Mouse: Pipe Flow
Uploaded by
Colleen MurphyMouse: Pipe Flow
Uploaded by
Colleen MurphyMOUSE
Pipe Flow
Reference Manual
MIKE 2017
2
PLEASE NOTE
COPYRIGHT This document refers to proprietary computer software which is pro-
tected by copyright. All rights are reserved. Copying or other repro-
duction of this manual or the related programs is prohibited without
prior written consent of DHI. For details please refer to your 'DHI
Software Licence Agreement'.
LIMITED LIABILITY The liability of DHI is limited as specified in Section III of your 'DHI
Software Licence Agreement':
'IN NO EVENT SHALL DHI OR ITS REPRESENTA-TIVES
(AGENTS AND SUPPLIERS) BE LIABLE FOR ANY DAMAGES
WHATSOEVER INCLUDING, WITHOUT LIMITATION, SPECIAL,
INDIRECT, INCIDENTAL OR CONSEQUENTIAL DAMAGES OR
DAMAGES FOR LOSS OF BUSINESS PROFITS OR SAVINGS,
BUSINESS INTERRUPTION, LOSS OF BUSINESS INFORMA-
TION OR OTHER PECUNIARY LOSS ARISING OUT OF THE
USE OF OR THE INABILITY TO USE THIS DHI SOFTWARE
PRODUCT, EVEN IF DHI HAS BEEN ADVISED OF THE POSSI-
BILITY OF SUCH DAMAGES. THIS LIMITATION SHALL APPLY
TO CLAIMS OF PERSONAL INJURY TO THE EXTENT PERMIT-
TED BY LAW. SOME COUNTRIES OR STATES DO NOT ALLOW
THE EXCLUSION OR LIMITATION OF LIABILITY FOR CONSE-
QUENTIAL, SPECIAL, INDIRECT, INCIDENTAL DAMAGES AND,
ACCORDINGLY, SOME PORTIONS OF THESE LIMITATIONS
MAY NOT APPLY TO YOU. BY YOUR OPENING OF THIS
SEALED PACKAGE OR INSTALLING OR USING THE SOFT-
WARE, YOU HAVE ACCEPTED THAT THE ABOVE LIMITATIONS
OR THE MAXIMUM LEGALLY APPLICABLE SUBSET OF THESE
LIMITATIONS APPLY TO YOUR PURCHASE OF THIS SOFT-
WARE.'
3
4 MIKE URBAN - © DHI
CONTENTS
MOUSE PIPE FLOW Reference Manual . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1 A General Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2 Modelling the Physical System . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1 The Model Elements - Inventory . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.1 General Description . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.2 Specification of a Link . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.1 General Description . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.2 Types and Definition of Nodes . . . . . . . . . . . . . . . . . . . . . 19
2.4 Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4.1 Overflow weirs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4.2 Orifice Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.3 Pump Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.4.4 Flow Regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.4.5 Non-return Valve . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.4.6 Combined Regulation (non-return valve + regulation) . . . . . . . . . 46
2.4.7 Long Weirs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.4.8 Valves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3 Description of Unsteady Flow in Links . . . . . . . . . . . . . . . . . . . . . . 51
3.1 Saint Venant Equations – General . . . . . . . . . . . . . . . . . . . . . . . . 51
3.2 Implementation of the Saint Venant Equations in MOUSE . . . . . . . . . . . . 52
3.3 Modelling The Pressurised Flow . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.4 Kinematic Wave Approximation . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.4.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.4.2 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.5 Diffusive Wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.6 Dynamic Wave Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.6.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.6.2 Supercritical flow simulations with dynamic wave approximation . . . 61
3.7 Flow Description in Links - Summary . . . . . . . . . . . . . . . . . . . . . . 61
3.7.1 Inventory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.7.2 Which Flow Description . . . . . . . . . . . . . . . . . . . . . . . . . 62
4 Numerical Solution of the Flow Equations in MOUSE Link Networks . . . 63
4.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.2 Computational Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5
4.3 Numerical Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3.1 Continuity Equation . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3.2 Momentum equation . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.4 The "Double Sweep" Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.4.1 "Branch matrix" . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.5 Stability Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.6 Optimising the Simulation Time Step t . . . . . . . . . . . . . . . . . . . . . 71
4.6.1 Automated, Self-adaptive Time Step Variation . . . . . . . . . . . . . 71
4.6.2 Criteria Controlling the Self-adaptive Time Step Variation . . . . . . . 72
4.7 Mass Continuity Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.7.1 Improved Continuity Balance for Links . . . . . . . . . . . . . . . . . 75
4.7.2 User Defined Minimum Water Depth . . . . . . . . . . . . . . . . . . 76
5 Initial Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.1 Default Initial Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.2 Initial Conditions provided by Hotstart . . . . . . . . . . . . . . . . . . . . . . 78
6 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
7 Flow Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.1 Friction Losses in Free-Surface Flow Links . . . . . . . . . . . . . . . . . . . 83
7.1.1 Numerical Description . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.1.2 The Friction Resistance Described by the Manning Formula . . . . . 84
7.1.3 Depth-variable Manning coefficient . . . . . . . . . . . . . . . . . . 85
7.1.4 Colebrook - White Formula for Circular Pipes . . . . . . . . . . . . . 87
7.1.5 Hazen-Williams Equation . . . . . . . . . . . . . . . . . . . . . . . 88
7.2 Head Losses in Manholes and Structures - Introduction . . . . . . . . . . . . 88
7.3 ‘Standard’ MOUSE Solution (F.A. Engelund) . . . . . . . . . . . . . . . . . . 89
7.3.1 Head Loss at the Node Inlet . . . . . . . . . . . . . . . . . . . . . . 89
7.3.2 Head losses at the outlet from a node . . . . . . . . . . . . . . . . . 89
7.3.3 Implementation of the Total Energy Loss Computation . . . . . . . . 92
7.4 An Alternative Solution Based on Weighted Inlet Energy Levels . . . . . . . . 93
7.5 Selecting an Appropriate Local Head-loss Computation . . . . . . . . . . . . 93
7.5.1 Constitutive Parameters of Head Loss Computation Options . . . . . 94
7.5.2 Default Computational Options . . . . . . . . . . . . . . . . . . . . . 95
7.5.3 Example: Node Outlet Head Losses Variation as Function of Head Loss Co-
efficient Mode 96
7.5.4 Implementation of Head Loss Description in Kinematic Wave Simulations
99
8 Some Special Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.1 Surface Flooding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.2 Sealed Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.3 Spilling Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
8.4 Pressure Mains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
8.5 Dry Conduits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
8.6 A note on flooding and spilling . . . . . . . . . . . . . . . . . . . . . . . . . 104
6 MIKE URBAN - © DHI
9 Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
10 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
7
8 MIKE URBAN - © DHI
MOUSE PIPE FLOW
Reference Manual
9
10 MIKE URBAN - © DHI
1 A General Description
The MOUSE Pipe Flow Model is a computational tool for simulations of
unsteady flows in pipe networks with alternating free surface and pressurised
flow conditions. The computation is based on an implicit, finite difference
numerical solution of basic 1-D, free surface flow equations (Saint Venant).
The implemented algorithm provides efficient and accurate solutions in multi-
ply connected branched and looped pipe networks.
The computational scheme is applicable to vertically homogeneous flow con-
ditions which occur in pipes ranging from small-profile collectors for detailed
urban drainage, to low-lying, often pressurised sewer mains, affected by the
varying water level at the outlet. Hydrodynamics of prismatic open channels
can also be simulated.
Both sub-critical and supercritical flows are treated by means of the same
numerical scheme that adapts according to the local flow conditions. Natu-
rally, flow features such as backwater effects and surcharges are precisely
simulated.
Pressurised flow computations are facilitated through implementation of a
narrow 'slot', as a vertical extension of a closed pipe cross section. Free sur-
face and pressurised flows are thus described within the same basic algo-
rithm, which ensures a smooth and stable transition between the two flow
types.
The complete non-linear flow equations can be solved for user-specified or
automatically supplied boundary conditions. In addition to this fully dynamic
description, simplified flow descriptions are available.
Within the Pipe Flow Model, advanced computational formulations enable
description of a variety of pipe network elements, system operation features
and flow phenomena, e.g.:
flexible cross-section database, including standard shapes;
circular manholes;
detention basins;
overflow weirs;
pump operation;
passive and active flow regulation;
constant or time variable outlet water level;
constant or time variable inflows into the sewer network;
head losses at manholes and basins;
depth-variable friction coefficients.
11
A General Description
The features implemented in conceptualisation of the physical system and
the flow process enable realistic and reliable simulations of the performance
of both existing sewer systems and those under design.
12 MIKE URBAN - © DHI
The Model Elements - Inventory
2 Modelling the Physical System
2.1 The Model Elements - Inventory
Elements available for definition of a numerical model in MOUSE are:
1. Links
pipes - standard and arbitrary cross-sections,
open channels - arbitrary cross-sections.
2. Nodes
manholes,
basins (structures),
storage nodes,
outlets.
3. Functions, for description of certain physical components of sewer sys-
tems, including:
overflow weirs,
orifices,
pumps,
non-return valves,
flow regulators.
4. Controllable structures, for the simulation of reactive or time dependent
operation real-time control, including:
rectangular underflow gate with movable blade,
rectangular overflow weir with changeable crest elevation.
Principles underlying the concept of controllable structures are described in
the "MIKE URBAN Collection System User Guide".
2.2 Links
2.2.1 General Description
Links in MOUSE Pipe Flow Model are defined as one-dimensional water con-
duits, connecting two nodes in the model. The link definition allows that the
dependent flow variables (e.g. water levels and discharges) can be uniquely
described as functions of time and space.
13
Modelling the Physical System
A link is featured by constant cross-section geometry, constant bottom slope
and constant friction properties along the entire length. A straight layout is
assumed.
MOUSE supports two classes of links:
closed conduit links (pipes),
open channel links.
Closed conduits under certain hydraulic conditions may become pressurised.
In such a case, the confinement of the flow fundamentally changes the envi-
ronment in which the flow process takes place, but the MOUSE Pipe Flow
Model continues to perform the computations using the same flow description
as for open channel flow. This is possible, because MOUSE furnishes actu-
ally closed conduits (pipes) with a fictitious slot (Preissmann slot) on the top
of the cross section, thus replacing a pipe with an open channel, featuring a
cross section shaped to approximate the hydraulic behaviour of a pressurised
pipe.
2.2.2 Specification of a Link
Specification of a Link requires specification of the associated nodes (see
paragraph 2.3 Nodes (p. 19)), the link material, longitudinal parameters and
the cross-section definition (shape and size).
Link Material
The parameter which characterises the link material is the link friction,
expressed as Manning's number (M or n = 1/M). The link can be defined as
14 MIKE URBAN - © DHI
Links
constituted of one of 8 predefined material types. Table 2.1 lists the available
link materials with MOUSE default values for Manning's number.
Table 2.1 Manning’s Numbers - MOUSE Default Values
Mouse Default Value
MOUSE Material M N=1/M
Code
1 Smooth Concrete 85 0.0118
2 Normal Concrete 75 0.0133
3 Rough Concrete 68 0.0147
4 Plastic 80 0.0125
5 Iron 70 0.0143
6 Ceramics 70 0.0143
7 Stone 80 0.0125
8 Other 50 0.0200
Fig 0.0.1
The default values can be edited by the user. The modified default values are
associated with the current project only; i.e. will affect any simulation carried
out with the MOUSE project file. Also, the default Manning number for any
individual link can be overwritten by a user-specified, link-specific value.
Longitudinal Profile
A link is longitudinally defined by bottom elevations of the upstream and
downstream end. By default, link bottom elevations are assumed to be equal
to the adjacent node's bottom elevations. The default setting can be over-
ruled by specification of the actual link end elevations, but not below the node
bottom.
Normally, length of a link is calculated on the basis of the nodes co-ordinates.
The length computation will take into account if the link between the nodes is
not straight. Optionally, for links connected to circular manholes, it is possible
to calculate the length from the manhole perimeter.
In cases where actual link length significantly deviates from the calculated
value, a user-specified length can be supplied instead.
Longitudinal slope of a link is assumed constant. It is calculated using link
end elevations and the link length.
15
Modelling the Physical System
Specification of a node as "upstream" or "downstream" has, in principle, only
a declarative meaning and does not affect the computations. An exception is
if the functions located in the link (see sections 2.4.2 Orifice Function (p. 30),
2.4.3 Pump Function (p. 44) and 2.4.4 Flow Regulation (p. 45)) are present in
the model. In the flow regulation (restriction), only positive flow is affected by
the regulation. Similarly, the non-return valve function allows only positive
flow.
By convention, positive flow values represent the flow in the direction from
"upstream" to "downstream" node.
Link Cross-Sections
As a built-in feature, MOUSE supports four different pipe cross-section types.
Any other "non-standard" pipe, tunnel or open channel can be described
through the Cross-section database facility (by specifying the geometric
shape of the cross section or a table of geometrical parameters.
MOUSE includes the following standard pipes:
1. Circular pipe,
2. Rectangular pipe (B H),
3. O-shaped pipe (H/B = 1.125/1),
4. Egg-shaped pipe (H/B = 1.5/1),
Figure 2.1 MOUSE egg-shaped cross-sections. Note the difference in selection of
the characteristic dimension D.
Any of the four "standard" pipe cross sections is fully defined by specifying
the pipe type, and characteristic dimension(s). While for the circular and rec-
tangular shape, this is straightforward, attention should be paid for the defini-
tion of the egg-shaped cross sections. For the O-shaped pipe the dimension
to be specified is the width (D, [m]), and for the Egg-shaped pipe the dimen-
sion to be specified is the cross-section height.
16 MIKE URBAN - © DHI
Links
The non-standard link cross-sections can be specified and maintained
through the Cross-section Editor. Cross-sections are distinguished as
"opened" and "closed", i.e. open channels on the one side, and pipes and
tunnels on the other.
The data required for description of a non-standard cross-section can be
entered in a raw form, either in a X-Z or in Height-Width format (please refer
to the user guide), which gives six options in total.
Figure 2.2 X-Z types of cross-sections
17
Modelling the Physical System
Figure 2.3 H-W types of Cross-sections
The raw geometrical data are then automatically processed, in order to create
tables with parameters suitable for flow computations. Such a table contains
50 data sets, covering the range from the lowest to the highest point speci-
fied, in equal increments.
The parameters in the table are:
W = surface width, [m]
L = height (relative depth), [m]
A = cross-section area, [m2]
R = A/P = hydraulic radius, [m]
In case of a closed link, MOUSE automatically provides an appropriate slot
for pressurised flow computations, see section 3.3 Modelling The Pressurised
Flow (p. 54).
Intermediate values are linearly interpolated.
The first set of values is associated with depth equal to zero (y = 0), and the
last set with the maximum specified value (relative to the bottom).
For open channels, MOUSE will compute the flow as long as the water level
is below the lower end of the cross-section. If this level is exceeded, the com-
putation will be stopped unless extrapolation of cross-section is specified in
the DHIAPP.INI file. For closed conduits, MOUSE allows an unlimited raise of
pressure; i.e. Preissmann slot is extended indefinitely in the height.
Processed data for a cross section is specified as a table with depth (Y),
width (B), area (A) and hydraulic radius (R). Conveyance is computed auto-
matically by MOUSE as C = AR2/3.
18 MIKE URBAN - © DHI
Nodes
The processed cross-section data table for an open cross-section should
cover the whole range of the expected oscillation of the water surface. If the
water surface exceeds the maximum specified elevation in the table, the
computation is stopped.
For closed cross-sections, the processed data table has to cover the entire
range from the bottom to the top of the cross-section. MOUSE adds the Pre-
issmann slot (see ref. /4/) automatically.
To ensure the computational stability, the cross-section conveyance should
be maintained monotonously increasing (or at least constant) with increase of
water level. This is normally not the case with closed conduits, where the
value of conveyance drops in the region near the top of the section. For such
cases, when raw data are input, MOUSE adjusts the hydraulic radius so that
the limiting conveyance for the cross-section corresponds to the actual con-
veyance value for the full profile. When closed cross-section data are input in
the processed form, attention should be paid in the upper region of the profile,
so that decreasing conveyance is avoided.
2.3 Nodes
2.3.1 General Description
Points associated with link ends and junctions are called nodes. Each link is
actually defined with exactly two nodes. Depending on the position in a net-
work layout, a node is associated with one or more links. In the later case, a
node is called a junction. An arbitrary number of links can be attached to a
junction, thus allowing construction of arbitrary network layouts.
2.3.2 Types and Definition of Nodes
Every node in a network is defined by its identification (max. 25 characters)
and its x and y co-ordinates, [m]. Exception is storage nodes, which do not
require co-ordinates. Further, according to the type of node, an appropriate
set of parameters is required.
Circular Manholes
Circular manhole is a vertical cylinder, defined by the following parameters:
Hbott = bottom elevation, [m]
Htop = surface elevation, [m]
Dm = diameter, [m]
K = outlet shape, types 1-9
Definition of the outlet shape is connected with calculation of head losses in
nodes (see section7.2 Head Losses in Manholes and Structures - Introduc-
tion (p. 88)).
19
Modelling the Physical System
Figure 2.4 MOUSE manhole
Flow conditions in a manhole are an important element of the overall flow
description. The following parameters are calculated:
Hm=water level in a manhole, [m]
vm= velocity, calculated per default as:
Q
v m = -------------------------------------------- (2.1)
H m – H bott D m
i.e. uniform velocity distribution is assumed.
