Even sewage tunnels can benefit from CFD simulations

1 2 2
Y Alostaz , A Carciumaru , T Werner
1.
AECOM USA, Inc, USA
2.
AECOM Canada Ltd., Canada

ABSTRACT

This paper describes the mechanical ventilation schemes of a large sewer tunnel located
in the Canadian York and Durham Regions. One of the major requirements for this
sewer tunnel is to eliminate odor release to the ambient atmosphere. To satisfy this
requirement, a ventilation system is developed in which odorous air is moved, within the
sewer trunk headspace, to an Odor Control Facility (OCF) for treatment before being
released to the atmosphere. The OCF is located at an intermediate location along the
alignment of the trunk sewer; hence, odorous air is moved in co-current manner within
the sewer line located upstream from the OCF. On the other hand, odorous air is moved
in counter current manner within the sewer line located downstream from the OCF.

1. INTRODUCTION

The new Southeast Collector Trunk Sewer (SeC) is approximately 15 km in length and
about 3 m in diameter. The trunk sewer will convey wastewater from existing York
collection system in the Town of Markham to Duffin Creek Wastewater Treatment Plant
in the Town of Pickering. The trunk sewer tunnel will be excavated using a tunnel
boring machine and precast liner will be installed inside the tunnel. The SeC will include
several access shafts, two drop structures with baffles, connection chambers, and a
metering facility.

One of the major requirements for the SeC is to prevent odor release to the ambient
atmosphere. To satisfy this requirement, a ventilation system is developed in which
odorous air is moved, within the sewer trunk headspace, to an Odor Control Facility
(OCF) for treatment before being released to the atmosphere. The OCF is located in the
vicinity of access shaft 9, almost midway between the start and end points of the SeC.

Due to the location of the OCF, the odorous air upstream of access shaft 9 is moved in
the same direction as the liquid flow, while the odorous air downstream of access shaft 9
is moved in the opposite direction to the liquid flow. The goal of the ventilation system
is to maintain a negative pressure, relative to ambient atmosphere, within the air
headspace inside the tunnel. Several sensors are added along the length of the tunnel in
order to monitor various parameters inside the tunnel; those parameters are used to adjust
the odorous air extraction rate at the OCF.

The proposed ventilation system was studied using Computational Fluid Dynamics
(CFD) modeling. The three-dimensional CFD models utilized multiphase flow in order

© BHR Group 2011 ISAVT14 493

to account for the interaction between the liquid and the air within the tunnel space. The
results of the CFD simulations were used to validate the ventilation concept and guide
the development of the ventilation scheme.

2. ODOROUS AIR AND SEWER TUNNELS

2.1. Air Movement Dynamics in Sewer Tunnels

Gravity sewers are designed to flow partially full in a similar manner to an open channel,
and are typically designed for projected peak liquid flow with a minimum velocity
exceeding 0.75 m/s at a flow depth (d) to sewer diameter (D) ratio (d/D) of 0.5 to 0.75.
The minimum liquid velocity is crucial in preventing deposition of solids from the
wastewater, and promotes entrainment of fresh air into the wastewater. A well mixed
and oxygenated wastewater remains aerobic [1, 2] and discourages anaerobic liquid
phase hydrogen sulfide (H2S) generation. Hydraulic turbulence of wastewater rich in
dissolved H2S results in release of H2S gas to the sewer headspace, and this gas might
leak to ambient atmosphere.

Airflow in sewers is governed by several mechanisms related to how the air space in the
sewer reacts to and interacts with its environment. Despite recognition of these
phenomena, there has been little formal work to characterize or quantify sewer
ventilation or the mechanisms involved. A report published by the Water Environment
Research Foundation (WERF) [3] states: In spite of its importance to wastewater odor
and corrosion, wastewater collection system ventilation is generally understood at only a
crude level among wastewater professionals.

Existing research on sewer ventilation dynamics had focused on typical sewers where air
moves naturally co-current to wastewater flow, and where there are numerous airflow
pathways for air to be either introduced or to escape, depending on the sewer s pressure
relative to atmosphere. These airflow pathways are primarily air introduced from
connected laterals and other trunk sewers, and air that can leak in or out from seams and
gaps at access manholes. Current literature [3, 4, 5, 6, 7, 8, 9, 10, 11] suggests that sewer
ventilation mechanisms presented below do exist in a typical sewer line.

2.1.1. Friction Drag by Flowing Wastewater

Drag generated by the liquid movement is most likely to move odorous air in the same
direction as the liquid (co-current flow). The drag forces are generated at the liquid-gas
interface, hence, the larger the interface area the larger the drag forces. For a circular
trunk sewer, this condition occurs when the sewer tunnel is half full. The induced air
flow caused by friction drag is the main parameter that is modeled to arrive at estimates
of the natural flow in a sewer headspace.

Liquid flowing through drop structures tend to entrain significant amounts of air, and that
air is moved downstream of the drop structure causing pressurization and potential for
odor release to atmosphere. As the sewage falls through the drop structure, liquid
droplets form, and that increases the air-liquid interface surface. Hence, more air is
entrained into the sewage thus causing pressurization in the downstream area of the drop
structure.

2.1.2. Changes in water velocity and depth

Changes in water depth and water velocity create a transient condition in which
increasing flow into large trunk sewers causes displacement of air out of the sewer. This
process can be quite pronounced on unventilated sewers. In ventilated sewers, the

494 © BHR Group 2011 ISAVT14

ddisplacement is ussually slow relativve to the amount of air removed by
b odor control
s
system fans.

