Enrico Remigi WEST Product Owner

[quote author=Enrico Remigi link=topic=2471.msg5708#msg5708 date=1432627810]
.. Sensitivity Analysis, like any of the Virtual Experiments, can only use [b]model parameters[/b] as degrees of freedom.
[/quote]
This was not entirely true, sorry.

Also [b]input variables[/b] and [b]initial conditions[/b] can be used as 'variables' in the scope of a Virtual Experiment.

I don't understand why you would like to use [b]top-level parameters[/b] in this context.
Top-level parameters are generally used a) to 'promote' parameters that are common to multiple blocks, to the top-level or b) in the scope of a "coupled model", to expose sub-model parameters to the top-level.

What would you like to achieve exactly in your case?

No. According to the IWA report, the TSS in ASM2 is to be calculated as follows:
[b]TSS[/b] = i_TSS_X_I*(X_I)+i_TSS_X_S*(X_S)+i_TSS_X_BM*(X_H+X_AUT+X_PAO) [b]+ 3.23 * (X_PP) + 0.6 * (X_PHA) + (X_MEOH) + (X_MEP)[/b]

So your calculation is essentially correct (except these additional terms).
It's a known issue that in ASM2(d) the "component" X_TSS and the algebraic state variable TSS (combination of other components, according the equation above) are redundant and in fact de-coupled. In other words, X_TSS evolves independently from its elementary components - which can evidently result in these discrepancies.

The safest way to compute TSS is therefore using the equation above.

[quote author=ivarurdalen link=topic=2471.msg5704#msg5704 date=1432579542]
Is it possible to do a sensitivity analysis or parameter estimation of X_I in the influent?
[/quote]
Not directly on X_I because this is a variable - and Sensitivity Analysis, like any of the Virtual Experiments, can only use [b]model parameters[/b] as degrees of freedom.
So you could define a (fractionation) parameter "[b]f_XI[/b]" (i.e. X_I fraction of ..) and use this for the Sensitivity Analysis.

Correct: the scaling factor is set to the initial value of the parameter. This ensures that two parameters that vary over different ranges (e.g. 0-1 vs. 0-1000) will be equally affected by the optimiser at every integration step.

In your screenshot though, it seems that the lower and upper bounds are disabled: this is because you have not set the "[b]constrained optimisation[/b]" option in the Solver tab page of the Analysis Properties.

You are right.
The initial volume is implicitly set by the [b]initial mass of water[/b] (the mass vector M is a derived state variable: you can set initial states in the Variables tab page of the Block Details).
Conversely, calculating the initial mass of water from the volume (which volume: min or max?) would also not be ideal - unless of course one would add a 3rd parameter, next to "V_Max" and "V_Min", e.g. "V_0". In which case, you could add the following [b]initial equation[/b]:
[code]state.M[IndexOfSolvent] = parameters.V_0 / parameters.WWTPSpecificVolume[IndexOfSolvent][/code]

You could use the following combination of blocks:
- a (multi)[b]probe [/b][after the influent block] that measures the influent flow rate
- a [b]2-way fraction splitter[/b] [after the probe] that either directs the flow to the plant or by-passes it
- a [b]"threshold" control model[/b] that receives the signal from the probe and sets the split fraction of the splitter: 0 (= no by-pass) when flow is lower than a given threshold; 1 (= full by-pass) when flow is greater than the threshold

The latter control model needs to be implemented in MSL - and associated to one of the existing control icons, e.g. the "onoff" icon.

The ANAMMOX model that is implemented in WEST (as of release 2012) derives from ASM1:
[list]
[li]S_NO is split into S_NO2 and S_NO3[/li]
[li]X_BA is split into X_NH, X_NO with the specific X_AN (the ANAMMOX biomass)[/li]
[/list]
The processes included in the conversion model are:
[list]
[li]Hydrolysis[/li]
[li]Aerobic growth of heterotrophs[/li]
[li]Decay of heterotrophic biomass[/li]
[li]Anoxic growth of heterotrophs by reducing nitrate to nitrite[/li]
[li]Anoxic growth of heterotrophs by reducing nitrite to nitrogen gas[/li]
[li]Aerobic growth of ammonium oxidisers (AOB)[/li]
[li]Decay of ammonium oxidisers[/li]
[li]Aerobic growth of nitrite oxidisers (NOB)[/li]
[li]Decay of nitrite oxidisers[/li]
[li]Anaerobic growth of Anammox biomass[/li]
[li]Decay of Anammox biomass[/li]
[li]Aeration[/li]
[/list]

This would in principle enable you to [b]create the Gujer Matrix for the new category[/b].
But the details of the conversion model, i.e. stoichiometry and kinetic expressions, as well as some interactions with the rest of the library, are difficult to explain here.

