916 © IWA Publishing 2020 Journal of Hydroinformatics | 22.

4 | 2020

Optimization of dam’s spillway design under climate

change conditions
Ahmad Ferdowsi, Sayed-Farhad Mousavi, Saeed Farzin and Hojat Karami

ABSTRACT

The present research introduces a model to find the best shape of a dam’s spillway under climate Ahmad Ferdowsi
Sayed-Farhad Mousavi
change impacts, considering a benchmark problem (i.e., Ute Dam’s labyrinth spillway in the Canadian Saeed Farzin (corresponding author)
Hojat Karami
River watershed, New Mexico, USA). A spillway design is based not only on historical data but also on Department of Water Engineering and Hydraulic
Structures, Faculty of Civil Engineering,
the future hydrologic events. Climate variables were predicted for the years 2021–2050 based on
Semnan University,
three representative concentration pathway (RCP2.6, RCP4.5, and RCP8.5) scenarios of the general Semnan 35131-19111,
Iran
circulation model from the fifth phase of the coupled model intercomparison project (CMIP5) using E-mail: [email protected]

the statistical downscaling model. Streamflow at the USGS 07226500 streamgage was simulated by a
rainfall–runoff model with predicted data. Instantaneous peak flow was estimated using an empirical
method. Flood frequency analysis was used for the estimation of the design flood. The shuffled frog-
leaping algorithm (SFLA) is used to optimize a labyrinth spillway design and its results were
compared with two other nature-inspired algorithms: invasive weed optimization (IWO) and cuckoo
search (CS). The spillway was optimized once with the actual design flood (16,143 m3/s) and again
with the design flood under climate change (12,250 m3/s). Results revealed that optimization with
realistic design flood reduced the concrete volume of the spillway by 37% and under climate change
by 43% using the SFLA.
Key words | Canadian River watershed, climate change, construction cost, labyrinth spillway design,
optimum design, rainfall–runoff model

HIGHLIGHTS

• Proposing a new framework for the evaluation of climate change impacts on dams’ spillways
optimum design.

• Reducing the construction cost of a dam spillway using metaheuristics.

• Comparing the performance of three nature-inspired algorithms (SFLA, IWO, and CS).

• Investigation of changes in rainfall, maximum, and minimum air temperatures in the Canadian
River watershed during a 30-year period (i.e., 2021–2050).

INTRODUCTION

It is worth starting this optimization study with a very walk or search is inevitable. There are some keys to perform
charming explanation by Yang (), which represented well and find the ultimate treasure (global optimum), includ-
an optimization problem as treasure hunting. In this treasure ing a search by a group of best hunters.
hunt, treasure is hidden in a wide landscape, and like any Optimization may be needed in many problems in our
real problem there are limitations such as time. As hunters everyday life. In today’s competitive world, humans try to
do not know where to look for the treasure, a random maximize efficiency from a limited number of available
doi: 10.2166/hydro.2020.019

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 917 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