The flow area calculated as above gives a very conservatively low estimate of
the velocity head and hence a conservative energy loss in the manhole,
causing higher water levels in the manholes than observed in reality.
An alternative formula for a more realistic calculation of the flow area in man-
holes is also available – however only for “flow-through” manholes with one
inlet pipe and one outlet pipe. The alternative formulation is based on the
assumption that the inflow behaves like a submerged jet, which entrains
water from the ambient fluid and increases the discharge through the man-
hole. The angle of entrainment is approximately 6.8. The cross-section area
of the jet thus depends on the distance from the inlet.
As a generalisation, it is assumed that the effective flow area in the manhole
equals the cross-section area of the jet at the outlet. This is valid in the case
of no change in direction from inlet to outlet. It is calculated as:
Dm 6.8 2
A flow = A jet = --- D in 1 + 2 -------- tan --------- 2
2
(2.2)
4 D in 360
where Din is the diameter of the inlet pipe.
20 MIKE URBAN - © DHI
Nodes
So far, the alternative formula is only applicable in MOUSE for manholes with
one inlet and one outlet. However, the implementation includes the possibility
for a change in elevation and a change in flow direction from inlet to outlet.
Figure 2.5 Manhole with one inlet, one outlet and a change in flow direction.
In the case of a change in flow direction the effect of the jet at the outlet will
gradually diminish with increasing angle. The effective flow area is therefore
linearly interpolated between the full cross-section area of the manhole, Am,
and the area of the jet, Ajet, as the angle increases.
The distance, a, from the point where the jet intercepts the manhole to the
centreline of the inlet, see Figure 2.5, is conservatively calculated as half the
diameter of the inlet, Din, thus neglecting the entrainment angle of 6.8
D in
a = -------
- (2.3)
2
The distance, b, from the point where the side of the outlet enters the man-
hole to the centreline of the inlet is approximated with
Dm D out
- tan --------- 2 – ----------
b = ------- - (2.4)
2 360 2
where is the angle between the centrelines of the inlet/outlet and Dout is the
diameter of the outlet pipe.
In the case of a change in elevation the effective flow area is diminished with
a factor, drop_factor, which is equal to 1 when the inlet flows directly into the
outlet and 0 when there is no interception between the incoming jet (calcu-
lated conservatively without the entrainment angle) and the outlet. In between
these two conditions the drop factor is interpolated linearly.
21
Modelling the Physical System
The effective flow area is then interpolated as:
b b
A flow = A jet 1 – --- drop_factor + 1 – drop_factor 1 – --- A m (2.5)
a a
For a straight inlet/outlet with no change in elevation the formula gives that
the effective flow area equals the jet area.
If the jet area is affected by the bottom level of the manhole the jet area is
reduced correspondingly.
If the effective flow area is greater than the full cross-section area of the man-
hole, Am, then the full cross-section area will be used.
The manhole volume contributes to the overall system volume and is
included in the computations.
If the water level exceeds the ground elevation Htop, then surface flooding
occurs, consequently followed by appropriate treatment by the model (see
section 8.1 Surface Flooding (p. 101))
Structures (basins)
This type of nodes is associated with arbitrarily shaped structures of signifi-
cant volume - non-circular manholes, tanks, reservoirs, basins and natural
ponds.
Structure geometry is defined by a table of data sets (min two) related to
monotonously increasing elevations, containing the following:
H =elevation, [m]
Ac =cross section area (used in calculation of the flow velocity in the
structure, assuming uniform velocity distribution), [m2]
As =water surface area (used for calculation of volume), [m2]
K =outlet shape, types 1-9
The first set of values corresponds to the structure bottom. The last set corre-
sponds to the surface level. Intermediate values are linearly interpolated. The
H-column can start at any value, e.g. 0.0 for interpretation of H as depth in the
basin. The MOUSE Engine will associate the first H-value to the bottom level
of the node.
Definition of the outlet shape is connected with calculation of head losses in
nodes (see section 7.2 Head Losses in Manholes and Structures - Introduc-
tion (p. 88)).
22 MIKE URBAN - © DHI
Nodes
A structure volume contributes to the overall system volume and is included
in the computations.
If the water level raises above the highest elevation value in the table describ-
ing the structure geometry, the program extends the basin geometry, follow-
ing the principle as described in section 8.1 Surface Flooding (p. 101).
An example of a definition of a basin is given in Figure 2.6.
Figure 2.6 Definition of a basin – an example
Storage Nodes
Purpose of storage nodes is a controlled simulation of the surface flooding,
i.e. controlled return of the water into the sewer system.
23
Modelling the Physical System
Storage nodes are fully defined with the identification string alone. The only
other parameter associated with a storage node is the content of water (the
capacity is not limited) currently stored in the storage node. Water enters a
storage node from any manhole or structure, either through a weir, gate/ori-
fice or a pump. A storage node may be emptied by an emptying function.
Outlets
Outlets are nodes specified at locations where the modelled system interacts
with receiving waters. External water volume is assumed so large that the
outlet water level is not affected by the outflow from the sewer system. As
such, outlets are appropriate for simulation of the sewer flow recipients (river,
lake, and sea).
An outlet can behave as an inlet, which depends on the flow conditions in the
link, attached to the outlet and the water level specified at the outlet. This
means that the flow in both directions can occur.
Outlets are defined with the following parameters:
Hbott = outlet bottom elevation, [m]
Hout = water surface elevation at outlet, [m]
Water surface elevation Hout can be specified as constant or as time depend-
ent (see section 6 Boundary Conditions (p. 79)).
Depending on the specified outlet water level, the model applies the following
elevation of the water surface H in the link adjacent to the outlet:
H out for H out H bott + min y c y n
h = else (2.6)
H bott + min y c y n
where:
yc = critical depth, [m]
yn = normal depth, [m]
In the later case, the outlet is considered to be a free outlet, meaning that the
outlet water level does not influence the flow in the adjacent link.
Otherwise, the model applies the specified water level, with the correspond-
ing backwater effect and a possibility for reverse flow.
24 MIKE URBAN - © DHI
Functions
2.4 Functions
Functions are used for the calculation of the flow between two nodes or in
specified links according to the functional relation and the hydraulic condi-
tions at relevant points in the system.
There can be more functions defined simultaneously between two nodes of
the network. One or more functions can be defined in a link between the two
nodes.
2.4.1 Overflow weirs
The overflow structures are normally found in sewer systems with purpose to
lessen the hydraulic load in the pipe system during extreme flow conditions
by allowing a part of the flow to be spilled to a recipient. Also, overflow struc-
tures can be used for internal distribution of the flow within the pipe system.
According to hydraulic conditions, two different types of overflow are possi-
ble:
free overflow
submerged overflow
The free overflow is the more frequent of the two types and the present con-
ceptualisation is therefore concentrated on this phenomenon. The computa-
tion of the submerged overflow is based on the same concept as the free
overflow and therefore inherently yields approximate results.
The following description covers weirs where it is acceptable to assume a
constant water depth at the weir crest.
MOUSE also supports a weir where this assumption is not acceptable. This
type of weir is called a “long” weir, please refer to the section concerning this
type of element.
Definition of an Overflow Weir – General
Overflow weirs (structures) can be specified in nodes defined either as man-
holes or as structures, but not at an outlet. A weir is topologically fully defined
with two node identifiers, defining the upstream node ('FROM') and down-
stream node ('TO').
The definition of the upstream and downstream nodes does not restrict direc-
tion of the flow because the weir function allows the flow in both directions,
depending on the current hydraulic conditions. Practically, this means that if
the water level in the downstream node is higher than the water level in the
upstream node, then the water flows 'backwards', i.e. the computed flow rates
are given a negative sign.
25
Modelling the Physical System
If an overflow structure discharges out of the contemplated pipe system then
the downstream node identifier is left unspecified (empty).
The relation between the water level in the structure or manhole and the
released discharge can be defined as a specific Q/H relation, or the built-in
overflow formula can be used. In the later case, the discharge is calculated
on the basis of a given structure geometry (crest elevation, structure width,
orientation relative to the flow, crest type). It is important that the width of the
overflow is realistic compared to the physical dimensions of the manhole or
structure. E.g., an overflow width of 10 m in a manhole having a diameter of 2
m will inevitably cause numerical problems when the overflow is in function.
Q/H Relation
The user-defined Q/H relation consists of at least 2 pairs of tabulated values
for water level above the weir crest, H [m], and corresponding discharge, Q
[m3/s]. Intermediate values are linearly interpolated.
The Q/H table has to fulfil certain conditions:
the first H-value has to be the overflow (weir crest) elevation;
the H-values have to be given in a monotonously increasing order;
the largest H-value given in the table shall not be less than the largest H-
value to be computed. The model does not extrapolate beyond the tabu-
lated values.
Built in Overflow Formula
MOUSE provides two different methods for the computation of the free over-
flow:
Flow computation based on the energy loss coefficient and weir orienta-
tion. This is applied if the field for the discharge coefficient on the
weir dialog is left empty.
Flow computation based on a standard rectangular overflow weir formula
with user-specified discharge coefficient. This is applied if a discharge
coefficient is specified.
Energy Loss Coefficient
In case of a free overflow, the water depth above the weir crest will be equal
to the critical water depth. Certain energy loss occurs, with a magnitude
depending on the structural configuration. The overflow situation is schema-
tised in Figure 2.7.
26 MIKE URBAN - © DHI
Functions
Figure 2.7 Free Overflow
In the critical flow section the Froude's number Fr equals to 1, and the critical
flow condition can be written as:
v
Fr = ------------- = 1 (2.7)
gy c
where:
v = mean flow velocity, [ms-1]
yc = critical depth, [m]
g = 9.81 [ms-2]
Conservation of energy between the upstream and critical cross section
yields with:
2
v
E = y c + ------ + E (2.8)
2g
with:
2
v
E = K c ------ (2.9)
2g
where:
E= energy level at the section just upstream the structure, [m]
E=entrance energy loss, [m]
Kc= energy loss coefficient
27
Modelling the Physical System
Based on the energy conservation and critical flow principles, discharge over
a plane overflow having a structure width B [m], is calculated for a weir
orthogonal to the flow axis (90o), as:
3
---
2 2
Q weir = B g --------------- H (2.10)
2 + Kc
and for a 'side' overflow weir (0o):
3
---
2 2
Q weir = B g --------------- H (2.11)
3 + Kc
where:
H = water depth above the weir crest level, [m]
Kc = energy loss coefficient associated with the outlet head-loss,
specified for the weir node (see sections
7.2 Head Losses in Manholes and Structures - Introduction
(p. 88)
to 7.5 Selecting an Appropriate Local Head-loss Computation
(p. 93)).
This actually corresponds to the standard overflow formula for a rectangular
notch:
2 32
Q weir = --- C d B 2g H (2.12)
3
where Cd is a discharge coefficient, expressed for an orthogonal weir as:
3
C d = --------------------------
- (2.13)
32
2 + Kc
and for a 'side' overflow weir:
3
C d = --------------------------
- (2.14)
32
3 + Kc
Where Kc is the head loss coefficient applied for the upstream manhole. E.g.,
this method if used with Kc = 0.5 (sharp-edged outlet), is equivalent to a
standard weir formula with Cd = 0.7589 and Cd = 0.4582 for orthogonal and
for side weir, respectively. Please note that the crest type - sharp or broad
crested has no influence on the calculations.
28 MIKE URBAN - © DHI
Functions
The side overflow yields a smaller discharge for the same overflow level,
because in this case the kinetic energy of the approaching flow is excluded
from the computations.
User-Specified Discharge Coefficient
If the method with default energy loss coefficient is not applicable for a par-
ticular weir, the standard overflow formula Equation (2.12) is applied with a
user-specified "level" discharge coefficient CH = 2/3Cd, which gives:
3
---
2
Q weir = C H B 2g H (2.15)
This implies that the head-loss coefficient specified for the weir node and the
weir orientation are ignored in the weir computation.
Submerged Overflow
The model calculates the flow rate for the submerged overflow using the
same critical depth formulation in the case of a submerged overflow. In this
situation, the head that is driving the flow is expressed as the difference
between the upstream and downstream water surface elevations.
Figure 2.8 Principle of submerged overflow.
The submerged weir flow is then (with user-specified "level" discharge coeffi-
cient) approximated as:
3
---
H 2
Q weir = C H B 2g -------- H (2.16)
H
or with energy loss coefficient, for orthogonal overflow weir:
--3-
2 2 H
Q weir = B g --------------- H -------- (2.17)
2 + Kc H
29
Modelling the Physical System
and for a 'side' overflow weir (0o):
3
---
2 2 H
Q weir = B g --------------- H -------- (2.18)
3 + Kc H
2.4.2 Orifice Function
Orifice is an opening of any shape, allowing water passage between other-
wise separated parts of the network. Usually, an orifice represents a flow
restriction.
Like an overflow weir, orifice is defined in MOUSE as a function between two
nodes.
MOUSE supports the computation of flows through orifices of any shape, in
all possible flow regimes. Further, a rectangular orifice with moveable top is
used for the simulation of controllable rectangular sluice gates.
Orifice functions can be specified in nodes defined either as manholes or as
structures, but not at an outlet. An orifice is topologically fully defined with two
node identifiers, defining the orifice upstream node ('FROM') and the orifice
downstream node ('TO').
Basic Geometrical Assumptions
Bottom is considered horizontal, both in the sections upstream and down-
stream from the orifice.
The upstream overflow crest height w1 is calculated as the distance between
the orifice invert level and the bottom level of the upstream node. Similarly,
the overflow crest height from downstream, w2 is given as the distance
between the orifice sill level and the bottom level of the downstream node.
Other parameters are described in the following text or illustrated on draw-
ings.
Approximation of Arbitrary Geometrical Shapes
An orifice opening is defined as a closed polygon through the MOUSE cross-
section editor. Any form of convex and concave shapes is allowed, as long as
there are no intersected arcs (see Figure 2.9).
30 MIKE URBAN - © DHI
Functions
Figure 2.9 Examples of an "illegal" (left) and correct definition of an orifice polygon.
For the computational purpose, a polygon is cut into a number of narrow rec-
tangles ('slices') which approximate the shape of an orifice (see Figure 2.10).
Figure 2.10 Illustration of a general shape orifice.
For the given upstream and downstream water levels, flow through the orifice
is computed as an integral of the flows through individual slices, with the total
flow corrected for lateral contraction.
Orifice Flow Regimes
Basically, there are four different types of flow regimes through an orifice (i.e.
for individual slice) for the approaching flow in sub-critical regime. These are
classified as:
Free overflow,
Submerged overflow,
31
Modelling the Physical System
Free underflow, and
Submerged underflow.
A definition sketch of the four types of flow regimes is shown on Figure 2.11.
Further, the theory distinguishes different forms of overflow jets, depending
on the geometrical and hydraulic relations. In the current implementation,
equations for the ventilated jet for the free overflow and the momentum equa-
tion for the filled jet (with a simplified correction for the downstream pressure)
for the submerged case have been adopted. These types are the most com-
mon.
The solution for the approaching flow in super-critical regime has not been
implemented.
Figure 2.11 Flow regimes through an orifice.
Free Overflow
This flow regime is identified when the downstream water level has no influ-
ence on the discharge over the weir. The water surface is free, and the solu-
tion is therefore a pure free-overflow weir solution.
The weir is considered to be ventilated and sharp-crested. The discharge
over a unit width of a weir for a given water level is given by:
q = C H H 2gH (2.19)
32 MIKE URBAN - © DHI
Functions
q = the discharge pr. m of the weir (m2/s).
H = upstream water level above the crest (m).
CH = the "level" discharge coefficient for the sharp-crested weir,
obtained as CH = 2/3Cd (see section 2.4.1 Overflow weirs
(p. 25)).
The coefficient CH is given in Table 2.2 for different values of the weir height
divided by the water level above the crest (w1/H).
Table 2.2 Variation of CH for different values of w1/H
CH Value of w1/H
0,0533
0.407 + ---------------- > 0.2
w1 H
0.673 0.2
0.757 0.1
0.761 0.05
w 1,5
0.707 1 + -----1- <0.05
H
In the interval from w1/H = 0.05 to 0.1 the coefficient CH is interpolated line-
arly between 0.761 and 0.757. In the interval from w1/H = 0.1 to 0.2 the coef-
ficient is interpolated linearly between 0.757 and 0.673.
Ignoring the energy loss from the upstream section to the weir section, the
energy equation reads:
2 2
q q
E = H + ------------------------------2- = y + -----------------------2- (2.20)
2g H + w 1 2g v y
where:
y = the distance from the sill level to the surface at the weir crest
(m),
the Coriolis factor;
v = the vertical contraction coefficient;
E = the energy level (m).
The depth at the weir crest is considered to be critical, i.e. y = yC = 2/3E. This
assumption is very rough, because the streamlines are curved. As a conse-
quence, the depth over the crest will be less than the critical depth. In the
context of the present implementation, curvature of the streamlines is
ignored, since the expression is only used to evaluate the effect the velocity
33
Modelling the Physical System
term have on the coefficient CH. The effect of curved streamlines is indeed
incorporated in the coefficients CH and CE.
By inserting y = 2/3E on the right hand side of Equation (2.20), the following
relation is obtained:
2
2 q 4
E = --- E + ----------------------------2- q = v ---------- E 2gE = C E E 2gE (2.21)
3 2 2 27
2g v --- E
3
where:
CE the energy discharge coefficient for the sharp-crested weir.