The "conversion model" for [b]ANAMMOX [/b]was first included in WEST in [b]release 2012[/b].
I am afraid your only option is to implement it yourself through the Model Editor.

This is an extremely appropriate and relevant topic!

[quote author=vbakos link=topic=2454.msg5660#msg5660 date=1431358781]
.. If I have an inlet file (e.g. txt), containing different analytical data (e.g. COD, TKN, TP, etc.), I can open this file during the inlet setting, fix the parameters measured (COD, TKN, TP) and choose the appropriate model (e.g. ASM2d) and the simulator generates the fractionated inlet quality automatically (fractionation which refers to the model chosen). For me it seems to be a "rough black-box" method ..[/quote]
I don't mean to be pedantic, but, since I believe this is a key-subject, I would like to be very clear.
[list]
[li]following the modelling terminology, COD, TKN, etc. are not '[b]parameters[/b]' (which are constant by definition) but '[b]variables[/b]' (which are time-dependent)[/li]
[li]I would not call this methodology '[b]black-box[/b]', quite on the contrary! In WEST, there's absolutely nothing that may be called 'black-box' and the influent model/fractionation is no exception: you can indicate any number of measured (custom) variables; you can define a custom fractionation model, starting from the default one for that category, or starting from scratch; and you can save your modified fractionation model[/li]
[/list]

[quote author=vbakos link=topic=2454.msg5660#msg5660 date=1431358781]
.. based on I do not know what - lot of experiences?, literary ratios for municipal wastewater ww quality?) because carrying out an adequate fractionation from this "narrow spectrum" of measured parameters (i.e. COD, TKN, TP), that is quite a big challenge... [/quote]
You are very right: the influent ('[b]fractionation[/b]') model is usually the single, most crucial model in a whole-plant model.
The least measurements (COD, ..) are available, the larger is the [b]uncertainty[/b] of the fractionation model (and, as a consequence, of the overall model). Thus, the first step in the calibration of the whole-plant model will be the calibration of the fractionation model.
And yes: the default values of the default ASM- fractionation models are based on literature values.

[quote author=vbakos link=topic=2454.msg5660#msg5660 date=1431358781]
1. If I have only some measured parameters (COD, TKN, TP), the automatic fractionation may be inaccurate, my calculations may be based on data charged by huge uncertainty, so the results of the simulations may fail. Is that a right suggestion?[/quote]
Precisely!

[quote author=vbakos link=topic=2454.msg5660#msg5660 date=1431358781]
2. If I .. have a broader spectrum of measured parameters (e.g. tot COD, diss COD filtered through 1.2 micron filter, filtered-flocculated COD through 0.45 micron, ..), wide enough for an accurate fractionation, how can I deal with these data? How can I import them appropriately into the WEST fractionation panel?
[/quote]
First: you're very lucky!  :)
Second: the list of four components (Water, COD, TKN, TSS) that you find in the General tab page of the Influent Tool is only a 'suggestion'.
You can add your own 'components' (i.e. measurements) to the list. As a result, the Fractionation page will be populated with all the corresponding [b]input blue blocks[/b] - that are to be connected to the relevant [b]output (green) blocks[/b].
As I mentioned above, you can start from one of the default fractionation models; or, more likely, if you have so many custom measurement, you can start from scratch and create your own, graphically (same principle as a layout in WEST main).

[quote author=vbakos link=topic=2454.msg5660#msg5660 date=1431358781]
3. Another basic technical question related to the previous: how can I / should I design the inlet file in order to be able to import the data in WEST? (is there any guideline for it? - which format (xls or txt), order of colomns, parameteres?)
[/quote]
See [b]User Guide, page 56[/b] onwards

[quote author=vbakos link=topic=2454.msg5660#msg5660 date=1431358781]
4. How can I discover what the "black-box" of WEST does with the parameters in the fractionaton step? (were are the background formules, methods?)
[/quote]
Again: no 'black-box'.
I am not sure I fully understand, but the fractionation 'step':
[list]
[li]consists in setting up of the fractionation model by graphically connecting the blue and the green blocks; optionally, defining 'work' variables ('V' blocks); and defining parameters (i.e. the 'weights' of the conversions from left- to right block along a connection)[/li]
[li]results in a [b]model [/b], written in MSL just like any other model in WEST and stored in the project folder[/li]
[/list]
Once you complete the Influent Tool (click 'Generate' in the last page), the fractionation model (the MSL file) is created and its details will be visible, like for any other model of your layout, in the Block Summary and Block Details panes.