Buletinul Ştiinţific al Universităţii POLITEHNICA Timişoara

Seria HIDROTEHNICĂ
TRANSACTIONS on HYDROTECHNICS
Volume 59(73), Issue 2, 2014

Numerical Limitations of 1D Hydraulic Models using

MIKE11 or HEC-RAS software
Robert F. Beilicci1 Erika Beilicci1

Abstract – MIKE 11 is a advanced hydroinformatic RAS one-dimensional approach. It is capable of

tools, professional engineering software package for modeling subcritical, supercritical, and mixed flow
simulation of one-dimensional flows in estuaries, rivers, regime flow along with the effects of bridges,
irrigation systems, channels and other water bodies. culverts, weirs, and structures.
MIKE 11 is a 1-dimensional river model. It was
developed by DHI Water • Environment • Health,
HEC-RAS is a computer program for
Denmark. The basic computational procedure of HEC- modeling water flowing through systems of open
RAS for steady flow is based on the solution of the one- channels and computing water surface profiles. HEC-
dimensional energy equation. Energy losses are RAS finds particular commercial application in
evaluated by friction and contraction / expansion. The floodplain management and flood insurance studies
momentum equation may be used in situations where to evaluate floodway encroachments. Some of the
the water surface profile is rapidly varied. These additional uses are: bridge and culvert design and
situations include hydraulic jumps, hydraulics of analysis, levee studies, and channel modification
bridges, and evaluating profiles at river confluences. studies. It can be used for dam breach analysis,
For unsteady flow, HEC-RAS solves the full, dynamic,
1-D Saint Venant Equation using an implicit, finite
though other modeling methods are presently more
difference method. The unsteady flow equation solver widely accepted for this purpose.
was adapted from Dr. Robert L. Barkau’s UNET HEC-RAS has merits, notably its support by the US
package. Fluid motion is controlled by the basic Army Corps of Engineers, the future enhancements
principles of conservation of mass, energy and in progress, and its acceptance by many government
momentum, which form the basis of fluid mechanics agencies and private firms. It is in the public domain
and hydraulic engineering. Complex flow situations and peer-reviewed. The use of HEC-RAS includes
must be solved using empirical approximations and extensive documentation, and scientists and
numerical models, which are based on derivations of the engineers versed in hydraulic analysis should have
basic principles (backwater equation, Navier-Stokes
equation etc). All numerical models are required to
little difficulty utilizing the software.
make some form of approximation to solve these Users may find numerical instability problems during
principles, and consequently all have their limitations. unsteady analyses, especially in steep and/or highly
Key words: Basic equations, numerical modelling, dynamic rivers and streams. It is often possible to use
limitations HEC-RAS to overcome instability issues on river
problems. HEC-RAS is a 1-dimensional
I. INTRODUCTION hydrodynamic model and will therefore not work
well in environments that require multi-dimensional
The basic computational procedure of HEC-RAS for modeling. However, there are built-in features that
steady flow is based on the solution of the one- can be used to approximate multi-dimensional
dimensional energy equation. Energy losses are hydraulics.
evaluated by friction and contraction / expansion. MIKE 11 is a advanced hydroinformatic tools,
The momentum equation may be used in situations professional engineering software package for
where the water surface profile is rapidly varied. simulation of one-dimensional flows in estuaries,
These situations include hydraulic jumps, hydraulics rivers, irrigation systems, channels and other water
of bridges, and evaluating profiles at river bodies.
confluences. MIKE 11 is a professional engineering software
For unsteady flow, HEC-RAS solves the full, package for the simulation of flows, water quality
dynamic, 1-D Saint Venant Equation using an and sediment transport in estuaries, rivers, irrigation
implicit, finite difference method. The unsteady flow systems, channels and other water bodies. MIKE 11
equation solver was adapted from Dr. Robert L. is a user-friendly, fully dynamic, one-dimensional
Barkau’s UNET package. modelling tool for the detailed analysis, design,
HEC-RAS is equipped to model a network of management and operation of both simple and
channels, a dendritic system or a single river reach. complex river and channel systems. With its
Certain simplifications must be made in order to exceptional flexibility, speed and user friendly
model some complex flow situations using the HEC- environment, MIKE 11 provides a complete and
1 Politehnica University of Timisoara, Faculty of Civil Engineering, Department of Hydrotecnnics, Timisoara, Str. George Enescu, no. 1/A,
Timisoara, Romania, www.hidro.upt.ro
11
