Numerical modelling of air-water flow in a vertical drop

manhole
V. Sousa*; I. Meireles**, J. Matos* and M. C. Almeida ***
* Department of Civil Engineering, Architecture and Georesources, Technical
University of Lisbon IST, Av. Rovisco Pais 1049-001 Lisbon, Portugal
(E-mail: [email protected]; [email protected])
** Department of Civil Engineering, University of Aveiro, Campus Universitrio de
Santiago, 3810-193 Aveiro, Portugal
(E-mail: [email protected])
*** National Laboratory for Civil Engineering, Avenida do Brasil, 1700-066 Lisbon,
Portugal
(E-mail: [email protected])
Abstract This paper presents results of a first step aiming at modelling the
transfer of oxygen in manhole drops. An experimental study was previously
conducted at IST, Lisbon, focusing on the hydraulics and reaeration on a vertical
drop followed by a hydraulic jump. In the present research, a CFD code was used
to replicate the experimental tests. The scope of the present paper is to evaluate
the performance of the CFD model to capture the overall hydraulic features of the
flow. The computed water depths were found to agree with the experimental
measurements, and water velocity profiles were successfully compared with
theoretical velocity laws. At last, an analysis was performed to study the influence
of the air-entrainment model included in the CFD code on the pressure head along
the invert of the outlet pipe downstream the free overfall. The activation of the
model was found to improve the prediction of the pressures downstream of the
free fall.
Keywords computational fluid dynamics (CFD); drainage system; manhole;
plunge flow; vertical drop
INTRODUCTION
Dissolved oxygen concentration is one of the most relevant parameters used in water quality
assessments, both in natural and artificial streams. Within these dynamic environments,
dissolved oxygen concentration is a result of the balance, on the one hand, of the diverse
chemical and biological processes taking place within the water body that involve oxygen
(i.e., bacterial activity, photosynthesis, etc.) and, on the other hand, of the natural exchange of
gas through the water surface/atmosphere boundary and due to local turbulence (occurring in
singularities such as drops). This is not an easy equation, since the former are also controlled
by dissolved oxygen concentration in the water. Gameson (1957) was the first to report on the
aeration potential of weirs in rivers and, since then, several other studies on local turbulence
aeration in rivers have been presented (e.g., Apted and Novak 1973; Avery and Novak 1978;
Nakasone 1987). Research on the aeration performance of existing hydraulic structures were
reviewed by Wilhelms et al. (1992) and Gulliver et al. (1998), and the studies on air
entrainment by water jets in general, and weirs in particular, is a topic thoroughly reviewed by

Bi (1993) and Chanson (1996), among others. In sewer drops, the presence, or absence, of
dissolved oxygen is of significance in relation to the build-up or persistence of sulphides. As
long as aerobic conditions are maintained, which have to be sustained by aeration, no sulphide
build-up will occur (Thistlethwayte 1972). Therefore, local turbulence, such as that promoted
by sewer drops, has been recommended for use in places where sulphides are not present as a
measure to prevent their formation (EPA 1985). Despite its health, structural and
environmental relevance, only a limited number of prototype or model studies involved the
use of vertical drops in circular channels, conveying either clean water, polluted river water or
wastewater (e.g., Pomeroy and Lofy 1977; Matos 1991; Almeida et al. 1999).
The use of numerical models to address the process of natural aeration due to local turbulence
pose a significant challenge. When a free nappe plunges into a downstream pool, turbulence
and air entrainment will contribute to the exchange of oxygen between water and air. Avery
and Novak (1978) found that a trade-off exists between bubble residence time, pressure, and
turbulence level. However, the mechanisms of bubble generation are still an issue of debate
and the characteristics of the generated bubbles are yet not totally correlated with the water
flow properties. Up to date, numerical modelling of turbulence and air entrained generated
aeration has been mainly focused on artificial aerators, such as bubble plums (e.g.,
Bombardelli et al. 2007).
EXPERIMENTAL STUDY
The experiments were conducted in a hydraulic recirculation flume composed of two circular
pipes in acrylic PVC, with an internal diameter of 153.6 mm, assembled at the Laboratory of
Hydraulics and Water Resources of the Technical University of Lisbon, IST (Figure 1), in the
framework of the research work by Sousa and Lopes (2002) and Soares (2003).

