I've not used it - had the spreadsheet version. But I am curious as to what is the problem that you are encountering, and which IFAS model you are using. It used to support at least two, one based on empirical correlations, the other on a modified ASM with a layered biofilm diffusion model.

The correlation-based model I have implemented, and had produce 'funny' results; but always explicable if you back to the original correlations, as they are based on low-loaded systems, and break on high-loaded systems.

The temperature is not low; the sludge age is.

The biofilm *may* be modelled as shedding viable autotrophs into the mixed liquor, which may explain why you get activity there. When I implemented the equations I did this originally, but my customer complained that the results did not match with his data, and we modified the code so that biofilm sheeding was was dead bacteria (particulate, bioavailable COD) rather than as viable bacteria.

Denitrification could always take place - the model does not have any special details about the nature of the biofilm support, other than its specific surface area.

I'll see what notes I can pull out on the work I did. May take a few days.

Dear BendelBoy,

The first two expressions you are mentioning are indeed the equations used for the calculation of the maximum biofilm nitrification rate. My problem lies in the predicted amount of nitri and deni incurred by the MLSS (in my opinion, it 's way too high to be attributed to biofilm slough, however I maybe wrong). According to the model developer (Mr. Dipankar Sen) "at a 2-day MLSS MCRT, 50% COD removal and 80% nitrification is on the biofilm; the rest is in the MLVSS (15oC)". In the MLE-IFAS scenarios I have run, the ammonium nitrogen loading rates may sometimes exceed 1.0 g/m2/d, nonethelles I still get almost full nitrification.  

You asked why you can get funny results at high COD loadings - the equations explain why, because at high COD loading they predict high r ates of ammonia removal - there is no explicit handling of COD/ammonia competition in the biofilm.

As to the issue of your high activity in the MLSS, the best that I can offer is that there is excessive viable biomass being shed from the biofilm into the mixed liquor. The Aquifas model *should* (I don't have access to it at the moment) allow you to see at least the state variables, to see if there is a high concentration of autotrophs/nitrifiers in the mixed liquor. If this is the case you either need to find a model coefficient to reduce the viability of shed biomass; reduce the activity of biofilm nitrification; or contact Aquifas to ask them to fix the model.

You state that you *assume* that the sloughed biofilm does not contribute viable heterotrophs/autotrophs. Check to see if the autotroph concentration predicted by the model is high, when compared with a test run where you have no media in the reactor.

If MLSS autotrophs are much higher with media than with no media, then the model is probably sloughing viable biomass. See above.

if MLSS auotrophs are comparable between media & no media, then either there is a bug in the MLSS model component; or you are not modelling what you think you are. It is more likely that the problem would be an error somewhere in your data entry. Double check that temperaures, flows, volumes, wastage rates are what you wanted, and that there are no unit conversion errors.

Plot the curve. You will see a large step change with your numbers at the 10 mg/l sCOD limit. Change ammonia to 20 mg/l. You still get a large step change, but now ammonia-limited nitrification is much greater than COD-limited nitrification. At 4 mg/l ammonia you get a curve that looks 'sensible.' If you ignore the limiting-nutrient switches you can easily see that the rate equation is

IF sCOD > 10 THEN A sCOD / (sCOD - B) else C NH3

and hence you cannot really choose sCOD and NH3 independently, as you want that for sCOD = 10 A sCOD / (sCOD - B) = C NH3. Your statement ' I would expect that when sCOD < 10 ppm, the resulting biofilm NITRI rate should always be higher than the corresponding NITRI rate for all sCOD > 10 ppm cases.' will fail whenever NH3 < A sCOD / (sCOD - B) / C

This is a feature of the correlation. In the 'real' world there is ammonia removal taking place with COD removal, so using a fixed ammonia value with varying COD values is implying that the wastewater is different between the three cases.

So, really you want that when you have reduced sCOD (strictly, degradable sCOD) to less than 10 mg/l that ammonia is still above 4 mg/l. And continue to apply the correlation gingerly.

Sen's diffusional model is intended to provide a better response; but the exposition of it in the WERF journal was highly confusing. The user guide for Aquifas, when I last looked it, was all over the place, and equally confusing.

But used carefully, ensuring that you are in the valid region for the correlation  for which you would need to go back to the original papers - you should get good results.

Effectively, 'funny' results usually mean you are too far from the correlation confidence region; and (IFAS) experience is required to decide when the results are 'funny.'

e-mail sent

e-mail address you used on your profile. Will try again.