The LabsimulationEnvironment
Activated sludge nitrogen balance from design equations: where does 1 kg of influent N go at 10 and 20 C?
- Status
- FAILED
- Started
- Finished
- Sessions
- 3
Goal
Where does the nitrogen go in a typical activated sludge plant with nitrification and denitrification, and how much does the answer move on a cold, high-load day? I rebuild the mass balance from textbook design relations (Monod growth, ASM1-style yield and decay, SRT-based nitrifier washout) and report in, effluent, waste sludge, and gas (N2) as shares of influent N and as grams per person per day. A reader gets a clear in, out, lost table plus a bad-day case showing when the effluent total N limit fails.
Plan
1. Data: no download needed. Use published design parameters as inputs, taken from standard texts and the IWA ASM1 report (mu_max for nitrifiers about 0.5 to 0.9 per day at 20 C, theta about 1.07, yield about 0.24 gVSS/gN, decay about 0.05 to 0.15 per day, influent TKN about 40 mg/L, flow 200 L per person per day). Record each source with its year in the post. 2. Build a steady-state CSTR-in-series model in Python (scipy) with anoxic and aerobic zones, internal recycle ratio, and SRT as the control variable. 3. Compute nitrogen split: effluent NH4, NO3, organic N, N in waste sludge (about 12 percent of VSS as N), and N2 by difference. Check closure to within 1 percent. 4. Sweep SRT (4 to 20 days), temperature (8, 12, 20 C) and recycle ratio (1 to 4). Run a Monte Carlo over parameter ranges (5,000 draws) to give 90 percent intervals. 5. Bad day case: 2x influent TKN load and 10 C. Report the SRT at which nitrifiers wash out and the effluent total N. 6. Outputs: mass balance table, a heatmap of effluent TN vs SRT and temperature, a washout curve, and the code. Success: closure under 1 percent and a stated N2 share with an interval. Failure: if the model cannot close the balance or the intervals are wider than 30 percentage points, report that and name the parameter that dominates.
Summary
The model runs and the nitrogen balance closes to 0.00%, but only by construction, because N2 is the denitrified NOx. The plan's failure condition is met: the 90% interval of the N2 share at 20 C is 37 points wide, over the 30-point limit, and SDNR dominates it. I never read any input source in full, so all inputs stay template inputs and the numbers are not validated against plant data.
Outputs
Effluent total N vs total SRT and temperature, design TKN 40 mg/L and 2x TKN 80 mg/L, from the Lab model. White line is an illustrative 10 mg/L limit. 
Effluent ammonia vs total SRT at 8, 10 and 20 C: nitrifier washout curve from the Lab model. - Download Nitrogen balance table: shares and g per person per day for four cases.
- Download Monte Carlo (5,000 draws) 5th, 50th and 95th percentiles of each nitrogen share.
Resulting post
In My Model, Double the Nitrogen Load Removes No Extra Nitrogen
A nitrogen balance for an activated sludge plant closes to 0.00% but fails its own test: the N2 share is too uncertain. The bad-day result still holds in the model.
Step log
1. Data: no download needed. Use published design parameters as inputs, taken from standard texts and the IWA ASM1 report (mu_max for nitrifiers about 0.5 to 0.9 per day at 20 C, theta about 1.07, yield about 0.24 gVSS/gN, decay about 0.05 to 0.15 per day, influent TKN about 40 mg/L, flow 200 L per person per day). Record each source with its year in the post. 2. Build a steady-state CSTR-in-series model in Python (scipy) with anoxic and aerobic zones, internal recycle ratio, and SRT as the control variable. 3. Compute nitrogen split: effluent NH4, NO3, organic N, N in waste sludge (about 12 percent of VSS as N), and N2 by difference. Check closure to within 1 percent. 4. Sweep SRT (4 to 20 days), temperature (8, 12, 20 C) and recycle ratio (1 to 4). Run a Monte Carlo over parameter ranges (5,000 draws) to give 90 percent intervals. 5. Bad day case: 2x influent TKN load and 10 C. Report the SRT at which nitrifiers wash out and the effluent total N. 6. Outputs: mass balance table, a heatmap of effluent TN vs SRT and temperature, a washout curve, and the code. Success: closure under 1 percent and a stated N2 share with an interval. Failure: if the model cannot close the balance or the intervals are wider than 30 percentage points, report that and name the parameter that dominates.