Figure 1. Recirculation flume at the Laboratory of Hydraulics and Water Resources at IST
(Soares 2003).
The length of the inlet and outlet pipes was 1.50 m and 3.50 m, respectively. The height of the
drop, between the inlet and the outlet pipes, was adjustable up to 0.60 m and both pipe slopes
were adjustable up to 5%. A vertical lift gate installed at the end of the flume allowed
establishing a fully-developed hydraulic jump in the outlet pipe. Recirculation of the flow was
done through a pressure conduit, connecting the tail water tank (capacity of 0.30 m3) and the

feeding reservoir (capacity of 0.07 m3), fed by a small pump (discharges up to 4 l/s). To
control and measure the hydraulic characteristics of the flow, the flume was equipped with: i)
a flow rate sensor (Georg Fischer SIGNET 515), previously calibrated with a triangular weir;
ii) a flow control valve, installed in the pressure conduit; iii) a manually operated point gauge,
both in the inlet and outlet channels; and iii) piezometers installed along the invert centre line
of the outlet channel. The reaeration (not addressed herein) was determined based on the
measurements of two portable dissolved oxygen and temperature meters (YSI 556 MPS), one
installed in the inlet channel and the other installed in the outlet channel. Further details
regarding the experimental study can be found in Sousa and Lopes (2002), Sousa et al.
(2003), Soares (2003) and Soares et al. (2004).
NUMERICAL CODE
The commercial code FLOW-3D, developed by Flow-Science, Inc, was selected for this
research due to its particularly efficient, robust and accurate method to simulate free-surface
flows. The finite volume/finite differences methods are used to solve the equations of motion
in a Cartesian, staggered grid. Single- or multi-block grids can be used to define the domain,
whereas the geometry can be incorporated through (Flow Science 2008): i) a solid
modeller, which allows the use of general quadratic functions; ii) Computer-Aided-Design
(CAD) files; or iii) topographic data. After both the geometry and the grid are defined, the
Fractional Area-Volume Obstacle Representation (FAVOR) method (Hirt and Sicilian
1985) automatically embeds the obstacles in the computational mesh by computing the
fraction areas and volumes blocked to flow. This feature makes grid generation and geometry
definition separate tasks, allowing for independent modifications in each one. More
information can be found in Flow Science (2008).
General flow model and boundary conditions
The mixture equations for an air-water flow were the base for the theoretical model. For a 3-D
dilute flow, the Reynolds-Averaged Navier-Stokes (RANS) mixture equations are as follows
(Bombardelli et al. 2011):
um 0

(1)

T
0 u m
'
'
0 u m u m B p u m u m 0 u m u m
t

(2)

where u m refers to the timeaveraged mixture velocity vector; 0 indicates the reference
density (water); B is the vector of body forces; p denotes the timeaveraged, modified
pressure; refers to the dynamic viscosity; t is the time coordinate; and u 'm indicates the
fluctuating mixture velocity vector. In turn, refers to the tensor product; T denotes the
transpose of a tensor; and the underline indicates vectors. For this specific problem the body
forces are composed only by gravity. The Reynolds stresses, last term of eq. (2), correlate the
velocity fluctuations and can be interpreted as a mechanism of momentum exchange between
the mean flow and turbulence. This is an additional unknown term to the original NavierStokes equations. To close the problem, the Reynolds stresses are modelled using the eddy