22.1.3. Temperaturre Gradients and Buoyancy

B
S
Some of the earliest work in sewer ventilation
v suggesteed the use of inducced draft stacks
t promote better natural
to n ventilation of sewers [12]. Laarge and tall stackss could produce
r
relatively good airrflow if sufficient temperature diffeerentials exist betw ween the sewer
a
atmosphere and thee outside air. How wever, in recent yeaars, many deep trunk sewers have
b
been categorized as
a closed systems with fully covereed and closed mannholes and very
f
few branch sewers. Hence, there is little
l opportunity for
fo significant odorr leakage due to
b
buoyancy and temmperature gradientss. It is interestinng to note, howevver, that closed
s
systems tend to agggravate pressurizaation from inducedd airflow, becausee the sewer has
f
fewer significant aiir leakage points.

22.1.4. Wind Educction

W
Wind eduction pheenomena occur as a result of wind moovement around seewer ventilation
s
stacks. Modern trunk sewers donn t add air stackss and induction stack s heads to
d
deliberately dischaarge sewer air to thee atmosphere. Thee majority of the veentilation stacks
a fitted with onee way valve mechaanisms that would let fresh air into the
are t trunk sewer
a prohibit releasee of odorous air intto the atmosphere.
and

22.1.5. Ventilation n Bottlenecks

V
Ventilation bottlennecks are constricctions in the air headspace
h of the sewer. These
b
bottlenecks can rannge from a slight reeduction in air headdspace area, to a to
otal blockage of
a
airflow from a full pipe. Ventilationn bottlenecks estabblish the tendency y of induced air
f
flow to cause prressurization of the t headspace. Fugitive emissions arise from
p
pressurization of thhe sewer
h
headspace resultinng from
f
frictional drag impposed on
t
the air space by flowing
w
wastewater. Heeadspace
c
constrictions deveelop as
l
liquid depth increases
w
which could resullt in the
d
development of positive
p
pressures that increeases the
p
potential for odor release.
A
An illustration of the
a
airflow in a typicaal sewer
i shown in Figure 1.
is Figure 1. Typical
T sewer system

22.1.6. Mechanicaal Ventilation

M
Mechanical ventilaation of sewer tunnnels can be accompplished using fans that extract the
o
odorous air from thhe tunnel headspacce. The intent herre is to maintain sllightly negative
p
pressure, relative to the ambient atmosphere, withiin the tunnel heaadspace. This
t
technique can be implemented effiiciently in tight sewer systems with w limited or
c
controlled vent loccations. Air extraaction tests were performed
p [8] on large diameter
s
sewers in the Uniteed States, and it waas shown that:
a
a. Negative presssures could be creeated within the seewer headspace wiith the use of a
fan unit,
b
b. Negative presssures could propaagate for long disstances in the sew wer, particularly
when the seewer had relatively well sealed manholesm and no major lateral
connections.

© BHR Group 2011 ISAVT14 495

As a result of the SeC s tight system, the impacts of temperature gradients, stack
effects and buoyancy, and wind eduction become far less important than accounting for
friction drag due to the movement of wastewater and assessment and handling of
ventilation bottlenecks. The SeC trunk sewer has a constant wastewater flow rate
throughout the system, few changes in sewer alignment, and constant slope in the sewer
tunnel. The result is a consistent headspace in which to convey air in co-current and
counter-current directions to the OCF. Hence, the relative importance of the typical
sewer ventilation mechanisms for the new SeC design can be summarized as follows:
Friction drag by flowing wastewater is the primary control for design,
Changes in water velocity and depth are less significant than friction drag, and should
not play a major role in the ventilation design,
Temperature gradients and buoyancy effects are negligible but will be used to
establish the maximum negative pressure to be maintained inside the trunk sewer,
Wind eduction should have negligible effect on the sewer tunnel ventilation,
Ventilation bottlenecks will be located at specific locations in order to control
odorous air movement inside the sewer tunnel.

2.2. Sewer Tunnel Ventilation Models

The primary motive force for air movement in sewers is air drag at the liquid surface.
Air velocity contours are typically assumed to be greatest and close to the liquid velocity
at the liquid surface, and decreasing quickly away from the liquid surface. Pressure
development in the sewer headspace air is mainly due to air velocity, volume of sewer
headspace air, and the overall air-tightness of the trunk sewer. Sewer pressures are
naturally low when manhole covers have open pick holes and/or several connecting
sewer branches. Such conditions are analogous to developing pressure in an air duct with
holes in it. Sealed covers and few connected laterals tend to allow pressure to increase
and extend throughout the system for several miles.

The published WERF research [3] evaluated three models currently used by designers to
predict the sewer ventilation rates:
Empirical extrapolation model by Pescod and Price (1981). This model was derived
from a 300 mm diameter laboratory scale gravity pipe.
Computational fluids dynamic (CFD) derived model by Edwini-Bonsu and Steffler
(2004). This model was derived from CFD analysis and is based on two dimensional
pipe geometry and wastewater flow.
Thermodynamic models by Olson et al. (1997). This model is based on work done
on an industrial sewer.

One of the earliest and more detailed researches on sewer ventilation was carried out by
Pescod and Price [9]. Most of the current research work relies on their work as the basic
foundation for further analysis. Pescod and Price conducted relatively small-scale
experiments with low liquid flow speed on induced air flow in sewers, and that was the
major limitation in the applicability of their work to large sewer tunnels. Fast flowing
wastewater and vertical drops would greatly increase drag-induced flow in the
surrounding air. Conversely, slow-moving laminar streams will exert minimal drag on
the air and move relatively small volumes of air.

While the empirical model by Pescod and Price proved to be most accurate of the three
models referenced in the WERF research, no model matched field data particularly well.
The Pescod and Price model tended to overestimate sewer ventilation rates but was very
easy to use. The WERF study concluded that more research and model development was

496 © BHR Group 2011 ISAVT14

needed. There has been no work that investigated the parameters of moving odorous air
in a counter-current manner.