Since the discharge q can be expressed either via the water level above the
crest upstream of the weir, or the energy level at the upstream section, the
following relation between the level discharge coefficient and the energy dis-
charge coefficient can be derived:
3
---
2 2
q
C H = C E 1 + -----------------------------------2- (2.22)
2gH w 1 + H
As it can be seen from the relation above, the coefficient CH takes several
effects into account. One effect is the change of the velocity term in the
energy equation (v2/2g). For large values of w1 is the upstream energy level E
approximately equal to the depth over the crest H and CE is equal CH. For
smaller values of w1, the upstream velocity term becomes more important
and CE and CH will deviate from each other. The other effects are the curved
streamlines, the change in the Coriolis coefficient (), the vertical contraction
coefficient (v), the surface tension and the friction. The latter effects influ-
ence both CE and CH.
By moving from a (q,H) relation to a (q,E) relation, the variation in the dis-
charge coefficient should be expected to be smaller.
The energy level is given as:
2 2
v = + ------------------------------
q
E = H + --------
- H - = 1 H (2.23)
2g 2g H + w
2
1
where:
1 Coefficient of the relation between energy and water level.
34 MIKE URBAN - © DHI
Functions
By combining Equations (2.23) with (2.22), 1can be expressed as:
2
2 1
1 = 1 + C H --------------------- (2.24)
w
1 + -----1-
H
and the coefficient CE can be expressed as:
CH
C E = ---------
32
- (2.25)
1
The table below shows the relation between CH, CE, 1 and q (for H = 1) for
different values of w1/H, showing indeed that the coefficient CE shows less
variation than the values for CH.
Table 2.3 Relations between CH, CE, , q, E for different values of w1/H
w1/H CH CE Q for H=1 E 1.71E3/2
0.407 0.407 1.00 1.81 1.000 1.71
1 0.460 0.426 1.053 2.04 1.053 1.84
0.1 0.757 0.423 1.474 3.35 1.474 3.05
0.05 0.761 0.404 1.525 3.37 1.525 3.22
0 0.707 0.385 1.500 3.13 1.500 3.15
Starting with values for H and w1 given, the energy level can be derived by
iteration. The iteration implemented in the program is based on a Newton-
Raphson technique.
The discharge over the weir can then be determined by inserting the energy
level into equation (2.21).
Submerged Overflow
The submerged overflow is identified when the downstream water level influ-
ences the discharge over the weir, and water surface is free (i.e. the upper of
the gate is not in contact with the water surface, as can be seen in
Figure 2.11). The submerged overflow case will be applied, when the w0/H
1.0 and H/H 1/3, where wo is the height of the orifice and H is the water
depth above the sill level.
The submerged overflow case is illustrated in Figure 2.12, also giving the
meaning of the geometrical parameters used in the sequel.
35
Modelling the Physical System
Figure 2.12 Definition fo submerged overflow
Since the energy loss from section 1 to 2 is much smaller than the energy
loss from 2 to 3, the energy loss is neglected (i.e. E1 = E2). The energy equa-
tion now reads:
2 2
q
y 1 – w 1 + ------------- q
2
- = y 2 – w 2 + -----------
2
- (2.26)
y 1 2g y 2g
The momentum equation from section 2 to 3 can be written as:
2 2 2 2
q y2 q y3
- = --------- + ------
------ + ------ - (2.27)
gy 2 gy 3 2
where the shear stress on the bottom between section 2 and 3 is neglected.
The contracted overflow area can be expressed by applying the vertical con-
traction coefficient given as y = v (y2-w2).
There are two unknowns in these two equations. By rearranging the equa-
tions and substituting the q (actually q2/2g) from one of the equations into
another, the remaining unknown in the obtained equation is y2.
36 MIKE URBAN - © DHI
Functions
The equation can be transformed into a 4th degree polynomial, of a general
form:
4 3 2
C4 y2 + C3 y2 + C2 y2 + C1 y2 + C0 = 0 (2.28)
The polynomial is solved iteratively, applying the Newton-Raphson principle.
The initial value of y2 applied in the iterations is y2 = y1-w1+w2- (1/2)H.
The iterative process terminates when y2 converges within the specified
threshold, or if the number of iterations exceeds the specified number. If the
convergence is achieved, the discharge can then be derived from equation
(2.26).
The value of y2 is rejected, if the maximum number of iterations is exceeded
or in the following cases:
If y2 gives a negative argument to the square-root for the discharge,
If y2 > y1-w1+w2 or y2 < w2.
The equations applied above have some shortcomings. At first, the effect of
the curved streamlines is not taken into account properly, in contrast to the
free overflow case (which is derived from empirical expressions). The curved
streamlines will in this case give a different pressure distribution over the
crest deviating from the hydrostatic pressure and the pressure will be smaller.
The curved streamlines will become less and less important, the smaller the
values of H/H are. Secondly, the contraction coefficient has a significant
effect on the discharge, e.g. this approach is very sensitive to the choice of
the vertical contraction coefficient.
The submerged overflow solution must be compatible with the free overflow
at the transition between the two flow regimes. In other words, introducing the
submerged solution at H/H = 1/3 requires that the submerged discharge for
this water level difference is equal to the free flow discharge. This is not
achievable in all cases and sometimes another pragmatic solution must be
adopted for the transitional regime.
Following the approximate rule as for the flow over a broad-crested weir, a
flow reduction is introduced as soon as the difference between upstream and
downstream water level is less than one third of the upstream water level.
The remaining submerged discharge is proportional to the square root of the
difference in upstream and downstream water levels above the weir crest.
The free flow is taken from the “sharp-crested” case, as described above.
The flow in the submerged flow can be approximated as:
q H 1
q = ------f- H
-------- ;for -------- --- (2.29)
1 H H 3
---
3
37
Modelling the Physical System
where:
H the water level difference between the upstream and down-
stream section,
qf the free flow at the level where H= 1/3.
The implemented algorithm includes a combination of the parabolic and the
momentum solution. The parabolic solution is applied, if the combined energy
and momentum equation does not give applicable solutions for the given
H/H (i.e. if y2 is rejected). The discharge is solved for decreasing values of
H/H and for each value of H/H < 1/3 is the combined energy and momen-
tum equation evaluated. As soon as the combined energy and momentum
equation begin to give applicable solutions, a swap from the approximate par-
abolic solution to the combined energy and momentum solution is performed.
The contraction coefficient will in this case be based on the criteria, that the
discharge applying the combined momentum and energy solution is the same
as from the parabolic solution, at the point of intersection.
Free Underflow
The underflow is free, if the issuing jet of the supercritical flow is open to the
atmosphere and is not overlaid or submerged by tail water. Following an
approach similar to the one developed in the section related to free overflow,
the discharge through the opening (e.g. gate) can be expressed as:
q = C E w o 2gE 1 (2.30)
where:
q the specific discharge,
E1 the energy level upstream of the opening,
wo the gate opening,
CE the discharge coefficient with respect to the energy level.
The energy level at the upstream side can be expressed as:
2
q - = 2 H
E = H + ------------------------------ (2.31)
2
2g H + w 1
where:
H the upstream water level measured from the crest of the weir,
q the discharge,
w1 is the weir height at the upstream side.
38 MIKE URBAN - © DHI
Functions
Usually, discharge is given as a function of the upstream water depth above
the crest, rather than by energy level:
q = C H w o 2gH (2.32)
with:
Cc Cc
C H = -------------------------------------
- = ------------------------------------
- (2.33)
wo w0
1 + C c ----------------- ------
H + w1 H
1 + C c ----------------
w
1 + -----1-
H
where Cc is a constant representing the contraction coefficient of the jet.
Substitution of equation (2.33) into equation (2.32) leads to the expression:
wo 2
------
2 H
2 = 1 + C H ---------------- (2.34)
1 + -----w 1
-
H
Further, the relationship between CE and CH may be derived as:
CH
C E = --------- (2.35)
2
From above equations, the underflow discharge can be computed. However,
it should be noted that the compatibility of discharge values at the transition
from the free overflow equation to the free underflow equation must be
secured. Theoretically, this transition should take place at the moment where
the upstream water level touches the top of the gate. This point is difficult to
define as the water level is drawn down towards the contracted section.
Another complication is the fact that the underflow equation is accurate only
for upstream depths considerably exceeding the depth of the gate opening.
For this reason, the transition is simply assumed to take place at an upstream
water level equal to the top of the gate, while the difference between overflow
and underflow equations is fully corrected in the underflow computation at
that level. This requires a correction in the free underflow equation, through
the use of a correction coefficient.
39
Modelling the Physical System
For increasing upstream water levels, this correction coefficient is gradually
reduced as follows:
C new = C c – C c – C E (2.36)
where:
q overflow
= C c – ------------------------------- Cc – CE (2.37)
w 0 2gE
with E and CE taken at the top of the gate level. For increasing upstream lev-
els the discharge coefficient approaches the constant value Cc, usually taken
as 0.608.
The free flow equations require a further correction based on the pressure
distribution at the outflow side.
There are two extreme cases, the jet can either emanate surrounded by free
atmosphere (like an orifice), or it can have full contact with the bottom on the
downstream side (the vertical sluice gate).
In the first case the pressure over the height of the jet is approximately
atmospheric. In the other case the pressure follows a hydrostatic distribution.
The real situation usually is somewhere in between these two extreme cases,
and the flow through the gate is corrected for the influence from the pressure
on the downstream side.
The underflow equation has been derived on the basis of experiments where
the downstream bottom level is the same as the sill level of gates (w2 = 0).
This implies a hydrostatic pressure distribution in the contracted flow section.
With positive values for w2 (drop structure), however, these pressures drop to
lower values, with nearly atmospheric pressure over the height of the jet. In
this case the discharge will be higher due to the lower counter pressure.
Comparison of the orifice flow equation and the underflow equation reveals
that this difference may be up to 9 %. The same reasoning applies, to some
extent, for the case of overflow, where the discharge equation for the case of
a free overfall (w1 = 0) is also based on hydrostatic pressure distribution
assumption.
To cover most cases in a reasonable way, therefore, the free flow discharges
are increased by 5 % for the case where the downstream water level is found
below the crest level of the gate. For the range of downstream water levels
between the crest level and the upstream level, the correction applied is
reduced quadratically as the downstream water level is increasing. The quad-
ratic reduction follows from the quadratic relation between the integrated
hydrostatic pressure force and the water depth. Although the matrix of free
flow discharges is set up for the complete range of downstream water levels
40 MIKE URBAN - © DHI
Functions
up to the level which equals the upstream level, it should be realised, that
some of these corrected discharges are overwritten by new values for the
submerged flow case.
Submerged Underflow
The submerged underflow is identified when the upstream water level is
above the gate level and the downstream water level influences the dis-
charge through the gate. The threshold for swapping from free underflow to
submerged underflow is, for the simplification purpose, defined at H/H = 1/3.
This ensures that the same criterion is applied both in the overflow and
underflow cases and a consistency of the solution is maintained when w0/H
approaches unity.
A definition sketch of the submerged underflow is shown in Figure 2.13.
Figure 2.13 Definition of submerged underflow.
A combined energy and momentum formulation is applied (the same principle
as for the submerged overflow). If the energy loss from section 1 to 2 is
ignored, the energy equation reads:
2 2
q q
y 1 – w 1 + -------------- = y 2 – w 2 + ------------ (2.38)
2 2
y 1 2g y 2g
41
Modelling the Physical System
The momentum equation from section 2 to 3 can be written as:
2 2 2 2
q y2 q y3
------ + ------
- = --------- + ------
- (2.39)
gy 2 gy 3 2
where the shear stress on the bottom from section 2 and 3 is neglected. The
contracted overflow area can then be expressed by applying the vertical con-
traction coefficient given as y = v w0.
By rearranging the two equations and eliminating one of the two unknowns (q
and y2), the combined equation reads:
1 2 2 1 1 - = 0
y 2 – w 2 – y 1 + w 1 – --- y 3 – y 2 ----
- – ------------------- (2.40)
4 2
y w 2
1 v 0
Introducing the constants C1=(1/(vw0)-1/y3) and C2=(1/y12-1/(vw0) 2), the
equation can be reduced to a second degree polynomial in the form:
1--- 1
C y + C 1 y 2 + C 1 w 1 – w 2 – y 1 – --- C 2 y 3 = 0
2 2
(2.41)
4 2 2 4
Introducing:
A = (1/4)C2, B = C1 , C = C1(w1-w2-y1) - (1/4)C2y32 (2.42)
it can be shown that the only realistic solution for the second-degree polyno-
mial, is the negative one. So, y2 can be expressed as:
2
– B – B – 4AC
y 2 = ------------------------------------------- (2.43)
2A
and the discharge can then be derived from equation (2.38).
The solution is sensible to the selection of the vertical contraction coefficient.
The contraction coefficient must be determined, so that smooth transition
between free and submerged underflow is maintained.
For a certain range of contraction coefficient values, only imaginary solutions
to the Equation (2.43) exist. In such cases, i.e. as long as the combined
energy and momentum equation fail to deliver reasonable results, the para-
bolic solution is applied, similarly as for the transition between free and sub-
merged overflow.
42 MIKE URBAN - © DHI
Functions
Practical Computational Aspects
Computation of the flows through an orifice is based on a pre-processed 4-D
table, containing the flows through a vertical slice of unit width, computed as
a function of four dimensionless parameters: w0/H, w1/H, w2/H and H/H, and
using the equations described in previous paragraphs. The 'unit' flows are
computed at discrete points determined by the following set of the dimension-
less parameter values:
w0/H = {0.00, 0.05, 0.10, 0.30, 0.50, 0.80, 1.00}
w1/H = {0.00, 0.05, 0.10, 0.30, 1.00, 5.00, 100.00}
w2/H = {0.00, 0.05, 0.10, 0.30, 1.00, 5.00, 100.00}
H/H ={0.00, 0.01, 0.04, 0.09, 0.16, 0.25, 0.36, 0.49, 0.64, 0.70, 0.80,
0.85, 0.90, 0.95, 1.00}
This table is stored in a binary file MOUSE650.ORI and is supplied as a part
of MOUSE installation.
At the simulation start, MOUSE generates a structure-specific 3-D table for
each orifice, where actual flows to be applied in the computation are stored.
This table, of the size 28 x 28 x 10 contains discharges for all the combina-
tions of 28 upstream and downstream water levels, covering the full range of
possible water levels. When the algorithm is used for a gate, the third dimen-
sion is used for 10 different gate openings.
A non-equidistant scaling (approximating logarithmic scaling) is applied for
the water levels, while the scaling of the gate position is linear.
During the pre-processing, the following operations are executed:
Grids for the full range of upstream and downstream water levels are
generated. The grid spacing depends on the local geometrical parame-
ters.
Discharge from the dimensionless 4D-table, for the given upstream and
downstream water level and, if relevant gate position, are read and inter-
polated.
The 'unit' discharges are scaled, by multiplying the discharge by the
upstream depth above the crest (i.e. slice bottom) to the power of 1.5
(H1.5).
The discharge is corrected (reduced) for the effect of lateral contraction.
The discharge for entire orifice is summed up.
The actual flow through an orifice in a given hydraulic situation is obtained
during the simulation, by interpolating the flow derivatives with respect to h1,
h2 and w0 in the 3-D table, and inserting these directly into the MOUSE pipe
flow algorithm. By these means, accuracy and stability of the computation is
preserved, even with very rapid water level changes and fast movement of
the gate.
43
Modelling the Physical System
2.4.3 Pump Function
Pump functions can be specified in nodes defined either as manholes or as
structures, but not at an outlet. A pump is topologically fully defined with two
node identifiers, defining the pump-sump basin node ('FROM') and the down-
stream (recipient) node ('TO').
If the pump discharges out system then the downstream node identifier ('TO’)
is left unspecified (empty).
The pump operation is specified by defining the range of operation: 'start
level' Hstart, [m], and 'stop level' Hstop, [m], and one of the two available rela-
tions in a form of tabulated pairs of values: (1) H [m] and Qpump [m3/s], or (2)
H [m] and Qpump [m3/s]. The Qpump/H table consists of min two data sets,
there is no upper limit. Intermediate values are linearly interpolated. Variables
H, Hstart and Hstop denote water level in a pump sump basin (pump wet well
node).
Relation 1 correlates water level in the pump-sump basin and the pump dis-
charge:
Q H if H stop H
or if H start H
Q pump = (2.44)
else
0
Relation 2 defines the pump performance as a function of the water level dif-
ference between the two nodes:
Q H if H stop H
or if H start H
Q pump = (2.45)
else
0
A number of pumps with different operation strategies can be simultaneously
defined between the two nodes.
As the pump performance can be quite significant even during the start-up, it
has been necessary to dampen the pump dynamics in order to sustain the
numerical stability. The dampening is obtained by centring the pump rate
backwards in time so that the pump performance does not instantaneously
reach the full capacity but instead the pump discharge is gradually increased
over some time steps.
44 MIKE URBAN - © DHI
Functions
If pumps are present in the model set-up it might be necessary to introduce
relatively small time steps (5 - 10 sec.).
2.4.4 Flow Regulation
In computational terms the flow regulation differ fundamentally from the weir,
orifice and pump function by the fact that the control is simulated within the
pipe connecting two nodes and NOT by replacing the pipe with a functional
relation. This means that the conduit connecting the two
specified nodes is treated by the algorithm as a normal link. The flow is con-
trolled by setting the general equation coefficients at the control location (first
upstream Q-point in the pipe).
