Can i perform "back analysis" within ... Log Out | Topics | Search
Moderators | Register | Edit Profile

FlexPDE User's Forum » User Postings » Can i perform "back analysis" within flexpde? « Previous Next »

Author Message
Top of pagePrevious messageNext messageBottom of page Link to this message

Jack Zhang (lovekk)
Member
Username: lovekk

Post Number: 4
Registered: 01-2007
Posted on Wednesday, February 07, 2007 - 11:49 am:   

Suppose i have built a groundwater flow model in flexpde, and i also have some measurement data from the real world. My objective is to find a set of model parameters, such as permeability, that can produce prediction that is closest to the measurement data. I wonder whether i can realize this in flexpde?
Besides, can i perform matrix inversion in flexpde? Thank you!
Top of pagePrevious messageNext messageBottom of page Link to this message

Robert G. Nelson (rgnelson)
Moderator
Username: rgnelson

Post Number: 760
Registered: 06-2003
Posted on Wednesday, February 07, 2007 - 01:11 pm:   

FlexPDE was used by Lawrence Livermore Laboratory in an extensive inverse modeling program in 1999. See the report at http://www-afim.llnl.gov/UCRL-ID-126377_Fin.pdf. See especially Appendix B for a theoretical description and Appendix D, which lists FlexPDE scripts.

A major problem with this kind of modeling is that the sparse sampling data do not uniquely define a solution. There may be many permeability patterns that will reproduce a scattered data set. The approach taken in the LLNL study was to impose a strong diffusion term in the K equation and to similarly filter representations of the head gradient, to filter out high frequency modes and select the lowest order spatial distribution that fits the scattered data.

As to the matrix inversion question, No, you cannot use FlexPDE as a generalized matrix inversion package. The solution methods are of course based on finding solutions to matrix equations. But the matrices are built dynamically from finite element models, and there an inverse matrix is never computed.
Top of pagePrevious messageNext messageBottom of page Link to this message

Jack Zhang (lovekk)
Member
Username: lovekk

Post Number: 6
Registered: 01-2007
Posted on Thursday, February 08, 2007 - 06:45 am:   

Nelson:
you information is very helpful. Thank you!
Top of pagePrevious messageNext messageBottom of page Link to this message

Julia Derx (juliamelie)
Junior Member
Username: juliamelie

Post Number: 3
Registered: 09-2008
Posted on Monday, March 23, 2009 - 05:46 am:   

I am also interested how one could implement inverse modeling with FlexPDE for groundwater problems. Unfortunately, the website link to the report cited above does not work. Is this report still available or is there any other documentary for inverse modeling?
Top of pagePrevious messageNext messageBottom of page Link to this message

Robert G. Nelson (rgnelson)
Moderator
Username: rgnelson

Post Number: 1225
Registered: 06-2003
Posted on Monday, March 23, 2009 - 03:57 pm:   

The report has apparently been moved. A Google search for "Gelinas Doss Ziagos" finds it at:
https://e-reports-ext.llnl.gov/pdf/236651.pdf
Top of pagePrevious messageNext messageBottom of page Link to this message

Robert G. Nelson (rgnelson)
Moderator
Username: rgnelson

Post Number: 1227
Registered: 06-2003
Posted on Monday, March 23, 2009 - 04:36 pm:   

We have added this report to our "Reprints" page, www.pdesolutions.com/reprints.html

Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action:

Topics | Last Day | Last Week | Tree View | Search | Help/Instructions | Program Credits Administration