viscosity concept (Rodi 1984), for which the RNG k model was applied to determine the
turbulent kinetic energy ( k ) and its rate of dissipation ( ). The previous equations are valid
in a domain limited by the incoming flow in the inlet pipe, the outgoing flow through the
outlet pipe, the pipe walls and the free surface. Water depth and an uniform velocity profile
were used to define the upstream boundary condition, whereas for the downstream boundary a
water depth was specified. An additional symmetry boundary condition was added along the
longitudinal axis of the geometry, allowing simulating only half of the physical model. Null
velocities normal to the pipe walls (impenetrability condition) and the usual wall functions for
the turbulence statistics were employed in the solid surfaces. The free surface is a particular
boundary condition, since it is a priori unknown in each time step. Free surface capturing in
FLOW-3D is accomplished using the TruVOF, a complete version of the Volume-of Fluid
method (Hirt and Nichols 1981). The TruVOF method is centred in the definition, use and
transport of a fraction of fluid function and application of boundary conditions at the free
surface. In this method, unlike others, the cells with gas are neglected and the flow is only
computed in cells with liquid. For this reason, the TruVOF combines the advantages of
minimum memory storage (only one variable has to be recorded), reasonable computational
cost, and satisfactory accuracy. A model of air-entrainment is also included in the FLOW3D, which allows for the simulation of air-water flows. This model is able to simulate selfaeration due to turbulence. When the disturbing energy due to turbulence exceeds the
stabilizing energy due to gravity and surface tension, a volume of air enters the flow, and is
subsequently transported within the flow and/or released to the atmosphere (see Bombardelli
et al. 2011 for details).
NUMERICAL MODEL IMPLEMENTATION
The geometry was generated using CAD software and then imported into FLOW-3D. It was
chosen to represent the flume from the feeding reservoir to the furthest downstream known
boundary condition, which is the water depth 3.0 m downstream from the drop. The resulting
geometry represents 4.8 m of the physical model, with an inlet pipe 1.5 m long (LI) and an
outlet pipe 3.0 m long (LO) and part of the feeding reservoir. The pipes are horizontal and
separated by a drop of 0.10 m. The domain was discretized using 4 main blocks covering: i)
the feeding reservoir; ii) the inlet pipe; iii) the drop region, including part of the inlet an outlet
pipes; and iv) the outlet pipe. Mesh refinement was performed until mesh-independent results
were achieved by reducing, progressively, the size of the cells within the defined mesh
blocks, particularly in the drop region. Table 1 details the simulations presented herein for
3.9 l/s discharge.
Table 1. Details of the simulations.
Turbulence model
Model of air-entrainment activated
Number of active cells
Minimum size of cells (mm)

RUN D
RNG
No
9.9E+05
2.5x2.5x5.0

RUN E
RNG
No
13.1E+05
2.5x2.5x2.0

RUN F
RNG
Yes
9.3E+05
1.9x2.5x2.0

Despite the smaller number of active cells of RUN F when compared with RUN E, the
resolution is not reduced since the domain was shortened using a grid overlay boundary
condition to increase resolution in the drop region. The proximity between velocity profiles,

water depths and pressures of RUNS D and E constitutes a check on the convergence of the
mesh (see Figures 3 to 5). A sensitivity analysis to the turbulence models was also performed
for simulations without the model of air-entrainment activated. The standard and the RNG
k models revealed to return similar results. Figure 2 provides a general overview of the
model configuration, where the mesh blocks limits are outlined. The mesh block of the outlet
channel is not completely shown to allow for a better view of the flow in the drop region.

Figure 2. Model configuration and velocity magnitude contours from RUN E (in m/s).
COMPARISON BETWEEN EXPERIMENTAL AND NUMERICAL RESULTS
Velocity profiles
The velocity profiles in the inlet pipe were checked against the theoretical log-law of the wall.
For this comparison, a section 1.3 m downstream from the feeding reservoir was selected, to
ensure that it was not affected by the inlet boundary condition nor by the brink. The number
of numerical points, in the profile, did not allow for an adjustment of the log-law in inner
variables, but in outer variables the adjustment was found to be good, with the shear velocity
(u*) stabilizing at an average value of 4.25E-02 m/s (Figure 3). In Figure 3, Uc is the critical
velocity, z is the transverse coordinate originating from the inlet pipe invert, and dc is the
critical water depth.
0.30
0.25

z/hc

0.20

Numerical results
RUN D

0.15

RUN E
Log Laws

0.10

RUN D u*=4.31E-02
0.05

RUN E u*=4.21E-02

0.00
0.4

0.6

0.8

1.0

1.2

u/U c

Figure 3. Adjustment of numerical velocity profiles in the inlet channel to theoretical log-laws
(at 1.3 m from the upstream reservoir).