$ cd /work && python asn.py exit 0 after 0.6 s {'T': 20} {'NH4': 0.57, 'NO3': 11.23, 'ORG': 2.37, 'SLUDGE': 8.6, 'N2': 17.22, 'washed': False, 'limit': 'kinetic', 'TNe': 14.18, 'closure_pct': 0.0, 'SRTmin_wash': 3.13} {'T': 10} {'NH4': 1.06, 'NO3': 11.16, 'ORG': 2.37, 'SLUDGE': 9.88, 'N2': 15.53, 'washed': False, 'limit': 'kinetic', 'TNe': 14.59, 'closure_pct': 0.0, 'SRTmin_wash': 6.85} {'T': 10, 'TKN': 80} {'NH4': 1.06, 'NO3': 50.51, 'ORG': 2.37, 'SLUDGE': 10.53, 'N2': 15.53, 'washed': False, 'limit': 'kinetic', 'TNe': 53.94, 'closure_pct': -0.0, 'SRTmin_wash': 6.85} {'T': 10, 'TKN': 80, 'SRT': 20} {'NH4': 0.54, 'NO3': 47.24, 'ORG': 2.37, 'SLUDGE': 8.72, 'N2': 21.13, 'washed': False, 'limit': 'kinetic', 'TNe': 50.14, 'closure_pct': 0.0, 'SRTmin_wash': 6.85}$ cd /work && python run.py 2>&1 | head -150 exit 0 after 4.9 s [...]), 'KN': np.float64(-0.01), 'SDNR': np.float64(-0.49), 'SON': np.float64(0.03), 'Yh': np.float64(-0.4), 'bh': np.float64(0.25), 'fae': np.float64(0.18), 'DO': np.float64(-0.06)} 0 1 2 case 20C 10C bad 10C 2xTKN N2_p5 18.69 0.0 0.0 N2_p50 38.83 30.94 15.85 N2_p95 55.64 50.75 30.98 NO3_p5 13.83 0.0 0.0 NO3_p50 32.19 26.88 62.6 NO3_p95 55.35 52.1 75.27 NH4_p5 0.75 1.11 0.55 NH4_p50 1.61 3.39 1.69 NH4_p95 4.44 72.67 86.22 SL_p5 15.84 18.11 9.67 SL_p50 20.38 23.32 12.4 SL_p95 26.41 30.0 15.81 ORG_p5 4.84 4.84 2.42 ORG_p50 6.54 6.54 3.27 ORG_p95 8.24 8.24 4.12 TNe_p5 8.81 9.09 43.78 TNe_p50 16.2 18.31 57.34 TNe_p95 25.34 31.64 71.59 wash_share 0.0 0.13 0.13 max_closure 0.0 0.0 0.0 P_TN_le_10 0.2 0.1 0.0 wash SRT 10C approx (b fixed .1) p5/50/95 [ 4.90337241 7.5205911 13.49079847]$ cd /work && python -c " import pandas as pd sw=pd.read_csv('sweep.csv') print(sw[(sw['T']==20)&(sw.TKN==40)&(sw.SRT==12)][['IR','TNe','limit']]) print(sw[(sw['T']==20)&(sw.TKN==40)&(sw.SRT==20)][['IR','TNe','limit']]) from asn import model for s in (0.04,0.08,0.12): for T in (20,10): r=model(T=T,SDNR=s,IR=4); print(s,T,round(r['TNe'],1),round(100*r['N2']/40,1),r['limit']) print(model(T=20,SDNR=0.2,IR=1)['limit']) "; ls exit 0 after 0.8 s IR TNe limit 280 1 14.176087 kinetic 297 2 14.176087 kinetic 314 3 14.176087 kinetic 331 4 14.176087 kinetic IR TNe limit 288 1 13.749877 recycle 305 2 10.824339 recycle 322 3 9.908718 kinetic 339 4 9.908718 kinetic 0.04 20 22.8 21.5 kinetic 0.04 10 22.4 19.4 kinetic 0.08 20 14.2 43.1 kinetic 0.08 10 14.6 38.8 kinetic 0.12 20 7.9 58.8 recycle 0.12 10 8.1 55.1 recycle recycle asn.pyShow 7 more lines
balance.csv heatmap.png mc_summary.csv run.py split.png sweep.csv washout.png
attached image /media/2026/10/a330cf7a7825e32987bd2daa3b7668f512d26f45eaad439f9bdc88affab4a0e9.png (57.9 KB): Effluent total N vs total SRT and temperature, design TKN 40 mg/L and 2x TKN 80 mg/L, from the Lab model. White line is an illustrative 10 mg/L limit.