resources. For this purpose, artificial intelligence or compu- Meanwhile, the use of fossil fuels continued due to
tational intelligence is used with optimization techniques world population and industrial growth, on the one hand,
(Erdik & Savci ; Erdik ; Erdik et al. ; Kaveh and the destruction of agricultural lands and reduction of
; Bozorg-Haddad ; Ehteram et al. ). Normally, forest areas, on the other hand, have increased greenhouse
an optimization problem requires sophisticated optimiz- gas emissions over the past decades. Climate change is a
ation tools. Over the last decade, a diverse spectrum of term used in scientific writings for the effects of greenhouse
algorithms, from traditional gradient-based algorithms and gases on the world climate and hydrological events. Several
simplex methods to evolutionary algorithms and nature- studies have warned about the impacts of climate change on
inspired metaheuristics (Yang ), have extensively been hydrological variables, e.g. increasing evapotranspiration
used for dealing with highly nonlinear and tough optimiz- due to rising temperatures, fluctuating rainfall patterns,
ation problems, in particular, problems in water and seawater-level rise (Raju & Kumar ; Xing et al.
engineering (Hassanvand et al. ; Monsef et al. ). ).
Dams, as crucial infrastructures for water supply and Climate change affects design and operation of water
flood mitigation, require ancillary structures and facilities infrastructures such as flood control systems, hydropower
(such as spillways) to enable them to pass design floods plants, irrigation and drainage systems, water distribution
(Novak et al. ). In terms of engineering, spillways must networks, as well as water management practices (Bates
be able to meet the basic requirements such as hydraulic et al. ). The impacts of climate change have been
performance, structural stability, environmental impacts, studied in some of the water-related items such as urban
and safety aspects (Ghare et al. ). A labyrinth spillway runoff (Zahmatkesh et al. ), irrigation (Ashofteh et al.
is one of the nonlinear spillways that due to its geometric ), drainage systems (Karamouz et al. ), hydropower
shape can increase discharge capacity. The use of this spill- plants (Sarzaeim et al. ), streamflow (Nazif & Karamouz
way increases the crest length without increasing the width ), water diversion systems (Karamouz et al. ), floods
of the overflow span (Khatsuria ). The popularity of (Dong et al. ), evapotranspiration (Safavi et al. ), and
labyrinth spillways has considerably increased over the last reservoir operation (Ehteram et al. a).
decade, notably in the USA (Khode & Tembhurkar ). A US Army Corps of Engineers’ survey on more than
While more than 100 years have passed from the introduc- 80,000 dams has shown that almost 36% of the existing
tion of labyrinth spillways (Hager et al. ), few studies have dams are unsafe due to various reasons, of which nearly
been carried out on optimum design of these structures using 80% of insecurity is due to insufficient spillway capacity
intelligent methods. Hosseini et al. () proposed an optimal (Mirnaseri & Emadi ). Also, the international
model for labyrinth spillways. An adaptive neural fuzzy infer- commission on large dams (ICOLD) has declared that
ence system (ANFIS) model was used to calculate the approximately 30% of the dam failures are due to the lack
discharge coefficient of the spillway and differential evolution of sufficient spillway capacity (Hosseini et al. ). Another
(DE) and genetic algorithm (GA) for determining the best geo- major issue of the spillways is their high cost of construction.
metry of the labyrinth. Kardan et al. () optimized the shape About 20% of the construction cost of small dams and 80%
of a trapezoidal labyrinth spillway by the evaluation of the of large dams are allocated to the construction of spillways
number of cycles using the GA. In the study of Tabari & (Bozorg-Haddad et al. ). Therefore, on the one hand,
Hashempour (), GWO-DSO (grey wolf optimization- the spillway design with sufficient capacity and also with
direct search optimization) and PSO-DSO (particle swarm the least cost is necessary and inevitable. On the other
optimization-direct search optimization) hybrid algorithms hand, climate change, as a prominent hydrologic uncer-
were used to optimize labyrinth spillway dimensions. Ferdowsi tainty, should be considered in the design of present water
et al. () studied the effect of spillway’s crest shape on opti- resource systems.
mizing of labyrinth geometry. The hybrid of bat algorithm In the current research, a novel optimization-based
(BA) and particle swarm optimization (PSO) algorithm was model is proposed in order to investigate the impact of cli-
used and its results were compared with BA, PSO, and GA. matic change on a dam’s spillway design. This study aims

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 918 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

to propose a framework to reduce the construction cost of a on design discharge under climate change. The performance
labyrinth spillway and its design considering future climatic of SFLA is also compared with invasive weed optimization
changes or uncertainty. A real benchmark design problem (IWO) and cuckoo search (CS) algorithms. The results are
(i.e., labyrinth spillway of Ute Dam, which is built in the obtained by the use of the MATLAB (R2013a) software. In
Canadian River watershed, New Mexico, USA, which is Figure 2, the main steps of the proposed methodology are
an impressive labyrinth spillway with 1,024.5 m length and presented.
9.14 m height) is used as a case study. The current frame-
work can also be used in the future designs of other water
Labyrinth spillway design
infrastructures.

