Sei sulla pagina 1di 8

7th International Conference on Hydroinformatics

HIC 2006, Nice, FRANCE

NUMERICAL MODELLING OF THE MORPHOLOGICAL


BEHAVIOR OF RIVER ROCHA FOR FLOOD MITIGATION

ROMERO MAURICIO
Department of Civil Engineering, Hydraulics Laboratory (LHUMSS), San Simón University,
Avenida Petrolera Km 4.2, Cochabamba, Bolivia

URQUIETA ALEJANDRO
UNESCO-IHE Institute of Water Education, Hydraulic Engineering and River Basin Development,
Mina Krusenmastraat 192a, 2614 JK Delft, The Netherlands

Models HEC-RAS and BRI-STARS have been hydrodynamically and


morphodynamically calibrated and validated respectively for river Rocha, a 60-Km-long
river running across the city of Cochabamba, Bolivia. Both models were successfully
used to assess water depth flows, sediment transport rates, river bed variations and to
identify zones with flood risk according to the risk classification proposed by the Basic
Planning Guidelines of Spain [1] for a number of hypothetical flow scenarios. A
simulation scenario with modified riverbed slopes, new cross-sections and a maximum
historical discharge produced significant reductions in water depths. In general, with
these improvements, the simulation results with BRI-STARS show a significant
increment in the river conveyance capacity, a decrease of flood risks and a more dynamic
stability of river Rocha.

INTRODUCTION

Since a number of years numerical modelling in river engineering has been a valuable
tool to assess complex hydro and morphodynamic processes in rivers as well as
environmental and related man made problems. This technique also proved to be very
useful to evaluate the impact of mitigation measures to those problems upon the
morphodynamic behaviour of the studied river.
Some of the aforementioned problems are present at river Rocha, a 60-Km-long river
running across the city of Cochabamba, Bolivia, where severe floods occur during almost
each rainy season in the southeast part of the city due to the river insufficient cross –
section area and low conveyance around this sector (see Figure 1). In general, river
Rocha presents accelerated processes of bed and lateral erosion, large accumulated debris
deposits near the shores and dense vegetated zones obstructing the main reach affecting
potentially the lives and safety of almost 860,000 inhabitants.
Hence, in order to improve the current situation, some mitigation measures were
thought to be suitable for the southeast reach of river Rocha like the implementation of
new riverbed slopes and new cross-sections.

1
2

Figure 1. Flood at a zone called Cotapachi (southeast Cochabamba) due to river Rocha
levees overtopping, March 2001.

Given complexity of the problem mainly due to the large sediment transport rates present
during floods, the non – prismatic river trajectory, complex cross-section shapes and
largely varying bed slopes and Manning’s friction coefficients, it was decided to
implement for the study zone two numerical models: HEC-RAS and BRI-STARS for the
respective hydrodynamic and morphodynamic assessment of river Rocha behaviour with
and without the aforementioned flood mitigation measures.

MODELS HEC-RAS & BRI-STARS

Model HEC-RAS, version 3.1, from de Hydrologic Engineering Center (US Army Corps
of Engineers) allows the user the calculation of water surface levels y gradually varying
flow. The package can run for just one river or a number of interconnected rivers in
subcritical o supercritical flow regime.
The flow equation to be solved is the 1D energy equation (see Eq. 1). Energy losses
he are evaluated by means of Manning’s equation and appropriate expressions for
contraction and expansion coefficients, as shown in Eq. 2. from Brunner [2].

α 2 v22 α1v12 (1)


Y2 + Z 2 + = Y1 + Z1 + + h1
2g 2g

⎛ α v2 α v2 ⎞
he = L S f + C ⎜⎜ 2 2 + 1 1 ⎟⎟ (2)
⎝ 2g 2g ⎠
3

where Y1,2 = water depth at control sections 1 and 2, Z1,2 = bed elevations at control
sections 1 and 2, v1,2= average velocities at control sections 1 and 2, α1,2 = velocity
coefficients at the borders of control sections 1 and 2, g = acceleration due to gravity, he =
energy losses, L = reach length, Sf = Energy line slope and C = coefficient of losses due
to expansion and/or contraction. When the flow is rapidly varied, HEC-RAS uses the
momentum equation to calculate water levels. One significant advantage is the
possibility of having flood maps by means of an option called “floodplain analysis”.
BRI-STARS (Bridge Stream Tube Alluvial River Simulation), version 5.03 is a
quasi-2D flow propagation model based upon a model of stream tubes of fixed widths
including a sediment transport module suitable for solving complex river engineering
problems with scarce data. It has the capability to simulate erosion and deposition
processes in subcritical and/or supercritical flow regimes using the concept of the stream
power minimization, as illustrated in Molinas & Wu [5].
In order to generalize the procedure of minimization of such stream power for
channels experiencing gradually varying flow, the stream power γQS has to be integrated
along the channel. This expression is defined as the total stream power given by (Chang
[3]):

Φ T = ∫ Φ S ( x )dx (3)
x

where ФS(x) = γQS is the stream power at the location x along the river. This
expression can be discretized as (Chang [4]):

N −1
⎛ Q S + Qi +1Si +1 ⎞
ΦT = ∑ γ ⎜ i i ⎟ ΔX i (4)
i =1 ⎝ 2 ⎠

where N = number of cross-sections a long the reach, ΔXi = reach length between
stations i and i+1, Qi = flow discharge at station i and Si+1 = slope at station i.
The model provides the following total sediment transport equations: (a) Fine sand
equations: Yang, Molinas & Wu (1996) (b) Sand equations: Engelund & Hansen (1972),
Ackers & White (1973) and Yang (1973) (c) Gravel equations: Meyer-Peter & Müller
(1948) and Yang (1984).

