8/3/2019 Optimizing Nitrogen Removal in a MBR
1/20
OPTIMIZING NITROGEN REMOVAL IN A MBR
UNDER ABNORMAL OPERATING CONDITIONS
John R. Holland, Dr. John Bratby, Frank Gall
Brown and Caldwell
110 South Church Avenue Suite 2300Tucson, Arizona 85614
Dr. John Bratby Brown and CaldwellFrank Gall Pima County Wastewater Management Division
ABSTRACT
During startup and commissioning of a modified Ludzack-Ettinger (MLE) membrane bioreactor
(MBR), Pima County Wastewater Management Division (PCWMD) staff were concerned withthe total nitrogen concentration (TN) in the plant effluent. Whereas in the traditional MLE
process, increasing recycle ratio increases the TN removal, in this particular process increasingthe recycle ratio proved to actually decrease TN removal. Influent five-day biochemical oxygen
demand (BOD5) was also found to be only 50% of design. The process was found to be verysensitive to recycle ratio due to oxygen entering the anoxic zones. Increased process control
testing provided insights into the process operation. Based on these data, significant changes
were made to equipment operation to lower the aeration basin dissolved oxygen (DO)concentration. In addition, the recycle ratio was managed to minimize oxygen transfer to the
anoxic zones. These operational changes improved TN removal beyond that predicted in the
Biowin model used for design. This paper describes the findings related to the operation ofthis MBR facility under abnormal conditions and some of the factors that were found to impact
TN removal.
KEYWORDS
Modified Ludzack-Ettinger, membrane bioreactor, total nitrogen, recycle ratio, dissolved oxygen
INTRODUCTION
The Randolph Park Water Reclamation Facility (RPWRF) is owned and operated by PCWMD
and is a scalping plant that is designed to deliver Arizona Class A+ reclaimed water to Tucson
Water, although permitted for Arizona Class A reclaimed water at the present time. ArizonaClass A+ reclaimed water is defined as wastewater that has undergone secondary treatment,
filtration, nitrogen removal treatment, and disinfection. Chemical enhancement of filtration is
required under most circumstances. This particular MBR process has been determined to meetthe turbidity requirements without chemical addition. Key requirements for Arizona Class A+
effluent are as follows:
Turbidity < 2 nephlometric turbidity units (NTU) daily average
Turbidity < 5 NTU average up to a 2 minute period
Fecal coliform < 23 per 100 milliliters (ml) maximum
2655
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
2/20
Fecal coliform = non-detect (ND) in 4 out of 7 consecutive samples
Total Nitrogen < 10 milligrams per liter (mg/L) geometric average
Arizona Class A reclaimed water has the same quality requirements as Class A+ reclaimed
water, with the exception of the total nitrogen limitation.
The Randolph Park MBR facility was constructed on the site of an abandoned air-activated
sludge treatment plant that had previously provided Arizona Class B reclaimed water at a flowrate of up to 1.5 million gallons per day (MGD). Arizona Class B reclaimed water is defined as
wastewater that has undergone secondary treatment and disinfection. Key requirements for
Arizona Class B effluent are as follows:
Fecal coliform < 800 per 100 ml maximum
Fecal coliform < 200 per 100 ml in 4 out of 7 consecutive samples
There were site constraint issues relative to increasing the plant capacity from 1.5 MGD of
Arizona Class B reclaimed water to 3.0 MGD of Arizona Class A+ reclaimed water. Dataindicated that the BOD5 concentration in the remote sewer would be up to 300 mg/L. With the
design intent of providing 3.0 MGD average daily flow (ADF) of Class A+ wastewater from thisexisting site, the MBR process was selected due to its small footprint and effluent quality. Since
neither secondary clarifiers nor filters would be needed, the MBR process would release valuable
real estate for anoxic process tanks and additional aeration and MBR basins.
Figure 1 Aerial View of Randolph Park Facility Site
2656
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
3/20
An MLE process modified for use with the MBR system was selected as the optimum process
within available site constraints based on an un-calibrated BIOWIN model. The differencebetween the selected process layout and the traditional MLE process is reflected in Figure 2 and
Figure 3.
Figure 2 Typical MLE Process
Figure 3 MLE Process as Implemented at Randolph Park
The process design resulted in retrofitting the facility to have eight anoxic tanks in series,followed with six parallel trains of aeration and MBR basins. Effluent disinfection by ultraviolet
light (UV) provides the required degree of disinfection in a small footprint.
