Skip to main content
eScholarship
Open Access Publications from the University of California

Scholarly Works (6 results)

  • Thesis
  • Peer Reviewed
UC Berkeley Electronic Theses and Dissertations (2022)

Fluid flows are non-intuitive. Even with years of experience, non-intuitive behavior of fluids can mean the optimal geometry of fluid machinery is surprising or even extreme (consider, for instance, the bulbous bow of a ship). Finding the optimal design of a hydrodynamic or aerodynamic surfaces is often impossible due to the expense of evaluating the cost functions (say, with computational fluid dynamics) needed to determine the performances of the flows that the surface controls. In addition, inherent limitations of the design space itself due to imposed geometric constraints, conventional parameterization methods, and user bias can restrict all of the designs within a chosen design space regardless of whether traditional optimization methods or newer, data-driven design algorithms with machine learning are used to search the design space.

This dissertation presents two methodologies to address these difficulties: (1) Design-by-Morphing (DbM), a novel strategy for creating a design search space by morphing homeomorphic shapes to create a continuous and constraint-free design search space that can produce radical extrapolated shapes, something which is unique from existing design strategies; and (2) an optimization algorithm to search that space that uses a novel Mixed-variable, Multi-Objective Bayesian Optimization that we call MixMOBO, that can optimize such expensive, black-box problems with minimum number of functions calls. We apply these methodologues for optimization of several problems and present shape optimization of airfoils, draft-tubes for hydrokinetic turbines, and architected meta-materials. In all cases, we show significantly improved and radical designs.

Chapter One of this thesis focuses on the details of the MixMOBO algorithm, the first mixed-variable, multi-objective Bayesian optimization algorithm. MixMOBO outperforms existing algorithms for mixed-variable problems. It details HedgeMO strategy for hedging acquisition function portfolios for multi-objective problems. MixMOBO is then applied for optimization of strain energy density of an architected meta-material structure with categorical variables. From a design space of 8.5 billion possible candidates, our algorithm is able to optimize the design space with only 250 function evaluation and achieve 10^4 times improvement in strain energy density over existing structures [1,2]. Chapter Two focuses on applying MixMOBO for design of Cauchy-Symmetric architected meta-material structures. With only 69 function calls, MixMOBO is able to find such a structure from a design space of 10^7 possible structures. Chapter Three demonstrates the use of Design-by-Morphing for optimization of airfoils. We show that with just 25 baseline shapes, we are able to reproduce the UIUC airfoil database with high fidelity and optimize this space to create aerodynamically superior and safer airfoils [3]. Chapter Four focuses on application of design of a draft tube for a hydrokinetic turbine to maximize pressure recovery at the exit of the turbine [4].

