Parameter calibration: comparison function

Good Morning
My project is about an activated sludge system in WEST.

A) I adjust the effluent fractionation so that COD and TSS values ??can be found in the treated effluent close to the observed values, using the WEST standard kinetic and stoichiometric parameters.

However, when I do the calibration of the model (with the fractionation established in a) used comparison by Mean difference, or Maximum difference or thel's inequality coeficient, I find results quite far from reality (higher values).

But when using the comparison by end value difference, I find quite satisfactory results. (I noticed that this function succeeds in having unsatisfactory results of the kinetic and stoichiometric parameters but can reproduce very good treated effluent results)

There is no problem using the end value difference in my project, however I have to ensure that the kinetic and stoichiometric parameters have correct values ???????

Thank you,
Luís

Ideally, if your model is well-defined (well-constrained), irrespective of the optimisation method, you should get the same results.
In practice, due to a variety of possible reasons, e.g. the initial conditions, some may be more effective than others; some may completely fail.
It's up to you as the modeller to decide which results are meaningful: e.g. if a certain method suggests growth rates, yields, .. that are totally irrealistic, they should be discarded.

It is normal using a comparison value difference at the end, the values of the parameters YH and muH, are approximately equal for the modeling intervals (a temperature of 16 and 19 ° C) ????

Thank you,
Luís

[quote author=Luís Roque Silva Lopes link=topic=20182.msg26093#msg26093 date=1497356230]
.. for the modeling intervals (a temperature of 16 and 19 ° C) ..
[/quote]
I don't understand this.
Do you mean you are testing two temperature scenarios, at 16 and at 19 Celsius; and you get similar values for YH and muH ?

Regarding [b]muH[/b] (parameter, i.e. max growth rate at reference temperature which is 20 Celsius): this is to be calibrated only at 20 Celsius. The actual growth rate is temperature-dependent (muH_Temp), according to the Arrhenius law and hence a state variable.
Regarding the yield, yes: strictly speaking, [b]YH[/b] is a stoichiometric coefficient and as such constant (time- and temperature invariant).