The control function is specified as a function of water level in a control node
A. The control is applied only within the specified range of water levels, and if
the water level is outside the specified range, an unregulated flow applies.
Therefore, it is important that the specified range covers all expected water
levels at point A. Otherwise, a sharp transition between the Q defined by the
control function, and 'natural' (unregulated) discharge would occur at the
range bounds causing numerical instabilities.
The following expression determines the flow:
min Q H A Q nat for H min H A H max
Q reg = else (2.46)
Q nat
where:
Qreg = applied (regulated) discharge, [m3s-1]
Q(HA) = discharge defined by the regulation function, [m3s-1]
Qnat = "natural" (unregulated) discharge, obtained as an explicit
estimate, based on the known water levels in the previ-
ous time step on each side of the regulation point, [m3/s]
HA = water level at the control point A, [m]
Hmin,Hmax = water levels at the control point A defining the range in
which the regulation is to be applied, [m].
2.4.5 Non-return Valve
The function for simulation of non-return valves is included into the model
structure identically as the flow regulation function.
45
Modelling the Physical System
The flow is applied according to the following:
Q for H up H down
Q reg = else (2.47)
0
where:
Q = calculated discharge, [m3s-1]
Qreg = applied discharge, [m3s-1]
Hup, Hdown = water levels at the computational points upstream and
downstream, respectively, [m].
2.4.6 Combined Regulation (non-return valve + regulation)
A combination of the two previous functions results with:
min Q H a ,Q nat for H min H A H max and H up H down
Q reg = else (2.48)
0
where:
Qreg = applied discharge, [m3s-1]
Q(HA) = discharge defined by the regulation function, [m3s1]
Qnat = "natural" discharge, obtained as an explicate estimate,
based on the known water levels in the previous time
step on each side of the regulationpoint,[m3s1]
HA = water level at the control point A, [m]
Hmin,Hmax = water levels at the control point A defining the range in
which the regulation is to be applied, [m]
Hup,Hdown = water levels at the computational points upstream and
downstream, respectively.
2.4.7 Long Weirs
A long weir is an element, which is able to simulate variations in the water
depth above the weir crestalong the weir itself.
The long weir must be defined as a link between two channels of the “natural
channel” type. The weir is topographically defined by the two links defining
the upstream link (“Source channel”) and the downstream link (“Destination
channel”). The “Location” field is the upstream node of the source channel
and the “To” field is the upstream node of the destination channel.
46 MIKE URBAN - © DHI
Functions
Discharge over the weir is calculated at each h-point in the upstream/down-
stream channel. Thus an internal weir, linking two branches require that the
number of computational points in the two branches is the same.
The weirs must always link two open channels.
Description of the Flow - the Long Weir
For free flow conditions the weir discharge per length of weir will be calcu-
lated as:
2 32
q s = --- C d 2gH (2.49)
3
where
H = the water level above the weir crest,
Cd = the weir coefficient
g = the gravity constant
For submerged weir the flow conditions are defined as
2h 32
q s = --------- C d 2gH (2.50)
3h d
where
H = the different between upstream and downstream water levels
h = the upstream depth over crest level
hd = height of the weir.
The Saint Venant Equations are solved for conservation of mass and conser-
vation of momentum. The equations are rewritten and solved for q and h
points.
Conservation of mass (h-point):
Q
------- + A
------- = 0 (2.51)
x t
Conservation of momentum (q-points):
2
Q
-------
Q A - + gA y
------- + ------------------- ------ + gAI f = gAI 0 (2.52)
t x x
47
Modelling the Physical System
where
Q = the flow
A = the flow cross section
y = the depth of water
If = the friction slope
I0 = the bottom slope of the canal
Taking into account the continuous discharge over the weir, Equation (2.52) is
modified to
Q
------- + A
------- = q s (2.53)
x t
In branches with long weirs the pair of equations (2.52) and (2.53) is solved.
The implementation of long weirs will include the possibility of specifying a
variation of the weirs crest along the weir, for example a sloping weir crest.
For this type of weirs, MIKE URBAN will divide the weir into a series of
smaller sections (equal to the number of computational points) and apply an
average crest level in each of these sections.
2.4.8 Valves
Implementation in MIKE URBAN
It is possible to define a valve between any internal nodes, but not at an out-
let. The valve will be topologically fully defined with two node identifiers,
defining the upstream node ('FROM') and downstream node ('TO').
The definition of the upstream and downstream nodes does not restrict the
direction of the flow because the valve function allows the flow in both direc-
tions, depending on the current hydraulic conditions. Practically, this means
that if the pressure level in the downstream node is higher than the pressure
level in the upstream node, then the water flows 'backwards', i.e. the com-
puted flow rates are given a negative sign. All valves can be defined either to
be non return valves, meaning that only flow in the positive direction is
allowed, or non restricted valves allowing flow in both directions.
The valve will apply the flow equation(2.54)
Q = A 2gH
--------------- (2.54)
k
where
48 MIKE URBAN - © DHI
Functions
g = the gravity constant
k = the flow factor which depends on the opening of the valve.
A = full open flow area of the valve
H = the energy drop over the valve.
An assumption for this equation is that the valve is located in a pipe which is
running under pressure. However, should the system run under non-pressur-
ized flow conditions the flow area A in equation (2.54) of the valve is reduced
by a linear reduction for non pressurized flow conditions:
d
A d = A ---- (2.55)
d f
where
d = the depth of the flow depth
df = the pressurized flow depth
This means that the velocity head upstream and downstream of the valve is
equal and equation (1) can be rewritten to
Q = A 2gh
---------- (2.56)
k
where
h = the drop in pressure over the valve.
MIKE URBAN supports full RTC control features of the new valve, which
means that it will be possible to define control algorithm for the opening of
valve.
49
Modelling the Physical System
50 MIKE URBAN - © DHI
Saint Venant Equations – General
3 Description of Unsteady Flow in Links
3.1 Saint Venant Equations – General
Computations of the unsteady flow in the links MOUSE Pipe Flow Model,
applied with the dynamic wave description, performs by solving the vertically
integrated equations of conservation of continuity and momentum (the 'Saint
Venant' equations), based on the following assumptions:
the water is incompressible and homogeneous, i.e. negligible variation in
density;
the bottom-slope is small, thus the cosine of the angle it makes with the
horizontal may be taken as 1;
the wavelengths are large compared to the water depth. This ensures
that the flow everywhere can be regarded as having a direction parallel
to the bottom, i.e. vertical acceleration can be neglected and a hydro-
static pressure variation along the vertical can be assumed;
the flow is sub-critical (Super-critical flow is also modelled in MOUSE,
but using more restrictive conditions).
The general form of the equations takes the form as follows:
Conservation of Mass ('continuity equation')
Q A
------- + ------- = 0 (3.1)
x t
Conservation of Momentum ('momentum equation')
2
Q
-------
Q A y
------- + -------------------- + gA ------ + gAI f = gAI 0 (3.2)
t x x
where:
Q = discharge, [m3s-1]
A = flow area, [m2]
y = flow depth, [m]
g = acceleration of gravity, [ms-2]
x = distance in the flow direction, [m]
t = time, [s]
= velocity distribution coefficient
I0 = bottom slope
If = friction slope
51
Description of Unsteady Flow in Links
The derivation of these equations is described in a number of textbooks and
scientific papers.
The general flow equations are non-linear, hyperbolic partial differential equa-
tions. The equations determine the flow condition (variation in water depth
and flow rate) in a pipe or channel when they are solved with respect to
proper initial and boundary conditions. Analytical solutions are only possible
in special cases with a rather limited number of applications, therefore the
general equations have to be solved numerically.
3.2 Implementation of the Saint Venant Equations in MOUSE
The Saint Venant equations can be rewritten as follows:
Q + A
------- ------- = 0 (3.3)
x t
and
2
Q
-------
Q A y
------- + -------------------- + gA ------ = gA I 0 – I f (3.4)
t x x
with the same nomenclature as for Equations (3.1) and (3.2). The sketch of
the system being described by the equations is presented in Figure 3.1.
Figure 3.1 Sketch of the pipe section.
52 MIKE URBAN - © DHI
Implementation of the Saint Venant Equations in MOUSE
The equations above are valid for free surface flow only. They can, however,
be generalised to include flow in full pipes (pressurised flow) as discussed in
section 3.3 Modelling The Pressurised Flow (p. 54).
The continuity equation expresses that the volume of water, Q, which is
added in pipe section of length x, is balanced by an increase in cross-sec-
tional area A (storage).
The first two terms on the left side of the momentum equation represent the
inertia forces (local and convective acceleration), while the third term repre-
sents pressure forces. The two terms on the right hand side of the equation
represent gravity and friction forces, respectively.
The velocity distribution coefficient accounts for an uneven velocity distribu-
tion across a section and corresponding difference in the actual momentum,
compared to those obtained with an average velocity. It is defined as:
A 2
= ------2- v dA (3.5)
Q A
Assuming that the bottom slope Io is small (~ 0), then Io can be expressed
as a function of the water depth and water surface gradient, i.e.:
y h
I 0 ------ – ------ (3.6)
x x
It is thus possible to use the height, h, above a certain reference level, as the
dependent variable instead of the water depth, y. The equation of momentum
can hence be written as:
2
Q
-------
Q A h
------- + -------------------- + gA ------ = gAI f (3.7)
t x x
Pressure and gravity forces can be expressed in one term only as:
h
gA ------ (3.8)
x
The friction slope If is equal to the slope of the energy grade line and is intro-
duced into the equation using the Manning's formulation (for more details see
section 4 Numerical Solution of the Flow Equations in MOUSE Link Networks
(p. 63)).
53
Description of Unsteady Flow in Links
3.3 Modelling The Pressurised Flow
The full flow capacity of a closed conduit (pipe) can be defined as a discharge
at which the flow depth is equal to the conduit height. Any further increase of
discharge fundamentally changes the conditions of flow, i.e. basic assump-
tions for the derivation of the Saint Venant equations are not valid. Namely,
the flow changes from the free surface flow to the pressurised flow.
However, it is possible to generalise the equations for free surface flow, so
that the pressurised flow in closed conduits is covered. This is done by intro-
ducing a fictitious slot in the top of the conduit, see Figure 3.2.
The idea of introducing a fictitious slot was first presented by Preissmann and
Cunge, 1961, and has since been used by Cunge and Wegner, 1964 (see ref.
/4/).
The derivation can be obtained from the continuity equation which can be
written as:
Q A
– ---------------- dxdt = --------------- dxdt (3.9)
x t
assuming the density of water constant over the cross section.
54 MIKE URBAN - © DHI
Modelling The Pressurised Flow
Figure 3.2 Pipe with a fictitious slot
By partial differentiation is found:
Q + Q
---- ------ + A
------- + ----
A
------- ------ = 0 (3.10)
x x t t
For a circular pipe, it can be shown that the density of the water can be
approximated as:
g y – D
0 1 + ---------------------
2
(3.11)
a0
where:
o = the density of water for a free surface flow [kgm-3]
ao = the speed of sound in water [ms-1]
y = the water depth [m]
D = the pipe diameter [m].
55
Description of Unsteady Flow in Links
Furthermore, it can be shown that the cross-sectional area in the case of the
excess pressure g(y-D) approximately equals to:
g y – D
A A 0 1 + ---------------------
2
(3.12)
ar
where:
Ao=the area without excess pressure and ar is given as:
Er e
ar = -------------
- (3.13)
0 D
with:
Er = the Young`s modulus of elasticity, [Nm-2]
e = the pipe wall thickness, [m].
The ar has the dimension [ms-1] and is in the order of 1400 [ms-1] for most
concrete pipes.
Combining these equations yields:
A A y 1 1 gA 0 y
- ------
------- + ---- ------ = g A ------ ----- + ----- = --------- (3.14)
t t t a a 2 2
a
2 t
r 0
where:
a0
a = ---------------------- (3.15)
2 2
1a 0 a r
represents the speed of sound in water considering the compressibility of
water and the deformation of the pipe wall.
Equation (3.10) can now be written as:
Q Q gA 0- ----- y
------- + ---- ------ + --------- - = 0 (3.16)
x x a
2 t
The analogy with the continuity equation can thus be maintained in case that
the fictitious slot width bslot is specified as:
A
b slot = g -----20- (3.17)
a
56 MIKE URBAN - © DHI
Kinematic Wave Approximation
a is in the order of 1000 [ms-1] for most pipes.
In order to obtain a smooth transition between the free surface flow computa-
tions and pressurised flow computations, it is required to apply a "soft" transi-
tion between the actual pipe geometry and the fictitious slot. Such a "smooth"
transition has been designed based on a series of tests with various slot con-
figurations. The slot configuration thus obtained ensures stable computations
without affecting the accuracy significantly. The applied slot width is larger
than the theoretical value. The default relation between relative depth and the
slot width as implemented in MOUSE is given in Table 3.1.
Table 3.1 Relation between relative depth and slot width
y/D Bslot/D(D=1m)
0.98 0.36
1.00 0.19
1.10 0.0166
1.20 0.0151
1.50 0.0105
>1.50 0.0100
The default slot width can be modified for individual links through the *.ADP
file.
3.4 Kinematic Wave Approximation
3.4.1 General
The flow conditions in steep, partly full pipelines are mainly established by the
balance between gravity forces and friction forces. Consequently, the inertia
and pressure terms in the momentum equation are less dominant. Accelera-
tions are comparably small and the flow is almost uniform, so that the kine-
matic wave approximation is a reasonable approach.
The momentum equation reduces to:
gAI f = gAI 0 (3.18)
i.e. the friction slope is equal to the bottom slope (uniform flow conditions). In
MOUSE, the Manning's formula for uniform flow is used and the momentum
equation reads:
23 12
Q = MAR I0 (3.19)
The kinematic wave is independent of the downstream conditions, meaning
that disturbances only propagate downstream. The kinematic wave descrip-
57
Description of Unsteady Flow in Links
tion can therefore only be applied in cases when the flow is independent of
the downstream conditions which is the case in supercritical flow (Froude's
number Fr > 1).
The analysis of the characteristics of the kinematic wave approximation
reveals that a solution obtained for partly filled pipes is physically unrealistic,
as the characteristic wave speed Q/A increases with increasing depth in a
circular pipe filled for less than 60% and decreases with increasing depth
when the pipe is filled for more than 60%. This points that an uncritical use of
the kinematic wave approach can lead to incorrect results, caused by an
unrealistic deformation of the propagating wave.
The kinematic wave is by nature undamped. The flow rate and the water
depth will therefore remain unchanged for an observer moving downstream
with the velocity Q/A.
Generally, it is not realistic to neglect pressure and inertia terms in the
momentum equation in most real flow situations. Therefore, the kinematic
wave approximation has to be used with care.
3.4.2 Implementation
The computations of the kinematic wave approximation in MOUSE are facili-
tated with the so-called 'degree-of-filling' function.
The 'filling' function can be determined from the Manning's formula assuming
uniform flow conditions, i.e. If = I0:
23
y Q MAR
F ---- = ----------- = ---------------------------------------- (3.20)
D Q full 23
M full A full R full
where suffix 'full' indicates values corresponding to a filled pipe and y/D indi-
cates the degree of filling.
This theoretically determined "filling" function has an over-capacity at y/D >
0.9.
The "filling" function applied in MOUSE does not include this over-capacity
but follows the Manning function up to a value of y/D = 0.8, see Figure 3.3.
According to the kinematic wave theory, Q/Qfull will not increase further after
the pipe runs full, as the pressure grade line is assumed to remain parallel to
the pipeline. In reality, however, pressurised flow often gives rise to an
increased pressure gradient and thus an increased flow rate. The
kinematic wave theory is therefore not suitable for computations of pressur-
ised flow without special adaptations.
58 MIKE URBAN - © DHI
Kinematic Wave Approximation
Figure 3.3 The "Degree-of-Filling" function applied in MOUSE
In order to make an approach to pressurised flow, the following assumption
has been made: An increase in pressure gradient gives rise to an increased
flow rate according to:
Q y–D
----------- = 0.1 ------------- + 1 (3.21)
Q full lr I0
lr is the remaining part of the pipe length.
This correction corresponds to an empirical deviation from the kinematic
wave theory, so that the pressure grade line is no longer parallel to the pipe
slope. 10% of the excess pressure is now used to increase the pressure gra-
dient, see Figure 3.4.
59
Description of Unsteady Flow in Links
Figure 3.4 The assumption that 10% of the excess pressure is used to increase
the pressure gradient
3.5 Diffusive Wave
It was mentioned above that numerical errors in connection with the numeri-
cal solution of the kinematic wave equations produce a diffusive (dampened)
wave motion. If the pressure term is included in the equation of momentum
then a damping term will automatically be included in the equations (the cor-
rect solution is a dampened wave motion).
The momentum equation for diffusive wave approximation reads:
gA y
------ + gAO f = gAI 0 (3.22)
x
By retaining the pressure term (h/x) in the computation, it is possible to
implement the downstream boundary conditions and thus consider backwater
effects. The diffusive wave approximation is therefore from a theoretical and
practical point of view a better approach than the kinematic wave approxima-
tion.
The computational basis for the diffusive wave approximation is in principle
identical to the one applied for the dynamic wave approximation for Froude
number Fr > 1 (supercritical flow). Further more, for stability reasons a mov-
ing average in time is applied to the slope of the water surface h/x in order
to dampen the short periodic fluctuations. This means that only relatively
steady backwater phenomena (compared to the time step) are resolved.
