New Virtual Chassis Method Targets Safer Car Handling Before a Single Prototype Is Built

Chassis engineers have long faced an uncomfortable truth: the way a car steers, grips and recovers at the edge of its physical limits cannot be fully understood until a real prototype is pushed hard on a test track. A new open-access study published in Automotive and Engine Technology by researchers at the FKFS Research Institute for Automotive Engineering and Powertrain Systems, the University of Stuttgart’s Institute of Automotive Engineering and Volvo Cars now offers a way to quantify those elusive handling qualities much earlier, entirely in the virtual world, and far faster than conventional simulation allows.

The research, led by Pascal Ambrosoli with co-authors Carl Sandberg, Werner Krantz, Jens Neubeck and Andreas Wagner, addresses a central problem in modern vehicle development. As automakers shift toward systems engineering approaches built around the well-known V-model, the earliest phases of a program require target values for overall vehicle characteristics to be fixed at the top of the V, long before hardware exists. Those targets must then be cascaded step by step down to subsystem level, so that suspension kinematics, tire selection, steering tuning and electronic controls can be designed in a decoupled way while still guaranteeing the behavior of the finished car. For this cascade to work, the top-level characteristics must be measurable in a precise sense: they must be physically grounded quantities that a virtual vehicle model can actually compute.

The team focuses on lateral dynamics, the domain that governs cornering behavior and, crucially, driving safety. Within this domain they define a family of objective metrics covering both controllability and stability. These include the steering wheel gradient, which describes how much steering angle is needed per unit of lateral acceleration in the linear range from zero to four meters per second squared; the steady-state and dynamic lateral acceleration peaks, which mark how much cornering power the vehicle can generate; the yaw acceleration peak, which captures how quickly the car can be rotated into a corner; and stability and controllability measures evaluated both in straight-ahead driving and near the limit area, defined as eighty percent of the steady-state lateral acceleration peak. Side-slip angles at the center of gravity and at the rear axle, along with slip angles at the front and rear axles, form the physical backbone of these quantities.

Why does the nonlinear part of lateral dynamics deserve special attention? Because that is where accidents happen. In the linear regime, a car responds predictably: double the steering input and the lateral response roughly doubles. But as tires approach the limit of their grip, their force potential saturates and rolls off, and the vehicle’s behavior becomes strongly nonlinear. Small changes in design or tuning can produce disproportionate shifts in how the car behaves when a driver demands maximum cornering. A chassis development process that only characterizes the benign, linear middle of the handling envelope leaves exactly the region that matters most for safety unmeasured, and therefore unmanaged.

The conventional way to extract such metrics is time-domain simulation: excite a full vehicle model with steering inputs, sweep the maneuvers, and post-process the resulting trajectories. The problem is computational cost. A target-oriented development process built on the solution space method is inherently iterative. Design variables are varied, targets are checked, the design is adjusted, and the cycle repeats, often thousands of times across many subsystems. If every evaluation of a stability or controllability metric requires a lengthy transient simulation, the iteration grinds to a halt. The Stuttgart team’s answer is to stop simulating time altogether for these evaluations and instead compute the metrics from partial equilibria of the vehicle’s equations of motion.

At the heart of the methodology lies the Newton-Raphson method, a classical numerical root-finding technique that iteratively drives the residuals of a nonlinear system to zero. Instead of integrating the vehicle model forward in time, the researchers formulate the steady-state and dynamic partial equilibrium conditions of the lateral dynamics and solve them directly. The Jacobian matrix of the model states, which contains the partial derivatives of the state derivatives with respect to the states themselves, guides each Newton step toward the equilibrium point. Inputs such as the steering wheel angle are varied systematically, and at each operating point the equilibrium states, including lateral velocities at the rear axle, side-slip angles and lateral tire forces at the front and rear axles, are recovered analytically rather than simulated. Tire relaxation length, which describes the lag with which a tire builds up lateral force, is retained in the formulation so that dynamic partial equilibria remain physically meaningful.

This equilibrium-based approach makes the defined characteristics computable automatically and efficiently, which is precisely what the iterative solution space method demands. Within a target cascading workflow, each subsystem design proposal can be checked against the top-level handling targets in seconds rather than hours, and the robustness of a design, meaning how much its metrics degrade as parameters vary, becomes visible early. The decoupling that target cascading promises, in which a suspension engineer and a tire engineer can work independently yet converge on a coherent vehicle, only functions if the shared top-level metrics are cheap enough to evaluate on every iteration. The Newton-Raphson methodology supplies that missing computational ingredient.

To demonstrate that the new metrics are not merely convenient but meaningful, the authors compare them against full time-domain simulations of the vehicle. The comparison shows that characteristics derived from steady-state and dynamic partial equilibria capture the relevant features of the vehicle’s lateral behavior, including the location of the lateral acceleration peak and the yaw acceleration response near that peak. The metrics thus serve as faithful proxies for quantities that would otherwise require expensive transient maneuvers, while remaining stable and well-defined across the design space, a property that time-domain peak values, which can jump discontinuously as designs change, often lack. The work also connects to established tools of the vehicle dynamics trade, including the Milliken moment method, whose diagram conventions for mapping stability and controllability the new characteristics are designed to complement, and alternative formulations such as the multibody system transfer matrix method.

The implications reach beyond the specific metrics. Battery electric vehicles, with their heavy floor-mounted packs, low centers of gravity and high curb masses, are reshaping the chassis design space, and development cycles are compressing as competition intensifies. A characteristics-based, V-model-oriented process in which handling targets are defined, monitored and tracked from the first concept sketch to production validation offers a way to keep safety-critical qualities under control amid that turbulence. Because the metrics are physically based, they can be carried consistently from early idealized models through to detailed multibody simulations and, ultimately, to test-track confirmation, giving program managers a single quantitative thread through the entire development process.

The study, received in March 2026 and accepted in August 2026, was published open access on 18 September 2026 with funding enabled by Projekt DEAL. Its authors report no conflict of interest. For an industry in which a single physical prototype iteration can cost millions and months, the ability to determine controllability and stability metrics automatically, efficiently and reliably in the virtual domain marks a practical step toward chassis development that is simultaneously faster, more systematic and more safety-aware. The nonlinear edge of the handling envelope, long the blind spot of early-phase engineering, is now a quantity that can be targeted, cascaded and verified before rubber ever meets asphalt.

Subject of Research: Efficient determination of vehicle lateral controllability and stability metrics for target-oriented virtual chassis development

Article Title: Determination of vehicle controllability and stability metrics in a target-oriented virtual chassis development process

Article References: Ambrosoli, P., Sandberg, C., Krantz, W., Neubeck, J., & Wagner, A. (2026). Determination of vehicle controllability and stability metrics in a target-oriented virtual chassis development process. Automotive and Engine Technology, 11(1), Article 15. https://doi.org/10.1007/s41104-026-00179-9

Image Credits: AI Generated

DOI: 10.1007/s41104-026-00179-9

Keywords: vehicle dynamics, chassis development, lateral dynamics, Newton-Raphson method, solution space method, V-model, driving safety, steering wheel gradient, yaw acceleration, virtual vehicle development, automotive engineering, target cascading