    Cover page: Mixed-Variable Multi-Objective Bayesian Optimization, Design-by-Morphing and their Applications
    • Article
    • Peer Reviewed
    UC Berkeley Previously Published Works (2023)
    Design-by-Morphing (DbM) is a novel design methodology that creates a search space for topology optimization. Traditional design techniques often impose geometric constraints and, sometimes, the designer’s biases on the design space, which restricts the novelty of the designs and allows for only small local changes. On the contrary, we show in this paper that DbM does not impose such restrictions on the design space, thus allowing for a radical and expansive search space with only a few design parameters. We compare DbM with other methods in the case of design space generation for 2D airfoils and find that DbM can reconstruct the entire UIUC database with >99.5% accuracy. Furthermore, using a bi-objective genetic algorithm, we optimize the airfoil designs created by DbM to maximize both the lift-over-drag ratio, CLDmax, and stall angle tolerance, ∆α, which results in a Pareto-front of innovative airfoils that exhibit substantial improvements in both objectives.
      Creative Commons 'BY' version 4.0 license
      • Article
      UC Berkeley Previously Published Works (2022)
      We present Design-by-Morphing (DbM), a novel design methodology applicable to creating a search space for topology optimization of 2D airfoils. Most design techniques impose geometric constraints and sometimes designers' bias on the design space itself, thus restricting the novelty of the designs created, and only allowing for small local changes. We show that DbM methodology does not impose any such restrictions on the design space and allows for extrapolation from the search space, thus granting truly radical and large search space with a few design parameters. In comparison to other shape design methodologies, we apply DbM to create a search space for 2D airfoils. We optimize this airfoil shape design space for maximizing the lift-over-drag ratio, $CLD_{max}$, and stall angle tolerance, $\Delta \alpha$. Using a bi-objective genetic algorithm to optimize the DbM space, it is found that we create a Pareto-front of radical airfoils exhibiting remarkable properties for both objectives.
        Creative Commons 'BY-NC' version 4.0 license
        • Article
        • Peer Reviewed
        UC Berkeley Previously Published Works (2024)
        Abstract: A computational approach is presented for optimizing new riblet surface designs in turbulent channel flow for drag reduction, utilizing Design-by-Morphing (DbM), Large Eddy Simulation (LES), and Bayesian Optimization (BO). The design space is generated using DbM to include a variety of novel riblet surface designs, which are then evaluated using LES to determine their drag-reducing capabilities. The riblet surface geometry and configuration are optimized for maximum drag reduction using the mixed-variable Bayesian optimization (MixMOBO) algorithm. A total of 125 optimization epochs are carried out, resulting in the identification of 3 optimal riblet surface designs that are comparable to or better than the reference drag reduction rate of 8 %. The Bayesian-optimized designs commonly suggest riblet sizes of around 15 wall units, relatively large spacing compared to conventional designs, and spiky tips with notches for the riblets. Our overall optimization process is conducted within a reasonable physical time frame with up to 12-core parallel computing and can be practical for fluid engineering optimization problems that require high-fidelity of computational design before materialization.
          Cover page: Bayesian-Optimized Riblet Surface Design for Turbulent Drag Reduction via Design-by-Morphing with Large Eddy SimulationCreative Commons 'BY-NC-ND' version 4.0 license
          • Article
          • Peer Reviewed
          UC Berkeley Previously Published Works (2021)
          We use a previously unexplored Bayesian optimization framework, “evolutionary Monte Carlo sampling,” to systematically design the arrangement of defects in an architected microlattice to maximize its strain energy density before undergoing catastrophic failure. Our algorithm searches a design space with billions of 4 × 4 × 5 3D lattices, yet it finds the global optimum with only 250 cost function evaluations. Our optimum has a normalized strain energy density 12,464 times greater than its commonly studied defect-free counterpart. Traditional optimization is inefficient for this microlattice because (i) the design space has discrete, qualitative parameter states as input variables, (ii) the cost function is computationally expensive, and (iii) the design space is large. Our proposed framework is useful for architected materials and for many optimization problems in science and elucidates how defects can enhance the mechanical performance of architected materials.
            Cover page: Strength through defects: A novel Bayesian approach for the optimization of architected materials
            • Article
            • Peer Reviewed
            UC Berkeley Previously Published Works (2022)
            Using a new Bayesian Optimization algorithm to guide the design of mechanical metamaterials, we design nonhomogeneous 3D structures possessing the Cauchy symmetry, which dictates the relationship between continuum and atomic deformations. Recent efforts to merge optimization techniques with the design of mechanical metamaterials has resulted in a concentrated effort to tailor their elastic and post elastic properties. Even though these properties of either individual unit cells or homogenized continua can be simulated using multi-physics solvers and well established optimization schemes, they are often computationally expensive and require many design iterations, rendering the validation stage a significant obstacle in the design of new metamaterial designs. This study aims to provide a framework on how to utilize miniscule computational cost to control the elastic properties of metamaterials such that specific symmetries can be accomplished. Using the Cauchy symmetry as a design objective, we engineer structures through the strategic arrangement of 5 different unit cells in a 5×5×5 cubic symmetric microlattice structure. This lattice design, despite constituting a design space with 510 3D lattice configurations, can converge to an effective solution in only 69 function calls as a result of the efficiency of the new Bayesian optimization scheme. To validate the mechanical behavior of the design, the lattice structures were fabricated using multiphoton lithography and mechanically tested, revealing a close correlation between experiments and simulated results in the elastic regime. Ultimately, a similar methodology can be utilized to design metamaterials with other material properties, aspiring to control properties at different length scales, an endeavor that requires inordinate computation cost.
              Cover page: Systematic design of Cauchy symmetric structures through Bayesian optimizationCreative Commons 'BY' version 4.0 license
              Top