Session: 07-13-01 Optimization, Uncertainty and Probability I

Paper Number: 77207

Start Time: Tuesday, 04:30 PM

77207 - Multi-Objective Design Optimization of a Passenger Quarter Car With Gradient-Based and Gradient-Free Algorithms 

The design of the suspension system from a vehicle dynamics point of view is very crucial, yet difficult. With many fluctuating parameters, complex objectives, and stochastic disturbances to consider, obtaining an optimum design is challenging and often application-specific. In this work, we aim to find the best optimization technique to obtain the design parameters for the suspension system of a passenger quarter car.

One of the primary functions of the suspension system of a vehicle is to reduce the vibrations transmitted from the road onto the body of the vehicle and hence the passenger which is a direct measure of ride comfort. Also, the suspension system must ensure that the tire is in contact with the road at all times whilst traveling on a range of terrains, which is quantitatively measured as road holding. These two parameters – ride comfort and road holding are conflicting in nature and make it even more challenging to find a sense of balance. Additionally, when designing for passenger cars, ISO standards recommend limits on parameters like suspension travel, tire deflection, ride frequency, and jerk. These constraints are mainly governed by the threshold after which the passenger will feel discomfort. We aim on optimizing both the ride comfort and the road holding of a vehicle simultaneously while considering the above constraints. The sprung and unsprung masses have been pre-set in this study. The design variables are the stiffness and damping coefficients of the sprung and the unsprung masses.

A quarter car model with a tire, suspension system, and passenger seat has been used to understand the car dynamics from the suspension perspective. The quarter car is to be modeled using state-space formulation to calculate all the above-mentioned parameters. The variation of the dynamic load because of the vehicle traveling on a bump is used as the criterion to measure the performance of the suspension system.

A great emphasis has been given to the optimization techniques used. Three major techniques, sequential quadratic programming (SQP), pattern search algorithm (PSA), and genetic algorithm (GA) have been implemented to obtain the optimum solution. These gradient-based and gradient-free methods have been compared to understand the advantages and challenges faced. Possible workarounds are assessed to ensure the algorithm converges to the best possible solution.

There are three problem statements solved. First, single-objective problems with ride comfort and road holding are solved individually taking the respective objective function. Further, the multi-objective problem with road holding and ride comfort as conflicting objective functions are also explored.  Additionally, the procedure of converting a single objective problem to a multi-objective problem is examined. We will define a composite objective function that is developed as a weightage combination of the two design objectives. The Pareto optimal solutions are to be obtained to do a trade-off analysis. Lastly, we aim to compare three solutions and rank the algorithms based on the results.

Future work focuses on reliability-based design optimization to consider the effects of the uncertainties in tire deflection on the design constraints and optimum solutions.

Presenting Author: Varsha S Swamy Virginia Tech

Authors:

Varsha S Swamy Virginia Tech
Yashasvi Achanta Virginia Tech
Pinar Acar Virginia Tech