A linear ASM1 based multi-model for activated sludge systems

Figures & data

Table 1. Overview of formation and removal of all components in the ASM1 model.

Component		Formation	Removal
Soluble inert organic matter	S I	–	–
Readily biodegradable organic matter	S S	Hydrolysis of X S	Growth of X BH
Particulate inert organic matter	X I	–	–
Slowly biodegradable organic matter	X S	Decay of X BA or X BH	Hydrolysis to S S
Heterotrophic biomass	X BH	Growth	Decay
Autotrophic biomass	X BA	Growth	Decay
Particulate products	X P	Decay of X BA or X BH	–
Dissolved oxygen	S O	–	Aerobic growth
Nitrate nitrogen	S NO	Nitrification	Denitrification
Ammonium nitrogen	S NH	Ammonification of S ND	Growth of X BA or X BH
Soluble organic nitrogen	S ND	Hydrolysis of X ND	Ammonification to S NH
Particulate organic nitrogen	X ND	Decay	Hydrolysis to S ND
Alkalinity	S ALK	(proton removal)	(proton formation)

Figure 1. Benchmark configuration: 2 anoxic and 3 aerated tanks (with oxygen setpoint control) followed by a sedimentation tank and coupled via an external and internal recycle stream.

Table 2. Influent and operational classes considered in this paper.

COD sol,in	[mg L^−1]	9.5 → 189.5
SS in	[mg L^−1]	11.27 → 411.27
TN in	[mg L^−1]	25 → 75
[Q in]low	[m^3 d^−1]	1500 → 9750
[Q in]medium_low	[m^3 d^−1]	9750 → 18000
[Q in]medium_high	[m^3 d^−1]	18000 → 26750
[Q in]high	[m^3 d^−1]	26750 → 35500
T	[°C]	9 → 24

Figure 2. Comparison of the two linearization techniques for the first anoxic tank (Q _in = medium high, T = 22°C). Left: ASM1 data (black) versus Taylor series based linearization (grey). Middle: ASM1 data (black) versus the weighted linear combination based linearization (grey). Right: residuals for the Taylor series based linearization (left) and the weighted linear combination based linearization (right). S _S is the readily biodegradable organic matter, X _S is the slowly biodegradable organic matter, X _BH the heterotrophic biomass, S _NO the nitrate concentration and S _NH the ammonium concentration.

Figure 3. Comparison of the two linearization techniques for the first aerated tank (Q _in = medium high, T = 22°C). Left: ASM1 data (black) versus Taylor series based linearization (grey). Middle: ASM1 data (black) versus the weighted linear combination based linearization (grey). Right: residuals for the Taylor series based linearization (left) and the weighted linear combination based linearization (right). S _S is the readily biodegradable organic matter, X _S is the slowly biodegradable organic matter, X _BH the heterotrophic biomass, S _NO the nitrate concentration and S _NH the ammonium concentration.

Figure 4. Improvement of identification when avoiding extreme weights (Q _in = medium low, T = 11°C). Evolution of ASM1 (black) and model predicted (grey) profiles of X _BA and S _NO. Left: no constraints. Right: constraints on the identified values of the weights.

Figure 5. Scheduling functions for the influent flow rate Q _in: λ_mlow (–·–), λ_high (–), λ_mhigh (––).

Figure 6. Scheduling functions for temperature T.

Figure 7. Inflow rate Q _in test profile.

Figure 8. Standard (Flemish) mean temperature profile in the aerated tanks over a period of one year.

Figure 9. Influent flow rate interpolation at 22°C. Evolution of ASM1 (black) and model predicted (grey) profiles of S _S, S _NO and S _NH. The right columns depict the respective residuals.

Figure 10. Influent flow rate interpolation and temperature interpolation. Evolution of ASM1 (black) and model predicted (grey) profiles of S _S, S _NO and S _NH in the second anoxic tank. Left: former strategy. Middle: refined strategy. The right columns depict the respective residuals.

Figure 11. Influent flow rate interpolation and temperature interpolation. Evolution of ASM1 (black) and model predicted (grey) profiles of S _S, S _NO and S _NH in the third aerated tank. Left: former strategy. Middle: refined strategy. The right columns show the respective residuals.