Sensitivity analysis of a nonlinear Newton-Krylov solver for heat transfer with phase change.

PDF Version Also Available for Download.

Description

Development of a complex metal-casting computer model requires information about how varying the problem parameters affects the results (metal flow and solidification). For example, we would like to know how the last point to solidify or the cooling rate at a given location changes when the physical properties of the metal, boundary conditions, or mold geometry are changed. As a preliminary step towards a complete sensitivity analysis of a three-dimensional casting simulation, we examine a one-dimensional version of a metal-alloy phase-change conductive-heat-transfer model by means of Automatic Differentiation (AD). This non-linear 'Jacobian-free' method is a combination of an outer Newton-based ... continued below

Physical Description

9 p.

Creation Information

Henninger, Rudolph J.; Knoll, D. A. (Dana A.); Kothe, D. B. (Douglas B.) & Lally, B. R. (Bryan R.) January 1, 2002.

Context

This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided by UNT Libraries Government Documents Department to Digital Library, a digital repository hosted by the UNT Libraries. More information about this article can be viewed below.

Who

People and organizations associated with either the creation of this article or its content.

Provided By

UNT Libraries Government Documents Department

Serving as both a federal and a state depository library, the UNT Libraries Government Documents Department maintains millions of items in a variety of formats. The department is a member of the FDLP Content Partnerships Program and an Affiliated Archive of the National Archives.

Contact Us

What

Descriptive information to help identify this article. Follow the links below to find similar items on the Digital Library.

Description

Development of a complex metal-casting computer model requires information about how varying the problem parameters affects the results (metal flow and solidification). For example, we would like to know how the last point to solidify or the cooling rate at a given location changes when the physical properties of the metal, boundary conditions, or mold geometry are changed. As a preliminary step towards a complete sensitivity analysis of a three-dimensional casting simulation, we examine a one-dimensional version of a metal-alloy phase-change conductive-heat-transfer model by means of Automatic Differentiation (AD). This non-linear 'Jacobian-free' method is a combination of an outer Newton-based iteration and an inner conjugate gradient-like (Krylov) iteration. The implicit solution algorithm has enthalpy as the dependent variable from which temperatures are determined. We examine the sensitivities of the difference between an exact analytical solution for the final temperature and that produced by this algorithm to the problem parameters. In all there are 17 parameters (12 physical constants such as liquid density, heat capacity, and thermal conductivity, 2 initial and boundary condition parameters, the final solution time, and 2 algorithm tolerances). We apply AD in the forward and reverse mode and verify the sensitivities by means of finite differences. In general, the finite-difference method requires at least N+1 computer runs to determine sensitivities for N problem parameters. By forward and reverse, we mean the direction through the solution and in time and space in which the derivative values are obtained. The forward mode is typically more efficient for determining the sensitivity of many responses to one or a few parameters, while the reverse mode is better suited for sensitivities of one or a few responses with respect to many parameters. The sensitivities produced by all the methods agreed to at least three significant figures. The forward and reverse AD code run times were similar and were approximately 34% faster than those of the finite-difference sensitivities. Real problems in three dimensions will certainly have many more parameters describing mold geometry and pouring conditions. If the trend seen here holds true reverse mode AD is favored since the computational time increases only slightly for additional parameters.

Physical Description

9 p.

Source

  • Submitted to: WCCM V - Fifth World Congress on Computational Mechnanics, Vienna, Austria, July 7-12, 2002

Language

Item Type

Identifier

Unique identifying numbers for this article in the Digital Library or other systems.

  • Report No.: LA-UR-02-2478
  • Grant Number: none
  • Office of Scientific & Technical Information Report Number: 976161
  • Archival Resource Key: ark:/67531/metadc935378

Collections

This article is part of the following collection of related materials.

Office of Scientific & Technical Information Technical Reports

What responsibilities do I have when using this article?

When

Dates and time periods associated with this article.

Creation Date

  • January 1, 2002

Added to The UNT Digital Library

  • Nov. 13, 2016, 7:26 p.m.

Description Last Updated

  • Dec. 12, 2016, 6:17 p.m.

Usage Statistics

When was this article last used?

Congratulations! It looks like you are the first person to view this item online.

Interact With This Article

Here are some suggestions for what to do next.

Start Reading

PDF Version Also Available for Download.

Citations, Rights, Re-Use

Henninger, Rudolph J.; Knoll, D. A. (Dana A.); Kothe, D. B. (Douglas B.) & Lally, B. R. (Bryan R.). Sensitivity analysis of a nonlinear Newton-Krylov solver for heat transfer with phase change., article, January 1, 2002; United States. (digital.library.unt.edu/ark:/67531/metadc935378/: accessed August 21, 2017), University of North Texas Libraries, Digital Library, digital.library.unt.edu; crediting UNT Libraries Government Documents Department.