Posted Mon, 22 May 2017 10:54:41 GMT by Luís Roque Silva Lopes
good afternoon
I am calibrating the following parameters of the activated sludge model: YH, Ks and aH for heterotrophic biomass and YA and umaxA for autotrophic biomass.

The input data entered were the daily mean values of: TKN, COD, water and BCOD.

I intend to calibrate the parameters considering the mean difference between the measured and simulated values of the treated effluent (TKN and COD).

The inserted TKN and COD data of the treated effluent in the property analysis sheet in the parameter estimation experiment (steady-state conditions) corresponds to a series of data of daily values for 90 days.

The problem that arises is that the values of the calibrated parameters are equal to the initial values of the same (indipendently of the initial values assumed by the parameters).

What could be happening?

Thank you, Luís Lopes
Posted Mon, 22 May 2017 11:42:28 GMT by Filip Claeys
Which optimization solver algorithm are you using? Did you supply a Step size and Scaling factor (if Auto-scale is off) for each tuned parameter?
Posted Mon, 22 May 2017 12:58:52 GMT by Luís Roque Silva Lopes
In the property analysis sheet I have:
Auto-scale: is selected;
Step size = 0;
Solver optimizer of the Simplex type
Posted Mon, 22 May 2017 13:42:41 GMT by Luís Roque Silva Lopes
Now it is working, I have changed the step size field to a value other than zero (= 0,3)

Thank you
Posted Wed, 24 May 2017 14:39:55 GMT by Filip Claeys
Yes, Step size determines the magnitude of the changes of the tuned parameters, so a Step size of 0 will lead to values that don't change at all.

You must be signed in to post in this forum.