effective design environment for engineering, water draw attention to some of the more common
resources, water quality management and planning limitations that are implicit to this operation, in the
applications. The Hydrodynamic (HD) module is the hope that this may encourage these models to be used
nucleus of the MIKE 11 modelling system and forms in (and only in) the manner for which they are
the basis for most modules including Flood intended.
Forecasting, Advection-Dispersion, Water Quality
and Non-cohesive sediment transport modules. The II. FLUID MECHANICS
MIKE 11 HD module solves the vertically integrated
equations for the conservation of continuity and Fluid mechanics is the study of fluids statics or
momentum, i.e. the Saint Venant equations. dynamics. As a branch of mechanics, fluid flow is
The MIKE 11 is an implicit finite difference model governed basic principles. However, application of
for one dimensional unsteady flow computation and these principles is far easier in theory than in practice
can be applied to looped networks and quasi-two due to the complexity of fluid flow. It is therefore
dimensional flow simulation on floodplains. The necessary to make assumptions that simplify the
model has been designed to perform detailed application of the controlling equations, and use
modeling of rivers, including special treatment of numerical modelling techniques to obtain solutions of
floodplains, road overtopping, culverts, gate openings complex problems. The principles of fluid mechanics
and weirs. MIKE 11 is capable of using kinematic, can be found in many textbooks. Simple descriptions
diffusive or fully dynamic, vertically integrated mass of the basic concepts and some of the more common
and momentum equations. Boundary types include simplifications, necessary for understanding how
Q-h relation, water level, discharge, wind field, dam these principles are used by numerical modelling
break, and resistance factor. The water level software, are provided below.
boundary must be applied to either the upstream or The mechanics of fluid flow is governed by three
downstream boundary condition in the model. The basic principles of conservation:
discharge boundary can be applied to either the Mass – The Lomonosov-Lavoisier law states that the
upstream or downstream boundary condition, and can mass of a closed system remains constant. If a
also be applied to the side tributary flow (lateral system is open then the rate of increase in the mass
inflow). The lateral inflow is used to depict runoff. within the control volume is equal to the cumulative
The Q-h relation boundary can only be applied to the mass flowrate into the control volume:
downstream boundary. MIKE 11 is a modeling dM
package for the simulation of surface runoff, flow,  mdS  dt (1)
S
sediment transport, and water quality in rivers,
If the fluid is incompressible then the equation can
channels, estuaries, and floodplains. The
be simplified by replacing mass with volume. For
computational core of MIKE 11 is hydrodynamic
steady-state conditions, this further simplifies to Q =
simulation engine, and this is complemented by a
constant.
wide range of additional modules and extensions
Momentum – Newton’s second law states that the
covering almost all conceivable aspects of river
rate of change of momentum of a body is
modelling.
proportional to the resultant force acting on the body
Fluid motion is controlled by three basic principles:
and is in the same direction. For a control volume,
conservation of mass, energy and momentum.
this may be written in differential vector format as:
Derivatives of these principles are commonly known 
as the continuity, energy and momentum equations.
d
mxv    F (2)
These principles are among the first taught in basic dt
fluid mechanics, and they form the foundation of the For a Newtonian fluid and assuming constant density
field of hydraulic engineering. However, as situations and viscosity, the equation of motion may be written
become increasingly complex, we lose track of these in the x-direction as:
essential principles. Basic equations are replaced by  v v v v 
empirical approximations, and mathematical   x  v x x  v y x  v z x  
 t x y z 
calculations with numerical models. Determining an (3)
 2 
equivalent surface roughness of a floodplain is far P   vx  2v x  2v x 
more difficult than estimating an equivalent g x      
x  x 2 y 2  z 2 
roughness height or a Manning’s roughness  
coefficient; solving a backwater equation for an Similar equations derived for the y and z directions
irregular channel would be an arduous task without are collectively known as the Navier-Stokes
the assistance of a numerical model. equations.
Numerical models come in a wide range of shapes Energy – The first law of thermodynamics states that
and flavours – one, two or three dimensions, steady the net energy supplied to a system is equal to the
or unsteady flow conditions etc. All are based on increase in energy of the system and the energy that