The CFD model developed by Edwini-Bonsu and Steffler [5] was based on two-
dimensional simulations normal to the direction of the liquid flow, and the drag at the
liquid/air interface is viewed as a Couette Flow . The Couette flow refers to the laminar
flow of viscous fluid between two parallel plates where one plate is moving relative to
the other one. The ability of the model to predict air flow in various sewer lines is
limited due to its inability to account for various conditions that might occur inside a
trunk sewer.

3. PROJECT
DESCRIPTION

3.1. General Description

The new SeC Trunk
passes through the
Regional Municipalities
of York and Durham,
refer to Figures 2 and 3.
The sewer tunnel starts at
the diversion facility in
York Region and ends at
the connection chamber in
Durham Region. The
existing York-Durham
Sewage System (YDSS) Figure 2. Trunk sewer plan
connects to the new SeC
at the diversion facility and at the connection chamber. The gravity sewer tunnel has a
constant slope of 0.18% between the diversion facility and access shaft 4; however, the
slope of the tunnel is reduced to 0.1% downstream of access shaft 4. The new SeC does
not connect to any sewer branches along its entire length. Only one minor sewer branch
is connected in the vicinity of access shaft 1. The main components of the SeC sewer
line are described below.

Figure 3. Trunk sewer profile

© BHR Group 2011 ISAVT14 497

33.2. Diversion Faccility
T diversion facillity consists of acceess shafts 13A, 13B
The B and 13C. The access shafts are
8 to 9.6 m in diaameter, and are aboout 27 m deep. Thhe diversion facilityy is designed to
8.8
s
split the flow between the existingg and the new SeeC sewers at norm mal operational
c
conditions. The facility also has the capability
c to divertt upstream flows innto either one of
t sewers at speciific operational andd flow conditions. The diversion faciility also allows
the
f the isolation of either sewer for maintenance
for m by clossing sluice gates thaat are located at
t facility. This facility
the f represents the upstream end of the SeC sewer tunnel.
t Access
s
shafts 10 through 12 are located between
b the diverssion facility and access
a shaft 9,
d
described below.

33.3. Access Shaft 9 and Metering Chamber

C
A
Access shaft 9 is designed
d with an aiir dam that is intennded to restrict or prevent
p air flow
b
between the upstreeam and downstreaam sides of the acccess shaft. For thiis purpose, four
p
proprietary flexiblee valves, TideFlex valves, are installed within the shaft.

The Odor Control F

T Facility (OCF), in the vicinity of shaft 9, extracts odoroous air from the
s
sewer headspace on either side of shhaft 9. The air dam
m at the bottom of shaft 9 isolates
t air flow from the
the t upstream segm ments (shafts 9 to 13)
1 and the downsttream segments
(
(shafts 9 to 7).

The presence of the

T t air dam at acccess shaft 9 causeed the liquid levell to rise in the
u
upstream sewer tunnnel leaving only about
a 10 to 15 cmm of air headspace. Hence, it was
n
necessary to place the air extraction point
p as far upstreaam of shaft 9 as posssible. Two air
e
extraction manholees measuring 1.0 m in diameter were placed upstream and a downstream
o shaft 9 at 124 annd 30 m, respectiveely.
of

A metering facilitty is located between access shafts 6 and 7. The opperation of the
m
metering facility requires
r that the pipes within the facility be surchargeed at all times.
H
Hence, the meterinng facility providees an air dam that prevents air moveements between
u
upstream and downnstream sewer tunnnels.

33.4. Drop Shafts 6 and 4

T
Through a seriess of baffles, the
l
liquid is dropped vertically
v about 20
m between the inlets and outlets of
s
shafts 6 and 4. The reinforced
c
concrete baffles are spaced at 2.8 m
v
vertically. The droop shafts are about
1 to 12 m in diameter
10 d with an
o
overall depth of about 33 m. Shaft 4
c
consists of two shhafts 4E and 4W;
s
shaft 4E, locatted immediately
d
downstream of shaaft 4W, is a drop
s
structure with a series
s of baffles,
d
described above. Figure 4 depicts
t drop structure at
the a shaft 4E. Figuree 4. Shafts 4E and
d 4W

33.5. Connection Chamber

C
T connection chhamber, located in the vicinity of shaft 1, joins the floow between the
The
e
existing and the neew SeC sewers, andd conveys the liquid to the Duffin Creeek Wastewater
T
Treatment Plant. The connection chhamber represents the downstream end e of the new
S tunnel.
SeC

498 © BHR Group 2011 ISAVT14

4. PROPOSED TUNNEL VENTILATION

The overall odor control process strategy reduces the creation of odors, limits corrosion
in the sewer and treats odor emissions that are generated. The odor control process
operates for the area between the diversion facility in York Region to the connection
chamber in Durham Region.

4.1. Tunnel Ventilation Components

4.1.1. Corrosion Control Facility
The Corrosion Control Facility (CCF), located near access shaft 13, reduces the release
of odors from the wastewater by injecting hydrogen peroxide primarily to oxidize
hydrogen sulfide from upstream sources. This chemical reduces the dissolved sulphide
concentration in the wastewater which is expected to be a principal odorous compound in
the SeC trunk sewer. Other odorous compounds that would be oxidized if present are
mercaptans, amines and aldehydes. Additional information on hydrogen peroxide
reactions can be found in the Water Environment Federation, Manual of Practice (No.
25) reference book, Control of Odors and Emissions from Wastewater Treatment Plants.
The amount of hydrogen peroxide injected into the tunnel is based on flow rate of the
wastewater.

