Search

This work includes a complete optimization, design, sizing and validation application of density-based topology optimization for a single stiffened panel from the upper skin of a commercial aircraft wing. In this work, topology optimization is carried with Optistruct from Altair. The initial values of the density variables are shown to have the biggest impact on the optimized material distribution. This effect is used in a random restart approach to explore multiple local minima, and thus to find different material distributions. Different material distribution obtained with topology optimization have shown the same buckling resistance within the implicit model used for optimization. However, the stiffened panel models created from the different material distributions show very different strength when they are sized with a subsequent optimization and validated using quasi-static non-linear analysis. Besides, the optimization results are almost all simple orthogrid solutions. Finding these layouts with topology optimization has proven to be difficult and highlights the fact that the technology is not mature for the design of new layouts for stiffened panels.

Machine learning-accelerated topology optimization faces the challenge of generating large amounts of optimal topologies for supervised learning on a training dataset. This data generation step is challenging for the method of Moving Morphable Components (MMC) which presents an oscillatory behavior near local optimum that negatively affects its convergence speed and therefore present a high data generation cost. This paper presents a comparative study of the most-used first-order optimizers applied to minimum compliance problems under the MMC framework for topology optimization. The Method of Moving Asymptotes (MMA), its Globally Convergent version (GCMMA) and the hybrid MMA-GCMMA optimizer are compared using their final compliance and the total number of iterations until convergence as performance metrics. An extensive set of diversified boundary conditions for a rectangular beam minimum compliance problem is used as case study. The method of performance profiles is utilized to provide for each solver a probability distribution of outperforming its counterparts. Numerical results show that using a hybrid optimizer can accelerate the convergence speed of the MMC framework while still producing equally compliant topologies when compared to the MMA optimizer.

The development of unmanned aerial manipulators (UAMs) allows a novel class of flying robots to carry out a wide variety of tasks in difficult environments due to their versatility and autonomy. However, the different tasks that can be carried out might call for different control strategies. To this end, one needs to categorize the possible tasks accomplishable by UAMs. This paper proposes a novel taxonomy, which is the result of a video information acquisition methodology combined with a review of research works in the literature. The different elements of the taxonomy are separated using a higher level of abstraction in a way that the general description of the tasks are considered and not its operational details. To illustrate the fact that algorithms must adapt to different tasks, a description of the usual UAM architecture is carried out. Four categories of criteria are used in the taxonomy to differentiate all possible tasks. These categories are the interaction type, the actual task definition, the environment condition and the time sensitivity of the task. This taxonomy forms the basis for possible machine-learning-based task classifiers that could be used in autonomous UAMs control and mission planning. Multiple tasks defined in the taxonomy can be combined to accomplish complex missions.

Reducing the weight of structures remains a major challenge in the aviation industry in order to reduce fuel consumption. The stiffened panel is the main assembly method for primary structures in aircraft, e.g. fuselage or wing. Density-based topology optimization has been used in research and in industry as a tool to help create new stiffening patterns for aircraft components, such as ribs, spars, bulkheads or even floor design. One critical aspect of stiffened panel design for wing structures is the buckling resistance. However, most work found in the literature does not include buckling analysis during optimization which leads to sub-optimal results when the stiffening layout is validated for buckling. Including buckling as a constraint for the density-based topology optimization has proven to be a complex task, mainly caused by the fact that the buckling of the stiffeners is not captured accurately. As such, this work presents an optimization method for stiffened panels based on the ground structure approach usually used for truss topology optimization. The main novelty of the method is the use of a stiffener activation variable (SAV) to activate only one variable at a time, either the height or density variable associated with each stiffeners of the ground structure. This work shows that while ground structure topology optimization requires more setup time and limiting the degrees of freedom of the optimization, it finds the best stiffening layout efficiently when compared to the density method.

Topology optimization is a tool that supports the creativity of structural-designers and is used in various industries, from automotive to aeronautics, to reduce design iterations towards obtaining the optimal structure layout. However, this tool requires both time and experience to interpret the results into manufacturable and reliable structure layout. To improve this aspect, an interpretation support tool is being developed by our research team based on industrial knowledge and axiomatic design principles. This design tool will be very useful for aircraft structure development, for instance, as it aims to help the structural designer in the conception of stiffened panels. The tool has been divided into three modules: feature recognition, feature analysis and design support. This paper presents the first of the three modules that identifies the trusses of the optimized topology in terms of truss recognition algorithms. The purpose of the truss recognition algorithm is to translate the densities of the element of the optimized topology (low-level abstraction) to a skeletal structure (high-level abstraction) that contains nodes and branches that describe the same topology as the optimized topology. It should also ensure that the structural skeleton retains connectivity with loads and boundary conditions. The information may then be used by the design support tool for analysis, comparison, decision-making, design and optimization purposes. To do so, a novel image-based method using a binary skeleton is proposed. For this work, we identified multiple limitations existing in similar solutions and we mitigated them. Therefore, a new skeletonization method is proposed, which is specifically designed for truss recognition in the optimized topology. The capabilities of the skeletonization method are demonstrated by comparing it with existing methods, and the truss recognition algorithm is used with a test case exhibiting the algorithm’s capabilities on an airplane wing box rib.