IMPLEMENTATION OF MODELS

Calibration of models
The geometry of the river was fully defined by means of 1,423 cross-sections, initial
Manning coefficients n were adopted for the main channel (from 0.03 to 0.065 m1/3s-1),
lateral shores (from 0.045 to 0.08 m1/3s-1) and flood plains (from 0.05 to 0.1 m1/3s-1) from
4

field work. Additionally, 15 bridges, 4 weirs and a lateral intake were taken into account
for this stage.
The available data for the calibration stage consisted in flow discharges measured at
5 control sections at the river from 1999 to 2002. The parameter chosen for calibration
was the Manning roughness coefficient by means of trial-error method. In order to assess
the accuracy of the procedure, the following errors were calculated (Romero & Heredia
[6]):
N

∑ (X obsi − X calci ) (5)


M .E . = i =1

N
N

∑X obsi − X calci (6)


A.E. = i =1

where M.E. = medium error, A.E = absolute error, Xobsi and Xcalci are the observed and
calculated water depths respectively. Figure 2 shows the results of the calibration
scenario with HEC-RAS, where it can be noticed that there is a good agreement between
the observed and calculated values of water depths (R2 = 0.9854) with average values of
M.E = -0.03 m and A.E = 0.08m. Calibrated values of Manning vary from 0.032 to 0.055
m1/3s-1 depending on the observed reach.

2,00
1,80
1,60
Calculated water levels (m)

1,40
1,20
1,00
0,80
0,60
0,40
0,20
0,00
0,00 0,50 1,00 1,50 2,00
Observed water levels (m)

Figure 2. Calculated and observed flow depths at control sections with 95% confidence
intervals with HEC-RAS related to a perfect calibration line (R2 = 1).
5

BRI-STARS calibration involved two stages: the first one for the hydrodynamic module
and the last one for the morphologic module. As well as in the first calibration scenario
with HEC-RAS, the same 7 control sections were observed for comparison purposes. For
the morphologic calibration, two additional control sections were considered (stations
Mesadilla and La Cabaña) where curves of flow discharge vs. sediment concentration
were available from sediment measurements from 1975 to 1976 (37 campaigns).
Results show that for the hydrodynamic module the calibration was satisfactorily
achieved with values of R2 = 0.9925, M.E = -0.038 m and A.E = 0.065 m. For the
morphologic module, the objective was to determine which of the aforementioned
sediment transport formulas was the most appropriate for the study. Five calibration
scenarios were chosen for stations Mesadilla and La Cabaña. Average results show good
agreement between calculated and observed values of sediment concentration for the
formulas of Molinas & Wu (1996) and Meyer-Peter & Müller (1948) for station
Mesadilla : R2 = 0.9683, M.E. = 0.102 mg l-1 and A.E. = 1.646 mg l-1, for station La
Cabaña: R2 = 0.9911, M.E. = -0.17 mg l-1 and A.E. = 0.886 mg l-1. The average value of
R2 for both transport models is 0.9797.

Validation of models
Model HEC-RAS was validated by means of other measured events at the 7 control
sections. Found Manning’s roughness coefficient values were assigned spatially
according to the results of the previous stage. Results show values of R2 = 0.9700. M.E.
= 0.03 m and A.E. = 0.10 m. For BRI-STARS model the validation for the morphologic
module considered other 6 measured scenarios, where formula of Molinas and Wu (1996)
and Meyer-Peter & Müller (1948) were applied. Results show average values of R2
=0.9730, M.E. = -0.066 mg l-1 and A.E. = 1.61 mg l-1 for station Mesadilla. For station
La Cabaña the obtained results were: R2 =0.9960, M.E. = -0.23 mg l-1 and A.E. = 2.32 mg
l-1 which were considered satisfactory. The average value of both R2 is 0.9845.
At this stage it was not considered to validate the hydrodynamic module of BRI-
STARS because this model accepts considerably less number of cross-sections than
HEC-RAS, and due to the fact that both models present very similar results during the
calibration phase (R2 = 0.9889), as shown in Figure 3.