3.6 Dynamic Wave Approximation
3.6.1 General
The general flow equations form the best theoretical foundation for a flow
model because the full equation of momentum makes it possible to describe
60 MIKE URBAN - © DHI
Flow Description in Links - Summary
all forces affecting the flow conditions. However, larger computational load in
comparison with the kinematic and diffusive wave approximations involves
correspondingly larger CPU time for the same analysis. Additionally, difficul-
ties are present when simulating the supercritical flow conditions.
3.6.2 Supercritical flow simulations with dynamic wave approximation
The full Saint Venant equations (3.1 and 3.2) are applicable in the dynamic
wave approximation only for sub-critical flow conditions, i.e. for Froude num-
ber Fr < 1. In supercritical flow conditions, the equations are reduced to the
diffusive wave approximation. In the sub-critical regime, the contribution of
2
Q A is gradually
the inertia terms (Q/t and --------------------------
-
x
taken out by a reduction factor, according to Figure 3.5.
Figure 3.5 Gradual reduction of momentum terms during transition to
supercritical flow
Similarly, the differential equation is gradually centred upstream (as the influ-
ence of the upstream conditions increases) according to the same function.
3.7 Flow Description in Links - Summary
3.7.1 Inventory
The MOUSE Pipe Flow Model provides a choice between 3 different
levels of flow description approximations:
1. Dynamic wave approach, which uses the full momentum equation,
including acceleration forces, thus allowing correct simulation of fast
transients and backwater profiles. The dynamic flow description should
be used where the change in inertia of the water body over time and
space is of importance. This is the case when the bed slope is small and
bed resistance forces are relatively small.
2. Diffusive wave approach, which only models the bed friction, gravity
force, and the hydrostatic gradient terms in the momentum equation.
61
Description of Unsteady Flow in Links
This allows the user to take downstream boundary conditions into
account, and thus simulate backwater effects. The diffusive wave
description ignores the inertia terms and is therefore suitable for backwa-
ter analyses in cases where the link bed and wall resistance forces dom-
inate, and for slowly propagating waves where the change in inertia is
negligible.
3. Kinematic wave approach, where the flow is calculated on the assump-
tion of a balance between the friction and gravity forces. This means that
the kinematic wave approach cannot simulate backwater effects. Thus
this description is appropriate for steep pipes without backwater effects.
3.7.2 Which Flow Description
Depending on the type of problem, the most appropriate description can be
selected. All three approaches simulate branched as well as looped net-
works.
The dynamic wave description is recommended to be used in all cases
except where it can be shown that either the diffusive or kinematic descrip-
tions are adequate.
The diffusive and kinematic wave approximations are simplifications of the
full dynamic descriptions. They are implemented to offer improved computa-
tional efficiency, but should only be used when the omitted terms have insig-
nificant influence. When there is any doubt it is better to use the full dynamic
description or trials should be undertaken to establish the difference between
the alternative methods, and advice sought from experienced persons. It is
very important to have a solid understanding of the influence of the different
terms.
None of the three wave descriptions includes detailed hydraulic descriptions
of hydraulic jumps. However, the chosen formulations ensure a correct
description upstream and downstream of the jump.
62 MIKE URBAN - © DHI
General
4 Numerical Solution of the Flow Equations in MOUSE
Link Networks
4.1 General
The implemented algorithm solves the flow equations by an implicit finite dif-
ference method. Setting the numerical scheme into the frame of the Double-
Sweep algorithm ensures preservation of the mass continuity and compatibil-
ity of energy levels in the network nodes.
The solution method is the same for each model level (kinematic, diffusive,
and dynamic).
4.2 Computational Grid
The transformation of Equations (3.1) and (3.2) to a set of implicit finite differ-
ence equations is performed on a computational grid consisting of alternating
Q- and h-points ("staggered" grid), i.e. points where the discharge Q and
water level h, respectively, are computed at each time step (see Figure 4.1).
The computational grid is generated automatically by the model, or with user
specified number of grid points.
The computational grid for a conduit contains an odd number N of Q and h
points, with h points at both ends. The minimum number of computational
points N in a conduit is 3, i.e. two h points and one Q point in between. The
points are all equally spaced, with a distance x equal to:
l -
x = ------------ (4.1)
N–1
where l is the conduit length.
On the basis of the input data and the specified time step the model automat-
ically generates a complete computational grid, based on the velocity condi-
tion (see section 4.5 Stability Criteria (p. 70)). The velocity used in the
calculation is a full-flow velocity, obtained from the Manning formulation
assuming completely filled conduit.
If the velocity condition can not be satisfied for the specified simulation time
step, which often happens with short and steep pipes, then the model issues
a warning, with proposal for a shorter time step, required for the condition to
be satisfied.
The grid generated by the model can be altered individually for each conduit,
i.e. can be made more dense or sparse, according to the needs of the current
application (see documentation on *.ADP file).
63
Numerical Solution of the Flow Equations in MOUSE Link Networks
Figure 4.1 A section of the network with Computational Grid.
4.3 Numerical Scheme
The implemented numerical scheme is a 6-point Abbott-scheme (see ref. /2/).
The scheme for the method is shown in Figure 4.2.
Figure 4.2 Centred 6-point Abbott scheme
The flow equations are approximated by finite differences.
4.3.1 Continuity Equation
In the continuity equation the storage width, bs, is introduced as:
A h
------- = b s ------ (4.2)
t t
giving
Q b ----- h
------- + s - = 0 (4.3)
x t
64 MIKE URBAN - © DHI
Numerical Scheme
As only Q has a derivative with respect to x, the equation can be centred at
an h-point (see Figure 4.3).
Figure 4.3 Centring of the continuity equation in the Abbott scheme (a generalised
scheme). Note that in MOUSE xj and xj+1 are always equal
The individual derivative terms in Equation (4.3) are expressed by finite differ-
ence approximations at the time level, n+½, as follows:
n+1 n n+1 n
Qj + 1 + Qj + 1 Qj – 1 + Qj – 1
------------------------------------- – -------------------------------------
Q 2 2
------- --------------------------------------------------------------------------------- (4.4)
x 2x j
n+1 n
h- h j – hj
----- -------------------------- (4.5)
t t
bs is approximated by:
A o j + A o j + 1
b s = -------------------------------
- (4.6)
2x j
where:
A0,j = the surface area between grid points j-1 and j
A0,j+1 = the surface area between grid points j and j+1
xj = distance between points j-1 and j+1
65
Numerical Solution of the Flow Equations in MOUSE Link Networks
Substituting for the finite difference approximations in Equation (4.3) and
rearranging gives a formulation of the following form:
n+1 n+1 n+1
j Qj – 1 + j hj + j Qj + 1 = j (4.7)
where , and are functions of bs and and, moreover, depend on Q and h
at time level n and Q on time level n+½.
4.3.2 Momentum equation
The momentum equation is centred at Q-points as illustrated in Figure 4.4.
The derivatives of Equation (3.7) are expressed as finite difference approxi-
mations in the following way:
n+1 n
Q Qj – Qj
------- ----------------------------
- (4.8)
t t
1 1
2 n + -- -
2 2 n + --2-
2 Q ------
- –
Q
------
-
Q
------- A j + 1 A j–1
A
-------------------- --------------------------------------------------------------- (4.9)
x 2x j
n+1 n n+1 n
hj + 1 + hj + 1 hj + 1 + hj – 1
----------------------------- – -----------------------------
h 2 2
------ -----------------------------------------------------------------
- (4.10)
x 2x j
Figure 4.4 Centring of the momentum equation in the Abbott scheme
66 MIKE URBAN - © DHI
The "Double Sweep" Algorithm
For the quadratic term in Equation (4.9), a special formulation is used to
ensure the correct sign for this term when the flow direction is changing dur-
ing a time step:
2 n+1 n n n
Q = f Qj Qj – f – 1 Qj Qj (4.11)
where
n+12 n+12 n n
Qj Qj – Qj Qj
f = ------------------------------------------------------------------
n n+j n
(4.12)
Qj Qj – Qj
As standard, f is set to 1.0.
With all the derivatives substituted by finite difference approximations and
appropriately rearranged, the momentum equation can be written in the fol-
lowing form:
n+1 n+1 n+1
j hj – 1 + j Qj + j hj + 1 = j (4.13)
where
j = f A
n
j = f Q j ,t ,x ,M ,A ,R
(4.14)
j = f A
n n+1 n n n+1
j = f A x t q v h j – 1 Q j Q j h j + 1 Q j + 1
4.4 The "Double Sweep" Algorithm
4.4.1 "Branch matrix"
As shown earlier, the continuity equation and momentum equation can be for-
mulated in a similar form (compare Equation (4.7) and Equation (4.13).
Using, instead of h and Q, the general variable, Z, (which thus becomes h in
grid points with odd numbers and Q in grid points with even numbers), the
general formulation will be:
n+1 n+1 n+1
j Zj – 1 + j Zj + j Zj + 1 = j (4.15)
Writing the appropriate equation for every grid point, a system of equations is
obtained for each conduit (branch) in the network, constituting the 'branch
coefficient matrix', as illustrated in Figure 4.5.
67
Numerical Solution of the Flow Equations in MOUSE Link Networks
Applying a local elimination, the branch coefficient matrix can, in principle, be
transformed as shown in Figure 4.6 below. It is thus possible to express any
water level or discharge variable within the branch as a function of the water
levels in the upstream and downstream nodes (e.g. manholes) H1 and H2,
i.e.:
h = h H 1 H 2 (4.16)
and similarly:
Q = Q H 1 H 2 (4.17)
Figure 4.5 “Branch” matrix, with coefficients derived from the node energy level,
momentum and continuity equations
Figure 4.6 “Branch” matrix after local elimination
The continuity equation around a node can in principle be expressed as:
n+1 n+1 n+1 n+1 n+1
ah node + bh branch + cQ branch + dh branch + eQ branch + .... = z (4.18)
1 1 2 2
68 MIKE URBAN - © DHI
The "Double Sweep" Algorithm
where, a..z are quasi-constants. If Equation (4.15) is substituted herein, a
global relation can be obtained:
AH 1 + BH 2 + .... = Z (4.19)
where A, B, .... to Z are quasi constants.
Equation (4.19) shows that the water level in a node can be described as a
function of the water levels in the neighbouring nodal points. It is therefore
possible to set up a 'nodal point matrix' at each time step using the coeffi-
cients from Equation (4.19) and the solution to the matrix yields, by backward
substitution, the water levels in all nodal points at the next time step.
Figure 4.7 shows an example with 8 nodal points and 9 branches.
Figure 4.7 Principle of a "nodal" matrix for a system with 8 nodes and 9 branches
The crosses in the matrix symbolise coefficients, meaning that, for instance,
the water level in node 4 can be expressed as a function of the water levels in
nodes 1, 5 and 6. When the nodal point matrix has been solved, the solution
in the branches is found by backward local elimination.
The bandwidth of the nodal point matrix, as indicated by the stippled lines,
depends on the order in which the nodal points are defined. The bandwidth of
the matrix in Figure 4.7 is equal to 5. The computational time required for
solution of the nodal point matrix depends on the bandwidth size, and sharply
increases with increasing bandwidth.
69
Numerical Solution of the Flow Equations in MOUSE Link Networks
In order to minimise the computational time, an automatic minimisation of the
bandwidth is performed by internal perturbation of the nodal points. The
bandwidth displayed in Figure 4.7 (for the network with 8 nodal points and the
9 branches) could be reduced to 4 as shown in the matrix in Figure 4.8.
Figure 4.8 Minimised matrix band width
4.5 Stability Criteria
A criterion for a stable solution of the finite difference scheme is given by the
Courant condition:
C r = t
v + gy -
------------------------------ (4.20)
x
where:
v=mean flow velocity, [ms-1]
t=time step, [s]
x=distance between computational points in the conduit, [m]
y=water depth, [m].
Theoretically, the implemented numerical scheme is unconditionally stable for
all Courant numbers. In practice, however, this is restricted, because the
numerical implementation and the accuracy criteria impose some additional
limitations.
The most conservative condition for a correct and stable solution of the imple-
mented finite difference scheme is the velocity condition:
v t x (4.21)
The automatically generated computational grid fulfils this condition.
70 MIKE URBAN - © DHI
Optimising the Simulation Time Step t
4.6 Optimising the Simulation Time Step t
The computational efficiency of any discrete-time numerical simulation algo-
rithm is highly dependent on the time step applied in the simulations. In turn,
the feasible time step in a concrete situation depends on, apart from the
inherent performance properties of the computational scheme, the dynamics
of the flows in the simulated network. It is therefore desirable to optimise the
algorithm so that in conditions of variable flow dynamics as usually occur dur-
ing the simulated interval, the total computational effort is minimised, while
preserving stable and accurate computations.
MOUSE optimises the simulation time step by:
The automated self-adaptive time step variation. This is controlled by the
actual hydraulic and operational conditions within the entire model area
throughout the numerical simulation
This concept can be applied in connection with the Dynamic and Diffusive
flow descriptions while it cannot be used with the Kinematic flow description.
In this context, it is important to note that a constant time step is simply a
restricted case of these concepts.
4.6.1 Automated, Self-adaptive Time Step Variation
The automated, self-adaptation of the simulation time step is performed dur-
ing the running simulation. Such on-the-fly calculation of the time step is per-
formed through a three-step procedure:
Before the actual time step is taken, a preliminary value of the time step
is calculated, on the basis of the following:
The instantaneously time step is increased by a user-specified fraction
(the time step acceleration). Acceptance of this time step is validated
through checking the resolution of boundary conditions and pump
operations (see below). Finally, the suggested time step is validated
with respect to user-specified minimum and maximum values. The
minimum and maximum values and acceleration factors are specified
as a part of the simulation configuration. If the maximum and minimum
values of dt are equal, the program will use a constant time step.
The preliminary hydrodynamic solution is calculated with the preliminary
time step value.
71
Numerical Solution of the Flow Equations in MOUSE Link Networks
Based on an assessment of the preliminary solution, a judgement is
made whether the used time step is acceptable or not. The user has the
opportunity to specify numerous different limitation factors, such as a
maximum allowed variation in the water level in grid points, a maximum
allowed variation in the courant number, etc. A list of all possible limiting
factors is given below. If any of these criteria are exceeded, i.e. if the
generated variation is too large, then a revised solution is calculated. The
revised solution is obtained as a linear interpolation between the last two
simulation results (the previous time step solution and the solution with
preliminary time step), so that all specified criteria are fulfilled.
The different criteria which control the variation in the time step are outlined
below. The user has the option to modify the individual criteria through varia-
tion in the parameters. All of these parameters must be defined in the
DHIAPP.INI file.
4.6.2 Criteria Controlling the Self-adaptive Time Step Variation
Resolution of the Boundary Conditions
The time step is limited by the excessive errors generated due to the differ-
ence in the boundary time series resolution. In case of relatively fine resolu-
tion of boundary time series, application of long time steps may e.g. cause
volume errors. The maximum allowed error in the boundary conditions is
given by :
QacceptLimitRel Bvar (4.22)
where is the largest error between the given and simulated boundary condi-
tions (see Figure 4.9), Bvar is the value of the given boundary conditions and
QacceptLimitRel is a user specified value given in the DHIAPP.INI file.
Figure 4.9 Resolution of the boundary conditions
The boundary resolution criteria is tested on all time series defined as bound-
ary or results from a runoff simulation (the *.CRF file). However, the test is
72 MIKE URBAN - © DHI
Optimising the Simulation Time Step t
only applied to boundary conditions which are larger than QlowLimitM3s , a
minimum flow threshhold value.
The default value of QacceptLimitRel is 0.1 and QlowLimitM3s 0.01.
Variation in the Operation of the Pump Flow
The variation in the pump flow through one time step is limited by:
Q MaxPumpFlowVar Q (4.23)
where Q is the variation in the pumped flow, Q is the current value of the
pumped flow and MaxPumpFlowVar is the user specified maximum relative
variation.
The default value of MaxPumpFlowVar is 0.1 which corresponds to a 10%
maximum variation in the pumped flow during one time step.
It should be noted that this test also implies that the simulation is always
decelerated down to the minimum time step whenever a pump is switched
ON or OFF.
Variations in the water level in grid points
The variation of the water level in all H-grid points is limited by the following
functions:
H WaterLevDiffMaxRel H (4.24)
for H WaterLevDiffMaxRel H1 H0
H H for H WaterLevDiffMaxRel H1 H0 (4.25)
H WaterLevDiffMaxRel for H WaterLevDiffMaxRel (4.26)
where H is the relative depth (the water depth divided by the height, e.g. by
diameter for circular pipes), H0 is the relative depth before the attempted time
step and H1 is the relative depth at the end of the time step. H is the differ-
ence in the relative depth through the time step. The WaterLevDiffMaxRel
value can be user-controlled from DHIAPP.INI file. If limitation is violated at
any H-point in the model, then the obtained solution is scaled down with
respect to dt.
The default value of WaterLevDiffMaxRel is 0.3, which corresponds to a max-
imum relative change of 30 %.
Variation of Cross-Section Parameters
The variation of cross-section parameters A, R2/3 and B, where A is the cross
section area, R is the hydraulic radius and B the width of water surface can
be included as additional criterion for limiting the simulation time step.
Whether the check on the cross-section parameters is to be activated or not
73
Numerical Solution of the Flow Equations in MOUSE Link Networks
is specified through the variable Crosscheck in the DHIAPP.INI file (the value
0 means that this is de-activated, while the values 1 or 2 mean that the check
is activated in one of the two available variants). If the check on the cross-
section parameters is activated, then it is carried out in all H-grid points.
