How virtual-first development is transforming powertrain engineering
Mahle Powertrain explains how simulation, model-based development and hardware-in-the-loop testing are reducing development risk, cutting costs and accelerating powertrain commissioning
The shift from a physical-first to a virtual-first approach in powertrain development is less a single technological leap and more an accelerating evolution. In a recent interview with Richard Mead, Senior Principal Controls Engineer and Mark Underwood, Controls Engineer from Mahle Powertrain, they unpacked how the combination of established simulation tools, model-based development workflows, and targeted hardware-in-the-loop (HiL) testing is changing programme planning, reducing risk, and compressing commissioning timelines – without sacrificing quality or safety.
"We're looking to compress times, and to save costs … That drives you to compress your hardware testing to the minimum to cover your essentials for validation, for development you might have planned
Why virtual-first now?
Two broad forces are pushing the industry toward virtual-first practices: market pressures to reduce cost and schedule, and the maturation of software and simulation environments that let engineers answer hard questions earlier. Mead framed the motivation pragmatically:
"We're looking to compress times, and to save costs … That drives you to compress your hardware testing to the minimum to cover your essentials for validation, for development you might have planned,” Mead explained.
That compression is not about replacing physical testing. As Mead emphasised, virtual methods are used to make the physical phases more efficient and less risky:
“I don't think we're at that place where we would choose to do virtual validation over trying to do tests on the engine. I think it's more just making the programme efficient and having confidence that when you go to test on your hardware, you've already covered the fundamentals including areas which might be difficult to explore in hardware.”
For Mahle Powertrain, the impetus is also commercial: as a consultancy working on low-volume, high-value programmes, protecting limited prototype hardware and delivering predictable timelines are essential to remain competitive. Underwood explained how the scarcity of prototype parts elevates the value of virtual work:
“Some of the programmes we're dealing with are very high-value, low-volume production … Those prototype parts are in serious demand across many different teams … Anything that we can do to offload our work from being intensive on those limited supply of production parts is helpful.”
XiL and the “right tool for the right job”
Mahle Powertrain’s approach is pragmatic and layered: model-in-the-loop (MIL) and software-in-the-loop (SIL) are used where they add the most value; HIL is the critical integration step that validates ECU behaviour in a whole-system context. Underwood summarised the philosophy:
“It's having the right tool for the right job … If we're developing a new function to control a brand-new system with new logic and algorithms in it, the obvious thing to do is test it at the earliest possible opportunity in the model environment.”
Mead expanded on how they leverage and combine standard tools but add company-specific value:
“We start with off-the-shelf packages, but for our ECU software, we have some in-house developed wrappers to add to that standard software that enhance the usability, and calibration of the software …”
That combination – commercial toolchains (MATLAB/Simulink, auto-code generation) plus Mahle Powertrain-specific libraries and analysis-fed plant models – enables engineers to tailor simulation fidelity and translate virtual results into robust hardware behaviour.
From simulation data to calibration-ready controllers
One of the most concrete benefits of a virtual-first strategy is how it front-loads calibration and integration work. By connecting controller models to high-fidelity plant representations, Mahle Powertrain can begin developing and tuning calibration long before the engine is fired.
Mead reinforced how fast iterations in a simulated environment compress downstream troubleshooting:
“Having some really detailed plots of information flow through the strategy, can really save time and de-risk the development because you're only waiting for a few seconds for it to run through.”
The outcome is not theoretical: in a recent new-engine and control project Mahle Powertrain achieved what the team termed “record time” commissioning on the dyno because of careful virtual preparation. Mead shared more details:
Where we are essentially pulling together code from our library of control functions … we are able to get a new engine set-up and running on the dyno in 2 weeks
“With the engine rigged on the dyno … we were at first fire of the engine in a matter of hours rather than days or weeks … And really, after a few hours, everything looked like it was doing the right thing. So spun the engine up, fired it, and everything just worked.”
Underwood added that reusing tested modular functions from Mahle Powertrain’s libraries also accelerates initial deployment:
“Where we are essentially pulling together code from our library of control functions … we are able to get a new engine set-up and running on the dyno in 2 weeks.”
This combination of modular software reuse and virtual validation is a clear demonstration of how digital tooling shortens the critical path to meaningful physical testing.
HiL: the last word in integration
While MiL and SiL are powerful for early development, Mahle Powertrain views HiL as the final arbiter of system readiness. HiL tests the ECU with realistic I/O, exercising sensor and actuator interactions and revealing integration issues that smaller-scale simulations can miss. Mead described HiL’s role in the validation chain:
“HiL is extremely useful because that's the last point when ECU is in the loop, where you really get to test all the flow through the control system.”
