-
A Method of Sequential Log-Convex Programming for Engineering Design
Authors:
Cody Karcher,
Robert Haimes
Abstract:
A method of Sequential Log-Convex Programming (SLCP) is constructed that exploits the log-convex structure present in many engineering design problems. The mathematical structure of Geometric Programming (GP) is combined with the ability of Sequential Quadratic Program (SQP) to accommodate a wide range of objective and constraint functions, resulting in a practical algorithm that can be adopted wi…
▽ More
A method of Sequential Log-Convex Programming (SLCP) is constructed that exploits the log-convex structure present in many engineering design problems. The mathematical structure of Geometric Programming (GP) is combined with the ability of Sequential Quadratic Program (SQP) to accommodate a wide range of objective and constraint functions, resulting in a practical algorithm that can be adopted with little to no modification of existing design practices. Three test problems are considered to demonstrate the SLCP algorithm, comparing it with SQP and the modified Logspace Sequential Quadratic Programming (LSQP). In these cases, SLCP shows up to a 77% reduction in number of iterations compared to SQP, and an 11% reduction compared to LSQP. The airfoil analysis code XFOIL is integrated into one of the case studies to show how SLCP can be used to evolve the fidelity of design problems that have initially been modeled as GP compatible. Finally, a methodology for design based on GP and SLCP is briefly discussed.
△ Less
Submitted 20 January, 2022;
originally announced January 2022.
-
Data Fitting with Signomial Programming Compatible Difference of Convex Functions
Authors:
Cody Karcher
Abstract:
Signomial Programming (SP) has proven to be a powerful tool for engineering design optimization, striking a balance between the computational efficiency of Geometric Programming (GP) and the extensibility of more general optimization methods like Sequential Quadratic Programming (SQP). But when an existing engineering analysis tool is incompatible with the mathematics of the SP formulation, option…
▽ More
Signomial Programming (SP) has proven to be a powerful tool for engineering design optimization, striking a balance between the computational efficiency of Geometric Programming (GP) and the extensibility of more general optimization methods like Sequential Quadratic Programming (SQP). But when an existing engineering analysis tool is incompatible with the mathematics of the SP formulation, options are limited. Previous literature has suggested schemes for fitting GP compatible models to pre-computed data, but no methods have yet been proposed that take advantage of the increased modeling flexibility available in SP. This paper describes a new Soft Difference of Max Affine (SDMA) function class that is constructed from existing methods of GP compatible fitting and the theory of Difference of Convex (DC) functions. When a SDMA function is fit to data in log-log transformed space, it becomes either a signomial or a set of signomials upon inverse transformation. Three examples of fitting are presented here, including simple test cases in 2D and 3D, and a fit to the performance data of the NACA 24xx family of airfoils. In each case, RMS error is driven to less than 1%.
△ Less
Submitted 22 October, 2021;
originally announced October 2021.
-
Logspace Sequential Quadratic Programming for Design Optimization
Authors:
Cody Karcher
Abstract:
A novel approach to exploiting the log-convex structure present in many design problems is developed by modifying the classical Sequential Quadratic Programming (SQP) algorithm. The modified algorithm, Logspace Sequential Quadratic Programming (LSQP), inherits some of the computational efficiency exhibited by log-convex methods such as Geometric Programing and Signomial Programing, but retains the…
▽ More
A novel approach to exploiting the log-convex structure present in many design problems is developed by modifying the classical Sequential Quadratic Programming (SQP) algorithm. The modified algorithm, Logspace Sequential Quadratic Programming (LSQP), inherits some of the computational efficiency exhibited by log-convex methods such as Geometric Programing and Signomial Programing, but retains the the natural integration of black box analysis methods from SQP. As a result, significant computational savings is achieved without the need to invasively modify existing black box analysis methods prevalent in practical design problems. In the cases considered here, the LSQP algorithm shows a 40-70% reduction in number of iterations compared to SQP.
△ Less
Submitted 21 December, 2021; v1 submitted 30 May, 2021;
originally announced May 2021.