The variation in the three cross-section parameters is limited by:
X MaxVarCrossCons tan t Max X (4.27)
where the variable X is one of the three cross-section variables and the
meaning of Max(X) depends on the value of Crosscheck. If Crosscheck is
given as 1 then Max(X) is the maximum value of the actual parameter over
the cross section, while a value of Crosscheck which is equal to 2 means that
Max(X) is given as the actual value of the respective cross-section parame-
ters. However, the check is carried out only if the relative depth in the cross-
section is larger than the variable CrossLowDepthLimit. The check of these
limitations is carried out at the end of a time step simulation. If limitations are
violated then the solution is scaled down with respect to dt.
The default value of MaxVarCrossConstant is 0.03, if Crosscheck is 1 and
CrossLowDepthLimit is 0.04.
Variation in Courant Number
In the dynamic flow conditions, the Courant number (see section 4.5 Stability
Criteria (p. 70)) is continuously changing from time step to time step. In order
to avoid stability and accuracy problems, the Courant number is limited by:
dt
C MaxCourant where C = V
------------- (4.28)
dx
V is flow velocity and dx the distance between two computational grid points.
Check of this limitation is carried out after the simulation of a time step. If the
limitation is violated the solution is scaled down with respect to dt.
Recommended value of MaxCourant (specified in DHIAPP.INI file) is 20 – 60.
Weir oscillations
If the storage volume in one of the nodes connecting a weir is small, weir
oscillations might occur for free flow conditions. This phenomenon results in a
continuous change in flow direction over the weir until the instability is damp-
ened. In order to avoid this situation, a criterion related to the change in water
levels between up- and downstream nodes around the weir is implemented.
The criterion relates to dt by:
n n+1 n
dt possible H + H = dt H – 0.02 (4.29)
74 MIKE URBAN - © DHI
Mass Continuity Balance
where H is the difference in water level between the two nodes connected to
the weir and n corresponds to the time step level. The absolute allowed
change of 0.02 m is hard-coded in the program and cannot be controlled by
the user.
4.7 Mass Continuity Balance
Theoretically, what concerns the mass continuity balance, the applied compu-
tational scheme is inherently conservative for prismatic conduits with vertical
walls. In practical applications, the continuity balance may be jeopardised in a
number of situations, such as:
Relatively sharp changes of surface width due to rapid changes of water
depth or a sharp change of cross-section shape with depth. This may be
e.g. the case at relatively small depths in circular pipes and in arbitrary
cross sections
Sharp changes in surface area of basins
Surcharge of manholes
Etc.
The scale of the problem is usually related to the length of the simulation time
step.
4.7.1 Improved Continuity Balance for Links
In order to reduce the amount of water generated in conduits due to the
changes of surface width as function of water depth, i.e. to improve the conti-
nuity balance, the Taylor expansion of the general continuity equation (3.1)
has been applied. Since the surface width is assumed to be constant during
two time steps, the continuity equation can be rearranged as:
1 Q h
---- ------- + ------ = 0 (4.30)
w x t
where h is the water level (m) and w is the surface width (m).
1 Q
The term ---- ------- in the equation above can be expanded in a Taylor series as:
w x
Q ---
1--- ------- 1 Q n 1 n w Q n 1 n Q n + 1 Q n
- ------- – -----2- ------- ------- h + ---- ----------------- – ---------- (4.31)
w x w x w h x w x x
where represents the time centering of the numerical scheme and n and
n+1 refer to the simulation time steps.
This modification is applicable only for conduits with relatively smooth
changes of the surface width. As the width for arbitrary pipes and pipes from
75
Numerical Solution of the Flow Equations in MOUSE Link Networks
the cross-section database may vary in a very unpredictable way, the Taylor
expanded equation is only applied to standard pipes.
4.7.2 User Defined Minimum Water Depth
Further means of controlling the volume continuity balance for links with no or
little water are provided as user-controlled minimum water depth for links run-
ning dry or with very little flow. The default minimum water depth can be
modified in the DHIAPP.INI file. In this file two parameters can be changed:
BRANCH_MIN_H_REL=20. This is the relative minimum water depth (
in promille of the characteristic dimension) in a link
BRANCH_MIN_H_ABS=20. This is the absolute minimum water depth
(mm) in a link.
The minimum water depth in a link will be set to BRANCH_MIN_H_REL, cal-
culated as promille of the link size (e.g. pipe diameter or height of the open
channel), but never larger than BRANCH_MIN_H_ABS mm. In the pre-
sented example, the minimum water depth is set to 20 promille of the link
size, but with a maximum of 20 mm. This means that for links smaller than 1
meter the minimum water depth is set to 20 promille of the link size. For links
larger than 1 meter the minimum water depth is kept at 20 mm, independently
of the link dimensions.
76 MIKE URBAN - © DHI
Default Initial Conditions
5 Initial Conditions
The hydrodynamic computation is started from the flow conditions in the sys-
tems specified for time t = 0. MOUSE provides two different options for estab-
lishment of proper initial conditions.
5.1 Default Initial Conditions
MOUSE automatically specifies the initial conditions, establishing a default
initial water depth equal to 0.5% of the characteristic dimension of the conduit
(diameter for circular pipes), but not more than 0.005 m, and flow rates are
calculated based on the Manning formulation for uniform flow.
In case of dry weather flow applications, the volume of artificially generated
water may be significant compared to the dry weather flows. This may com-
promise the volume balance the analysis. For such cases, the default initial
depth can be reduced by setting the parameters BRANCH_MIN_H_REL
(controls the initial depth relative to the conduit size) and
BRANCH_MIN_H_ABS (controls the absolute depth of the initial water depth)
to appropriate values (in the DHIAPP.INI file).
If there are outlets in the system with initial water level specified higher than
the outlet bottom, a horizontal water surface is assumed extending inside the
system, until the point in the pipe system where the water level coincides with
the bottom level (see Figure 5.1).
Figure 5.1 Initial conditions with backwater outlet.
77
Initial Conditions
5.2 Initial Conditions provided by Hotstart
Realistic initial conditions can be specified by taking the water levels and dis-
charges from previously calculated result file. Flow conditions at any time
level contained in the interval covered by the result file can be chosen as ini-
tial condition.
The result file used as a HOTSTART file has to be complete, i.e. water levels
and flows at all computational points have to be saved.
78 MIKE URBAN - © DHI
Initial Conditions provided by Hotstart
6 Boundary Conditions
Unique solution of the flow equations requires appropriate set of boundary
conditions. Flow equations are solved for each conduit between two nodes,
and the boundary conditions are required at both end of the conduit, at each
time step throughout the computation.
In some situations boundary conditions are specified as unique relations of
two flow variables (e.g. stage/discharge relation), i.e. as hydraulic boundaries
in certain points. These are defined as functions, i.e. as a part of the system
description.
In other cases, proper boundary conditions are constructed by the model as a
consequence of current flow situation and of various user-specified distur-
bances in form of e.g. adding or extracting water, controlling the flow, adding
energy (pumping) or as effect of external water level. These disturbances
may be constant (stationary) or time-variable.
By default, MOUSE supplies all necessary boundary conditions, founded on
the topology and geometry of the system. Therefore, the simulations can be
run even if no boundary conditions of the other type are specified by the user.
With respect to the volume balance in the system, two groups of boundary
conditions can be distinguished:
1. External boundary conditions, describing the interaction of the
modelled system with its surroundings
2. Internal boundary conditions, describing relations between certain parts
of the model.
The external boundary conditions comprise the following:
At manholes and structures:
1. Constant inflow or extraction, Q = const.,
2. Time variable inflow or extraction, Q = Q(t),
3. Computed inflow hydrograph, Q = Q(t),
4. Weir discharging out of the system, Q = Q(H),
5. Pump discharging out of the system, Q = Q(H).
Application of negative inflows (extraction) should be done with due care,
because extraction of more volume than the system can supply would end up
with error in computations.
At outlets:
1. Constant outlet water level, H = const.,
2. Time variable outlet water level, H = H(t),
79
Boundary Conditions
3. Q/H relation at the outlet, Q = Q(H).
Internal boundary conditions can be defined as follows:
At manholes and structures:
1. Weir discharging to another manhole or structure, Q = Q(H), where H
stands for energy level above the weir crest in case of a free overflow,
and for difference of energy levels upstream and downstream of the weir
in case of a submerged overflow.
If an alternative formulation for the weir is specified with a user-specified Q/H
relation, such conditions should be provided that the overflow is always free,
i.e. that holds the unique relation between the water level and the flow.
2. Pump discharging to another manhole or structure, Q = Q(H) or Q =
Q(H), where H stands for water level in the manhole or structure, and
H level difference between the two manholes or structures associated
with the pump.
Some of the listed boundary conditions are illustrated in Figure 6.1.
80 MIKE URBAN - © DHI
Initial Conditions provided by Hotstart
Figure 6.1 Supplying Boundary Conditions – Examples (to be continued)
81
Boundary Conditions
Figure 6.2 Supplying Boundary Conditions – Examples (continued)
82 MIKE URBAN - © DHI
Friction Losses in Free-Surface Flow Links
7 Flow Resistance
7.1 Friction Losses in Free-Surface Flow Links
7.1.1 Numerical Description
Head losses caused by the resistance in free-surface flow links are intro-
duced as a friction slope term into the momentum equation (see section 3.2
Implementation of the Saint Venant Equations in MOUSE (p. 52)). The friction
slope If is equal to the slope of the energy grade line and is defined as:
I f = ----------- (7.1)
gR
where:
= tangential stress caused by the wall friction, [Nm-2],
= density of water, [kgm-3],
R = hydraulic radius, [m], (= A/P, where P is the wetted
parameter).
The friction slope can be derived as a function of an appropriate combination
of the flow parameters (Q, A and R) and the water and conduit wall properties
(, k).
Generally, the friction slope can be expressed as:
2
If = f Q (7.2)
where f is a generalised friction factor. By these means, the friction slope is
explicitly determined as a function of instantaneous values of local flow
parameters.
A more stable formulation is achieved through an implicit description of the
friction term. It is derived from a variational principle at a grid point j as:
n+ n n n+1
I fj = I fj + dI fj = 1 – I f + I f (7.3)
83
Flow Resistance
This results in:
n+ n
If j = I f j + dI f j (7.4)
n n 2
= f j Q j + dI f j
n
n 2 df
= f j Q j + 2f j Q j + Q j ------- h j
n n 2 n n 2
dh j
The coefficient determines the time weighting of the scheme. For stability
reasons the coefficient should be above 0.5. The recommended (also default)
value is 1.0, i.e. a fully forward time weighting of the scheme.
MOUSE provides an optional choice between the explicit and implicit flow
resistance description through the DHIAPP.INI file (see relevant documenta-
tion). The explicit description is selected per default.
7.1.2 The Friction Resistance Described by the Manning Formula
The "classic", explicit application of the Manning's formula reads as:
QQ
I f = ------------------------- (7.5)
2 2 43
M A R
with the friction factor:
1
f = ------------------------- (7.6)
2 2 43
M A R
where M is the Manning number, A the area, and R the hydraulic radius.
Usage of the QQ|, instead of Q2 facilitates computations of the reverse flow.
The Manning's number M (or n = 1/M) is the parameter used as a measure of
the conduit's wall roughness. Default values are given in section 2.2.2 Speci-
fication of a Link (p. 14).
The implicit formulation of the Manning's formula is obtained by the differenti-
ation of f with respect to h, which results in:
f f M f A 4 f R
------ = – 2 ----- -------- – 2 ---- ------- – --- ---- ------- (7.7)
h M h A h 3 R h
and substituting the derivative into the Equation (7.4).
84 MIKE URBAN - © DHI
Friction Losses in Free-Surface Flow Links
7.1.3 Depth-variable Manning coefficient
Per default, MOUSE assumes a constant Manning’s number over the link
section height. However, in real situations conduit wall roughness often
changes with water depth, because different parts of the link cross section
are exposed to quite different flow conditions during its lifetime. This intro-
duces difficulties in fitting the computed stage-discharge curve, based on a
single M value specified for a link, with the actual, measured stage-discharge
relation. This is usually related to old systems, where significant sediment
deposits and pipe-wall erosion are present.
The MOUSE Pipe Flow Model accepts a specification of a non-linear
variation of Manning number with relative elevation (water depth in the con-
duit). Three parameters define the Manning's number variation: bottom value,
full flow value and a non-linear exponent. Intermediate values are calculated
by a general expression:
y exp
M act = M bott + M top – M bott ---- (7.8)
D
where:
Mact = calculated Manning's number,
Mbott, Mtop = Manning's numbers specified for the conduit bottom and
top, respectively,
exp = Manning's number variation exponent, default,
y/D = the relative water depth in a conduit.
The formula is used for relative depths h/D in the interval 0.0 - 1.0. For rela-
tive depth > 1.0 the Manning number is set to the Manningtop value.
The variation between Manningbot and Manningtop is controlled by the "Varia-
tion Exponent". The variation of the Manning number in relative terms is illus-
trated in Figure 7.1. An example of the variation is shown in Figure 7.2 with
Manning M values.
85
Flow Resistance
Figure 7.1 Relative variation of the Manning number with relative depth.
h/D Manning
1.100 90.000 90.000 90.000 90.000 90.000 90.000
1.000 90.000 90.000 90.000 90.000 90.000 90.000
0.900 89.686 89.374 88.461 87.000 84.300 81.870
0.800 89.338 88.691 86.833 84.000 79.200 75.360
0.700 88.949 87.934 85.100 81.000 74.700 70.290
0.600 88.506 87.086 83.238 78.000 70.800 66.480
0.500 87.991 86.117 81.213 75.000 67.500 63.750
0.400 87.373 84.977 78.974 72.000 64.800 61.920
0.300 86.597 83.580 76.432 69.000 62.700 60.810
0.200 85.540 81.743 73.416 66.000 61.200 60.240
0.100 83.830 78.929 69.487 63.000 60.300 60.030
0.000 60.000 60.000 60.000 60.000 60.000 60.000
Exponent: 0.1 0.2 0.5 1.0 2.0 3.0
Figure 7.2 Variation of the Manning M for Manningbot=60 and Manningtop=90 with
different values of the variation exponent.
The Manning number variation is specified through the ASCII file *.ADP. The
specified Manning numbers in the *.ADP file must follow the selected option
for the Manning number convention. Syntax of the format of the '*.ADP' files
must be as shown in the 'DHIAPP.INI and *.ADP Reference Manual'.
The lines of the '*ADP' file related to the Manning number variation may be
easily compiled by copy-and-paste operations from the MOUSE650.OUT file.
This ASCII file is generated by every computation with the MOUSE Pipe Flow
Model.
86 MIKE URBAN - © DHI
Friction Losses in Free-Surface Flow Links
Before using the '*.ADP' file, the Manning number parameters for the
selected lines must be modified, i.e. values for bottom and top of pipe Man-
ning numbers and possibly the variation exponent must be adjusted for the
pipes or canals where varying Manning numbers are to be used.
7.1.4 Colebrook - White Formula for Circular Pipes
In 1939, Colebrook and White derived an approximate formula which unifies
the description of the turbulent flow in both rough and smooth circular pipes.
This formula is extensively used for the computation of flow resistance in pre-
dominantly full-flowing pipe networks.
According to Colebrook and White, the friction factor f is computed iteratively,
using one of the several formulations known from the literature. The formula
implemented in MOUSE reads:
2 cw 2 k
------- = cw 1 + cw 2 ln ---------3- ------- + cw 4 ---- (7.9)
f cw Re f cw R
where:
k = the equivalent wall roughness [m],
R = the hydraulic radius,
Re = the Reynolds number,
cw1, cw2, cw3, cw4 = empirical constants.
The default values of the constants cw1-cw4 are:
cw1 = 6.4
cw2 = -2.45
cw3 = 3.3
cw4 = 1.0
The default values can be modified through DHIAPP.INI file.
The actual friction slope is calculated by using the following relation:
2
Q f cw
I f = f cw -----------------
2
- f = -----------------
- (7.10)
2
2gA R 2gA R
The Colebrook -White friction resistance can only be used if an implicit friction
formulation is activated.
Use of the Colebrook-White formula must be restricted to circular pipes only.
Also, the Colebrook-White formula is fully valid for full-flowing pipes.
87
Flow Resistance
7.1.5 Hazen-Williams Equation
The Hazen-Williams equation is an empirical formula which relates the flow of
water in a pipe with the physical properties of the pipe and the pressure drop
caused by friction. It is used in the design of water pipe systems such as fire
sprinkler systems, water supply networks, and irrigation systems. The Hazen-
Williams equation has the advantage that the coefficient C is not a function of
the Reynolds number, but it has the disadvantage that it is only valid for water
and it not able to account for temperature or viscosity.
0 63 0 54
Q = kCAR S (7.11)
where:
k = is a conversion factor for the unit system (k = 1.318) for US
customary units, k = 0.849 for SI units),
C = is the roughness coefficient,
R = is the hydraulic radius,
S = is slope of the energy line (head loss per length of pipe)
7.2 Head Losses in Manholes and Structures - Introduction
The general flow equations are valid only for continuous conduits where, in
principle, the only resistance to the flow originates from the bottom and side-
wall friction. Hydraulic conditions in nodes, i.e. at manholes and structures
take the role of boundary conditions for computation of the flows in the con-
duits. In turn, hydraulic conditions in a node depend on the flows in the inlet
and outlet conduits.