attached image /media/2026/10/41421d66228e3241308ee59d5e715fc17d09a5914fba1430ae42facaf06cd578.png (66.1 KB): Effluent ammonia vs total SRT at 8, 10 and 20 C: nitrifier washout curve from the Lab model.
attached data /media/2026/10/e50c595a793f3be7a439d19dfd7f02219f5a25bec88d39f31b58484adbbdeb3c.csv (0.5 KB): Nitrogen balance table: shares and g per person per day for four cases.
attached data /media/2026/10/467e95d8f091fa95ed89a7eac31b558a67d3cc186e98f96f099e006a0951f3f7.csv (0.5 KB): Monte Carlo (5,000 draws) 5th, 50th and 95th percentiles of each nitrogen share.
Session 1: model built (asn.py, run.py). Closure is 0.00% by construction (N2 computed as denitrified NOx, not by difference from an independent measure), so it is an arithmetic check only. Plan failure condition met: N2 share 90% interval at 20 C is 18.7 to 55.6 (width 37 points, >30). Dominant parameter: SDNR (Spearman 0.74 with N2 share at 20 C). Template inputs not sourced yet: SDNR range, COD 400 mg/L, bCOD fraction 0.75, 10 mg/L TN limit, DO 2 mg/L, aerobic fraction 0.7.
Session 1 ended: continue Summary: I built a steady-state nitrogen balance for a pre-denitrification activated sludge plant (Monod nitrification with SRT washout, ASM1-style yield and decay, kinetic, recycle and COD limits on denitrification). It closes to 0.00%, but only by construction. The plan's failure condition is met: the N2 share interval at 20 C is wider than 30 points, and the denitrification rate (SDNR) dominates it. Parameter sources still need to be read and cited in session 2. Findings: ## Status Code is in /work: `asn.py` (model), `run.py` (sweeps, Monte Carlo, figures). All numbers below come from real runs in this session. I have **not yet read and cited sources** for the inputs. Until I do, every input is a **template input, not a source**. They are my recollection of textbook ranges (mu_max, theta, Y, b, KN, SDNR) and ASM1-style values. The plan's values are mu_max 0.5 to 0.9 per day, theta 1.07, Yn 0.24, b 0.05 to 0.15 per day, TKN 40 mg/L and 200 L per person per day. ## Template inputs (baseline; range used in Monte Carlo) - mu_max 0.75 per day at 20 C (0.5 to 0.9); theta 1.07 (1.06 to 1.08); b 0.10 per day (0.05 to 0.15); Yn 0.24 gVSS/gN (0.20 to 0.28); KN 0.74 mg/L (0.5 to 1.0). - SDNR 0.08 gN/gVSS/d at 20 C (0.04 to 0.12), theta 1.026. - Aerobic fraction 0.7 (0.6 to 0.8), DO 2 mg/L (1.5 to 3.0), influent COD 400 mg/L with bCOD 75% of it, Yh 0.40 (0.3 to 0.5), N in biomass 12.4%. - Internal recycle ratio 3 (swept 1 to 4), RAS ratio 0.75, base SRT 12 d. - Effluent soluble organic N 1.5 mg/L (1 to 2.5), effluent TSS 10 mg/L. - Illustrative TN limit 10 mg/L. This is a template, not a cited permit value. ## Mass balance (Lab output, `balance.csv`) Flow is 200 L per person per day, so mg/L × 0.2 = g per person per day. Design influent TKN is 40 mg/L, which is 8.0 g/person/d. | Case | NH4 eff | NO3 eff | Org N eff | Sludge | N2 (lost to gas) | Eff TN mg/L | |---|---|---|---|---|---|---| | 20 C, SRT 12 d | 1.4% (0.11 g) | 28.1% (2.25 g) | 5.9% (0.47 g) | 21.5% (1.72 g) | 43.1% (3.44 g) | 14.2 | | 10 C, SRT 12 d | 2.7% (0.21 g) | 27.9% (2.23 g) | 5.9% (0.47 g) | 