derivations of the basic principles. All are required to leaves the system as work is done:
make some form of numerical approximation to solve dE dQh dW
these principles. All have their limitations.   (4)
dt dt dt
The objective of this paper is to promote a basic
awareness of how numerical models operate and to
12
The ‘Backwater Equation’ is a derivative of the One-dimensional models make a number of
Energy Equation for steady flow along a streamline: approximations in line with their simplistic nature.
Some are so obvious that they (hopefully) cannot be
H 1 v2
 S f   f (5) missed, while others are not so well recognised. Flow
S DH 2 g properties must be calculated based on characteristic
properties of the cross-section (eg hydraulic diameter,
average velocity). Some software packages try to
III. 1D NUMERICAL MODELS provide greater flexibility by dividing each cross-
section into sub-areas (such as the main channel, left
Computational fluid dynamics can be defined as a and right overbanks), then applying various
branch of fluid mechanics that uses numerical weighting factors to the flow distribution between the
methods and algorithms to solve and analyse sub-areas and the travel distance of each component.
problems involving fluid flows. The term “CFD More complex software packages can simulate quasi-
model” is commonly used to refer to a high-order 2D situations as a series of inter- linked channels,
numerical model capable of solving complex flow however the definition of flow path and length is
situations with relatively few simplifications. inflexible. Even with these abilities, one-dimensional
In reality, all numerical models are CFD models modeling is only appropriate for modelling well-
(even a simple spreadsheet solution of the backwater defined and constant flowpaths; the model cannot
equation). There are generally considered to be two match the flexibility of two- and three-dimensional
methods of analysing fluid motion: by describing the modelling necessary for representing complex
detailed flow pattern at every point in the flow field channel/floodplain interactions.
(small scale or differential analysis), or by examining A less obvious simplification common to many
a finite region and determining the gross effects of numerical models (HEC-RAS, MIKE 11) is to
and on the region (finite or control-volume analysis). assume that the grade of the channel is small,
The complexity of real fluid flow makes it impossible nominally less than 1:10, and therefore the sine and
to solve the governing equations without making cosine of the channel slope can be assumed equal to
some form of simplifying approximation, even with zero and unity respectively.
the use of complex models and fast computers. Rather than using a physically derived coefficient,
Common practices include: simplification of the such as the Darcy friction factor f, most numerical
spatial and geometric properties, assumption of models estimate friction losses using an empirical
steady or quasi-steady flow conditions, neglect of approximation such as Manning’s coefficient. While
fluid properties that would have negligible influence Manning’s equation has been in use for over 120
in the circumstances being investigated, and use of years, it is perhaps this universal acceptance that
empirical formulae to approximate flow has lead to evident ignorance about the limitations of
characteristics. the equation. A popular misconception is that
Hydraulic models may be categorized by the spatial Manning’s roughness coefficient is a dimensionless
and temporal simplifications that the model employs. constant, whereas in reality it has units (s/m1/3) and
Each category has associated with it a number of is dependent upon the hydraulic radius. Care must
fluid property and dynamic assumptions. therefore be taken, not only in the estimation of
One-dimensional models assume that the flow is in appropriate roughness coefficients, but to realise that
one direction only, and there is no direct modelling of although a model has been calibrated for one
changes in flow distribution, cross-section shape, particular discharge, the performance may be
flow direction, or other two- and three-dimensional different for other flow conditions. Additionally,
properties of the flow. The channel geometry is because there is no direct modelling of two- and
typically represented as a series of cross-sections at three-dimensional flow effects, the roughness
fixed (but not necessarily uniform) intervals. coefficient must account for the contribution of these
Although often considered to be relatively simplistic aspects to hydraulic losses in the channel.
by modern standards, one-dimensional modelling Numerical models usually solve the backwater
remains a useful and valid tool in many situations. equation between adjacent cross-sections using an
One-dimensional hydraulic models may be iterative procedure called the standard step method,
categorized as steady or unsteady. While these where the backwater equation is integrated as:

