Research On Modular Methodology And Multi-objective Optimization Of Automobile Passive Safety

According to the statistics provided by WHO in2004, people killed by trafficaccidents was up to1.2million, and the traffic accidents would be the thirdkiller to human being after ischemic heart disease and depression in2020.More than19million autos have been produced and sold in China and rankedfirst in the world in four consecutive years released by Association ofAutomobile Manufacturers which inducing a severe road traffic safetysituation. In2010, caused by the traffic accidents in China,65225people wereinjured and178people lost their lives in daily average. Chinese governmentnow is paying great attension to road traffic safety and enacting stricterregulations and requiring automobile companies to develop more secureproducts.New active safety product is developed by automobile manufactureenterprises and research institutions, as well as the passive safety. Generally,the crashworthiness of frontal, side and rear structure should be analyzedduring the product development in order to improve passive safety, and theanalysis goes with the whole product development. Once the design changedand structures optimized, the crashworthiness will be validated again to ensurethe crashworthiness stability. Therefore, there are several conditions andoptimization models should be rebuilt which resulted in huge modelingworkload. If using traditional modeling method, the modeling will occupy80%workload in analysis simulation. Improving modeling efficiency hasbecome a primary and importanr task to reduce repetitive works in analysissimulation.In this paper, modular strategy was introduced in the crash simulation modeling, and modular methodology was proposed. Latin hypercubeexperimental design, orthogonal design, polynomial response surface (PRS)and Kriging approximation were used in project analysis. Body structurescrashworthiness was improved by multi-objective optimization with thethird-order PRS, Kriging and adaptive simulated annealing (ASA). The maincontents in this thesis are showed as follows:Modular methodology was proposed in order to reduce repetitive works inmodeling and model updating. The vehicle was separated into severalindependent modules based on the principles of independent model, minimumassociation, function, and position. The principles were described as follows:1) to seperate the structures according to the principles of structuralsimplifying and element quality;2) to simulate the physical or chemicalconnection with correct element, number range for the node and element inthe modules, using the Rbody element to connect the module to body in white.Engine hatch is an example to show the modeling process.According to the collision vehicle configuration using modular methodology,3crash-condition models were built, including frontal impact, side impact,and rear impact model. The vehicle configuration is1.6L engine, MQ200manual transmission,16-inch tires and body without skylight. The moduleaccuracy was validated through comparison the acceleration, intrusion,intrusion velocity of simulation with test results. The accuracy reachedengineering requirements; the modular methodology and simulated methodare all validated as well. Based on the validated model, the body structureshould be optimized.Several configuration cars have been developed to meet customer for theproduct diversity. In generally, the least safe car should be selected as a basemodel to optimize. Based on the modular methodology, the frontal impactmodels with AQ250and MQ200transmission were built in order to assess theinfluence of crashworthiness on different modular. Comparing with the testand simulation results, the frontal crashworthiness of AQ250model is worsethan MQ200model, because the dimension and weight of AQ250transmission are larger and heavier than MQ200, which result in moreintrusion of firewall and more peak value of acceleration. The side impact results for the body with and without skylight were compared. The body sidecrashworthiness of the model with skylight is better than the model withoutskylight, because there is several reinforcing plate around the skylight.The model of1.6L engine, MQ200manual transmission,16-inch tires andbody without skylight was validated in chapter4. The crashworthiness ofbody side structures did not reach the engineering requirement. In order tocreat the third-order PRS, the part thickness and material was selected as adesign variable, the intrusion and intrusion velocity of B-pillar was selected asan objective function,200samples were created using Latin hypercubeexperimental design. Based on the limited number of design variablesengaged in the optimization, it is sufficient to obtain a good pareto frontdescriptive function using200input virtual data points and the ASA.Considering the minimum intrusion and velocity, a multi-objectiveoptimization scheme is the optimized point223. Comparing with the originalscheme, the intrusion in b4decreased62%, intrusion velocity in b3decreased29%. Although the mass increased7.7%, the side crashworthiness is improvedobviously. The scheme The results revealed that it is a good method toincrease the crashworthiness in the area of B-pillar and threshold used thehigh strength steel and increasing the thicknessIn order to get the more design environment to match the occupant restraintsystem and reduce the vehicle acceleration, the energy-box and frontallongitudinal beam as design variables were improved by multi-objectiveoptimization. The18test samples were built with the orthogonal design. Thefunction of the variables and response was simulated by Krigingapproximation, and to optimize with adaptive simulated annealing (ASA).Base on the better crashworthiness, the lower vehicle acceleration wasobtained at the cost of the higher intrusion and model quality.