RESULTS AND DISCUSION

For the simulation phase, several hypothetical flow scenarios were considered. The most
unfavourable one was a historic maximum discharge Q = 420 m3s-1 according to a
previous study carried out by Sevilla [7]. The reach to be studied was the one in between
two important locations at the city of Cochabamba: Sacaba (beginning of reach) and
Parotani (end of reach), with a length of 60 Km.
Figure 4 presents the result o the simulation with BRI-STARS model where zones of
aggradation (sedimentation) and degradation (erosion) were identified. Additionally
sediment transport rates were assessed as well as water depth flows and velocity fields.
6

2,00
Calculated water depths with BRI

1,80
1,60
1,40
STARS (m)

1,20
1,00
0,80
0,60
0,40
0,20
0,00
0,00 0,50 1,00 1,50 2,00

Calculated water depths with HEC-RAS (m)

Figure 3. Calculated water depths with HEC-RAS and BRI-STARS models at the
selected control sections during the calibration phase, including 95 % confidence interval.

Aggradation zones Degradation zones

Figure 4. Aggradation and degradation zones at the study area shown in a Landsat image
from BRI-STARS simulation.
7

According to the Basic Planning Guidelines of Spain [1] adapted to the present study,
three flood risk zones were identified at the study area based upon the obtained results
with BRI-STARS: low, significant and high. Hence, it was decided to propose two
mitigation measures for specific reaches of river Rocha: (a) cross-section enhancement
(b) cross-section enhancement plus riverbed slope modification (USACE [8]).
Since HEC-RAS accepts more number of cross-sections than BRI-STARS, it was
decided to test numerically the proposed mitigation measures with the first model,
considering again the maximum historic discharge of 420 m3s-1. Figure 5 presents the
obtained results where it was noticed that it was a 35 % reduction in average water depths
(from 6.0 to 3.9 m) if option (a) was implemented and 42 % if option (b) was considered
(water depths varying from 6.0 to 3.5 m). Additionally, with option (b), the flood risk
zone status lowered significantly in the whole study reach, and the river was able to
convey de maximum discharge without overtopping the river levees in any zone, and a
general morphodynamic stability was achieved (decrease of flow velocity and erosion
patterns).

Sacaba – Parotani reach

Water level with no mitigation measures

Mitigation measure (a)


Water depth (m)

Flow direction

Mitigation measure (b)

Distance (Km)

Figure 5. Maximum water depths for the Sacaba – Parotani reach with no mitigation
measures, options (a) and (b) simulated with HEC-RAS
8

SUMMARY AND CONCLUSIONS

Models HEC-RAS and BRI-STARS have been calibrated and validated for river Rocha,
where severe floods occur during the rainy season in the southeast part of the city due to
the river insufficient cross-section area and low conveyance around this sector. HEC-
RAS yielded satisfactory results for the calibration and validation of water depths at 7
control sections within the study zone (R2 = 0.9854 and 0.9700 respectively), while BRI-
STARS provided good agreement between observed and calculated sediment transport
rates for the empirical formulas of Molinas & Wu (1996) and Meyer-Peter & Müller
(1948), with average calibration and validation values of R2 of 0.9797 and 0.9845. The
aforementioned models were used to assess water depth flows, sediment transport rates,
river bed variations and to identify zones with flood risk. Hence, high flood risk zones
were accurately identified and mitigation measures were proposed and numerically tested
at those places. A simulation scenario with modified riverbed slopes, new cross-sections
and a maximum historical discharge Q = 420 m3s-1 (Sevilla, [7]), produced reductions in
water depths up to 35 % if only cross-section enhancement were considered and up to 42
% if also river bed slope modifications were also applied. In general, with these
improvements, the simulation results with BRI-STARS show a significant increment in
the river conveyance capacity, a decrease of flood risks and a more dynamic stability of
river Rocha.

REFERENCES

[1] Basic Planning Guidelines of Spain, “Civil protection against flood risks”, Basic
Planning Direction of Spain, Madrid, (1995).
[2] Brunner W., “HEC –RAS, River Analysis System user’s manual”, User Manual,
HEC US Army Corps of Engineers, (2001)
[3] Chang H., “Geometry of gravel streams”, Journal of the Hydraulics Division,
ASCE, Vol. 106, N° HY9, (1980), pp. 1443 – 1456.
[4] Chang H., “Mathematical model for erodible channels”, Journal of the Hydraulics
Division, ASCE, Vol. 108, N° HY5, (1982), pp. 678 – 689.
[5] Molinas A. and Wu B., “BRISTARS, User’s manual for BRISTARS, Bridge Stream
Tube Model for alluvial river simulations”, User Manual, US Department of
Transportation, (1998)
[6] Romero M. and Heredia M., “Assessment of the effect of stabilization works at the
river confluence Desaguadero-Jachajahuira, Bolivia” Proceedings of the XXXI
Congress of the International Association of Hydraulics Research (IAHR), Seoul,
South Korea, Vol I, (2005). pp 182 – 183.
[7] Sevilla R., “Hydrological study for river Rocha relocation”, Majorship of
Cochabamba, (1988). pp. 1-10, 46-50.
[8] US Army Corps of Engineers (USACE), “Hydraulic design of flood control
channels”. USCE, Reston, Virginia, (1998). pp. 34-40.

Potrebbero piacerti anche