4.1.2. Odor Control Facility

The OCF, located near access shaft 9, extracts the air from the sewer headspace.
Untreated air is routed through a two stage bioscrubber/biofilter system and carbon
adsorption unit before being discharged to the atmosphere. Induced draft fans will be
used to extract the odorous headspace air from the upstream and downstream sides of
access shaft 9. The induced draft fans are responsible for ventilating the headspace from
the diversion facility to the OCF on the upstream side, and from shaft 7 to the OCF on
the downstream side. The remaining headspace is ventilated using Air Handling
Facilities.

4.1.3. Air Handling Facilities at Drop Shafts 6 and 4

Shafts 6 and 4 are drop structures consisting of a series of baffles that severely restrict air
movements between the downstream and upstream sides of the drop structure. The Air
Handling Facility (AHF) at shafts 6 and 4 will bypass the drop structure by extracting the
air from downstream section of the drop structure and routing the air back to the
upstream section of the drop structure. The AHF provides ventilation for the headspace
of the SeC from shaft 4W to shaft 6 and from shaft 4E to shaft 1.

4.1.4. Pressure and H 2S Monitoring Stations

A system of static air pressure and H2S sensors will be installed at various shafts to
provide monitoring data to assess the overall ventilation performance. Data from
pressure and H2S monitoring stations will be used to control the speed of the induced
draft fans at the OCF and AHFs at shafts 6 and 4. The pressure sensors will use wireless
communication to transmit the data. Additional wired pressure sensors will be installed
at other location, as described below. The pressure and H2S monitoring stations are
located at shaft 12, 10, 8, 5, 2, 1 and at the connection chamber.

4.1.5. Air Intake Vents

Air intakes will be installed at the connection chamber, metering chamber inlet, shafts
13B and 13C to allow fresh outside air into the SeC trunk sewer headspace. The air
intakes will be a goose neck type made of stainless steel with an internal diameter
ranging from 750 to 800 mm. The air intakes will be equipped with gravity dampers,

© BHR Group 2011 ISAVT14 499

which will mechanically open and allow outside air into the shaft and sewer headspace
when under negative pressure. When under positive pressure, the damper is designed to
close, preventing the release of untreated air. The damper pressure set point is adjustable
and will be initially calibrated to close if the pressure inside the shaft is positive and to
open if the inside pressure is negative. A screen or mesh will be installed on the vent
opening to prevent birds, animals and debris from entering the shaft. An additional
butterfly damper will be installed to act as an isolation damper to further reduce the
potential leakage. The butterfly damper will be actuated based on the inside pressure.

Position indicators will be installed on the gravity and butter dampers and a local
pressure sensor will be installed to observe the static pressure inside the shaft and to
command the butterfly damper. Except at the connection chamber, the damper
positioners and pressure sensor will be wired to a local odor control facility (CCF, OCF,
AHF shafts 6 and 4).

4.2. Ventilation Design Criteria

To safeguard against accidental release of odorous air to the ambient atmosphere, it is
required that a negative pressure be maintained within the tunnel headspace. Initially, the
ventilation system was designed such that the maximum negative pressure in the tunnel
headspace was not to exceed -12.5 Pa. As the tunnel design progressed, it was clear that
some deep access shafts, in excess of 40 m, are required. The top of the access shaft is
exposed to the ambient atmosphere with air temperature as low as -20°C, and relative
humidity of about 25%. The sewage inside the access shaft is flowing at almost a
constant temperature of about 13°C, with the odorous air at 100% relative humidity. Due
to its higher temperature and humidity, the odorous air will have lower density compared
to the ambient air; hence generating stack/buoyancy effect phenomena. To overcome the
stack effect, the following condition should be satisfied:

(1)

Where: is the maximum negative pressure within the tunnel headspace,

H is the access shaft depth,
is the density difference between inside and outside air,
and is the density of the air inside the manhole.

Equation (1) indicates that for a 43 m deep manhole, the pressure difference between the
bottom and the top of the manhole is about 69.75 Pa. Note that the air density inside and
outside the manhole was adjusted to account for both of the air temperature and the
humidity level. The pressure calculated using Equation (1) is conservative; hence,
alternative approaches were investigated. The pressure difference across the depth of a
stack can be calculated based on the 2009 ASHRAE [13], as shown in Equation (2)

(2)

Where: is a constant = 0.00598,

is the ambient air density,
is the stack height,
and & are the ambient and inside air temperatures, respectively.

Equation (2) indicates a pressure difference of about 67.25 Pa due to stack effect. The
pressure calculated by equation (2) is slightly lower than that calculated by equation (1),

500 © BHR Group 2011 ISAVT14

and that is mostly likely attributed to the fact that the air density of Equation (1)
considered the humidity level in the air.

Unlike cooling towers and typical ventilation stacks, deep access shafts in sewer lines are
surrounded by earth that maintains almost a constant temperature and only the manhole
cover is exposed to the ambient atmosphere. Additionally, the drag generated by the
sewage flow would counteract the stack effect. Hence, deep access shafts are expected to
exhibit less buoyancy effects compared with above grade ventilation stacks. So in order
to account for the special conditions related to deep access shaft, a three dimension
multiphase CFD model was generated for a typical deep access shaft. The circular access
shaft was about 11 m in diameter and 43 m deep. The lower 40 m of the 1.1 m thick
shaft walls were kept at a constant temperature of 10°C, the sewage at the bottom of the
shaft had a temperature of 13°C, and the 1.2x1.2 m access hatch had a temperature of -
20°C. The results of the analysis indicated a differential pressure of about -0.0374 kPa
between the top and the bottom of the shaft. Therefore, it was recommended that a
maximum negative pressure of -37.4 Pa should be maintained inside the tunnel.

4.3. Induced Fan Sizing

In order to satisfy the tunnel ventilation criteria discussed above, it was crucial to
estimate the required induced fan size at access shafts 9, 6 and 4. The design approach
was to determine the fan size based on the amount of odorous air that needs to be moved
through the tunnel headspace, and then use CFD modeling in order to verify the
adequacy of the proposed fan size.

