The gap that would not close
Ask any chassis dynamics engineer what they do when a simulation requires wet road conditions. The honest answer is usually one of three things: they skip it and hope dry results are close enough, they apply rough manual corrections that are not grounded in physics (typically by scaling down the available grip and calling it a wet-road model..), or they commission a wet test campaign that takes weeks, it costs a significant budget, and still delivers incomplete coverage while being hardly repeatable.
None of these are real solutions. And everyone in the industry knows it.
Pacejka's “Magic Formula” is one of the most widely adopted tire models in vehicle dynamics. On dry ground, it works. Engineers trust it. It fits into simulation workflows, integrates with major vehicle dynamics tools, and delivers the accuracy teams need to make design decisions with confidence.
On wet ground, the story has always been different. For years, chassis and active stability systems engineers have faced the same frustrating reality: no clear standard for wet tire modeling has emerged. Not because people have not tried. But because the physics of wet ground behavior are genuinely harder to capture and because every measurement tool available has real limits.
That gap is precisely what MICHELIN SIMIX set out to close. For the first time, chassis teams have access to a wet Pacejka dataset that is physics-based, plug-and-play, and ready to use in their existing simulation workflow: without weeks of test campaigns, without homebrewed workarounds, and without disrupting how they already work.
Why the “Magic Formula” falls short on wet roads
To understand why the gap exists, you need to understand what changes when the road is wet.
On dry ground, contact patch length is relatively stable. The Pacejka model does not need to account for speed dependence in its Forces & Moments curves. That is fine for dry conditions and it is why MF-Tyre became the industry reference.
On wet roads, everything shifts. As speed increases and water film thickness grows, the contact patch length decreases significantly. That reduction has direct consequences on tire behavior:
- Longitudinal and cornering stiffnesses drop
- Peak grip forces decrease
- The aligning moment (Mz) is altered through a different effective trail
- Transient behavior (relaxation lengths) is impacted by the same mechanism
- Rolling resistance (My) increases
Because standard MF-Tyre does not model speed dependence in its F&M curves, it simply cannot reproduce these evolutions. The gap is not a flaw in Pacejka, it is a scope limitation. Dry was the starting point. Wet has remained an open problem.
For programs that include complex ADAS validation, active systems tuning, or any safety-critical scenario on wet roads, that open problem is a real bottleneck. You can not design for conditions you can not simulate.
The measurement challenge: why no single source is enough
The obvious answer to "we need wet data" is "go measure it." But wet tire testing is one of the most complex data acquisition problems in vehicle dynamics. Every method available has genuine strengths and genuine blind spots.
If you want to go deeper on why hybrid approaches are essential in tire modeling, our article Why tire model accuracy requires more than lab measurements covers the broader principles behind this challenge.
For wet ground specifically, here is what each source contributes and where it stops:
Lab tests (drum or flat-trac rigs) provide tight control over water depth, speed, slip angle, vertical load, and tire pressure. Repeatability is generally excellent. However, laboratory surfaces do not always represent real roads, and rapid surface wear during campaigns can cause grip conditions to drift mid-test. Wet testing also means deliberately splashing water onto highly instrumented test rigs worth millions of dollars—not an environment that engineers would choose if repeatability and equipment longevity were the only concerns. Valuable for isolating trends. Insufficient on their own.
Instrumented trailer tests bring real-road anchoring and a larger usable surface that reduces local wear effects. But precise measurement and control of slip angle is challenging, especially for small-signal responses. Environmental variability: temperature, sun exposure, contaminants; introduces noise that must be carefully characterized.
Vehicle tests are indispensable for assessing driver feel: grip/slip transitions, progressivity, steering torque linked to Mz. But they do not allow exhaustive analytical sweeps of the domain, and isolating individual parameters is difficult. They serve as a global coherence and subjective validation stage, not a primary data source.
Dynamic footprint and contact-area measurements (pressure mapping, high-speed imaging) provide direct access to how contact patch geometry evolves with speed and water height. They constrain the model on geometry and transients. But they do not directly measure forces or moments.