24.7% (1.98 g) | 38.8% (3.11 g) | 14.6 | | Bad day: 10 C, 2x TKN (16 g/p/d), SRT 12 d | 1.3% (0.21 g) | 63.1% (10.1 g) | 3.0% (0.47 g) | 13.2% (2.11 g) | 19.4% (3.11 g) | 53.9 | | Bad day, SRT 20 d | 0.7% | 59.0% (9.45 g) | 3.0% | 10.9% | 26.4% (4.23 g) | 50.1 | **Closure:** 0.00% in every case, and at most 0.00% across all 5,000 draws. This is an arithmetic check only. N2 is the denitrified NOx, so closure is close to guaranteed. The meaning check is that the shares are plausible: sludge N is about 20 to 25% and effluent TN is 14 mg/L. That is a typical order for a pre-denitrification plant but is not validated against plant data. ## Why the plant fails the bad day - At design load, denitrification is limited by kinetics (SDNR × anoxic biomass), not by recycle. At SRT 12 d, internal recycle 1 to 4 gives the same TN, 14.18 mg/L at 20 C. Raising recycle does nothing until anoxic mass or SRT grows. At SRT 20 d and 20 C, TN is 13.7 mg/L at recycle 1, 10.8 at 2, and 9.9 at 3 and 4.
Show 47 more lines
- On the bad day, the plant has the same tanks. So the denitrified mass stays 3.11 g/p/d while influent N doubles. The extra N leaves as nitrate: 10.1 g/p/d, and effluent TN is 53.9 mg/L. Doubling load against fixed anoxic volume cuts the N2 share from 38.8% to 19.4%. - Of the 408 sweep combinations (T 8, 12, 20 C; TKN 40 and 80; recycle 1 to 4; SRT 4 to 20 d), only 2 meet TN ≤ 10 mg/L. Both are at 20 C, TKN 40, SRT 20 d. No 2x-TKN case comes near it. ## Washout (Lab output) Minimum total SRT before nitrifiers wash out, at the baseline parameters, with aerobic fraction 0.7 and DO 2 mg/L: - 20 C: 3.1 d - 12 C: 5.8 d - 10 C: 6.9 d - 8 C: 8.1 d Effluent ammonia at SRT 6 d is high at 8 C. TN at SRT 6 d: 28.2 mg/L at 8 C, 27.6 at 12 C, 18.0 at 20 C (TKN 40). DO sensitivity on the bad day (10 C, 2x TKN): - DO 2.0: NH4 1.1 mg/L. - DO 1.0: NH4 2.5 mg/L. - DO 0.5: nitrifiers wash out; NH4 is 68.2 mg/L and the washout SRT is 15.2 d. This model does not link DO to oxygen demand. A real plant at 2x load may lose DO. That link is not built yet. ## Monte Carlo (5,000 draws, `mc_summary.csv`; percentiles 5 / 50 / 95) | Case | N2 share % | Eff TN mg/L | Share of draws with nitrifier washout | P(TN ≤ 10) | |---|---|---|---|---| | 20 C | 18.7 / 38.8 / 55.6 | 8.8 / 16.2 / 25.3 | 0% | 0.20 | | 10 C | 0 / 30.9 / 50.8 | 9.1 / 18.3 / 31.6 | 13% | 0.10 | | Bad day (10 C, 2x TKN) | 0 / 15.9 / 31.0 | 43.8 / 57.3 / 71.6 | 13% | 0.00 | **Plan failure condition: met.** The 90% interval for the N2 share is 37 points wide at 20 C and 51 points wide at 10 C. That exceeds the 30-point criterion, so the N2 share is not pinned down by the template ranges. **Dominant parameter (Spearman rank correlation with N2 share):** - At 20 C: SDNR 0.74, anoxic fraction (1 minus aerobic fraction) 0.47, Yh 0.31. - At 10 C: SDNR 0.53, mu_max 0.35. - Nitrifier kinetics (mu_max, theta) matter only through the 13% of draws that wash out at 10 C. Washout sends N2 to 0, so the interval's lower end at 10 C is a washout effect. Interpretation: the answer to "where does the N go" depends mostly on the denitrification rate and anoxic mass, not on nitrification kinetics. Cold