The amount of odorous air that needs to be moved was calculated based on two empirical
approaches developed by Thistlethwayte [14] and Pescod and Price [9]. Thistlethwayte
has recommended a minimum of 2.36x10-5 m3/s of natural air flow for each square foot
of sewage surface to control sewer ambient humidity levels to less than 85%, though he
states that ventilation alone may not prevent corrosion. In comparison, air flows
calculated using the Pescod and Price procedure have also generally proven to control
H2S concentrations to acceptable levels.

Given the lack of more accurate approaches, a comparison of odorous air flow quantities
based on Pescod and Price and the Thistlethwayte approaches was performed. It should
be noted that separate calculations of air handling requirements showed that induced air
flow movement produces a much larger airflow quantity than displaced airflow due to
wastewater flow increases. Thus, the governing airflow parameter will either be
entrained airflow or ventilation as dictated by Thistlethwayte whichever is higher
provides greater conservatism and protection from fugitive emissions.

The air handling requirements change with liquid level and velocity, therefore it was
necessary to consider various liquid flow scenarios shown below:
Minimum flow condition of 2.81 m3/s,
Half full tunnel with d/D ratio of 50%. This scenario corresponds to liquid flow rate
of 9.5 and 7.1 m3/s in the 0.18% and 0.1% slope sections, respectively. This scenario
produces the largest air entrainment into the trunk sewer, hence, the largest volume of
odorous air to be extracted,
Maximum flow of 14.2 m3/s which corresponds to d/D ratio of 65% and 80% in the
0.18% and 0.1% slope sections, respectively. This scenario produces the minimum
tunnel headspace for odorous air movement.

© BHR Group 2011 ISAVT14 501

The results of the comparison, notedd above, indicated a good agreement of air handling
T
r
requirements by thhe two methods. Use of the Thistllethwayte approachh results in air
h
handling volumes slightly
s higher thann Pescod and Pricee in most cases. Peescod and Price
r
requirements are hiigher for counter-current airflow undeer the half-full scen
nario. It should
b noted that a saafety factor has beeen added to calculated airflow req
be quirements in a
c
counter-current coondition since the Pescod and Pricee model was not developed for
c
counter-current floow conditions. Indduced airflow due to air drag, per Peescod and Price
m
model, and wastew water surface area,, per Thistlethwayte recommendationn, are typically
g
greatest in sewers flowing
f half full.

Sewers flowing neearly full restrict the

S t available headspace for air convveyance. Head
l
losses from frictionn can be evaluated using the Darcy Weisbach
W formula using
u estimated
c
coefficients for inteerface roughness. The friction coefficcient for the concreete tunnel walls
i well documenteed; however, the friction at the air-liquid interface depends
is d on the
p
presence of waveleets on the surface. An absolute wall roughness
r coefficieent of 3 mm for
c
concrete was usedd resulting in a D arcy Weisbach friiction coefficient of o 0.026. The
r
roughness coefficiient for the air-liqquid interface was estimated based on wind shear
s
stress formulas devveloped for marinne applications. Thhis approach resullted in a Darcy
W
Weisbach friction coefficient
c of 0.020, but given potenttial variability, a Darcy
D Weisbach
f
friction coefficientt equivalent to conncrete (0.026) was used throughout the analyses in
o
order to be conservvative.

5
5. CFD SIMUL
LATIONS

Three-dimensionall CFD models weree generated to simuulate the air and liqquid flow inside
T
t sewer tunnel; a sample CFD modeel is shown in Figuure 5. The commerrcially available
the
s
software ANSYS/C CFX was used to perform
p the CFD analyses. All CFD D models were
g
generated using thee ANSYS Workbennch software.

Figure 5.
5 Typical CFD Model
M

IIt should be noted that

t a separate set of hydraulic CFD analyses
a were carrried out in order
t verify the hydraaulic design of the various componennts of the SeC tunnnel. This paper
to
p
presents only the ventilation CFD anaalyses.

55.1. Modeling Ap pproach and Meth hodology

T CFD models utilized
The u a stratified multiphase flow of
o air and liquid; su uch assumption
w valid since thee majority of the tuunnel system exhibbited clear separatiion between the
was
a and the liquid. Due to the depth of
air o the tunnel below
w the ground surfacce, there was no
s
significant changee in sewage tempperature. The isoothermal nature of o the problem

502 © BHR Group 2011 ISAVT14

combined with low flow velocity justified the use of incompressible flow models for both
the air and the liquid. The standard ANSYS/CFX k- turbulence model was used. This
modeling approach, in addition to the modeling assumptions discussed below, was
implemented by many researchers modeling air-liquid interactions [15, 16, 17, 18].

5.2. Modeling Assumptions

5.2.1. Free Surface modeling
Free surface flow occurs in multiphase simulations where liquid flow is separated from
the air flow by a distinct surface. Examples of such flow include flow in open channels,
flow around ship hulls, and flow inside partially filled sewer lines. During turbulent
flow, parts of the liquid might rise or separate from the interface surface and would
entrain air into the liquid when they fell back into the liquid stream. This mechanism is
more pronounced in drop shafts where the interface surface is significantly increased due
to the formation of liquid droplets, hence increasing the amount of entrained air into the
liquid downstream from the drop shaft. Except at the drop shafts, i.e. drop shafts 4 and 6,
the majority of the liquid flow within the trunk sewer will be laminar. This justified the
use of homogeneous flow model where both fluids would have the same velocity field at
the interface surface.

The homogeneous flow model is typically recommended for gravity driven flow with co-
current air-liquid flow. However, for countercurrent air flow it was necessary to use a
non-homogenous flow model that allowed phase separation at the free surface interface.
The interface transfer rate was controlled using the drag coefficient Cd. Further, a
homogeneous turbulence model was used to improve numerical stability.

