A number of controllers have been designed and added to the benchmark plant one at a time. So far the controllers only have been evaluated individually and the effects on the already existing controllers when adding a new one have not been examined. Now when all control strategies work the next step is to run the plant with all controllers and
evaluate the overall plant performance as well as the performance of the individual controllers when they interact with each other. All three weather files will be run with noise and delay on the sensors, and for each of the files the controlled parameters will be presented together with their manipulated variables. The total effluent nitrogen
concentration and the effluent ammonia concentration will also be presented with their corresponding effluent threshold values. For these two effluent concentrations a
comparison will be made with the original benchmark control strategy, which consists of two controllers. The dissolved oxygen level in zone five is controlled by manipulation of the oxygen transfer coefficient and the nitrate level in zone two is controlled by
manipulation of the internal recycle flowrate. To make the comparison as fair as possible
the maximum value of the airflow rate is set to 240 [ m
3/h ] and the upper bound on the
internal recycle flowrate is set to 5 ⋅ Q
in0[ m
3/d ]. The aeration volume controller is not
included in the final plant configuration since the supervisory DO controller performed better. The performance of the sludge blanket height controller will not be presented in this section since the conditions in section 8 is similar to those in this section except for the upper bound of the airflow rate.
'U\ZHDWKHUFRQGLWLRQV
All controller parameters of the external carbon controller are the same as in section 4. If Figure 9.1 is compared to Figure 4.9 in section 4 it can be seen that the performance of the controller is almost the same in the two situations. In section 4 the internal recycle flowrate was kept constant at 5 ⋅ Q
in0[ m
3/d ], but when the internal recycle controller is active the flow rate can vary from zero up to 5 ⋅ Q
in0[ m
3/d ]. The deviations from the set point are slightly smaller in Figure 9.1, but on the other hand the peaks are somewhat wider compared to Figure 4.9.
0 2 4 6 8 10 12 14
0 0.5 1 1.5 2
SNO,2 [g/m3]
0 2 4 6 8 10 12 14
0 0.5 1 1.5 2 2.5
x 106
Time [days]
External carbon [g/d]
)LJXUH
Nitrate concentration in zone two in upper plot and external carbon flow rate in lower plot.The file dryinfluent is simulated with noise and delay on the sensors.
Next the supervisory DO controller is assessed. The upper bound of the slave controllers has been lowered from 500 to 240 [ m
3/h ], but else all other parameters are the same as in section 5. As seen in Figure 9.2 the controller does not manage to level out the peaks in the ammonia concentration, and compared to Figure 5.10 the peaks are roughly one to two [ g/m ] higher. In the lower plots of Figures 9.1 and 5.10 the DO concentration in
3zone five and its reference value are found. When the upper bound of the slave
controllers is 500 [ m
3/h ] the DO reference value can be kept, but when the bound is
lowered to 240 [ m
3/h ] the reference value is far from reached. This is the main reason
for the performance decrease of the supervisory DO controller. The positive influence of
the sludge blanket height controller on the ammonia concentration was illustrated in section 8, and without this controller active the difference would probably be even higher
0 2 4 6 8 10 12 14
0 2 4 6 8 10
SNH,5 [g/m3]
0 2 4 6 8 10 12 14
0 1 2 3 4 5
Time [days]
SO,5 and SO,ref [g/m3]
)LJXUH
Ammonia concentration in zone five in upper plot, and dissolved oxygen concentration and reference value in zone five in lower plot. The file dryinfluent is simulated with noise and delay on the sensors.Finally the internal recycle controller is presented. The lower limit for the integral part was increased from 5 ⋅ Q
in0/3 [ m
3/d ] to 5 ⋅ Q
in0/2 [ m
3/d ], since the nitrate concentration in section 7 seemed to be a little too high above the reference value on second thoughts.
The same simulation was run as in section 7 with the new value of the upper bound and the peaks were lowered at the expense of bigger deviations below the reference value.
This simulation result is not illustrated. However, the peaks were not lowered as much as
in Figure 9.3. The main cause of the improvement is probably due to the lower DO
concentrations in the recirculated water. The sludge blanket height controller might of
course also influence the performance.
0 2 4 6 8 10 12 14 4.5
5 5.5 6 6.5 7 7.5
SNO,5 [g/m3]
0 2 4 6 8 10 12 14
3 4 5 6 7 8 9 10
x 104
Time [days]
Qa [m3/d]
)LJXUH
Nitrate concentration in zone five in upper plot and internal recycle flow rate in lower plot.The file dryinfluent is simulated with noise and delay on the sensors.
As seen in upper plot in Figure 9.4 the total effluent nitrogen is well below the threshold value of 18 [ g/m ] when all controllers are active. The effluent total nitrogen
3concentration is much higher for the original plant setup and the concentrations are far above the threshold value, see upper plot in Figure 9.5. The effluent ammonia
concentration exceeds its threshold value at five different occasions during the second week, which is seen in lower plot in Figure 9.4.
0 2 4 6 8 10 12 14
6 8 10 12 14 16 18
NTOT,e [g/m3]
0 2 4 6 8 10 12 14
0 2 4 6 8
Time [days]
SNH,e [g/m3]
)LJXUH
Effluent total nitrogen concentration in upper plot and effluent ammonia concentration in lower plot. The effluent threshold values are marked with solid lines; 18 [g/m
3] for the total effluent nitrogen concentration and 4 [g/m
3] for the effluent ammonia concentration. The file dryinfluent is simulated with noise and delay on the sensors.As expected the difference is not as big when comparing the effluent ammonia concentrations. The original setup has a constant airflow rate of 240 [ m
3/h ] in zones three and four, which corresponds to maximum capacity for the supervisory DO controller, and a reference value of 2 [ g/m ] for the dissolved oxygen concentration in
3zone five. With the sludge blanket height controller active the sludge concentrations can be kept higher in the system, and this is probably crucial to the lower ammonia
concentrations in the lower plot in Figure 9.4 compared to the lower plot in Figure 9.5.
0 2 4 6 8 10 12 14
14 16 18 20 22
NTOT,e [g/m3]
0 2 4 6 8 10 12 14
0 2 4 6 8 10
Time [days]
SNH,e [g/m3]
)LJXUH
Effluent total nitrogen concentration in upper plot and effluent ammonia concentration in lower plot for the original plant setup. The effluent threshold values are marked with solid lines; 18 [g/m
3] for the total effluent nitrogen concentration and 4 [g/m
3] for the effluent ammonia concentration.The file dryinfluent is simulated with noise and delay on the sensors.