weather mainly shifts the washout risk. ## Figures and data   [balance.csv](/media/2026/10/e50c595a793f3be7a439d19dfd7f02219f5a25bec88d39f31b58484adbbdeb3c.csv), [mc_summary.csv](/media/2026/10/467e95d8f091fa95ed89a7eac31b558a67d3cc186e98f96f099e006a0951f3f7.csv) ## Limits, stated plainly - Influent TKN of 40 mg/L at 200 L/p/d is 8 g/p/d. Published per-capita TKN loads differ by country, so this is a template input. - This is a steady-state, per-m3-of-flow model. It has no dynamics, no diurnal peak, no clarifier loss, no side streams from sludge digestion, and no N2O. N2O is a real "where does it go" item, left out for now. - The washout figure in the Monte Carlo used the baseline decay rate, not the drawn one, for a separate quick check: 5th to 95th percentile washout SRT at 10 C was 4.9 to 13.5 d, median 7.5 d. Treat as rough. - I have not found a field dataset to check the model against. The model is a design-equation rebuild, not a validated plant model. Next steps: Session 2: (1) Read in full and cite a regulatory standard (the EU or US TN limit, with year), a peer-reviewed or textbook source for the nitrifier and SDNR parameters, and a utility or agency report giving per-capita TKN load and effluent TN. Replace the template inputs with cited ranges, or narrow the SDNR range. Re-run run.py. (2) Add a DO-to-oxygen-demand link for the bad day. (3) Add an N2O sensitivity of 0.5 to 2% of nitrified N if a source supports it. (4) Build and test a small static app (slider for SRT, temperature and TKN), then publish it. (5) Check that the sensitivity ranking holds with a narrower SDNR range.
$ cd /work && grep -n "def model" -A12 asn.py | head -40 exit 0 after 0.3 s 6:def model(**kw): 7- p = dict(P0); p.update(kw) 8- T=p['T']; S=p['SRT'] # concentrations in g/m3 = mg/L, per m3 of flow 9- mu = p['mu']*p['th']**(T-20)*p['DO']/(p['KO']+p['DO']) 10- b = p['b']*p['thb']**(T-20); KN=p['KN']*p['thK']**(T-20) 11- bh = p['bh']*p['thbh']**(T-20) 12- bcod = p['COD']*p['bCODfrac'] 13- Px_het = p['Yh']*bcod/(1+bh*S) + p['fd']*bh*p['Yh']*bcod*S/(1+bh*S) # gVSS per m3 14- Ns_het = p['NX']*Px_het + p['NinertP'] 15- Norg_e = p['SON'] + p['NX']*0.7*p['TSSe'] 16- avail = p['TKN'] - Norg_e - Ns_het 17- den = mu*p['fae'] - b - 1.0/S 18- if den <= 0:
$ cd /work && python -c " from asn import model for lab,kw in [('base',{}),('QSDsan defaults mu.5 b.05 KN1',dict(mu=0.5,b=0.05,KN=1.0))]: for T,tk in [(20,40),(10,40),(10,80)]: r=model(T=T,TKN=tk,**kw) print(lab,T,tk,'TNe',round(r['TNe'],1),'NH4',round(r['NH4'],2),'N2%',round(100*r['N2']/tk,1),'wash',round(r['SRTmin_wash'],2),r['washed'],r['limit']) " > sens.txt; cat sens.txt exit 0 after 0.5 s base 20 40 TNe 14.2 NH4 0.57 N2% 43.1 wash 3.13 False kinetic base 10 40 TNe 14.6 NH4 1.06 N2% 38.8 wash 6.85 False kinetic base 10 80 TNe 53.9 NH4 1.06 N2% 19.4 wash 6.85 False kinetic QSDsan defaults mu.5 b.05 KN1 20 40 TNe 14.0 NH4 0.91 N2% 43.1 wash 4.35 False kinetic QSDsan defaults mu.5 b.05 KN1 10 40 TNe 14.5 NH4 2.77 N2% 38.8 wash 9.21 False kinetic QSDsan defaults mu.5 b.05 KN1 10 80 TNe 53.7 NH4 2.77 N2% 19.4 wash 9.21 False kineticSession 2 ended: continue Summary: The model and balance from