Bi-directional Fixed Grid Evolutionary Structural Optimization Method And Its Engineering Application

Evolutionary structural optimization (ESO) method proposed in the last decadeis promising in engineering shape and topology design. Its further theoreticaldevelopment and introduction into new engineering fields are of apparent significance.This thesis developed a novel bi-directional fixed grid evolutionary structuraloptimization procedure (FG BESO), which efficiently solved the non-smooth zigzagboundary problems associated with most classic topology optimization methods likeESO. The thesis also explored two classes of important practical design problems on(1) the cutouts shape optimization in laminated composite plates and shells, (2) thedesign criteria for the tunnel stability and topology optimization for tunnel support.The major contributions in this study are as follows.1. As a typical geotechnical application, cross-sectional shape optimization ofgravity dam was performed by using nibbling ESO method. The results show thatgravity dam with triangular section has the minimal displacement in the top of damunder basic load including gravity, water pressure and uplift pressure.2. Based on a fixed grid finite element framework, a novel bi-directionalevolutionary optimization technique, namely FG BESO, is presented to generatesmoother boundary in fixed grid shape design. The solution produced by FG BESO isvalidated with analytic result, which demonstrates a higher computational precisionthan traditional ESO method. One of its significant advantages is having capability ofreducing the dependence of the optimal results on initial design, thus provides ameans to obtaining the global optimum.3. A uniform Tsai-Hill index criterion is presented to evaluate stressconcentration around cutout boundary. The FG BESO method is then applied tooptimize the cutouts shape in laminated composite plates and shells. The results showthat load is one of the decisive factors of shape optimization. In addition, thecomposite lay-ups can remarkably affect the optimal design. The study demonstratesthe capability of FG BESO in solving cutout design problems in curved shells.4. Fully reinforced rock is considered as artificial reinforcement material, andtopology optimization of tunnel support is transform to seeking the optimaldistribution of reinforcement material in original rock. The comparison with thehomogenization method proves that FG BESO is effective to the design of tunnelsupports. The optimal shape of tunnel support with the object function of the totalstrain energy approaches ellipse, whose principal axis is the same as that of theprincipal stress. The result indicates that the floor and sidewall corner reinforcementare significant for preventing floor and sidewall heaves, which is in a good agreementwith experimental observation and engineering practice. The effect of some factorssuch as different ground stress on the optimal reinforcement is explored. From thisstudy, it is also found that the optimal reinforcement of large-scale group ofunderground chambers can be quite different from that of simple tunnel. Finally, thethesis presents a new concept of "monitoring function" for dynamic design of tunnelsupports.