5.2.2. Boundary conditions

Due to the considerable length of the sewer tunnel, it was crucial from a computational
point of view to divide the sewer tunnel into several segments such that each segment
had known boundary conditions. Due to the absence of any branch sewer that may feed
the trunk sewer, the same liquid flow rate introduced at access shaft 13B should exit the
sewer domain at the connection chamber downstream from access shaft 1. The following
tunnel segments and boundary conditions were considered:
Tunnel section between the outlet of access shaft 13B and access shaft 9. Liquid
flow rate at atmospheric pressure was specified at the outlet of access shaft 13B. The
specification of atmospheric pressure at access shaft 13B is a conservative
assumption since the gravity damper at the air intake vent at access shaft 13B is
activated at a pressure value of 32.4 Pa. Due to the air dam at access shaft 9, air
flow between the upstream and downstream sides of access shaft 9 is prohibited.
Hence, the downstream side of this tunnel segment allows liquid flow only. The
odorous air was extracted by specifying the required air flow rate at the air extraction
location upstream of access shaft 9.
Tunnel section between access shaft 9 and access shaft 7. Downstream of access
shaft 9, liquid flow rate is identical to that specified at access shaft 13B, and air is
extracted at the specified air flow rate. Air movement is not possible through the
metering chamber downstream of access shaft 7. Hence, only liquid flow can pass
through the outlet of access shaft 7. The air intake vent at access shaft 7 was
assumed at ambient atmospheric pressure.
Tunnel section between access shaft 6 and access shaft 4E. The metering chamber,
located upstream of access shaft 6, prevented air movement into access shaft 6,
hence, only liquid flow rate was specified at the inlet to access shaft 6. The specified
air extraction flow rate was applied downstream of access shaft 6. The outlet of
access shaft 4E was assumed to be submerged, thus allowing liquid flow only.

© BHR Group 2011 ISAVT14 503

 Tunnel section between access shaft 4E and the connection chamber. For design
purposes it was assumed that there is no air interaction between the existing YDSS
and the new SeC at the connection chamber. Furthermore, due to the surcharge
conditions of the primary trunk, downstream of the connection chamber, it was
assumed that only fresh air can get into the new SeC through the air intake vent in the
vicinity of shaft 2.

Minor leakage was assumed at manhole covers. The amount of leakage was calculated
assuming a 3 mm gap between the manhole cover and the structure of the manhole.

5.2.3. Wall Roughness

The sewer tunnel ventilation required that odorous air be moved within the tunnel
headspace, and the volume of this headspace is dependent on the liquid level inside the
tunnel. According to the Manning equation, the liquid level can be determined based on
three parameters, namely: the liquid flow rate, the slope of the tunnel, and the wall
roughness. The liquid flow rate and the slope of the tunnel are design parameters that
can be specified with great confidence. However, the roughness of the tunnel walls is
open for interpretation. The hydraulic design of the tunnel utilized a Manning s
roughness coefficient of about 0.013.

The wall roughness height specified by the ANSYS/CFX software is the equivalent sand
grain roughness; this is not exactly equal to the real roughness height of the surface under
consideration. In order to determine the appropriate value of the equivalent sand grain
roughness, a short segment of the tunnel was modeled and the wall roughness was varied
until the liquid level was very close to that calculated using the Manning equation. The
equivalent sand grain roughness using in the CFD analyses was about 2 to 3 mm.

5.3. Modeling Verification

Two approaches were used to verify the CFD modeling methodology and assumptions.
In the first approach, several models were generated for a typical access shaft with short
segments of the sewer tunnel. The goal of the models was to investigate the sensitivity of
the entrained air volume to variations in the mesh size and other modeling parameters
such as drag coefficient, turbulence model, homogenous vs non-homogenous model,
pressure vs mass flow boundary conditions etc. The results from this series of
simulations helped in refining the CFD parameters that were used in the analyses.

The second approach utilized the CFD modeling that was performed on the existing
YDSS. The existing YDSS, built in the 1970s, has an odor control facility located
midway along its 10 km long tunnel. The odor control facility extracted the odorous air
in a countercurrent scheme only, and the sewer system has one fresh air intake located
downstream from the odor control facility. Field measurements of air pressure were
available at the fresh air intake point and the odorous air extraction shaft. A CFD model
was constructed for the existing YDSS, and the available field measurements were used
to calibrate the modeling parameters. Once the CFD approach was established for the
existing YDSS, the same approach was applied for the new SeC CFD analyses.

5.4. Results and Discussion

The main goals of the CFD simulations were to:
a. confirm that it is feasible to move odorous air within the tunnel headspace,
especially under high liquid flow rate that results in minimum tunnel headspace.
Under this scenario, it is crucial to adjust the rate at which the odorous air is being
extracted in order to reduce the air velocity above the liquid surface while maintain
the negative pressure within the tunnel headspace, and

504 © BHR Group 2011 ISAVT14

b
b. confirm that the
t proposed size of the induced fann units can maintaain the required
negative presssure within the tunnnel headspace. Thhis issue is crucial for liquid flow
rate that woulld result in maximuum entrained air innto the sewer tunnnel, i.e. half full
tunnel scenariio.

To that end, the presentation

T p of thee results will focuus on air speed ab
bove the liquid
s
surface and static pressure
p within the tunnel headspace.