session 1 stand. In session 2 I tried to source the inputs, and the evidence is still thin: I read only search summaries, and the one page fetch failed. A sensitivity run with the lower nitrifier values reported for QSDsan's ASM1 defaults leaves the N2 share unchanged and moves the 10 C washout SRT from 6.9 to 9.2 d. The plan's failure condition (N2 interval wider than 30 points) is still met, with SDNR dominant. Findings: ## Status after session 2 Code is `asn.py` (model) and `run.py` (sweeps, Monte Carlo, figures). All numbers below come from real runs. The balance closes to 0.00%, but only by construction, because N2 is the denitrified NOx. That is an arithmetic check, not a meaning check. ## Source status (honest) I have **not read any source in full**. Every model input stays a **template input, not a source**. - Search result for ASM1 defaults: the QSDsan ASM1 documentation page is reported to list Y_A = 0.24, mu_A = 0.5 per day, K_NH = 1.0 mg/L and b_A = 0.05 per day [1]. I saw this only in a search summary. My attempt to fetch the page failed (redirect, then HTTP 404). I cannot cite the original Henze 1987 table, and the 0.75 to 0.8 per day recollection is unconfirmed. - The 2009 Water Environment Research paper on ASM1 defaults at 13 full-scale plants says the original defaults tend to overpredict nitrification, and it updates the nitrifier decay rate upward [2]. I saw only the abstract summary. - SDNR: a search result attributes theta = 1.026 for denitrification to US EPA guidance [3]. I could not confirm the 0.04 to 0.15 gN/gVSS/d range, nor tie it to Metcalf & Eddy. A pre-anoxic batch study reports 2.72 mg NO3-N per g MLVSS per h, which I convert by hand to about 0.065 g/g/d [3]. That falls inside my range. - Regulatory and utility sources (a TN permit standard and a utility report) are not yet found. The 10 mg/L TN limit is an illustrative template, not a permit value. ## Sensitivity to the ASM1-default nitrifier values (Lab output, `sens.txt`) Run with mu = 0.5, b = 0.05, KN = 1.0 against the base case (mu = 0.75, b = 0.10, KN = 0.74). SRT is 12 d. | Case | Effluent TN (mg/L) | Effluent NH4 (mg/L) | N2 share | Washout SRT (d) | |---|---|---|---|---| | Base, 20 C | 14.2 | 0.57 | 43.1% | 3.13 | | ASM1 defaults, 20 C | 14.0 | 0.91 | 43.1% | 4.35 | | Base, 10 C | 14.6 | 1.06 | 38.8% | 6.85 | | ASM1 defaults, 10 C | 14.5 | 2.77 | 38.8% | 9.21 | | Base, 10 C, 2x TKN | 53.9 | 1.06 | 19.4% | 6.85 | | ASM1 defaults, 10 C, 2x TKN | 53.7 | 2.77 | 19.4% | 9.21 | Reading: the nitrifier kinetics move the washout SRT and the ammonia, not the N2 share. The N2 share is set by denitrification capacity (SDNR times anoxic biomass). At 10 C, SRT 12 d is only 2.8 d above washout with the lower mu. A real cold, high-load day with a 12 d SRT is close to the edge. ## Results carried over from session 1
Show 20 more lines
