Geometric optimization of radiative enclosures through non-linear programming

K. J. Daun*, J. R. Howell, D. P. Morton

*Corresponding author for this work

Research output: Contribution to journalConference articlepeer-review


Optimization routines have become a popular tool in the design of thermal systems. These routines dramatically reduce the time and computational effort needed to find the optimal design by reducing the number of iterations required by the forward design procedure. Also, these methods often find solutions that are not intuitive to the designer. Although these routines have been applied to solve conduction and convection problems, they have not been used to design radiant enclosures. This paper introduces a methodology for applying non-linear programming to design 2-D radiant enclosures. The process is facilitated by representing the enclosure surface parametrically with B-spline curves, and an infinitesimal-area analysis technique is then used to solve the radiosity distribution within the enclosure. The enclosure geometry is repeatedly adjusted with a gradient-based minimization algorithm, until the optimum solution is found. This technique is demonstrated by optimizing the geometry of a 2-D radiant enclosure, with the objective to obtain a specified radiosity distribution over a portion of the enclosure surface. The steepest descent, Newton's method, and quasi-Newton's method are used to find the optimum enclosure geometry, and the performance of these methods is compared.

Original languageEnglish (US)
Pages (from-to)227-236
Number of pages10
JournalAmerican Society of Mechanical Engineers, Heat Transfer Division, (Publication) HTD
Issue number5
StatePublished - Dec 1 2001
Event2001 ASME International Mechanical Engineering Congress and Exposition - New York, NY, United States
Duration: Nov 11 2001Nov 16 2001

ASJC Scopus subject areas

  • Mechanical Engineering
  • Fluid Flow and Transfer Processes

Fingerprint Dive into the research topics of 'Geometric optimization of radiative enclosures through non-linear programming'. Together they form a unique fingerprint.

Cite this