appear, superficially, to be similar and share many of  v 2   v 2 
the same limitations, the basic hydraulic principles to  d  z0     d  z0   
numerically solve these two situations are very  2g   2g 
 2  1
different. Steady-state numerical models are in most (6)
cases based on a derivative of the 'backwater' (or v22 v12
LS f  C 
Energy) equation, while unsteady models are based 2g 2g
on a derivative of the Saint Venant (or Momentum)
equation. Each solution has its advantages and where the losses are separated into friction and
disadvantages, and neither is appropriate in all contraction/expansion losses with Sf a representative
situations. The derivation and implicit limitations of friction slope between the two sections, L is the
these solutions are described in the sections below. distance between the sections, C is a contraction or

13
expansion coefficient, d is the flow depth, z0 is the algorithm that cannot accommodate two boundary
invert level. conditions at the same boundary. As a consequence
The primary assumption of the integrated backwater they cannot model supercritical flow, for which both
equation used in steady-state numerical modelling is discharge and water level are controlled by the
that the flow is gradually varied (Henderson 1966). upstream boundary. Instead, supercritical flow
This implies that changes along the channel, such as conditions are ‘solved’ by suppressing the convective
cross-section shape, invert level, flow depth and acceleration as the Froude number increases.
pressure distribution, are relatively small over short III. CONCLUSIONS
distances. The backwater equation has questionable
or no accuracy in: areas of rapid acceleration or The study of hydraulics and fluid mechanics is
deceleration, where the assumption of a hydrostatic founded on the three basic principles of conservation
pressure distribution is no longer valid, areas of large of mass, energy and momentum. Real-life situations
turbulence and/or energy loss, and areas of large are frequently too complex to solve without the aid of
change in cross - section property where the numerical models. There is a tendency among some
assumption that the representative friction slope and engineers to discard the basic principles taught at
contraction/expansion losses can be estimated by university and blindly assume that the results
some combination of the section properties at each produced by the model are correct. Regardless of the
end. complexity of models and despite the claims of their
Unlike steady-state modelling, which uses a solution developers, all numerical models are required to
of the continuity and energy equations, unsteady make approximations. These may be related to
modelling is based on a solution of the continuity and geometric limitations, numerical simplification, or
momentum equations. The derivation of these the use of empirical correlations. Some are obvious:
equations into a format suitable for one-dimensional one-dimensional models must average properties
modeling is complex but fairly well documented. The over the two remaining directions. It is the less
vertically integrated equations of continuity and obvious and poorly advertised approximations that
momentum, commonly known as the Saint Venant pose the greatest threat to the novice user. Some of
equations, may be presented as: these, such as the inability of one-dimensional
Q A unsteady models to simulate supercritical flow can
 q 0
x t cause significant inaccuracy in the model predictions.
(7)
Q Qv  z 
  gA  S f   0 REFERENCES
t x  x 
where Q is the discharge, q is the lateral inflow (per [1] Henderson, F.M. (1966). Open Channel Flow. MacMillan
unit length), A is the flow area, z is the free- surface Company, New York, USA.
elevation and v is the velocity. [2] Liggett, J.A. (1975). Unsteady Flow in Open Channels WRP
Publ. Vol. 1 Fort Collins, USA, pp. 29-62. Liggett, J.A. (1994).
Fluid Mechanics. McGraw-Hill, New York, USA.
While the backwater equation is based on a steady- [3] MIKE by DHI (2014) MIKE 11 A Modelling System for Rivers
state differential form of the Energy equation, the and Channels Reference Manual
Saint Venant equation based solutions can model [4] USACE (20138). HEC-RAS River Analysis System Hydraulic
Reference Manual, v4.0, California, USA
unsteady flow conditions. Many software packages,
including HEC-RAS and MIKE 11, adopt an

14