Water depths
The comparison between measured and computed water depths (d) showed reasonable
agreement (Figure 4). In Figure 4, the water depths are normalized by the critical depth (dc)
and xI and xO represent, respectively, the stream-wise coordinates originating at the upstream
end of the inlet and outlet pipes.
2.50

1.5
1.4

2.00

1.3

1.50
d/dc

d/dc

1.2
1.1
1.0
0.9

Exp.

Experimental
1.00 RUN D

RUN D

RUN E

RUN E

0.50

0.8
0.00

0.7

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.5

0.6

0.7

0.8

0.9

1.0

xO/L O

xI/L I

Figure 4. Water depth in the inlet (left) and in the outlet (right) pipes.
Pressure heads
The largest differences between experimental and numerical results for RUNS D and E are in
the pressure head (p), since the measured values were taken in the self-aerated flow region at
the impact zone downstream of the drop and the model of air-entrainment was not activated.
Nevertheless, the qualitative evolution of the pressure along the pipe is captured (RUNS D
and E of Figure 5). In Figure 5, the pressure head is normalized by the pressure head at the
critical section in the upstream pipe (pc). A significant improvement is observed with the
activation of the model of air-entrainment (RUN F of Figure 5).
3.0
2.5

p/pc

2.0
Exp.

1.5

RUN D
RUN E

1.0

RUN F
0.5
0.0
0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

xO/L O

Figure 5. Pressure along the invert of the outlet pipe.

CONCLUSION
This paper provides a first step on the numerical assessment of the hydraulic performance of
vertical drops in sewer manholes. The simulations performed in this study show that the use
of a RNG k model combined with the TruVOF method allows for an adequate overall
representation of the flow features in terms of water depth and pressure head. The activation
of the model of air-entrainment was found crucial for the accuracy of the simulation of the
pressure head of the air-water flow along the invert of the outlet channel. It is worth noting
that no significant difference was observed among the results of water depth and water
velocity with and without the activation of the model of air-entrainment, as expected, since
these respected to a water flow region. Further, the multi-block feature embedded in FLOW3D, which helps optimizing the mesh, was decisive for computational time savings.
Since the main objective of the ongoing research is to address numerically the topic of gas
transfer downstream of vertical sewer drops, the results presented in this paper are very
encouraging. They show that the code seems simulates air-water flows down vertical drops
correctly, in terms of pressure head. Future work will require a more extended research on the
features of the model of air-entrainment, in particular the air release through the free surface,
to address the issue of the volume of air entrained before studying oxygen transfer.
The model chosen for this stage of the research is a simplified representation of many
drainage systems where free drops exist at the manholes, particularly in older systems. The
experimental program includes, also, models with guided drops present in more recent
systems that will be studied in the future. Considering that many drainage systems in Portugal
and other countries are comprised of 200 mm (wastewater) and 300 mm (stormwater) pipes,
the scale of the physical model is a close representation of both the water and air flows and
the Froude similitude provide a good approximation for extrapolating the results. The main
simplification comparing to the prototypes was the assumption of zero slope for the inlet and
outlet pipes.
ACKNOWLEDGMENTS
Vitor Sousa and Ins Meireles were Visiting Scholars at the University of California, Davis,
to develop numerical research under the supervision of Prof. Fabin Bombardelli, and were
financially supported by Fulbright Research Grants. This support is gratefully acknowledged,
as well as the support of the ICIST Research Institute from IST, Technical University of
Lisbon. The research is being carried out under the scope of the project
PTDC/ECM/108128/2008 and the grant SFRH/BD/39923/2007 from FCT, Foundation for
Science and Technology, Portugal.
REFERENCES
Almeida M. C., Butler D. and Matos J. S. 1999 Reaeration by sewer drops. 8th Int. Conf. on
Urban Storm Drainage, Sydney, Australia, pp. 413-420.
Apted R. W., and Novak P. 1973 Some studies of oxygen uptake at weirs. Proc. XV IAHR
Congress, IAHR, Paper B23, Istanbul, Turkey, pp. 177-186.
Avery S. and Novak P. 1978 Oxygen transfer at hydraulic structures. J. Hydraul. Eng., ASCE
104 (HY11), 1521-1540.