Underwood reiterated the value of automated, continuous integration-style testing at the HiL layer:
“The key there is around continuous integration and more automated testing of that system to really test each new software build as it's delivered and find bugs that may have been introduced by new features or modifying functionality.”
The workflow is iterative: use MiL to validate new logic and get early calibration baselines, employ SIL to verify code translation, then validate full system behaviour in HIL and through regression testing.
Leveraging internal data and the caution around AI
Mahle Powertrain is also exploring ways to make the organisation’s historical models, test data, and simulation assets more discoverable. There’s recognition that a company’s internal knowledge is a powerful asset, and that AI-style search and data-sifting tools can help unlock that knowledge.
Underwood noted an existing group-wide tool, but Mahle Powertrain draws a careful line on applying generative AI to safety-critical code: the team is cautious about using AI-generated embedded code because the effort required to verify every line reduces the value proposition.
For now, the immediate AI value is in data analysis, reporting, and knowledge discovery – not in replacing the disciplined engineering practices needed for time and safety-critical systems.
Transferring methods across powertrain domains: ICE, hybrid and EV
A central question for many engineering teams is whether methods optimised for internal combustion engines translate to hybrid and electrified systems. Mahle Powertrain’s answer is a clear yes: the XiL approach and model-driven validation are transferable across domains.
Mead: “The XiL approach and exercising control strategies against virtual models is definitely transferable and applicable … being able to exercise edge cases and your responses to faults virtually is always going to be a benefit.”
Underwood highlighted concrete applications in motor control, battery safety analysis, and thermal management:
“We are using analysis tools to inform safety investigations into battery control systems … I've been involved in some model-in-the-loop work on motor control systems … There's a lot you can do in modelling the electrical dynamics of a motor and getting your inverter control correct before you meet the real hardware.”
Hybrid powertrains add layers of complexity – energy management, additional torque sources, and expanded degrees of freedom – but those are exactly the scenarios where virtual modelling offers the most utility. Underwood put it, as the degrees of freedom are increasing, modelling is a necessary countermeasure.
Balancing hardware and software innovation
A recurring theme is that software and hardware advances are complementary, not substitutive. Software opens new operating envelopes that hardware improvements make possible. Mead framed the relationship pragmatically:
“Hardware is kind of an enabler giving control strategies authority to enhance efficiency … Hardware developments enable these new strategies and new ways of looking at the system performance as a whole to extract more out of it.”
Underwood added an example where hardware innovation unlocks step-change performance and emissions capability, underscoring that hardware development remains central:
“If you take the Mahle Jet Ignition system that we're fitting to the latest V12 programme, that's a development of hardware … it's capable of delivering an enormously powerful engine that’s compliant with the latest emission standards … To the outside observer, it can seem quite straightforward, but to deliver those performance levels while still keeping a catalyst functioning properly is quite something.”
Cost, competitiveness and the consultancy perspective
For consultancies working in the low-volume, high-value space, the economic drivers are very real. Virtual-first methods are a route to preserved margins and competitive pricing:
“We do have to find the most cost-effective way to deliver it … as a business developing an engine from scratch, we will already have the 1D models … utilising that 1D model to do other things and test the control system and even calibrate the control system before we see any hardware is an obvious cost save if you can realise it,” said Underwood.
By reusing libraries and leveraging virtual platforms, Mahle Powertrain can shorten milestones and provide customers with faster time-to-insight, without compromising necessary validation and safety checks.
An accelerating evolution, not a revolution
Mahle Powertrain’s experience underscores a simple fact: virtual-first development is an accelerating evolution. It is built on decades of engineering experience, validated models, modular software libraries, and disciplined integration testing. The result is not replacement of physical testing, but smarter, earlier work that reduces risk and compresses the expensive, time-sensitive phases of hardware commissioning.
As Mead summed up the ethos driving the practice:
“You know, if you get a feel for how things are going to work virtually and then just go in, plug it in, and know that you can crack on and be productive, that's where we're aiming to be.”
And Underwood’s closing view reflects the balanced perspective required for modern powertrain development:
“Building confidence in virtual modelling is all about correlation...”
The path forward will continue to blend hardware innovation, modular software practices, and increasingly sophisticated virtual models. Where this leads is not the end of physical testing, but rather the more strategic, lower-risk use of hardware time to solve the problems virtual methods cannot fully emulate. For teams under commercial pressure, that is precisely the value proposition that makes the virtual-first approach compelling.