Figure 6. Headspace air pressure

55.4.1. Tunnel Seggment between Acccess Shafts 13B an nd 9

W the air dam located
With l at access shaft
s 9 and the odoorous air extraction point located
u
upstream of accesss shaft 9, the air within
w the tunnel headspace
h is moveed in co-current
m
manner. An air inntake vent, located at access shaft 133B, would allow frresh air into the
s
sewer tunnel. For this
t section of the tunnel,
t the followinng scenarios were examined:
e
High liquid floow rate of 14.2 m3/s, and odorous aiir is extracted at a rate of 17,670
m3/hr. Since thhe air extraction raate was calculated for
f the maximum air a entrainment,
the results off this analysis shhowed high negattive pressure withhin the tunnel
headspace, refeer to Figure 6, andd the peak air veloccity was almost 4 to t 5 m/s. Note
that the averagge liquid velocity is about 2.9 m/s. During normal operation, it is
anticipated thaat the induced fann speed will be reeduced, and that will w reduce the
amount of negaative pressure insidde the trunk sewer headspace.
h
Liquid flow ratte of 9.5 m3/s, andd odorous air is exxtracted at a rate off 17,670 m3/hr:
this liquid flow
w rate corresponds to d/D ratio of aboout 0.5, with the laargest liquid-air
interface surfacce, hence, the larrgest air entrainedd into the sewer trunk.
t The air
extraction rate was not sufficient to maintain negativve pressure at all manholes
m within
the sewer trunkk. So it was decided to increase thhe odorous air exttraction rate as
shown in the foollowing scenario.
Liquid flow ratte of 9.5 m3/s, andd odorous air is exxtracted at a rate off 22,000 m3/hr:
this analysis is similar to the anaalyses performed abbove, however, thee vent at access
shaft 13B was outfitted with a gravity
g damper thatt lets the air into the
t sewer trunk
when the presssure inside the acccess shaft drops beelow -32.4 Pa insiide the trunk at
access shaft 13B. The air extractiion rate was sufficient to maintain neegative pressure
at all manholes within the sewer trunk.
t

The results of the analyses indicate that

T t a fan extractioon rate of 22,000 m3/hr should be
u
used in order to maaintain a negative pressure
p within thee half-full trunk sew
wer. When the
l
liquid level rises, thhe fan speed should be lowered in order to maintain thee desired design
n
negative pressure.

© BHR Group 2011 ISAVT14 505

 5

-5

Theoretical
air velocity
-10
AS7

AS8 Bottom of Air Space

-15 Top of Air Space

AS9
Mid of Air Space

-20
0 500 1000 1500 2000 2500 3000 3500
Distance Along Tunnel Length, m

2 0.5
AS7
AS8

0 0.0

AS9
-2 -0.5

-4 -1.0

-6 -1.5

-8 -2.0

-10 -2.5

-12 -3.0
0 500 1000 1500 2000 2500 3000 3500
Distance Along Tunnel Length, m

Figure 7. Air pressure and velocity (Q=14.2 m3/s)

5.4.2. Tunnel Segment between Access Shafts 9 and 7

With the air dam located at access shaft 9 and the odorous air extraction point located
downstream of access shaft 9, the air within the tunnel headspace is moved in
countercurrent manner. The odorous air was extracted at 30 m downstream of access
shaft 9, while fresh air was allowed into access shaft 7 through an air intake vent.
Additionally, odorous air extracted from downstream area of access shaft 6 was injected
back into access shaft 7 bypassing the air dam created by the presence of the metering
chamber.

For the high liquid flow rate of 14.2 m3/s, the odorous air extraction flow rate was
ramped up gradually in increments of 20% of the proposed fan capacity at access shaft 9
(33,400 m3/hr), and the pressure was monitored inside the air headspace. At 80% of the
proposed fan capacity, negative pressure was maintained within the sewer tunnel. Hence,
it was determined that for a liquid flow rate of 14.2 m3/s, the air extraction rate of 26,720
m3/hr should be sufficient to generate negative pressure within the tunnel headspace, as
shown in Figure 7. Air was introduced into access shaft 7 at a rate of 18,800 m3/hr, the
remaining 7920 m3/hr of air came through the intake air vent at access shaft 7. The
distribution of the air velocity within the tunnel headspace, shown in Figure 7, indicates
that air located at the bottom of the headspace was flowing in the same direction of the

506 © BHR Group 2011 ISAVT14

lliquid; however, air
a located farther from the air-liquuid interface was flowing in the
o
opposite direction.

When the liquid floow rate was reduceed to 9.5 m3/s moree drag surface wass generated, and
W
t
that required increasing the odorous air extraction ratee up the proposed capacity of the
i
induced draft fan (333,400 m3/hr). Thee increased air extrraction rate was neeeded in order to
m
maintain a negativee pressure within thhe sewer tunnel.

55.4.3. Tunnel Seggment between Acccess Shafts 6 and 4

F this part of thee sewer tunnel, thee odorous air is mooved in a counterccurrent manner.
For
F the high liquuid flow rate of 14.2
For 1 m3/s, the annalysis results indicated that the
r
rectangular air extrraction pipe, as shoown in Figure 8, iss susceptible to being surcharged.
H
Hence, a shallowerr air duct was recoommended. Furtheermore, it was reco ommended that
t air duct be exttended up to 30 m from the outletss of drop shafts 4 and 6 in order
the
e
eliminate any poteential for surchargging at air extracttion points. The CFD analyses
c
confirmed that the odorous air can bee moved in counterrcurrent manner, an nd the proposed
i
induced fan size was
w adequate in maintaining
m a neggative pressure witthin the tunnel
h
headspace.

Figure 8. Air extraction

e pipe con
nfiguration

55.4.4. Tunnel Seggment between Acccess Shaft 4 and Connection

C Chamber
S
Similar to the tunnnel section betweenn drop shafts 6 andd 4, the odorous airr is moved in a
c
countercurrent mannner within this secction of the tunnel.

Under the high liiquid flow rate, the

U t
o
outlet of drop shaft 4W was almoost
s
surcharged; hence creating an air damm,
r
refer to Figure 9. Therefore, an air a
p
pipe jumper is added
a between thet
d
downstream and upstream
u sides of the
t
s
shaft 4W.