FEA/CFD simulations offer mechanistic insight into how water accumulates and how tread deformation interacts with the water film. When calibrated against measurements, they usefully complement physical understanding. But they do not directly produce Forces & Moments at the wheel center, and become costly to scale across tire variants.
The conclusion is clear: never rely on a single measurement source. Each method illuminates something the others can not. Relying on just one, as most approaches do, creates blind spots that show up late in the program, when they are most expensive to fix. This is also exactly what we explored in Simulation vs reality in vehicle engineering: the strongest engineering decisions come from combining physical and virtual worlds intelligently, not from choosing one over the other.
Our approach: from physics understanding to a predictive model
Rather than fitting each dataset independently, we took a different path.
We used all available measurement means to understand the physics of wet conditions first: how contact patch length evolves with speed and water height, and how that reduction cascades into stiffnesses, peak forces, and aligning moment behavior. That physical understanding became the foundation of the model, not just a calibration target.
On that basis, we built a physics-based predictive model. The key idea: instead of running full wet test campaigns for every tire, the model estimates tire behavior on wet roads from information that already exists on the tire design side:
- The dry-ground Pacejka dataset (existing .tir file)
- Tire profile dimensions and shape
- Tread-band parameters: materials, tread pattern, relevant geometric features
The model introduces an effective contact patch length as the central pivot variable, decreasing with speed and water height, to re-scale stiffnesses, force levels, and trail consistently. All functions are smooth and monotonic, dictated by physics. Continuity back to dry conditions is guaranteed: when water height approaches zero and speed is low, the model returns seamlessly to standard MF-Tyre behavior.
Validation, which is still ongoing, combines objective comparisons on representative maneuvers (constant-radius tests, braking at different water heights, combined slip scenarios...) and Driver-in-the-Loop evaluation to confirm grip/slip transition progressivity and coherent directional feel. Objective and subjective (both matter).
What this means for your programs, before talking about files and formats
Before getting into the technical delivery, here is what this approach actually changes for chassis teams:
You can run wet-road simulation earlier. No more waiting for a full wet test campaign before you can start exploring wet behavior in your models. From your existing dry .tir dataset and tire design parameters, you get a physics-grounded wet dataset ready to use.
You reduce cost and lead time significantly. Wet road test campaigns are complex, expensive, and time-consuming. This approach does not eliminate physical testing, it uses it smarter, combining sources to build a model that covers what no single campaign could.
You work within your existing tools, in Real-Time whenever possible. The dataset integrates via the Standard Tire Interface (STI), already supported by major vehicle dynamics platforms (ADAMS, CarSim, VI-CRT, IPG, MathWorks). No new workflow. No integration effort. Plug-and-play.
The solution is real-time compatible, including for Driver-in-the-Loop (DIL) and Hardware-in-the-Loop (HIL) applications. No numerical discontinuities. Leasable on the short or long-term , licensed via a dongle, a cloud service or a MAC address.
Use Cases
At its current stage, the wet handling tire model is not intended to discriminate between tire designs and should therefore not be considered a tire design tool…at least not yet.
Its primary value lies in reproducing realistic wet handling behavior for vehicle development and simulation activities, including:
- Driver training
- Chassis setup and tuning
- ADAS calibration and stability assessment
- Scenario de-risking and validation
This model is best suited for applications where realistic wet-handling characteristics are required to support vehicle dynamics studies, control-system development, and driver-in-the-loop evaluations.
The bottom line
Chassis development timelines are tighter. Validation requirements are broader. Active systems, ADAS, stability control, brake assist, increasingly need to be tuned and validated beyond dry conditions. Wet-road simulation is no longer optional.
Until now, the choice was between avoiding wet simulation altogether or investing in campaigns that were too costly, too slow, and still incomplete. Neither was a real answer.
This approach changes that. Physics-based, multi-source, validated objectively and in DIL and built on decades of Michelin tire research that no one else brings to the table.