As illustrated in Figure 3, the Randolph Park process design allows for the influent to eithercombine with the return activated sludge (RAS) prior to the anoxic tanks or to discharge into the
third anoxic tank. This designed feature was implemented to allow for the first two anoxic tanks
to be used to remove residual DO from the MBR basins. This process flexibility would proveessential during the process optimization efforts.
The influent to the RPWRF is scalped from a 42 inch interceptor, screened, degritted andpumped, all from a remote pumping station located approximately 2.5 miles away. An overview
of the process layout is presented in Figure 4.
2657
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
4/20
Figure 4 Randolph Park Process Flow Diagram Six MBR Trains in Service
As can be seen in Figure 4, the anoxic tank volume is fixed, whereas the number of MBR trains
can be varied to match the influent flow rate with the permeate pump capacity. The differencebetween the minimum permeate pumping rate and the maximum flux rate through the
membranes allows for a little over 2:1 turndown between minimum and maximum flow through
an individual MBR train. With the design allowing for the influent flow rate to vary between 0.5and 3.5 MGD, there could be as few as one MBR train or as many as six MBR trains on line,
depending on the influent flow rate.
When the plant is in service and producing up to 3.5 MGD for 18 hours per day, all six parallelMBR trains must be in service. That is, they must have mixed liquor suspended solids (MLSS)flowing through them and the permeate pump must be removing the liquid from the MBR basin.
The process design for this facility resulted in the treatment plant trains being paired up, that is,
two aeration basins would share a single blower and two MBR trains would share a singleblower. The surface foam removal system also was shared between a pair of trains. The
membrane vendor preferred this design, as it allowed them to alternate the agitation air blower
between each of the two trains in a pair. One MBR would have agitation air for 15 seconds, thenthe other one for 15 seconds and so on. This design did result in some programming changes
being required, as described later.
When the plant is in service and producing only 0.5 MGD, then only one of the six parallel MBR
trains needs to be in service. In this case, all six trains still have MLSS flowing through them,
but only one permeate pump will be removing liquid from an MBR basin. Refer to Figure 5 for
a schematic of how the RPWRF operates under this condition.
2658
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
5/20
Figure 5 Randolph Park Process Flow Diagram One MBR Train in Service
In the case of Figure 5, permeate pumps and UV systems for trains 2 through 6 have been turned
off. The air is still on to all aeration and MBR basins, and the RAS pumps for all trains are stillrunning and moving MLSS through the basins and back to the RAS box. This allows all six
trains to provide biological treatment, even though only one train is producing effluent.
IDENTIFICATION OF ISSUES AFTER STARTUP
The plan for the startup of the RPWRF was created to address several objectives:
Attain Class A+ effluent quality in as short a time as possible this was accomplished by
bringing in nitrifying seed
Process 2.0 MGD as soon as permit compliance was established this was accomplished by
bringing in sufficient seed to adequately treat 2.0 MGD
Reach full plant capacity of 3.0 MGD as quickly as possible this was accomplished bywasting only a nominal amount of biomass each day for the first 2 weeks of operation
Maintain optimum membrane flux by having a high quality mixed liquor this wasaccomplished by ensuring that the MLSS was greater than 1,200 mg/L, a value that the
membrane vendor agreed would produce a flocculating biomass
Avoid foaming problems common to activated sludge plant, and especially MBR plant
startups this was accomplished by bringing in enough seed to provide an MLSS of 1,500mg/L, a value that the startup team agreed would minimize the potential for foaming
The key component to the success of this plan was based on seeding the RPWRF with asufficient mass of nitrifying mixed liquor. The mass of seed trucked to the RPWRF on April 10
th
2005 produced a total suspended solids (TSS) concentration of more than 1,500 mg/L in the
process (a total of approximately 7,500 pounds of solids). For 14 days the plant was operated atan influent flow rate 0.8 MGD, the maximum rate that effluent could be discharged to the local
2659
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
6/20
sewer due to hydraulic constraints in the sewer. Only a nominal amount of biomass was wasted,
approximately 200 pounds per day. The effluent was monitored daily for compliance with theAquifer Protection Permit (APP) and the Arizona Pollutant Discharge Elimination permit
(AZPDES) and for TN.