These hydraulic conditions, expressed in terms of the energy conservation
principle, are calculated as water levels and velocity heads. The calculation is
based on the mass continuity and formulation of more or less advanced
energy relation between the node and the neighbouring links, with inclusion
of some energy losses caused by local flow disturbances at different loca-
tions in the node.
The implemented solution ensures that mutual dependence of the flows in
links and hydraulic conditions in nodes are correctly resolved, even for com-
plex branched and looped conduit networks.
Energy losses in junctions are of the same order of magnitude as those
caused by the pipe wall friction. Knowledge about the magnitude of these
energy losses, based on experimental data, is very limited, but some theoret-
ical results are available (e.g. ref. /3/). Importance of a detailed evaluation of
these losses is related to the relative length of the links (l/D), and grows with
relative shortening of the conduits.
88 MIKE URBAN - © DHI
‘Standard’ MOUSE Solution (F.A. Engelund)
7.3 ‘Standard’ MOUSE Solution (F.A. Engelund)
A simplified computational model for energy losses in junctions implemented
in MOUSE is based on F.A. Engelund's energy loss formulae (see ref. /5/).
Furthermore, a critical depth formulation, with approximation of critical flow
conditions, is used in MOUSE for simulation of a free inlet to a manhole.
7.3.1 Head Loss at the Node Inlet
It is assumed that the water levels in the inlet conduit and in the manhole or
structure are the same. This assumption implies that the energy loss of the
flow entering and expanding in the node amounts to the difference of the
velocity heads in the inlet conduit i and the node m, respectively:
2 2
vi – vm
E i = ------------------
- (7.12)
2g
Essentially one-dimensional analysis in MOUSE relies on this simplification
also in nodes with multiple inlet and outlet conduits, i.e. where mixing of flows
of different energy levels occurs. Therefore, in some extreme cases, where
head losses in nodes play a crucial role for the correct solution, it is advisable
to perform a more detailed analysis, in order to assess the approximation
errors inherent to this approach.
In a case of a free inlet of a sub-critical flow, i.e. when the water level in the
junction is lower than the critical depth level in the inlet link, the water level in
the link is assumed to be equal to the critical depth. For different cross sec-
tions, appropriate approximations are applied, e.g. for a circular pipe as fol-
lows:
Q
yc i = 0.32 --------i- (7.13)
Di
where:
Di=diameter of the circular pipe [m].
Similarly, in a case of a low water level in the junction with supercritical flow
(steep inlet links), the downstream water level is set equal to normal depth in
the link.
7.3.2 Head losses at the outlet from a node
All the individual losses in a node (except the inlet loss) calculated by the
model are added-up at the outlet, separately for each outlet link. The outlet
89
Flow Resistance
loss for the link j is assumed to be proportional to the velocity head in the out-
let link j:
2
v
E j = jk -----j- (7.14)
2g
k
where jk are individual head loss coefficients for link j, calculated on the
basis of geometrical set-up of the node and flow distribution among the links
attached to the node.
The model distinguishes among the following losses:
Change in flow direction,
Change in elevation,
Loss due to contraction at outlet.
Loss due to change in flow direction
This loss is a function of the angles between the inlet and outlet links and dis-
tribution of the discharge in the inlet and outlet links, as shown in Figure 7.3
and Figure 7.4.
Figure 7.3 Manhole consisting of 2 inlet links and 1 outlet link
Figure 7.4 Manhole consisting of 1 inlet link and 2 outlet links
90 MIKE URBAN - © DHI
‘Standard’ MOUSE Solution (F.A. Engelund)
Based on the generalised notation, the calculation of the head loss coefficient
is performed individually for each outlet link as follows:
n 2
Q i ij
dir j = ----- -----------
Q j 90 2
(7.15)
i=1
where i stands for inlet links, and j stands for outlet links.
Loss Due to Change in Elevation
Vertical changes in flow direction occur and cause energy losses if there is a
difference in elevation between inlet and outlet link. These losses are
described considering the magnitude of the difference in elevation, see
Figure 7.5.
Figure 7.5 Manhole with a difference in elevation between inlet and outlet pipe
The individual head loss coefficient is calculated according to the following
expression, where the weighting relative to the flow rates in the inlet links rel-
ative to the outlet link is also included:
n
Qi Zj – Zi Zj + Dj – Zi – Di
level j = -----
Qj
--------------------------------------------------------------------
Di Dj
(7.16)
i=1
If the calculated head loss coefficient is smaller than 0, a zero value is
assumed.
Loss Due to Contraction
The flow leaving the manhole and entering the outlet conduit is more or less
contracted and due to subsequent expansion there occurs an energy loss.
The outlet head loss coefficient depends on the shape of the manhole outlet,
manhole and the link cross sections and distribution of flow among multiple
inlet and outlet links.
91
Flow Resistance
MOUSE calculates the outlet head loss coefficient according to the following:
Aj
contr j = K m 1 – --------------------------- (7.17)
Qj
A m -------------
n
-
Q i
i=1
where:
Km = specified outlet 'shape' coefficient for the node. For relatively
large basins, Km approaches contr.
Am = flow cross sectional area in the node.
7.3.3 Implementation of the Total Energy Loss Computation
Theoretically, the total energy loss at the outlet from the node, expressed as a
function of the velocity head in the outlet pipe can be as high as the available
energy level in the node. The limiting case occurs e.g. with completely
clogged outlet (Km ->, with no flow in the outlet pipe.
However, in computational reality, in order to preserve a robustness of the
computation, various additional limitations could be introduced. With respect
to that, MOUSE offers two possibilities.
The first limitation relates the maximum head loss to the depth in the outlet
pipe:
2 2
v v
h j = min h j , 1 + out -----j- – -----m- (7.18)
2g 2g
It also introduces the limitation on the total head loss coefficient as:
out = j 1.0 (7.19)
These limitations have caused that the computed head losses and the corre-
sponding flow conditions around nodes in some cases were inexact.
92 MIKE URBAN - © DHI
An Alternative Solution Based on Weighted Inlet Energy Levels
Due to the advances in the computational implementation, the limitation from
Equation (7.19) could been removed, allowing the total energy loss for the
outlet pipe j being computed as:
2
v
E j = min dir j + level j + contr j ,1 -----j- (7.20)
2g
The limitation of the total head loss coefficient to 1.0 is, however, still present.
7.4 An Alternative Solution Based on Weighted Inlet Energy
Levels
The assumption applied in the MOUSE ‘standard’ solution that the water level
in the manhole and all downstream water levels of the inflowing conduits are
the same, often leads to overestimates of the energy loss at the inlet. In many
cases, the wetted cross-section area in the inlet pipe is smaller than in the
manhole, leading to almost entire loss of the kinetic energy of the incoming
flows, which is not the case. This problem is reduced by applying the effective
flow area in the manhole, but this is available in MOUSE only for circular
pipes and for the flow-through manholes (i.e. with one pipe in- and one pipe
outflow).
An alternative solution is available which fully ignores the energy loss at the
inlet. For a flow-through manhole, this practically means that the energy level
in the manhole is set to be equal as at the downstream end of the inlet pipe.
For manholes with multiple inlets, the energy level is calculated as the
weighted average of the inlet flows (i.e. large flows contribute most to the
energy level).
Thus, in this formulation, the total loss at the manhole is concentrated compu-
tationally at the outlet, and can be fully controlled by the user.
Without doubt, this approach proves valuable for some specific situations,
particularly for the flow-through manholes with ‘normal’ flow conditions. How-
ever, due attention must be paid for cases with high inlet energy
levels, e.g. a small pipe with high-velocity flow entering a large basin. In such
a case, the energy level of otherwise still water in the basin would be calcu-
lated as equal to the energy level of the approaching flow, i.e. much higher
than realistic, with erroneous results as a consequence.
7.5 Selecting an Appropriate Local Head-loss Computation
In some cases, results from using different approach for node head loss cal-
culation can be considerably different, and due attention must be paid to the
selection of the most appropriate approach.
93
Flow Resistance
The head loss calculation for individual nodes can be controlled by selecting
one of the available options - there are default five options in MIKE URBAN.
You can always create your own option.
Each option is characterised by the fundamental computational principle and
by a number of parameters which control the behaviour of the algorithm or
the size of loss coefficient. In MIKE URBAN, the existing options can be mod-
ified and new options created in "Outlet Head Loss" dialog. Furthermore, the
actual head loss calculation for individual nodes can be both in MOUSE and
in MIKE URBAN controlled by a local specification of various relevant param-
eters.
7.5.1 Constitutive Parameters of Head Loss Computation Options
The following parameters constitute a definition of head loss calculation
option:
Computation Method
Three different methods are avialable:
MOUSE Classic (Engelund) - described in section 7.3
Weighted Inlet Energy Method - described in section 7.4
No Head Loss Calculation
The first two are described in detail in respective chapters.
The third option ignores all local losses. Regardless of the shape of the out-
lets, geometrical set-up of the junction and distribution of flows among inlet
and outlet conduits, water levels in the junction and the outlet conduit are set
equal, as if there is no change of geometry and the flow conditions between
the junction and outlet conduit. This literary means that this option should be
applied only where there is no change in cross section. If inappropriately
applied, inconsistent results may be generated.
On the contrary, this option can be recommended for use if an artificial node
is introduced somewhere on a straight section of a conduit, where actually
no losses occur.
Maximum Loss Limit
This parameter is of relevance for both MOUSE Classic and Weigthed Inlet
Energy computation methods. It actually sets the limitation on the maximum
computed headloss to the water depth or the velocity head in the outlet pipe,
according to equations (7.18) to (7.20).
94 MIKE URBAN - © DHI
Selecting an Appropriate Local Head-loss Computation
Loss Coefficient
The available loss coefficient types distinguish three different interpretations
of the specified head loss coefficient.
Selection of "Km " (Type (a)) interprets the specified value as the outlet
'shape' coefficient Km (see Equation (7.17)).
Selection of "Contraction HLC" (Type (b)) interprets the specified value as the
outlet 'contraction' coefficient zcontr(j) (see Equation (7.17)). This means that
the model ignores the geometrical relations between the node and the outlet
links (outlet shape), and applies the specified value directly as the zcontr. The
contraction losses in the outlet links are then computed by multiplying the
velocity head in the respective link by the zcontr. The total head loss for an
outlet link is computed as a sum of the contraction, direction and elevation
loss.
Selection of "Total HLC" (Type c)) interprets the specified value as the total
outlet head loss. This means that the model completely ignores the geometry
of the node/links, and applies the specified value (Total HLC) directly as the
zout, the same for all outlet links at the node. The total head losses in the out-
let links are then computed by multiplying the velocity head in the respective
link by the specified zout.
Effective Node Area
This parameter is only relevant for MOUSE classic computational method
and for flow-through manholes in circular pipes. In all other cases, the default
total wetted node area is applied. The following choices are available:
Total wetted area: calculated as product of diameter and water depth for
manholes and red from the basin geometry table (Ac) for basins. Typically
results in overestimate of local loss in a node.
Calculated Effective Area: The effective area in a manhole is calculated on
the basis of empirical formula (see section 2.3.2 Types and Definition of
Nodes (p. 19)). This results in a significantly smaller area than full wetted
area and, consequently with a more realistic flow calculation.
Reduced Calculated Effective Area: The effective area in a manhole is further
reduced to 50% of the calculated effective area.
7.5.2 Default Computational Options
The following tables provide an overview of available head loss calculation
options in MIKE URBAN.
95
Flow Resistance
7.5.3 Example: Node Outlet Head Losses Variation as Function of Head Loss
Coefficient Mode
In this example a simple sewer system consisting of two pipes, two manholes
and one outlet is constructed. Tests for different head-loss types (a), (b) and
(c) have been performed with various modifications in flow direction or drop
height, or both. Table 7.1 shows a complete test matrix.
Four variants of the model setup have been constructed:
I)Straight sewer pipelines with no drops and no changes in direc-
tions.
II)A change in direction is introduced in variant I).
III)A drop is introduced in variant I).
IV)A drop and a change in direction are introduced in variant I).
A definition sketch of the setups I)-IV) is shown in Figure 7.6. The
manual calculation example corresponds to test No. 4 in the test matrix.
Figure 7.6 Example definition sketch
In the performed tests, the value of the HEADLOSS COEFFICIENT has been
set to 0.5 for all three types (a), (b) and (c). The head loss coefficients for
drop in the setup III) and IV) is 0.4 (inlet pipe is 0.6 m above the bottom in
manhole B). The head loss for direction in the setup II) and IV) is 0.25 (angle
between pipes are 45).
96 MIKE URBAN - © DHI
Selecting an Appropriate Local Head-loss Computation
The example also includes calculation of the friction loss in the downstream
pipe.
Manual Head Loss Calculation
Assumptions:
The water level in the inlet pipe is assumed equal to the water level in the
manhole. This implies that the expansion loss at the inlet is automatically
assumed. All calculated energy losses are assumed to occur at the outlet
pipe, i.e.:
E manhole = E pipe + H (7.21)
or expressed by using the notation in Figure 7.6:
2 2 2
vm v out v out
- + y out + Z out + --------
------ + y m + Z m = -------- - (7.22)
2g 2g 2g
= a coefficient expressing the total outlet energy loss (see
section 7.3.2 Head losses at the outlet from a node (p. 89)).
Data:
Discharge Q = 2.0 m3s-1
Diameter in outlet pipe Dout = 1.0 m
Diameter in manhole Dm = 1.5 m
Velocity in outlet pipe
(Q is capacity assumed) Vout ~ 2.55 ms-1
Length of outlet pipe L = 50.0 m
Manning number M = 70 m1/3s-1
Water level in outlet Hout = 15.0 m
Bottom level in manhole Zm = 14.0 m
Head “shape” loss coefficient Km = 0.5
Friction loss in outlet pipe from manhole to outlet:
2 2
Q L 2.0 50.0
h f = ------------------------- = -----------------------------------------------------
- (7.23)
2 2 43 2 2 43
M A R 70 0.7854 0.25
The water level in the manhole (H) can be found from:
2 2 2
v v out v out
H + -----m- = H out + h f + --------
- + --------- (7.24)
2g 2g 2g
97
Flow Resistance
vm is approximated by assuming that ym = yout:
v m = ------
Q
Q- = -------------- 2.0
- = ------------------------------------------------------------ = 0.94ms
–1
(7.25)
Am Dm ym 1.5 15.0 + 0.42 – 14.0
2
Q 2.0 –1 v out
v out = ---------- = ---------- = 2.55ms , --------
- = 0.33m (7.26)
A out 0.78 2g
2
vm
------ = 0.045 m (7.27)
2g
= contr + dir (7.28)
A out Q in Q in 2
= K m 1 – ----------
- -------- + ----------- --------
Q out A m Q out 90 2
2
0.78 45
= 1 – ---------------------- + -------2- = 0.567
1.5 1.42 90
Substituting values to the Equation (7.25) and calculating H yields:
H = 15.89 m (7.29)
The deviation between the MOUSE simulation and manual calculation result
is due to the fact that MOUSE calculates vout by using the following area in
the pipe:
2
D out
A out = -------------
- + Preismann slot area
4
The table shows which setups have been used for the calculation, and also
which head loss types are included. c0, c1, c2 are all head loss coefficients
due to contraction and correspond to the types (a) , (b) and (c).
Table 7.1 Test matrix for implementation fo Head Loss Type
Table 7.2
Contraction c
Test Setup Direction d Drop l (a) b)c1 (c)c0 Hcalc HM
c0
1 I) X 15.81 15.82
2 I) X 15.87 15.87
3 I) X 15.87 15.87
98 MIKE URBAN - © DHI
Selecting an Appropriate Local Head-loss Computation
Table 7.2
Contraction c
Test Setup Direction d Drop l (a) b)c1 (c)c0 Hcalc HM
c0
4 II) X X 15.89 15.90
5 II) X X 15.95 15.94
6 II) X X 15.87 15.87
7 III) X X 15.94 15.94
8 III) X X 16.00 15.99
9 IV) X X X 16.02 16.02
10 IV) X X X 16.03 16.02
7.5.4 Implementation of Head Loss Description in Kinematic Wave
Simulations
When applying the kinematic wave approximation, the head loss description
in nodes is based on the same equations as described above.
However, in order to reduce the computational time, the energy losses are
computed once for a number of different flow conditions and tabulated for use
during the simulation. In cases where there is more than one inlet link in a
manhole, the losses are calculated on the basis of the assumption that the
flow in each link (relative to the flow in the other inlet links) is proportional to
the corresponding full flow capacity. This assumption affects the energy
losses due to changes in elevation and direction only when these losses are
different for the different inlet links.
99
Flow Resistance
100 MIKE URBAN - © DHI
Surface Flooding
8 Some Special Techniques
8.1 Surface Flooding
If the water level in a manhole or a basin reaches the ground level, an artifi-
cial “inundation” basin is inserted above the node. The surface area of this
basin is gradually (over one meter) increased from the area in the manhole or
the basin to a 1000 times larger area, thus simulating the surface inundation.
The maximum level of inundation is 10 meter above the specified ground
level. When the outflow from the node surmounts the inflow, the water stored
in the inundation basin re-enters the system.
When the water level in the node increases and is above ground level, the fol-
lowing is assumed:
During a time step the surface area in the basin is calculated using the water
level from the start of the actual time step. A situation like this is shown in
Figure 8.1. If the water level passes through the transition region between the
actual manhole or structure and the artificial basin, this assumption leads to
generation of water. In Figure 8.1 the shaded area illustrates the generated
volume of water.
