Co-authors​ :

 Arne WULFF , Boyang CHEN (NETHERLANDS), Matthew STEINBERG (NETHERLANDS), Yinglu TANG , Matthias MOLLER (NETHERLANDS), Sebastian FELD  

Composite laminates offer a wide range of design possibilities: each ply can be oriented differently so as to tailor the resulting laminates for desired stiffness and strength along the critical loading directions. This perceived design advantage also comes with a price - the number of possible stacking sequences grows exponentially as we increase the number of plies in the stack. The practical design of such laminated structures often take a bi-level approach [1]: firstly the homogenized continuous properties, namely the lamination parameters, are obtained from a gradient-based approach; secondly, a manufacturable stacking sequence which produces the optimal lamination parameters is retrieved with a suitable combinatorial optimization algorithm. The second step often relies on heuristic algorithms such as Genetic Algorithms in the literature.
In this work, we focus our attention on the second step of the bi-level optimization approach, namely the stacking sequence retrieval problem for a given set of target lamination parameters. We define this problem as a quantum mechanics problem where the Hamiltonian derives from the loss function of the optimization and the ground-state encodes the stacking sequence we seek. We demonstrate here the incorporation of Density Matrix Renormalization Group algorithm, a tensor network method for finding the ground state of quantum many-body systems [2]. The stacking sequence for a laminate of 200 plies is successfully retrieved. Our results open up a new avenue of exploration – the use of quantum(-inspired) algorithms for the design optimization of laminated composite structures.