During this period of time the TSS in the process increased to almost 3,800 mg/L. On April 26,2005 the isolation valve to Tucson Water was opened and reclaimed water delivery commenced.
Once the connection to the Tucson Water distribution system was opened, the hydraulic capacity
of the local sewer to accept plant effluent was no longer the limitation to plant flow. Thelimitation to plant flow rate became the biological process and/or the number of MBR trains that
could be put into service. Tucson Water was more than willing to accept the entire plant output
of 3.0 MGD.
For a couple of months following startup, effluent TN values were at or lower than what had
been predicted in the Biowin modeling. However, as additional MBR trains were brought online, and especially if MBR trains were taken out of permeating service while still running the
RAS pumps, effluent TN values began to rise. This trend is shown in Figure 6.
Figure 6 RPWRF Effluent Total Nitrogen
RPWRF Effluent Total N
0
2
4
6
8
10
04/26/05
05/10/05
05/24/05
06/07/05
06/21/05
07/05/05
07/19/05
08/02/05
08/16/05
08/30/05
09/13/05
09/27/05
10/11/05
10/25/05
TotalN(mg/L)
Anticipated 7.5 mg/L
Numerous efforts were made to improve TN removal by increasing the recycle ratio up to more
than seven times the influent flow rate. When these efforts did nothing to reduce the TNconcentrations in the plant effluent, a systematic review of plant data was begun.
ANALYSIS OF PLANT DATA
The single issue with plant performance was effluent TN concentrations they were above what
the PCWMD staff was looking to achieve. All permitted effluent parameters were well belowthe limit. It was proposed that the combination of low influent BOD5 concentration, combined
with below design flows, was leading to lack of TN removal efficiency, even with the increased
recycle ratio. Influent BOD5 values during this time are shown in Figure 7.
2660
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
7/20
Figure 7 RPWRF Influent BOD5
RPWRF Influent BOD5
0
50
100
150
200
250
300
350
04/26/05
05/10/05
05/2
4/05
06/07/05
06/2
1/05
07/05/05
07/19/05
08/02/05
08/16/05
08/30/05
09/13/05
09/27/05
10/1
1/05
10/25/05
BOD5(mg/L)
Design = 300 mg/L
As can be seen in Figure 7, by the middle of July plant influent BOD5 concentrations had fallen
to roughly 50% of those assumed in the basis of design. This led to some staff within PCWMD
to consider supplementing the process with methanol to improve nitrogen removal. This ideawas later abandoned as the process control testing and changes began to improve nitrogen
removal.
As a result of equipment issues, the hydraulic loading to the RPWRF had also not reached thetarget of 3.0 MGD, even after 2 months of operation, although there was sufficient biomass
available to treat the design 3.0 MGD. A graph of plant flow is presented in Figure 8.
Figure 8 RPWRF Effluent Flow
RPWRF Effluent Flow
0
0.5
1
1.5
2
2.5
3
3.5
04/26/05
05/10/05
05/24/05
06/07/05
06/21/05
07/05/05
07/19/05
08/02/05
08/16/05
08/30/05
09/13/05
09/27/05
10/11/05
10/25/05
Flow
(MGD)
Design = 3.0 MGD
As can be seen in Figure 8, RPWRF flow did not reach values approaching design flow until
well into the month of October.
2661
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
8/20
The fact that both the influent BOD5 concentration and plant hydraulic loadings were less thanthe design basis resulted in the MLSS also being significantly less than the design basis, even at
a higher than design solids retention time (SRT), as shown in Figure 9.
Figure 9 RPWRF MLSS
RPWRF TSS
0
2000
4000
6000
8000
10000
04/26/05
05/10/05
05/24/05
06/07/05
06/21/05
07/05/05
07/19/05
08/02/05
08/16/05
08/30/05
09/13/05
09/27/05
10/11/05
10/25/05
TSS
(mg/L)
Design = 8,000 mg/L
As can be seen in Figure 9, the MLSS only increased to approximately 50% of the design value
of 8,000 mg/L, despite operating the plant at an SRT of between 15 and 20 days, versus a design
SRT of 7 to 10 days. This reduced MLSS concentration was identified as a contributing factor inthe reduced TN removal rates.
The data as described above confirmed that the RPWRF was operating far from its design basis.
Numerous process control changes related to the recycle ratio were proving to be unhelpful atbest and detrimental at worst. As a result of this, RPWRF staff began additional process control
testing to better understand the dynamics of nitrogen removal in the anoxic tanks. The results of
this testing and the conclusions from the analysis of the data are presented in the followingsection.
PROCESS CONTROL TESTING PROGRAM AND RESULTS
The process control testing program for this operating condition consisted of collecting grab
samples for dissolved oxygen concentration, nitrite/nitrate concentration and oxygen uptake rate
(OUR) in selected anoxic tanks. The tanks selected for testing were the RAS box and anoxic
tanks numbers 1, 3 and 7. Refer to Figure 4 for the location of these tanks within the process.The data collected during this period is presented in the following figures.
Figure 10 presents the DO concentrations measured in anoxic tank 1 on various shifts from
September 13th
through October 24th
2005.
2662
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
9/20
Figure 10 RPWRF Anoxic Tank 1 DO Concentration
Anoxic Tank 1 DO
0
0.2
0.4
0.6
0.8
1
1.2
9/13/05
9/16/05
9/19/05
9/22/05
9/25/05
9/28/05
10/1/05
10/4/05
10/7/05
10/10/05
10/13/05
10/16/05
10/19/05
10/22/05
DO
(mg/l)
Figure 11 presents the nitrite/nitrate concentrations measured in anoxic tank 1 on various shiftsduring the same period of time.
Figure 11 RPWRF Anoxic Tank 1 Nitrite/Nitrate Concentration
Anoxic Tank 1 NO2/NO3
0
2
4
6
8
10
12
14
16
9/13/05
9/17/05
9/21/05
9/25/05
9/29/05
10/3/05
10/7/05
10/11/05
10/15/05
10/19/05
10/23/05
NO2/NO3(m
g/l)
Figure 12 presents the DO concentrations measured in anoxic tank 3 on various shifts from
September 13th
through October 24th
.
2663
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
10/20
Figure 12 RPWRF Anoxic Tank 3 DO Concentration
Anoxic Tank 3 DO
0
0.2
0.4
0.6
0.8
1
1.2
9/13/05
9/16/05
9/19/05
9/22/05
9/25/05
9/28/05
10/1/05
10/4/05
10/7/05
10/10/05
10/13/05
10/16/05
10/19/05
10/22/05
DO
(mg/l)
Figure 13 presents the nitrite/nitrate concentrations measured in anoxic tank 3 on various shifts
during the same period of time.
Figure 13 RPWRF Anoxic Tank 3 Nitrite/Nitrate Concentration
Anoxic Tank 3 NO2/NO3
0
2
4
6
8
10
12
14
16
9/13/05
9/17/05
9/21/05
9/25/05
9/29/05
10/3/05
10/7/05
10/11/05
10/15/05
10/19/05
10/23/05
NO2/NO3(m
g/l)
Figure 14 presents the DO concentrations measured in anoxic tank 7 on various shifts from
September 13th
through October 24th
.
2664
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
11/20
Figure 14 RPWRF Anoxic Tank 7 DO Concentration
Anoxic Tank 7 DO
0
0.2
0.4
0.6
0.8
1
1.2
9/13/05
9/16/05
9/19/05
9/22/05
9/25/05
9/28/05
10/1/05
10/4/05
10/7/05
10/10/05
10/13/05
10/16/05
10/19/05
10/22/05
DO
(mg/l)
Figure 15 presents the nitrite/nitrate concentrations measured in anoxic tank 3 on various shifts
during the same period of time.
Figure 15 RPWRF Anoxic Tank 3 Nitrite/Nitrate Concentration
Anoxic Tank 7 NO2/NO3
0
2
4
6
8
10
12
14
16
9/13/05
9/17/05
9/21/05
9/25/05
9/29/05
10/3/05
10/7/05
10/11/05
10/15/05
10/19/05
10/23/05
NO2/NO3(m
g/l)
Based on the data presented in Figures 9 through 14, the nitrate was not being removed through
the anoxic zone, regardless of the time of day. The modeling data indicated that all nitrate wouldbe removed by the end of the 3
rdanoxic tank during much of the day, as shown in Figure 16.
2665
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
12/20
Figure 16 RPWRF Biowin Model Diurnal Nitrate Concentrations
Nitrates
Recycle NO3-diluted AX1-1 NO3-N AX1-2 NO3-N AX2-3 NO3-N
AX2-1 NO3-N AX2-2 NO3-N AX3-1 NO3-N AX3-2 NO3-N
AX3-3 NO3-N Filtrate-Effluent NO3-N
TIME0222018161412108642
CONC.
(mg/L)
14
12
10
8
6
4
2
0
The OUR in the various tanks was also reviewed and compared to the Biowin model results.
These data are presented in the following figures.
Figure 17 presents the measured OUR in the RAS box on various shifts from September 13th
through October 24th
.
Figure 17 RPWRF RAS Box Oxygen Uptake Rate
RAS Box OUR
0
20
40
60
80
100
09/13/05
09/16/05
09/19/05
09/22/05
09/25/05
09/28/05
10/01/05
10/04/05
10/07/05
10/10/05
10/13/05
10/16/05
10/19/05
10/22/05
OUR(mg/L/hr)
Anticipated 60 mg/L/hr
Figure 18 presents the measured OUR in anoxic tank 3 on various shifts for the same period of
time.
2666
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
13/20
Figure 18 RPWRF Anoxic Tank 3 Oxygen Uptake Rate
Anoxic Tank 3 OUR
0
20
40
60
80
100
120
140
09/13/05
09/16/05
09/19/05
09/22/05
09/25/05
09/28/05
10/01/05
10/04/05
10/07/05
10/10/05
10/13/05
10/16/05
10/19/05
10/22/05
OUR(mg/L/hr)
The OUR in the RAS box is representative of the treated effluent in the MBR tanks, as the RAS
is pumped from the end of the MBR tanks to the combined RAS box. As can be seen in Figure
17, the OUR in the RAS box was approximately 50% of that predicted in the design. With thelower OUR, the DO in the anoxic tanks takes longer to be consumed, delaying the onset of
anoxic conditions.
The impact of adding the raw influent to anoxic tank 3 can be seen in the increased OUR
observed in anoxic tank 3 over that of the RAS box.
DISCUSSION
During the time of these process control tests, all raw influent was being directed to anoxic tank
3. Based on Biowin model simulations under the actual operating conditions, it wasdetermined that the plant would achieve optimum nitrate removal if the raw influent were to
continue to be added to anoxic tank 3 rather than the RAS box. By adding the raw influent to theanoxic tank 3, the minimal carbon available in the influent would be directed towards nitrogen
removal, not removing dissolved oxygen.
Because of the plant design, with the RAS from the MBR tanks serving as the internal recycle as
well, the higher the recycle ratio, the higher the DO concentration in the anoxic tanks. Since the
RAS box has an on-line DO analyzer installed, the RAS box DO concentration was used as a
guide to the DO concentration in the anoxic tanks. Figure 19 shows the impact of changing therecycle ratio on the DO concentration in the RAS box.
2667
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
14/20
Figure 19 RPWRF Recycle Ratio and RAS Box DO Concentration
DO RAS Tank with Qr/Qi (May 2006)
0
1
2
3
4
5
6
7
0 20 40 60 80 100 120 140 160 180 200 220 240
Time (min)
mg/L
Qr/Qi
Figure 19 spans the period of time from 6:30 AM until 10:30 AM, a time when the influent flow
rate has risen from the diurnal low to the diurnal high. The normal response of the RAS box DOconcentration is to rise to maximum by around 7:00 AM and then drop back down 1 to 2 hours
later. As the DO concentration was beginning to fall at the 120 minute mark, the recycle ratio
was raised from 3.5:1 to 6:1. At this time the RAS box DO concentration began to rise again andsettle at about 1 mg/L. The recycle ratio was lowered to 3.5:1 at about the 190 minute mark and
the RAS box DO concentration fell to 0 mg/L. A similar pattern can be seen between the 200
minute mark and the end of the test. Note that the data presented in Figure 19 are recent data,which is the end result of many months spent optimizing the nitrogen removal process.
At the beginning of the process optimization, the DO concentration of the aeration basins wasmuch higher than design, even at minimum blower output. This is illustrated in Figure 20.
Figure 20 RPWRF Aeration Basin DO Concentration
DO Aeration basin (May 2005)
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hr)
DO
(m
g/L)
Design 1.5 mg/L
2668
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
15/20
In addition to the DO that exists in the aeration basin, where the higher oxygen demand exists,the MBR tanks are constantly being aerated to maintain the membranes. This only serves to
exacerbate the problem with high DO concentrations in the combined RAS/recycle stream.
The low influent loading, coupled with an increase in the oxygen transfer efficiency, wasdetermined to be the cause of the high DO. During the design process, the MLSS was projected
to be at 8,000 to 10,000 mg/L, a level high enough to significantly lower the oxygen transfer
efficiency. The impact of MLSS on oxygen transfer efficiency is illustrated in Figure 21.
Figure 21 Relationship of Oxygen Transfer and MLSS
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 5 10 15 20 25 30 35 40 45 50 55
MLSS, g/L
Alpha Fine Pore
Coarse Bubble
The RPWRF was operating at only about 4,000 mg/L instead of the design 8,000 mg/L. Thisresulted in a significant increase in the amount of oxygen transferred to the aeration and MBR
basins. While the air to the aeration basins could be controlled, the air to the MBR basins could
not due to the requirement for membrane agitation.
It was not an easy matter to further reduce the air supply to the aeration basins. Programming
changes were needed to allow only one of the three aeration basin blowers to deliver air to all sixaeration basins. However, these changes were implemented and the impact of reducing the
amount of air to the aeration basins on the aeration basin DO concentrations is illustrated in
Figure 22.
2669
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
16/20
Figure 22 RPWRF Aeration Basin DO Concentration
DO Aeration basin (July 2005)
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (hr)
mg/L
Once the plant staff was able to reduce the DO concentrations in the aeration basins, other issues
arose relative to MBR train availability. These issues negatively impacted RPWRF staff abilityto control the DO concentrations. This is reflected in Figure 23, which is a plot of aeration basin
DO concentrations during a 24 hour period in October 2005.
Figure 23 RPWRF Aeration Basin DO Concentration
DO Aeration basin (October 2005)
0
1
2
3
4
5
6
7
8
9
10
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0 1 2 3 4 5 6 7 8
Time (hr)
mg/L
The impact of having near saturation DO concentrations in the aeration basins on the ability to
run at high recycle ratios is clear. Yet, despite all of these issues, plant staff was able to maintain
effluent TN concentrations near the predicted design at all times. This required continualprocess control testing of DO and TN concentrations in the various anoxic tanks, as well as
experimentation with various recycle rates. As a result of these efforts, when the issues related
to train availability were resolved, the effluent TN was significantly lower than predicted duringdesign, even with much lower influent BOD5.
2670
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
17/20
CONCLUSIONS
The information garnered through the specialized process control testing program has proved
invaluable in helping the operations staff understand and control the processes that impact TN
removal. At the RPWRF, the primary process controls available to minimize TN concentrationsare the amount of air delivered to the aeration basins and the recycle ratio. And, the recycle ratio
is very sensitive to the DO concentration in the aeration basins as it impacts the onset of anoxic
conditions. With all the experience gained during these extremely abnormal periods of plantoperation, RPWRF staff can now maintain the process to produce TN concentrations lower than
predicted during the design.
In the time since the end of October 2005 and June 2006, the RPWRF has produced an effluent
that has a lower TN concentration than predicted in the design, despite having only 50 to 65
percent as much influent BOD5 as projected during design. Figure 24 presents the effluent TNconcentrations through May 2006.
Figure 24 RPWRF Effluent TN Concentration
RPWRF Effluent Total N
0
2
4
6
8
10
04/26/05
05/26/05
06/26/05
07/26/05
08/26/05
09/26/05
10/26/05
11/26/05
12/26/05
01/26/06
02/26/06
03/26/06
04/26/06
05/26/06
TotalN(mg/L)
Anticipated 7.5 m g/L
During the period of time represented in Figure 24, average daily plant flow has varied between
1.5 and 2.7 MGD, and yet no trains were taken out of service. That is, all six trains wereconstantly providing treatment, even if the permeate pumps were not taking any liquid for
discharge. Based on the process control testing program, RPWRF staff have determined that a
recycle ratio of 3.5 results in the optimum TN removal efficiency at their facility.
Getting to the point where TN removal can be consistently controlled has required that the DO
concentrations in the aeration basin be allowed to fall to levels well below that assumed in the
design. The current diurnal DO concentration in the aeration basins is shown in Figure 25.
2671
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
18/20
Figure 25 RPWRF Aeration Basin Diurnal DO Concentration
DO Aeration basin (May 2006)
0
1
2
3
4
5
6
7
8
9
10
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0 1 2 3 4 5 6 7 8
Time (hr)
DO
(mg/L)
Design 1.5 mg/L
The DO concentration in the aeration basins is now so low that plant staff must continually
monitor to ensure that ammonia nitrogen concentrations do not begin to increase in the planteffluent. Whenever process control testing indicates that effluent ammonia concentrations are on
the increase the plant staff will increase the amount of air being sent to the aeration basins.
One thing that may have contributed to the improved success of TN removal is the fact that the
influent BOD5 has increased during this period of time, although it is still nowhere near the
concentration used in the design. Figure 26 shows the influent BOD5 concentration throughMay of 2006.
Figure 26 RPWRF Influent BOD5 Concentration
RPWRF Influent BOD5
0
50
100
150
200
250
300
350
04/26/05
05/26/05
06/26/05
07/26/05
08/26/05
09/26/05
10/26/05
11/26/05
12/26/05
01/26/06
02/26/06
03/26/06
04/26/06
05/26/06
BOD5(mg/L)
Design = 300 mg/L
2672
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
19/20
The RPWRF plant staff continues to look for ways to improve TN removal in their facility.
Some of the things presently being considered are actually removing trains from treatmentservice when they are taken out of permeating service. This would amount to turning off the
RAS pumps and allowing a train to sit in a mode similar to an aerobic digester. This would
increase the effective loading to the other basins, and could both increase the OUR and decrease
the amount of DO carried back to the RAS box by the RAS pumps. This could allow for anincrease in the recycle ratio and further reduce effluent TN concentrations.
How has the RPWRF performed relative to some of its APP and AZPDES permit limits? Thefollowing Figures 27 through Figure 29 present some of those data.
Figure 27 RPWRF Effluent BOD5 Concentration
RPWRF Effluent BOD5
0
5
10
15
20
25
30
04/26/05
05/26/05
06/26/05
07/26/05
08/26/05
09/26/05
10/26/05
11/26/05
12/26/05
01/26/06
02/26/06
03/26/06
04/26/06
05/26/06
BOD5(mg/L)
Permit limit 30 mg/L
Figure 28 RPWRF Effluent Fecal Coliform
RPWRF Effluent Fecal coliform
0.0
5.0
10.0
15.0
20.0
25.0
04/26/05
05/26/05
06/26/05
07/26/05
08/26/05
09/26/05
10/26/05
11/26/05
12/26/05
01/26/06
02/26/06
03/26/06
04/26/06
05/26/06
Count(MPN/100ml)
Maximum 23 MPN/100 ml
2673
WEFTEC.06
Copyright 2006 Water Environment Foundation. All Rights Reserved
8/3/2019 Optimizing Nitrogen Removal in a MBR
20/20
Figure 29 RPWRF Effluent Turbidity
RPWRF Effluent Daily Average Turbidity
0.00
0.40
0.80
1.20
1.60
2.00
04/26/05
05/26/05
06/26/05
07/26/05
08/26/05
09/26/05
10/26/05
11/26/05
12/26/05
01/26/06
02/26/06
03/26/06
Turbidity(NTU)
Limit = 2 NTU
Perme ate coupling failure
The membranes themselves have performed well within their design capacity to date. The trans-
membrane pressures (TMP) have remained at about -1.5 psi at the maximum flux rates. The
maximum allowable TMP is -7.0 psi, where the associated permeate pumps will reduce theirtargeted flux rates to maintain no more than -7.0 psi TMP. At -8.0 psi TMP the associated
permeate pump will shut down.
ACKNOWLEDGEMENTS
Special acknowledgment goes to Mr. Frank Gall and the rest of the PCWMD outlying facilities
management and plant staff for their efforts to support and perform the additional process controltesting, as well as their willingness to share that information for inclusion in this paper. Dr. John
Bratbys efforts to provide Biowin model results and process insights during the startup period
are also greatly appreciated.
REFERENCES
Monitoring Data PCWMD Randolph Park WRF, April 2005 through May 2006Process Control Data PCWMD Randolph Park WRF, April 2005 through May 2006
WEFTEC.06