Posted Fri, 13 Sep 2019 09:19:01 GMT by Rodolfo Perego Egr. ing.
Good morning everyone. I am doing the calibration of a preliminary steady state regional model where a series of hydraulic head and temperature observations are available. I attached the set of parameters I chose to calibrate and the set of observations I obtained on field. I tried to calibrate the model for both flow and heat transport. The results are that the parameters for the flow are correctly calibrated, while the parameters for the heat transport are maintained unchanged. The sensitivities are calculated correctly for the head observations but for temperature observations is always calculated 0 as semsitivities. This is what I gest from parameter estimation with PEST:


ITERATION 1
    Sum of squared weighted residuals (ie phi)              =  219.34   
    Contribution to phi from observation group "heagp"      =  43.458   
    Contribution to phi from observation group "hetgp"      =  175.88   

ITERATION 2
OPTIMISATION ITERATION NO.        : 2
    Model calls so far            : 112
    Starting phi for this iteration                          :  203.56   
    Contribution to phi from observation group "heagp"      :  27.679   
    Contribution to phi from observation group "hetgp"      :  175.88 

OPTIMISATION ITERATION NO.        : 3
    Model calls so far            : 331
    Starting phi for this iteration                          :  197.18   
    Contribution to phi from observation group "heagp"      :  21.295   
    Contribution to phi from observation group "hetgp"      :  175.88   


The contribution to phi from observation group hetgp remains the same and the sensitivities for the temperature observations are 0. The model is nicely calibrated for the heads, but not for the temperatures. In the scatter plot, when the parameter estimation is calculated, the simulated temperature always stays the same as the initial one. I don't get what is going on. An help is very appreciated. If you need more information to have a clue don't hesitate to ask me.



Posted Fri, 26 Nov 2021 13:19:10 GMT by Rodolfo Christiansen RUB
Hi,

Did you find any solution? I tried with upwinding and everything I could but I don´t gent any results
Posted Tue, 14 Dec 2021 15:46:33 GMT by Carlos Andres Rivera Villarreyes Global Product Specialist - FEFLOW
Hi,
Without knowing the model settings, I think the problem can be managed by setting different weights for the observations. You can run FePEST with the NOPT option (Termination Criteria) to estimate one full Jacobian matrix. This can give you some insights about the sensitivities.

Cheers, Carlos

You must be signed in to post in this forum.