scholarly journals Minimum length scale control in a NURBS-based SIMP method

2019 ◽  
Vol 354 ◽  
pp. 963-989 ◽  
Author(s):  
Giulio Costa ◽  
Marco Montemurro ◽  
Jérôme Pailhès
2020 ◽  
Vol 143 (2) ◽  
Author(s):  
Chuan Luo ◽  
James K. Guest

Abstract Discrete material optimization (DMO) has proven to be an effective framework for optimizing the orientation of orthotropic laminate composite panels across a structural design domain. The typical design problem is one of maximizing stiffness by assigning a fiber orientation to each subdomain, where the orientation must be selected from a set of discrete magnitudes (e.g., 0 deg, ±45 deg, 90 deg). The DMO approach converts this discrete problem into a continuous formulation where a design variable is introduced for each candidate orientation. Local constraints and penalization are then used to ensure that each subdomain is assigned a single orientation in the final solution. The subdomain over which orientation is constant is most simply defined as a finite element, ultimately leading to complex orientation layouts that may be difficult to manufacture. Recent literature has introduced threshold projections commonly used in density-based topology optimization into the DMO approach in order to influence the manufacturability of solutions. This work takes this idea one step further and utilizes the Heaviside projection method within DMO to provide formal control over the minimum length scale of structural features, holes, and patches of uniform orientation. The proposed approach is demonstrated on benchmark maximum stiffness design problems, and numerical results are near discrete with strict length scale control, providing a direct avenue to controlling the complexity of orientation layouts. This ultimately suggests that projection-based methods can play an important role in controlling the manufacturability of optimized material orientations.


Author(s):  
Chuan Luo ◽  
James K. Guest

Abstract Discrete Material Optimization (DMO) has proven to be an effective framework for optimizing the orientation of orthotropic laminate composite panels across a structural design domain. The typical design problem is one of maximizing stiffness by assigning a fiber orientation to each subdomain, where the orientation must be selected from a set of discrete magnitudes (e.g., 0°, ±45°, 90°). The DMO approach converts this discrete problem into a continuous formulation where a design variable is introduced for each candidate orientation. Local constraints and SIMP style penalization are then used to ensure each subdomain is assigned a single orientation in the final solution. The subdomain over which orientation is constant is typically defined as a finite element, ultimately leading to complex orientation layouts that may be difficult to manufacture. Recent literature has introduced threshold projections, originally developed for density-based topology optimization, into the DMO approach in order to influence the manufacturability of solutions. This work takes this idea one step further and utilizes the Heaviside Projection Method within DMO to provide formal control over the minimum length scale of structural features, holes, and patches of uniform orientation. The proposed approach is demonstrated on benchmark maximum stiffness design problems in terms of objective function, solution discreteness, and manufacturability. Numerical results suggest that projection-based methods can play an important role in controlling the manufacturability of optimized material distributions in optimized design and that solutions are near-discrete with performance properties comparable to designs without manufacturing considerations.


2017 ◽  
Vol 145 (5) ◽  
pp. 1659-1678 ◽  
Author(s):  
Michael Battalio ◽  
Jamie Dyer

Abstract The minimum length scale to investigate quasigeostrophic (QG) vertical motion within a mesoscale operational model is determined using simulations of 28 baroclinic systems from the North American Mesoscale Forecast System (NAM) model. Two upscaling methods are tested to find the optimal QG characteristic length. The box method takes an average of each field before performing finite-differencing calculations. The cross method samples the data at increasing distances between finite-difference calculations. The traditional QG omega equation is evaluated with each upscaling technique and found to be reliable between 800 and 200 hPa. The minimum QG length scale is found to be L = 140 km considering correlations of QG omega back to operational model values, which are for both methods on an “extended” QG omega. The box method performs marginally better than the cross method due to a larger reduction of QG forcing in higher-order wavenumbers, but at the appropriate length scale, both methods have indistinguishable correlations.


1998 ◽  
Vol 13 (12) ◽  
pp. 961-971 ◽  
Author(s):  
D. H. COULE

In a recent paper1 Euclidean wormhole solution has been obtained with a vacuum Brans–Dicke theory with parameter ω=0. These wormholes suffer from unphysical values of the gravitational constant. One can relate the various known wormholes by means of conformal transformations; although one should not transform them directly to the Einstein frame as the gravitational constant there is "forced" physical: so removing the wormholes. However, by arguing for the existence of a fundamental minimum length scale such wormholes can now be considered as representative of quantum gravitational phenomena. One can also obtain wormholes as solutions of the quantum mechanical Wheeler–De Witt equation; now in Brans–Dicke for any ω>-3/2.


Sign in / Sign up

Export Citation Format

Share Document