Figure 8.1 Simulation of the surface flooding
When the increase of the water level during a time step is relatively small,
then the generated water volume is negligible. If the water level is changing
rapidly, the generated volume of water is important and due to that an appro-
priate correction is built in the program to ensure no generation of water.
An alternative to the assumption of constant surface area during a time step
is to introduce iterations in the simulation. Iterations would significantly
increase the simulation time.
8.2 Sealed Nodes
Any manhole or basin can be defined as 'sealed'. If a node is defined as a
sealed node, then the maximum water level at a node is set to the ground
101
Some Special Techniques
surface. In this case, the pressure will rise without any water on the ground
surface. The following relations are valid:
H m = P m for P m H top (8.1)
and
H m = H top for P m H top
where:
Hm is the water level in the node [m],
Pm is the pressure level in the node [m],
Htop is the ground level for the node [m].
8.3 Spilling Nodes
Any manhole or basin can be defined as 'spilling'. If the water level in a node
defined as a spilling node reaches the ground level, the water will start spilling
irreversibly out of the system. The flow will be computed using the free over-
flow formula, according to the following:
for H m H top + P : (8.2)
Q spill = 0;
for H m H top + P :
32
Q spill = Relative Weir Coefficient 0.63 B 2g H m – H top + P
where:
Qspill = the spill discharge [m3/s],
B = a conceptual spill width [m],
Hm = the water level in the manhole [m],
Htop = the ground level in the manhole [m],
P = the "Buffer Pressure Level" for the spill [m],
g = the acceleration of gravity [ms-2),
RelativeWeirCoefficient = the linear scaling coefficient for the spill.
The level (i.e. head) at which the spill starts can be controlled by optionally
specifying the 'Buffer Pressure Level' as a relative elevation above (or below)
the ground surface (default value = 0).
For circular manholes, the spill width B equals to 1.5 times the manhole diam-
eter for the water level Hm = Htop+P. With increasing water level, the spill
width B increases, following the same functional relation as used for the
"basin" area above surcharging nodes (i.e. increases exponentially to approx-
imately max. 1000 times the manhole diameter) - see paragraph 8.1. For
102 MIKE URBAN - © DHI
Pressure Mains
nodes defined as basins, the spill width B is set equal to the square root of
the basin surface area.
The spilling capacity of a spilling manhole can be controlled by specifying the
Relative Weir Coefficient (default value = 1).
8.4 Pressure Mains
The ‘pressure mains’ (also referred to as rising mains in earlier versions of
MOUSE) feature is intended for modeling the permanently pressurized indi-
vidual pipes or networks in connection to pumps. Computationally MOUSE
assumes that a rising main network always runs under pressure and there-
fore the reaction time within the rising main network is insignificant.
Solution in pressure mains is based on the two equations:
Q
------- = 0 (8.3)
x
and
Q
------- + gA y
------ = gA I 0 – I f (8.4)
t x
where
Q = discharge, [m3s-1]
A = flow area, [m2]
y = flow depth, [m]
g = acceleration of gravity, [ms-2]
x = distance in the flow direction, [m]
t = time, [s]
I0 = bottom slope
If = friction slope
All nodes within the pressure main networks are assumed to be sealed.
MOUSE supports modeling of an arbitrary number of pressure main networks
and there is no limitation on the number of elements in each sub network.
Pressure main networks must always converge down to one receiving man-
hole, which is called the tail node. The tail node is the point of transition
between domains where the hydraulic solution is based on the St. Venant
equation and the special pressure main model.
The computation of the special pressure main sub models uses the maximum
of the water level in the “St Venant” governed domain and the water level at
the tail nodes as downstream boundary conditions. As default it is assumed
103
Some Special Techniques
that the tail node water level is equal to the maximum of the up-vert level of all
pressure pipes attached to the tail node, but the user can change this default
value.
The “upstream” pressure main network must be linked with the St. Venant
controlled domain through pumps. The “pressure mains” feature can
handle an unlimited number of pumps attached to one pressure main net-
work, but the solution feature can only handle networks where the upstream
link to the St Venant domain is modeled by pumps.
8.5 Dry Conduits
If parts of the sewer system dry-out during the simulation, then the model arti-
ficially maintains a minimum water depth in those conduits, corresponding
(per default) to 5‰ of the characteristic dimension of the conduit (diameter for
circular pipes), or max. 5 mm. This is necessary with regards to the numerical
stability in the solution of the flow equations.
This correction practically means artificial generation of water, i.e. some
water volume is added to the system. As a consequence of that, the continu-
ity status report shown at the end of the simulation does not give a fair
impression of the accuracy of the simulation.
8.6 A note on flooding and spilling
The term "flooding" is used when water level is above ground level, and
"floodrate" is the volume of water per second running from below ground level
to above ground level.
Spilling is defined when water is running out of the system and "spilling rate"
is the volume of water per second running out of the system.
Flooding, floodrate, spilling, and spilling rate can all be reported in an .XRF-
file for MOUSE and in the .AddOut.res1d-file for MIKE 1D.
Note that floodrate can be both positive (water coming from below ground
level) and negative (water running into the ground level), while spilling rate
can only be positive.
Since both nodes and pipes can have Preismann slots that emulate pressur-
ized nodes or pipes, it often happens that there is flooding in the system but
NOT spilling. Thus, flooding does not necessarily mean that there is water
running out of the system onto the ground.
Water volume above ground level in nodes is calculated as total water vol-
ume on the node minus water volume below ground level. This means that
sealed nodes can also have some (small) amount of flooding due to the slot
which extends above ground level.
104 MIKE URBAN - © DHI
A note on flooding and spilling
Spilling nodes have two input parameters - buffer pressure and spilling coeffi-
cient. The buffer pressure represents the water depth above ground needed
before water is lost from the system. Hence, if you set the buffer pressure to
e.g. 10 cm then water will not spill before the water level is more than 10 cm
above ground level. In this case you will also see flooding because water is
between ground level and ground level + 10 cm, but it is not lost from the sys-
tem. Hence when the water level is reduced this volume of water is going
back into the network again. Hence some water is flooding and going back to
the network again while another volume is spilled and lost to the system.
The buffer pressure is used to model the effect of two physical processes: 1)
a cover that requires a certain pressure before it lifts and water is lost (in this
case the water level above ground is really a pressure and hence no real
water is on the ground), and 2) a sewer grate that is located in a small
depression in the road and it is not until the water level is above 10 cm that
water starts running down the street and is lost to the system (in this case
there actually is water on the ground even though it is not lost.)
When the buffer pressure is zero, the spilling coefficient is still used. Spilling
is calculated as water spilling over a weir. The weir coefficient is equal to the
spilling coefficient and the weir width is calculated from the circumference of
the manhole. In some cases the spilling weir becomes a restriction on the
flow and then the water level will build up behind the weir (i.e. in the manhole
itself) and the water level will become higher than the ground. Also this is
reported as flooding, because the water level is above ground level.
105
Some Special Techniques
106 MIKE URBAN - © DHI
A note on flooding and spilling
9 Nomenclature
a,b,c,..,zquasi-constants in a modified continuity equation around a
node
A,B,..,Zquasi-constants in a generalised continuity equation around
a
node
athe speed of sound in water, with actual pipe walls rigidity,
(ms-1)
avertical distance from the point where the jet intercepts the
manhole to the centreline of the inlet
aothe speed of sound in water for absolutely rigid pipe walls,
(ms-1)
ar the speed of sound along pipe walls, (ms-1)
Across-section area, (m2)
Afloweffective flow area in a manhole, (m2)
Ajetcrosssection area of the jet at the point of interception with the
manhole, (m2)
Am crosssection area of the wet part of the manhole, (m2)
Afullcross-section area at full pipe flow, (m2)
Aothe cross-section area without excess pressure, (m2)
Atstructure, wetted cross section area, (m2)
A0structure, water surface area, (m2)
A0,jsurface area between grid points j-1 and j
A0,j+1surface area between grid points j and j+1
b vertical distance from the point where the side of the outlet
enters the manhole to the centreline of the inlet
bsurface width, (m)
bsstorage width, (m)
bslotwidth of Preissmann slot, (m)
Boverflows, width, (m)
107
Nomenclature
C = AR2/3cross-section conveyance, (m8/3)
CrCourant number
CdCoefficient of discharge
CECoefficient of discharge (energy based)
CHCoefficient of discharge (level based)
Dpipe diameter, (m)
Din diameter of the inlet pipe, (m)
Dm diameter of the manhole, (m)
Dout diameter of the outlet pipe, (m)
drop_ factor factor diminishing the effective flow area in a manhole
due to
drop in elevation
ethe pipe wall thickness, (m)
Eenergy level just upstream overflow (m)
expManning's number variation exponent, default 1.00
Erthe Young's modulus of elasticity, (Nm-2)
fcoeff. for flow direction change (default f =1)
g = 9.81 constant, acceleration of gravity, (ms-2)
FrFroude's number
hwater level, (m)
Hcross sections - elevation relative to bottom, (m) pumps, water
level in a pump-sump, (m) overflows, water level just upstream
the overflow, (m)
Hoverflows, entrance energy loss, (m) pumps, level difference
between two nodes
HAregulation, water level at the control point A, (m)
Hbottnode, bottom elevation, (m)
Hmwater level in a node, (m)
Hmin,
108 MIKE URBAN - © DHI
A note on flooding and spilling
Hmaxregulation, water levels at the control point A defining the range
in which the regulation is to be applied, (m)
Houtwater surface elevation at outlet, (m)
Hstart,
Hstoppumps, start and stop level for a pump, (m)
Htopnode, surface elevation, (m)
Hup,Hdown or
H1,H2water levels at the computational points upstream and down
stream, respectively, (m)
I0bottom slope, (m-1)
Iffriction slope, (m-1)
kwall roughness (m)
Kcoverflows, energy loss coefficient
Kmspecified outlet 'shape' coefficient for a node
lconduit length, (m)
lrpipe length which gives rise to pressurised flow, (m)
MManning number, (m1/3s-1)
Mactcalculated Manning's number, (m1/3s-1)
MbottManning's numbers specified for the conduit bottom, (m1/3s-1)
MfullManning number at full pipe flow, (m1/3s-1)
ninvers of manning number (1/M)
Nnumber of grid points in a pipe
qspecific discharge, (m2s-1)
Qdischarge, (m3s-1)
Qfullfull pipe flow for uniform flow conditions, (m3s-1)
Q(HA)regulation, discharge defined by the regulation function, (m3s-
1)
Qnatregulation, "natural" discharge, (m3s-1)
109
Nomenclature
Qregregulation, applied discharge, (m3s-1)
Qweiroverflows, discharge, (m3s-1)
R = A/Phydraulic radius, (m)
Rfullhydraulic radius at full pipe flow, (m)
ttime, (s)
tcomputational time step, (s)
vmean flow velocity, (ms-1)
vmflow velocity in a node, (ms-1)
w0gate opening
w1distance from the overflow crest to the upstream bottom (m)
w2distance from the overflow crest to the downstream bottom (m)
xdistance in the flow direction, (m)
xdistance between two computational points, (m)
x,ynode co-ordinates, (m)
ydepth, (m)
ydepth in a contracted section (m)
y1 y2 y3 depth in upstream, central and downstream section, (m)
yccritical depth, (m)
ynnormal (natural) depth, (m)
y/Dthe relative water depth
Zgeneralised flow variable, substituting h and Q
Coriolis' velocity distribution coefficient
,‚, coefficients in finite-difference equations
(j)total calculated node head loss coefficient for outlet conduit j
dircalculated node head loss coefficient, due to change of direction
elevationcalculated node head loss coefficient, due to change of
elevation
110 MIKE URBAN - © DHI
A note on flooding and spilling
contr(j)calculated outlet contraction head loss coefficient, for outlet
conduit j
weighting coefficent of the numerical scheme
ijhorizontal angle between inlet conduit i and outlet conduit j
water density, (kgm-3)
odensity of water for a free surface flow, (kgm-3)
tangential stress caused by the wall friction, (Nm-2)
kinematic viscosity (m2s-1)
vvertical contraction coeff.
111
Nomenclature
112 MIKE URBAN - © DHI
A note on flooding and spilling
10 References
/1/MOUSE User Manual and Tutorial, DHI 1999
/2/ Abbott, M.B.: Computational Hydraulics, Elements of the Theory
of Free Surface Flows, Pitman, 1979.
/3/ Pedersen, F.B., Mark, O.: Head Losses in Storm Sewer Man-
holes: Submerged Jet Theory, Journal of Hydraulic Engineering,
Vol. 116, No. 11, November 1990.
/4/Cunge J.A. and Wegner M. (1964): Integration numerique des
equations d'ecoulement de Barre de Saint-Venant par un schema
implicite de differences finies: Application au cas d'une galerie tan-
tot en charge, tantot a surface libre. La Houille Blance, No 1
/5/ F.A. Engelund og Fl. Bo Pedersen: Hydraulik, Den Private
Ingeniørfond, Danmarks Tekniske Højskole, ISBN 87-87245-64-7
(In Danish)
113
References
114 MIKE URBAN - © DHI
You might also like
- Flow Theory and Capacity in Sewage SystemsNo ratings yetFlow Theory and Capacity in Sewage Systems20 pages
- EPANET Water Distribution Modeling GuideNo ratings yetEPANET Water Distribution Modeling Guide51 pages
- Openflows Sewerops Connect Edition: Operational Decision Support Software For Wastewater ConveyanceNo ratings yetOpenflows Sewerops Connect Edition: Operational Decision Support Software For Wastewater Conveyance2 pages
- Outflow Distribution in Multi-Port DiffusersNo ratings yetOutflow Distribution in Multi-Port Diffusers8 pages
- Openflows Waterops Connect Edition: Real-Time Operational Decision Support Software For Smart Water DistributionNo ratings yetOpenflows Waterops Connect Edition: Real-Time Operational Decision Support Software For Smart Water Distribution2 pages
- Cavitation and Hydrofoil CharacteristicsNo ratings yetCavitation and Hydrofoil Characteristics31 pages
- Ebook Let Your Saving Flow With Openflows en LRNo ratings yetEbook Let Your Saving Flow With Openflows en LR14 pages
- Ms Thesis - Water Hammer Analysis of Villalba-Coamo Potable Water Transmisson LineNo ratings yetMs Thesis - Water Hammer Analysis of Villalba-Coamo Potable Water Transmisson Line462 pages
- Understanding Waterhammer in Pump SystemsNo ratings yetUnderstanding Waterhammer in Pump Systems4 pages
- Manifold Flow in Pressure-Distribution SystemsNo ratings yetManifold Flow in Pressure-Distribution Systems7 pages
- Chlorine Dose & THM Prediction with EPANETNo ratings yetChlorine Dose & THM Prediction with EPANET8 pages
- Ghanim - Objectives of The Open Channel Flow PracticalNo ratings yetGhanim - Objectives of The Open Channel Flow Practical6 pages
- Water Hammer Arresters - Engineer Report PDFNo ratings yetWater Hammer Arresters - Engineer Report PDF48 pages
- Water Supply and Sanitary Engineering Seminar (Sewer Appurtenances)No ratings yetWater Supply and Sanitary Engineering Seminar (Sewer Appurtenances)20 pages
- Water Cad Training Manual AA - Docx (Final) .Docx Eneyew100% (1)Water Cad Training Manual AA - Docx (Final) .Docx Eneyew33 pages
- Selection of Transient Analysis SoftwareNo ratings yetSelection of Transient Analysis Software11 pages
- Direct Rainfall Modelling and Interflow: JacobsNo ratings yetDirect Rainfall Modelling and Interflow: Jacobs46 pages
- (Esch B.P.M. Van, (1997) ) Simulation of Three-Dime PDFNo ratings yet(Esch B.P.M. Van, (1997) ) Simulation of Three-Dime PDF152 pages
- Pump-as-Turbine For Energy Recovery Applications: The Case Study of An AqueductNo ratings yetPump-as-Turbine For Energy Recovery Applications: The Case Study of An Aqueduct8 pages
- Technology Selection Approaches: Will Sarni CEO Water FoundryNo ratings yetTechnology Selection Approaches: Will Sarni CEO Water Foundry16 pages
- Dynamic Programming and Taguchi Method Optimization of Water-Treatment-Plant DesignNo ratings yetDynamic Programming and Taguchi Method Optimization of Water-Treatment-Plant Design10 pages
- Water Treatment Plant Optimization GoalsNo ratings yetWater Treatment Plant Optimization Goals2 pages
- Please Read This Entire Document.: Static Control Targetpro ™ Software and Test Target FilesNo ratings yetPlease Read This Entire Document.: Static Control Targetpro ™ Software and Test Target Files3 pages
- SMART Product Drivers 12 and Smart Ink 5: Basic Installation and Administration GuideNo ratings yetSMART Product Drivers 12 and Smart Ink 5: Basic Installation and Administration Guide32 pages
- E1502-1000 - R2 PocketAE & Pocket AEwin Users ManualNo ratings yetE1502-1000 - R2 PocketAE & Pocket AEwin Users Manual40 pages
- Software License Agreement For Hexagon PPM ("PPM") Software ProductsNo ratings yetSoftware License Agreement For Hexagon PPM ("PPM") Software Products12 pages
- 989 592N TCM400 Operators Manual EnglishNo ratings yet989 592N TCM400 Operators Manual English128 pages
