induction motor – reducing diameter from a typical 230–260 mm down to 180 mm – which saves mass, improves packaging, and eliminates the supply chain and demagnetization risks associated with rare earth permanent magnet alternatives. A new flooded stator cooling system, using an integrated air gap sleeve to direct oil flow uniformly across the Copper windings, enables sustained motor output at more than double the current density of conventional cooling approaches. The result is continuous rated power above 225 kW – not a peak figure, but a sustained capability this single e-Beam axle can maintain through a full Davis Dam or Baker Grade simulation without thermal derating.

The inverter also benefits from innovative cooling technology. A folded fin heat sink design allows direct oil cooling of the silicon carbide switching devices, reducing junction temperatures and improving continuous current capability. Benchmarked against competing architectures, this configuration delivers more than 50% improvement over a conventional isolated back plate module – meaning equivalent continuous power with significantly less silicon area, reducing cost while improving thermal margin.

Real-world validation has confirmed what the simulations predicted. A prototype 2-speed HD e-Beam was integrated into a RAM Heavy-Duty BEV conversion vehicle and subjected to a full validation program spanning dynamometer testing, cold weather operation at Aumovio’s Brimley, Michigan development center, and hot weather SAE J2807 trailer tow simulation at the General Motors Yuma Desert Proving Ground in ambient temperatures exceeding 100°F. The vehicle successfully completed J2807 grade requirements at 26,000 lbs gross combined weight rating in two-wheel-drive – a demanding threshold that no comparable production-ready electric rear axle has achieved.

Thermal data captured during Davis Dam simulation at 45 mph in low range showed a 14–18% reduction in EDU operating temperatures compared to high range operation at the same load. That margin is not incidental – lower thermal load directly enables higher continuous power, longer towing duration, and greater calibration headroom. OEM customers present during vehicle demonstrations described the shift behavior as smooth and comparable to a traditional automatic transmission.

The technology is protected by a portfolio of 13 issued and pending patents, including two developed specifically for this application. Third-party freedom-to-operate searches have identified no blocking patents.

What this body of work ultimately demonstrates is that propulsion readiness can be removed from the list of barriers to HD electrification. The technical constraints that have prevented practical BEV and EREV adoption in the heavy-duty pickup segment – launch torque, thermal robustness under sustained load, production-viable shift quality – have been resolved. The remaining variables are market timing, energy storage and infrastructure solution readiness, and OEM program pacing. Those are real constraints, but they are external to the propulsion system. When the market conditions are right, Dauch is ready to provide an electrified HD solution.