Several locations were

S w considered for
f
t fresh air intake vent; those includded
the
a
access shafts 1, 2,, and the connection
c
chamber. The ressults of the analysses
i
indicated that duue to the potential
s
surcharge conditioons just downstreaam
o the connection chamber, the freesh
of
a intake vent shoould be placed in the
air t Figuree 9. Liquid level att shafts
v
vicinity of access shaft
s 2. 4E & 4W

© BHR Group 2011 ISAVT14 507

6. SYSTEM COMMISSIONING

The system commissioning will be performed by the Contractor, with the assistance of
all equipment Vendors, the Consultant and the York s operations staff. A two stage
commissioning process is planned for the ventilation system. The first stage will take
place after substantial construction is done, but before raw sewage is allowed into the
tunnel. During this stage, the induced draft fans at the OCF and the AHFs will be tested
in order to insure proper operation and communication between the various tunnel
ventilation components.

The second stage commissioning will take place after the sewage is diverted into the new
tunnel. This stage will involve the calibration of various components that control the
ventilation of the tunnel, such as the pressure sensors, the intake vent dampers, the
induced fan speed etc.

7. CONCLUSION

One of the major requirements for the new SeC sewer tunnel is to prevent odor release to
the ambient atmosphere. A typical tunnel ventilations scheme would not be able to
satisfy the ventilation requirements established for this trunk sewer. Hence, a mechanical
ventilation system was designed for the sewer tunnel in order to remove the odorous air
from the tunnel headspace and maintain a negative pressure within the headspace. The
ventilation design requires that odorous air be moved in co-current as well as
countercurrent scenarios.

This paper describes the analysis and design process for the sewer tunnel ventilation
system. The CFD analyses were able to confirm the feasibility of the proposed
ventilation system. Furthermore, the results of the analyses were used to refine some of
the ventilation design parameters. It is anticipated that the ventilation scheme described
in this paper can easily be adapted to comparable sewer tunnels.

8. REFERENCES

(1) D. Donaldson (1932), Sewer Ventilation, Why and How?, Sewage Works Journal,
Vol. 4, No. 1, pp. 108-113.
(2) Richard Pomeroy (1945), The Pros and Cons of Sewer Ventilation, Sewage
Works Journal Vol. 17, No. 2, pp. 203-208.
(3) Jay Witherspoon, Chris Easter, and Dirk Apgar (2009), Collection System
Ventilation, the Water Environment Research Foundation (WERF).
(4) Harvey Sorensen, Jim Joyce, David Day, C. Timothy Fallara (2000), Odor Control
for Large Diameter Deep Sewer Tunnels the City of Columbus Ohio, Water
Environment Federation, Odors and VOC Emissions.
(5) S. Edwini-Bonsu and P.M. Steffler, J. Environ (2004), Air flow in sanitary sewer
conduits due to wastewater drag: a computational fluid dynamics approach, Eng.
Sci. 3: 331 342.
(6) Harvey Sorensen, C. Timothy Fallara, David Day (2008), Evaluation of Upper
Scioto Large Diameter Deep Sewer Odor Control 1999 TO 2006 the city of
Columbus Ohio, Water Environment Federation, WEF/A&WMA Odors and Air
Emissions.

508 © BHR Group 2011 ISAVT14

(7) Punda Pai, Jim Joyce, Harvey Sorensen (200), Large Diameter Sewer Ventilation
Dynamics Require a Three Mile-long Odor Control Duct in Las Vegas, Water
Environment Federation, Odors and VOC Emissions.
(8) Dirk Apgar, Jay Witherspoon, and Chris Easter (2007), Minimization of Odors and
Corrosion in Collection Systems, , the Water Environment Research Foundation
(WERF).
(9) Pescod, M.B., Price, A.C. (1981), Fundamentals of Sewer Ventilation as Allied to
the Tyneside Sewerage Scheme, Water Pollution Control Vol 80, No 1, p 17-33.
(10) S. Edwini-Bonsu, A.M.A; and P. M. Steffler (2006), Dynamics of Air Flow in
Sewer Conduit Headspace, Journal of Hydraulic Engineering ASCE.
(11) James Joyce, Harvey W. Sorensen, Mark M. Smith (2000), Large Diameter Sewer
and Tunnel Ventilation Characteristics and Odor Control: Recent Development and
Case Histories, Water Environment Federation, WEFTEC.
(12) Sewer Ventilation, an article published in the New York Times on June 28, 1880.
(13) ASHRAE Handbook Fundamentals (2009).
(14) Bowker, R.P.G, G.A. Audibert, H.J. Shah and N.A. Webster (1992), Detection
control and correction of hydrogen sulfide in existing wastewater systems, a report
prepared for the U.S. Environmental Protection Agency.
(15) Francisco M., Lopes A.M.G, and Costa V.A.F. (2005), Numerical and
Experimental Optimization of a Gravity Driven Free-Surface Flow, World
Scientific and Engineering Academy and Society.
(16) Thomas Frank (2005), Numerical Simulation of Slug Flow Regime for an Air-
Water Two-Phase Flow in Horizontal Pipes, the 11th International Tropical
Meeting on Nuclear Reactor Thermal-Hydraulics, France.
(17) Thomas Hone (2009), Experimental and Numerical Simulation of Horizontal Two
Phase Flow Regimes, Seventh International Conference on CFD in the Minerals
and process Industries, CSIRO, Melbourne, Australia.
(18) P. Bhramara, V.D. Rao, K.V. Sharma, and T.K.K. Reddy (2008), CFD Analysis of
Two Phase Flow in a Horizontal Pipe Prediction of Pressure Drop, World
Academy of Science, Engineering and Technology 40.

© BHR Group 2011 ISAVT14 509