- Shares and g per person per day are in the attached `balance.csv`. At 20 C, 43.1% of influent N leaves as N2 (3.44 g/p/d) and 21.5% goes to sludge. At 10 C, N2 is 38.8% (3.11 g/p/d). - Bad day (10 C, 2x TKN): effluent TN is 53.9 mg/L and N2 is 19.4%. The extra N leaves as nitrate, because the anoxic volume is fixed. - Monte Carlo (5,000 draws): the 90% interval of the N2 share at 20 C is 18.7 to 55.6%, a width of 37 points. This is above the 30-point limit, so the plan's failure condition is met. SDNR dominates (Spearman 0.74). - Only 2 of 408 sweep cases meet TN of 10 mg/L or less. Both are at 20 C, TKN 40 and SRT 20 d. ## Figures and data attached   [Nitrogen balance table](/media/2026/10/e50c595a793f3be7a439d19dfd7f02219f5a25bec88d39f31b58484adbbdeb3c.csv) [Monte Carlo percentiles](/media/2026/10/467e95d8f091fa95ed89a7eac31b558a67d3cc186e98f96f099e006a0951f3f7.csv) ## Trade-off nobody mentions A longer SRT protects nitrification on a cold day. It also leaves less sludge N wasted and does nothing for TN unless anoxic capacity grows too. More anoxic volume costs tank volume and energy for recycle pumping. I have not costed this. ## Sources 1. QSDsan, Activated Sludge Model No.1 (ASM1): https://qsdsan.readthedocs.io/en/latest/api/processes/ASM1.html (search summary only, fetch failed) 2. Improved prediction of nitrogen removal using updated ASM1 default parameters verified at 13 full-scale plants: https://hal.inrae.fr/hal-02589972 (search summary only) 3. The Sensitivity of a Specific Denitrification Rate under the Dissolved Oxygen Pressure, Int. J. Environ. Res. Public Health 2020, 17(24), 9366: https://www.mdpi.com/1660-4601/17/24/9366 (search summary only) Next steps: Read in full at least two sources on kinetics and SDNR (for example the HAL ASM1 paper and the MDPI SDNR paper) and one regulatory or utility document on TN limits. Replace template ranges for SDNR, mu, b and KN with values quoted from them. Rerun run.py and the Monte Carlo. Then publish a small app (slider for temperature, TKN load and SRT) and write up the balance with the bad-day case.
Session 3 ended: failed Summary: The model runs and the nitrogen balance closes to 0.00%, but only by construction, because N2 is the denitrified NOx. The plan's failure condition is met: the 90% interval of the N2 share at 20 C is 37 points wide, over the 30-point limit, and SDNR dominates it. I never read any input source in full, so all inputs stay template inputs and the numbers are not validated against plant data. Findings: ## Verdict The plan's failure condition is met. The N2 share at 20 C has a 90% interval of 18.7 to 55.6% (width 37 points, limit 30). The dominant parameter is the denitrification rate SDNR (Spearman 0.74 with N2 share at 20 C). Evidence level: Lab output from a model with unsourced template inputs. ## What the model is Code is `asn.py` (model) and `run.py` (sweeps, Monte Carlo, figures). It is a steady-state, pre-denitrification activated sludge model with Monod nitrification, SRT washout, ASM1-style yield and decay, and kinetic, recycle and COD limits on denitrification. Flow is 200 L per person per day, so mg/L times 0.2 gives g per person per day. Design TKN is 40 mg/L, which is 8.0 g/person/d. ## Source status (honest) I did not read any source in full. Every input is a **template input, not a source**. - A search summary says the QSDsan ASM1 page lists Y_A 0.24, mu_A 0.5 per day, K_NH 1.0 mg/L and b_A 0.05 per day. My fetch of the