Modelling and Mass Optimisation of an Aluminium Honeycomb Impact Attenuator
Authors
Sørensen, Jens Bonderup ; Simonsen, Christian Dam ; Broberg, Nikolaj
Term
4. term
Education
Publication year
2020
Submitted on
2020-06-03
Pages
116
Abstract
Dette speciale undersøger, hvordan en stødabsorber (impact attenuator) til AAU Racings Formula Student-racerbil kan modelleres og masseoptimeres ved brug af en aluminiums-honeycombstruktur. Først sammenlignes flere materialer gennem analyser og quasistatiske kompressionstest, hvorefter aluminiumshoneycomb udvælges som det mest egnede. Dynamiske dropforsøg ved ca. 7 m/s viser, at dynamiske effekter er uden væsentlig betydning og derfor kan udelades i modellen. En numerisk model udvikles i LS-DYNA med afsæt i viden om plasticitet og stabilitet samt Wierzbickis hypotese om, at delvist brud af limfuger mellem celler kræver mindre energi end at deformere intakte cellevægge. Den centrale udfordring er at simulere det delvise brud i limsamlingen; kontaktformuleringen i LS-DYNA viser sig ustabil og afprøves derfor på enklere modeller for at forstå virkemåden. Herefter implementeres den på en forsimplet “Y”-sektion af honeycomben, hvilket – efter kalibrering – giver tilfredsstillende overensstemmelse med dropforsøg og datablad, og valideres yderligere mod en model med syv celler. Modellen indikerer desuden, at energien for deformation reduceres markant, når delvist limbrud medtages. På basis af modellen opstilles en metamodel (Response Surface Method) for middelkraften som funktion af geometriske parametre, der anvendes som begrænsning i en masseminimering under Formula Student-krav. Med MATLABs fmincon findes et optimalt design med en teoretisk masse på 1,29 kg, svarende til en reduktion på 62 % (2,13 kg) i forhold til det nuværende design.
This thesis investigates how to model and mass-optimise an impact attenuator for AAU Racing’s Formula Student car using an aluminium honeycomb structure. Several materials are assessed through analyses and quasi-static compression tests, leading to the selection of aluminium honeycomb as the most suitable option. Dynamic drop tests at about 7 m/s show that dynamic effects are insignificant and can be neglected in the model. A numerical model is developed in LS-DYNA based on knowledge of plasticity and stability and Wierzbicki’s hypothesis that partial failure of the adhesive joints between cells requires less energy than deforming intact cell walls. The key modelling challenge is to simulate this partial debonding; the chosen contact formulation in LS-DYNA initially proves unstable and is therefore explored on simpler models to build understanding. It is then implemented on a simplified honeycomb “Y”-section, which—after calibration—matches drop-test data and the manufacturer’s datasheet satisfactorily, and is further compared to a seven-cell model with good agreement. The model also indicates that including partial adhesive failure markedly reduces the energy required for deformation. Based on the model, a metamodel (Response Surface Method) of the mean crushing load as a function of geometric parameters is constructed and used as a constraint in mass minimisation under Formula Student requirements. Using MATLAB’s fmincon, an optimal design with a theoretical mass of 1.29 kg is found, representing a 62% reduction (2.13 kg) compared to the current design.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents