EVALUATING AND OPTIMISING THE PERFORMANCE OF CHLORINE CONTACT TANKS USING CFD Jonathan Church, Jason Colton – H2OPE Noel Roberts –Greater Wellington Regional Council ABSTRACT The bacteriological compliance criterion 2A in DWSNZ2008 states that chlorine contact time must be more than 30 minutes, taking into account short circuiting in the tank. There are three methods available to water suppliers to demonstrate this. The first is by using a baffle factor, which is a gross simplification. The second is by using tracer testing, which is very laborious and the third is by using computational fluid dynamic (CFD) modeling. Until recently CFD has been cost prohibitive, principally due to the licensing cost of CFD software. However, open source modeling suites are now available which in combination with computing advances now make CFD affordable for water suppliers. This allows the benefits of CFD to be fully realised in evaluating and optimising the performance of chlorine contact tanks. CFD can be used to accurately model the exit age distribution for a whole range of flow and level scenarios, giving accurate T 10 contact times. It can identify dead zones; it can be used to optimise baffle design, inlet and outlet arrangements and to select optimum dosing and sampling point locations. This paper will demonstrate the use of all these CFD capabilities on a full scale chlorine contact tank using an open source CFD tools suite called CAE-Linux. KEYWORDS Baffle Factors, Computational Fluid Dynamics, T 10 , Chlorine Contact Time, Water Treatment 1 INTRODUCTION In order to fully comply with criterion 2A of the New Zealand Drinking Water Standards 2005, revised 2008 (DWSNZ2008) the chlorine contact time must be more than 30 minutes, taking into account short circuiting in the chlorine contact tank. In order to do this a T 10 contact time has to be calculated for a range of conditions, in particular the worst case condition – minimum level and maximum flow. The T 10 contact time is defined in the USEPA Guidance Manual for Disinfection Profiling and Benchmarking (2003), as the minimum detention time experienced by 90 percent of the water passing through the tank. The T 10 can be estimated by using a baffle factor (Equation 1). T his factor is the ratio between the T 10 for a particular tank and the theoretical maximum detention time of that tank. A list of baffle factors and the corresponding baffle design is provided in Table 1. Equation 1 Where: Equation 2
Transcript
EVALUATING AND OPTIMISING THE PERFORMANCE OF CHLORINE CONTACT TANKS USING CFD
Jonathan Church, Jason Colton – H2OPENoel Roberts –Greater Wellington Regional Council
ABSTRACT
The bacteriological compliance criterion 2A in DWSNZ2008 states that chlorine contact time must be more than 30 minutes, taking into account short circuiting in the tank. There are three methods available to water suppliers to demonstrate this. The first is by using a baffle factor, which is a gross simplification. The second is by using tracer testing, which is very laborious and the third is by using computational fluid dynamic (CFD) modeling.
Until recently CFD has been cost prohibitive, principally due to the licensing cost of CFD software. However, open source modeling suites are now available which in combination with computing advances now make CFD affordable for water suppliers. This allows the benefits of CFD to be fully realised in evaluating and optimising the performance of chlorine contact tanks.
CFD can be used to accurately model the exit age distribution for a whole range of flow and level scenarios, giving accurate T10 contact times. It can identify dead zones; it can be used to optimise baffle design, inlet and outlet arrangements and to select optimum dosing and sampling point locations.
This paper will demonstrate the use of all these CFD capabilities on a full scale chlorine contact tank using an open source CFD tools suite called CAE-Linux.
KEYWORDS
Baffle Factors, Computational Fluid Dynamics, T10, Chlorine Contact Time, Water Treatment
1 INTRODUCTION
In order to fully comply with criterion 2A of the New Zealand Drinking Water Standards 2005, revised 2008 (DWSNZ2008) the chlorine contact time must be more than 30 minutes, taking into account short circuiting in the chlorine contact tank.
In order to do this a T 10 contact time has to be calculated for a range of conditions, in particular the worst case condition – minimum level and maximum flow. The T10 contact time is defined in the USEPA Guidance Manual
for Disinfection Profiling and Benchmarking (2003), as the minimum detention time experienced by 90 percent of the water passing through the tank. The T 10 can be estimated by using a baffle factor (Equation 1). This factor is the ratio between the T 10 for a particular tank and the theoretical maximum detention time of that tank. A list of baffle factors and the corresponding baffle design is provided in Table 1.
0.1 None, agitated basin, very low length to width ratio, high inlet and outlet flow velocities. Can be approximately achieved in flash mix tank.
Poor 0.3 Single or multiple un-baffled inlets and outlets, no intra-basin baffles.
Average 0.5 Baffled inlet or outlet with some intra-basin baffles.
Superior 0.7 Perforated inlet baffle, serpentine or perforated intra-basin baffles, outlet weir or perforated launders.
Perfect
(plug flow)
1.0 Very high length to width ratio (pipeline flow), perforated inlet, outlet, and
intra basin baffles.
The problem with using baffle factors is that they are a gross simplification and the resulting T 10 contact time is an assumption.
Greater Wellington Regional Council (GWRC) wished to determine an actual T 10 contact time for the chlorine contact tanks at their Te Marua and Wainuiomata water treatment plants (WTP) in order to demonstrate compliance with the requirements of Criterion 2A in the DWSNZ2008.
Two methods are available to provide a demonstrable test result for T 10 contact times, these are:
Tracer testing
Computational Fluid Dynamic (CFD) modeling
1.1 TRACER TESTING
Tracer testing is where a chemical is added to the water entering the chlorine contact tank and the change in concentration at the exit of the tank is measured over time. The shape of the resulting concentration versus time graph provides insight into the amount of short-circuiting and dead zones within the tanks and actual T 10 and baffle factors can be determined. There are two methods of tracer testing: the slug-dose method and the step-dose method. The easiest method to use is usually the step-dose method since chemicals that are already in use on the plant, for example fluoride, can be used as the tracer. The step-dose method entails dosing of a tracer
chemical at a fixed dose until the concentration at the exit of the tank reaches a steady-state level (the concentration dosed). A graph of tracer concentration at the exit of the tank (C) / dosed tracer concentration (C0) is plotted and from this the T10 can be identified (Figure 1). This type of graph is also known as cumulative distribution function curve.
0 20 40 60 80 100 120 140 160
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
C/C
0
Time (min)
T10
Figure 1: Step Dose Tracer Test - Cumulative Distribution Curve
Tracer testing is a proven technique for demonstrating T10 contact times. However, it is time consuming and can be expensive since a minimum of four tests are recommended - covering different flow and level conditions. It
can also have a significant impact on plant operations. Flow and level need to be fixed for each test, which can often mean inhibiting filter backwashing. Furthermore there may be some situations where a tracer is not readily available e.g. the fluoride dosing point may be after the chlorine contact tank. In which case an alternative tracer has to be used, increasing the cost.
1.2 COMPUTATIONAL FLUID DYNAMIC (CFD) MODELLING
Computational fluid dynamics (CFD) is a branch of fluid mechanics that uses numerical methods and algorithms to solve and analyse problems that involve fluid flows. Computers are used to perform the calculations required to simulate the interaction of liquids and gases with surfaces defined by boundary
conditions.
CFD is used widely overseas in a range of industries but its use in New Zealand, particularly in the water industry, has been very limited. The reason for this is both cost and resources. Up until recently CFD software had extremely high licensing costs which meant engineering and consulting companies were reluctant to invest in it. This in turn meant that there were very few people capable of running CFD simulations.
The recent release of an open source CFD software suite (CAE-Linux) allied to advances in computing power has changed this making CFD very price competitive compared to tracer testing (30-40% cheaper than tracer testing).
In order to use CFD to determine T 10 contact times a number of steps have to be followed:
1. Construct a 3D model of the chlorine contact tank;
2. Convert the model to a bubble mesh of the contact tank;
3. Run a steady state velocity solver;
4. Run a mass transport solver to simulate a step-dose tracer test
Once a model is constructed it is relatively straight-forward to run multiple flow scenarios.
1.3 T10 CONTACT TIME DETERMINATION FOR GWRC CHLORINE CONTACT TANKS
In 2007 step-dose tracer testing, using fluoride, was used to determine the T 10 chlorine contact time for a range
of level and flows at the Te Marua WTP (Figure 2). The intent was to do the same at the Wainuiomata WTP (Figure 3) but this proved more difficult because the fluoride was dosed after the chlorine contact tank. The project got “parked” and GWRC continued to rely on a baffle factor to estimate the T 10 chlorine contact time for the Wainuiomata WTP. The recent availability of CFD allowed GWRC to reinstate the project, using CFD rather than tracer testing to determine an actual T 10 chlorine contact time. A CFD model was also developed for the Te Marua WTP chlorine contact tank so that the CFD results could be compared to the tracer testing results.
Figure 2: Te Marua WTP – Chlorine Contact Tank Figure 3: Wainuiomata Chlorine Contact Tank
2 CFD METHODOLOGY
2.1 SELECTING SCENARIOS
There are two main variables to consider when selecting scenarios to model. The first is plant flow and the second is chlorine contact tank level. It is always necessary to model the maximum plant flow since this is the worst case for contact times. Similarly with chlorine contact tank level it is critical to select the worst case, which is the minimum operating level. This should be determined from SCADA trends.
2.1.1 WAINUIOMATA WTP
The plant maximum flow was 60ML/d and two other flow conditions - 20ML/d and 40ML/d were selected. The minimum operating level from SCADA data was found to be 51%. The maximum operating level of 96% was also selected along with a mid-point value – 76%. The resulting matrix of scenarios is shown in Table 2.
Table 2: Modeling Scenarios for Wainuiomata WTP Chlorine Contact Tank
Flow (Ml/d)Level (%)
60 40 2096 A1 A2 A3
76 B1 B2 B3
51 C1 C2 C3
2.1.2 TE MARUA WTP
The scenario selected for comparison with the tracer testing was 60ML/d and 75% level.
2.2 CONSTRUCTING THE MODELS
Models were constructed for both contact tanks from construction record drawings using Salome Meca software (part of the CAE-Linux suite). Operations staff confirmed that these were still an accurate record and that no
significant changes had been made to the contact tanks. A number of assumptions were used in the construction of the models. These were as follows:
The membrane baffle curtains were assumed to act as rigid fixed walls which allowed no short
circuiting;
Pressure was assumed to be constant at one atmosphere;
The scenarios assumed a steady in-out flow of water through the contact tank;
The level in the contact tank was fixed.
2.2.1 WAINUIOMATA WTP
The Wainuiomata WTP chlorine contact tank is a circular tank, 16.9m diameter and 6m tall with a volume of 5170m3. The reservoir is baffled by two parallel membrane curtains 25m long running from opposing walls. These curtains are fixed to the floor using a full length skirt which was factory welded to the main membrane curtain. Water enters on the right of the reservoir through a 1200mm CLS pipe angled at 45 degrees from underground towards the floor. The pipes are flush with the floor of the reservoir, providing an elliptical exit. The basic wireframe model was built in a simplistic form, shown in Figures 4 and 5.
The wireframe represented the edges of an available incompressible volume for the water to flow through. Three geometries were created with different heights to represent the reservoir at 96% (5.76m), 76% (4.56m) and 51% (3.06m) full.
Figure 4: Plan View of Wainuiomata WTP CCT Figure 5: Isometric View of Wa inuiomata WTP CCT
2.2.2 TE MARUA WTP
The Te Marua WTP chlorine contact tank is 5ML in volume. It is of rectangular design with a high level weir inlet and a low level pipe outlet. There is a single curtain baffle bisecting the tank. The level in the tank can vary from 75% to 99%.
Figure 6: Isometric View of Te Marua WTP CCT
2.3 MESH BUILD
Using the NetGen 3D method in Salome-Meca the two contact tanks were then meshed to represent individual ‘bubbles’ of volume within the reservoir, as shown in Figures 4 and 5. For the Wainuiomata WTP chlorine contact tank in the 96% full scenario this created 299,434 bubbles. Each bubble of volume is a cell in which the CFD mathematics are performed.
The meshed reservoirs were then exported to the OpenFoam software.
OpenFoam, an open source CFD software package was used for the calculations. The package uses fully customisable and modular mathematical solvers to determine velocities, chemical reactions, temperatures etc within a pre-defined mesh for given inlet, outlet and wall conditions.
The solvers can be modified as desired to solve specific mathematical problems. Once a solver has been created it is run against a set of initial start variables or conditions, where it solves each mesh ‘volume’ for a result at each time step.
For all scenarios the following steps were taken:
Each of the meshes was decompiled to convert the visual mesh into a list of coordinates.
Steady State Velocity: For each scenario the steady state velocity was determined using a Navier-Stokessolver, with the inlet velocity as the only starting condition.
Mass Transport: The resulting steady state velocities were used as a starting condition for a modified
turbulent convection-diffusion solver. A step tracer test was simulated by dosing 1mg/L of a tracer into the inlet of the tank at time (T) = 0.
2.4.1 STEADY STATE VELOCITY
An example of the steady state velocity profile for the Wainuiomata WTP chlorine contact tank is shown in
Figure 9. This was for the 60ML/d and 96% level scenario (A1). The profile shows higher velocities at the inlet and outlet pipe as would be expected but it also demonstrates the increase in velocity around the turns and clearly shows up some very low velocity areas. The CFD viewer can also show velocity lines. The same scenario is shown in this format in Figure 10.
Figure 9: Wainuiomata WTP CCT Velocity Distribution (60ML/d and 96% level - Scaled 0 to 0.1m/s)
Figure 10: Wainuiomata WTP CCT Velocity Lines(60ML/d and 96% level - Scaled 0 to 0.1m/s)
2.4.2 MASS TRANSPORT
A tracer test was simulated by adding 1mg/L of an inert tracer to the inlet of the tank. The simulation was run until the tank outlet reached steady state i.e. 1mg/L. A time sequence of the tracer concentration in the Wainuiomata chlorine contact tank for 60ML/d and 96% level is shown in Figure 11.
The simulated tracer concentration at the exit of the tank was used to plot a cumulative distribution function curve (C/Co vs time) from which the T 10 could be calculated. A baffle factor was then calculated by dividing the calculated T 10 contact time by the theoretical contact time.
1 Minute 10 Minutes 30 Minutes 60 Minutes
Figure 11: Examples of Concentration Change over time for Wainuiomata WTP CCT (60ML/d, 96%)
3 RESULTS
3.1 CFD VALIDATION
Despite the fact that CFD is a well proven technique it was desirable to validate the model outputs against actual tracer test results. This was done using results from the tracer testing at Te Marua WTP. The scenario used was a flow of 60ML/d at a chlorine contact tank operating level of 75%. The cumulative distribution function curves for the CFD model and the tracer test are shown in Figure 12. A C/C0 value of 0.1 gives the T 10 contact time.
0 20 40 60 80 100
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
C/C
0
Time (mins)
CFDTracer Testing
Figure 12: Te Marua WTP CCT – Cumulative Distribution Functions – Tracer Test & CFD Outputs
The T 10 contact time calculated from the tracer testing was 68.9 minutes and the T10 contact time calculated from CFD was 67.2 minutes. Regression analysis showed that the two plots had an R2 of 0.99.
3.2 WAINUIOMATA WTP CHLORINE CONTACT TANK EVALUATION
3.2.1 T10 CONTACT TIMES
The output of the tracer test simulations for all scenarios are shown as cumulative distribution function curves in Figure 13. The calculated T 10 contact times are shown in Table 3.
Table 3: Wainuiomata WTP CCT T10 Contact Times For All Scenarios
T10 Contact Time (mins)Level (%)
60 ML/d 40 ML/d 20 ML/d96 49 75 159
76 37 57 127
51 28 43 100
It can be seen from Figure 13 and Table 3 that when the reservoir is operated at maximum flow (60 ML/d) and
minimum operating level (51%) that the T10 contact time is less than 30 minutes. In order to more clearly demonstrate the range of conditions for which the T 10 is less than 30 minutes the T10 values from Table 3 were plotted against flow (Figure 14). These results indicated the following operational limits would be required to ensure compliance with the requirements Criteria 2A in DWSNZ2008:
When the tank is operating at minimum level the plant flow must be limited to 57ML/d.
When operating at maximum plant flow of 60ML/d the tank level must be greater than 60% full.
20 25 30 35 40 45 50 55 60
0
10
20
30
40
50
60
70
80
90
100
110
120
T1
0 (
min
)
Plant Flow (ML/d)
96% Level76% Level51% Level
Figure 14: Wainuiomata WTP CCT - T10 Contact Time Relationship to Flow
3.2.2 CALCULATED BAFFLE FACTORS
The calculated baffle factors for all the scenarios are shown in Table 4. This shows a range in baffle factors from 0.34 to 0.46. The baffle factors are also shown graphically in Figure 15 which is the cumulative distribution function normalized using the theoretical detention time.
Table 4: Wainuiomata WTP CCT Baffle Factors For All Scenarios
Baffle FactorLevel (%)
60 ML/d 40 ML/d 20 ML/d96 0.34 0.36 0.40
76 0.35 0.37 0.41
51 0.39 0.42 0.46
This data highlights two issues with using the standard baffle factor definitions shown in Table 1. The first is that when a baffle factor is selected it is applied to all flow and level scenarios. In reality the baffle factor varies with both flow and level. The second issue is that given the range in actual baffle factors it is quite difficult to select a baffle factor which will cover all scenarios. For example, the baffle factor that has been selected for the Wainuiomata WTP chlorine contact tank was 0.4. This was selected because it was felt that the tank fell between
the 0.3 and 0.5 definitions. It can be seen from Table 4 that under many operating scenarios the baffle factor was underestimating short-circuiting.
3.2.3 IDENTIFYING DEAD ZONES
The modeling exercise highlighted a number of dead zones within the chlorine contact tank, as demonstrated in Figure 16. Whilst there is not a lot that can practically be done to eliminate these dead zones in an existing tank CFD can be used to prevent excessive dead zones when designing new chlorine contact tanks.
Figure 16: Illustration of Dead Zones in Wainuiomata WTP CCT.
3.2.4 OPTIMISING SAMPLE POINT LOCATION
One area where dead zone identification is important for existing tanks is in the locating of sample points. If the tank is being sampled from a dead zone then the sample will not be representative under changing conditions. If the sample is being used for pH and or chlorine measurement and control then the process control will be very difficult to tune effectively. In order to achieve good process control performance a homogenous sample is
required. Using CFD a sample location that provides a homogenous sample with a delay time in the order of 3-5 minutes can be identified, giving the best chance of optimising process control. In order to demonstrate this the optimum sample point location (for control purposes) was identified for the Wainuiomata WTP chlorine contact tank (Figure 17). The sample would be collected using a pump mounted on the tank roof drawing from a water depth of 4.5m.
Figure 17: Illustration of Optimum Sample Location Wainuiomata WTP CCT.
3.2.5 EV ALUATING MODIFICATIONS
In Section 3.3.1 operating limits were identified that would ensure compliance with the DWSNZ2008. As an alternative to imposing operational limits it was decided to investigate modifications to the inlet to the chlorine contact tank at Wainuiomata using CFD to see if these would increase the T 10 contact time.
A number of options were considered and are demonstrated in Figure 18.
Current Baffle Plate Riser Side Entry
Figure 18: Schematic of Inlet Modifications Evaluated
The resulting T 10 contact time for the 60ML/d and 51% level scenario are shown in Table 5. This shows that simple modifications to the inlet would improve the T 10 contact time sufficiently to preclude the need to impose operational limitations. The option selected for implementation at the Wainuiomata WTP was the baffle platesince this was the easiest retrofit.
This work has shown that CFD is a viable alternative to tracer testing in order to demonstrate compliance with Criteria 2A of DWSNZ2008. In fact CFD has many advantages over tracer testing. It is now cheaper than tracer testing due to the availability of open source software, it has no operational impact and can be used to provide
additional value over and above the determination of T 10 contact times and baffle factors. It can be used to identify dead zones, to identify optimum sample locations and to evaluate and optimise tank modifications. Such modifications may preclude the need for alternative disinfection processes such as UV.
This work has also demonstrated that the use of standard baffle factors is a gross simplification and that their use can result in an underestimate of short-circuiting in chlorine contact tanks.
ACKNOWLEDGEMENTS
The authors with to thank Greater Wellington Regional Council for permission to publish this paper.
REFERENCES
Brown L., Jacobsen F. (2009) ‘Optimise Tank Design Using CFD’ Proceedings of 72nd Annual Victorian Water Industry Engineers and Operators Conference, Bendigo pp 89-95
Ministry of Health. (2008). Drinking water standards for New Zealand 2005 (Revised 2008).
USEPA. (2003). LT1ESWTR Disinfection Profiling and Benchmarking. Technical Guidance Manual.