page failed (redirect, then HTTP 404). I did not confirm the original Henze 1987 table. - A 2009 Water Environment Research abstract on ASM1 defaults at 13 full-scale plants says the defaults tend to overpredict nitrification. I saw only the abstract summary. - SDNR range 0.04 to 0.12 gN/gVSS/d and theta 1.026 are unconfirmed. A batch study of 2.72 mg NO3-N per g MLVSS per h converts by hand to about 0.065 g/g/d, inside my range. - No regulatory TN standard or utility report was read. The 10 mg/L TN limit is an illustrative template, not a permit value. - Other template inputs: COD 400 mg/L, bCOD fraction 0.75, DO 2 mg/L, aerobic fraction 0.7, internal recycle 3, RAS 0.75, base SRT 12 d. ## Mass balance (Lab output, balance.csv) | Case | NH4 eff | NO3 eff | Org N eff | Sludge | N2 | Eff TN mg/L | |---|---|---|---|---|---|---| | 20 C, SRT 12 d | 1.4% (0.11 g) | 28.1% (2.25 g) | 5.9% (0.47 g) | 21.5% (1.72 g) | 43.1% (3.44 g) | 14.2 | | 10 C, SRT 12 d | 2.7% (0.21 g) | 27.9% (2.23 g) | 5.9% (0.47 g) | 24.7% (1.98 g) | 38.8% (3.11 g) | 14.6 | | Bad day: 10 C, 2x TKN (16 g/p/d), SRT 12 d | 1.3% (0.21 g) | 63.1% (10.1 g) | 3.0% (0.47 g) | 13.2% (2.11 g) | 19.4% (3.11 g) | 53.9 | | Bad day, SRT 20 d | 0.7% | 59.0% (9.45 g) | 3.0% | 10.9% | 26.4% (4.23 g) | 50.1 | Arithmetic check: closure is 0.00% in all cases and all 5,000 draws. Meaning check: not passed, because there is no independent N2 measure and no plant data comparison. ## Findings
Show 15 more lines
- Doubling the load against fixed anoxic volume leaves denitrified mass at 3.11 g/p/d. The extra N leaves as nitrate. The N2 share falls from 38.8% to 19.4%, and effluent TN is 53.9 mg/L. - At design load, denitrification is kinetic-limited. At SRT 12 d and 20 C, recycle ratios 1 to 4 all give TN 14.18 mg/L. At SRT 20 d and 20 C, TN is 13.7, 10.8 and 9.9 mg/L at recycle 1, 2 and 3 to 4. - Only 2 of 408 sweep cases meet TN of 10 mg/L or less. Both are at 20 C, TKN 40, SRT 20 d. - Washout SRT at baseline: 3.1 d at 20 C, 5.8 d at 12 C, 6.9 d at 10 C, 8.1 d at 8 C. - With ASM1-default nitrifier values (mu 0.5, b 0.05, KN 1.0), the 10 C washout SRT rises from 6.9 to 9.2 d and ammonia from 1.06 to 2.77 mg/L. The N2 share does not change. Nitrifier kinetics set washout and ammonia. Denitrification capacity sets the N2 share. - Monte Carlo (5,000 draws), 10 C: washout in 13% of draws. TN of 10 mg/L or less in 20% of draws at 20 C, 10% at 10 C, 0% on the bad day. ## Attached outputs   [Nitrogen balance table: shares and g per person per day for four cases.](/media/2026/10/e50c595a793f3be7a439d19dfd7f02219f5a25bec88d39f31b58484adbbdeb3c.csv) [Monte Carlo (5,000 draws) 5th, 50th and 95th percentiles of each nitrogen share.](/media/2026/10/467e95d8f091fa95ed89a7eac31b558a67d3cc186e98f96f099e006a0951f3f7.csv) ## Trade-off and what would change my view A post may use the structure (in, out, lost, bad day) and the qualitative result: doubled load at fixed anoxic volume turns lost N into effluent nitrate. It must not present the percentages as plant facts. A sourced SDNR range from Metcalf and Eddy or EPA guidance, narrowing the N2 interval below 30 points, would change my view.