Bi A. 1993 Gas entrainment by plunging liquid jets. Chemical Engineering Science, 48 (21),
3585-3630.
Bombardelli F. A., Buscaglia G. C., Rehmann C. R., Rincn L. E. and Garca M. H. 2007
Modeling and scaling of aeration bubble plumes: a two-phase flow analysis. J. Hyd. Res.,
IAHR, 45 (5), 617-630.
Bombardelli F. A., Meireles I. and Matos J. 2011 Laboratory measurements and multi-block
numerical simulations of the mean flow and turbulence in the non-aerated skimming flow
region of steep stepped spillways. Environ. Fluid Mech. 11(3), 263-288.
Chanson H. 1996 Air bubble entrainment in free-surface turbulent shear flows. Academic
Press, London.
EPA 1985 Odor and Corrosion Control in Sanitary Sewerage Systems and Treatment Plants.
US EPA Report 625/1-85/018.
Flow Science 2008 FLOW-3D Users Manuals Version 9.3. Flow Science, Inc., Los Alamos,
New Mexico, USA.
Gameson A. L. H. 1957 Weirs and aeration of rivers. J. Inst. Water Eng. 11(5), 477-490.
Gulliver J. S., Wilhelms S. C. and Parkhill K. L. 1998 Predictive Capabilities in Oxygen
Transfer at Hydraulics Structures. J. Hydr. Eng., ASCE, 124 (7), 664671.
Hirt C. W. and Sicilian J. M. 1985 A Porosity Technique for the Definition of Obstacles in
Rectangular Cell Meshes. Proc. 4th Int. Conf. Ship Hydro., National Academy of Science,
Washington, DC, USA.
Hirt C. and Nichols B. (1981) Volume of Fluid (VoF) Method for the Dynamics of Free
Boundaries. J. of Comp. Physics, 39, 201-225.
Matos J. S. 1991 Aerobiose e Septicidade em Sistemas de Drenagem de guas Residuais.
Ph.D. Thesis, IST, Lisbon (in Portuguese).
Nakasone H. 1987 Study of aeration at weirs and cascades. J. Environ. Eng., ASCE 113 (1),
64-81.
Pomeroy R. D. and Lofy R. J. 1972 Feasibility Study on In-Sewer Treatment Methods. NTIS
Nr.271445, US EPA, Cincinnati, OH.
Rodi W. 1984 Turbulence Models and their Application in Hydraulics. State-of-the-Art Paper,
IAHR.
Soares A. A. 2003 Rearejamento em Quedas em Colectores de guas Residuais. M.Sc. thesis,
FCTUC, Coimbra (in Portuguese).
Soares A., Almeida M. C. and Matos, J. 2004 Measuring oxygen transfer in sewer drops.
Water and Environmental Management Series, IWA Publishing, pp. 86-93 (ISBN
1843395061), pp. 86-93.
Sousa C. M. and Lopes R. R. 2002 Hidrulica e Rearejamento em Quedas Verticais em
Colectores. Research report, IST, Lisbon (in Portuguese).
Sousa C., Lopes R., Matos J. and Almeida M.C. 2003 Reaeration by Vertical Free-fall Drops
in Circular Channels: A Model Study. Proc. XXX IAHR Congress, IAHR, Theme B,
Thessaloniki, Greece, pp. 361-367.
Thistlethwayte D. K. B. (Ed.) 1972 The Control of Sulphides in Sewerage Systems. Ann
Arbor: Ann Arbor Science.
Wilhelms S. C., Gulliver J. S. and Parkhill K. 1992 Reaeration at Low-head Hydraulic
Structures. Tech. Rep. HL-91, US Army Engineer Waterways Experiment Station, Vicksburg,
Mississipi, USA.