The primary hypothesis behind developing labyrinth spill-

ways was to increase the discharge capacity through the
CASE STUDY AND DATA COLLECTION
increase of crest length in a constant width. Labyrinth spill-
way’s cycles could have a linear configuration (e.g.,
The case study is the labyrinth spillway of Ute Dam
trapezoidal, triangular, and rectangular) or an arced con-
(Figure 1). This 36.88 m high embankment dam was built
figuration (Sangsefidi et al. ; Azimi & Seyed Hakim
on the Canadian River, Ute Creek watershed, near Logan,
; Safarrazavi Zadeh et al. ; Ghaderi et al. ). A
New Mexico, USA, in 1962 (Houston ). The Ute Creek
general form of the labyrinth spillway discharge equation
watershed is 5,335 km2 wide. Ute Creek, a tributary to the
is shown in the following equation:
Canadian River and the Ute Reservoir, provides most of
the inflow to the Ute Reservoir. The Ute Reservoir was
2 pffiffiffiffiffi 3=2
built for water storage for municipal, industrial, and agricul- Q ¼ Cd L 2gHt (1)
3
tural uses. Much of the runoff from the Canadian River
watershed is captured by the Conchas Reservoir, 65 km
where Q is the discharge over the spillway, Cd is the dimen-
west of the Ute Dam Reservoir. Mean daily streamflow
sionless discharge coefficient, L is the spillway length, g is
data were obtained from the USGS streamgage (USGS
the acceleration due to gravity, Ht is the total head on the
07226500, Ute Creek near Logan), located immediately
crest (Ht ¼ h þ V 2/2g), V is the mean cross-sectional water
upstream of the Ute Reservoir and at the outlet of Ute
velocity, and h is the piezometric head upstream of the spill-
Creek watershed (Figure 1). Daily rainfall and air tempera-
way relative to the spillway crest elevation. The calculation
tures of two COOP stations (Conchas Dam and
of spillway discharge depends on the exact determination
Pasamonte) are used to represent weather conditions of
of Cd value. The effects of labyrinth spillway geometry and
the Ute Dam watershed (Table 1 and Figure 1).
flow conditions on discharge are represented as Cd
(Crookston & Tullis ). To calculate the Cd, Equation
(2) is used (Crookston & Tullis ), which was extracted
METHODOLOGY
from physical models of non-vented trapezoidal labyrinths
with wall angles between 6 and 35 and without concern-
In the proposed approach, climatic parameters (minimum
ing the influence of abutments on flow. In summary, Cd is
and maximum air temperatures and rainfall) are predicted
determined according to different wall angles, half-round
in the catchment area for a period of 30 years (2021–
crest shape, and headwater ratio.
2050). Afterwards, streamflow is simulated using the pre-
H C
dicted data (rainfall and air temperatures). Design B t
Ht
 P

discharge, as a key parameter in water infrastructure Cd ¼ A × þD (2)

P
designs, is calculated under climate change conditions.
The shuffled frog-leaping algorithm (SFLA) is employed to where P is the crest height, and coefficients A, B, C, and D
determine the optimal shape of labyrinth spillway based are derived from Table 2. It is worth mentioning that Equation

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 919 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Figure 1 | Location of Ute Dam and its watershed, selected stations, dam’s labyrinth spillway, and its parameters.

(2) and the coefficients in this table are derived from a figure labyrinth spillway crest including flat, sharp, half-round
which is provided in Crookston & Tullis (). (HR), quarter-round (QR), ogee, and WES (truncated
A labyrinth spillway crest profile has a noteworthy ogee). Previous studies have proved that half-round and
impact on discharge coefficient and thus on discharge ogee crest profiles are more efficient crest shapes (Willmore
capacity. Various forms have been proposed for the ; Crookston ; Ferdowsi et al. ).

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 920 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Table 1 | Description of the selected stations

Station Data Latitude Longitude Datum Source

USGS 07226500, Ute Creek near Logan Streamflow 35.44 103.53 1,164.3 USGS
Conchas Dam Rain, Tmax, Tmin 35.41 104.19 1,294.4 NM Univ.
Pasamonte Rain, Tmax, Tmin 36.30 103.74 1,723.3 NM Univ.

Note: Tmax and Tmin, maximum and minimum air temperatures; USGS, US Geological Survey; NM Univ., Website of Climate Center, New Mexico University.

Figure 2 | Flowchart of the proposed methodology.

Table 2 | Curve-fit coefficients for half-round labyrinth spillway, validated for 0.05  Ht/ VW (volume of spillway wall), Ve (volume of the end walls),
P < 0.9 (Crookston & Tullis 2013)
and VS (volume of floor slab) (Falvey ) as follows:
α A B C D

6 0.009447 4.039 0.3955 0.1870 VW ¼ N × Lc × P × Tw (3)

8 0.017090 3.497 0.4048 0.2286
Ve ¼ 2 × (P þ Ht þ Fb ) × (B þ Ht ) × Tw (4)
10 0.029900 2.978 0.4107 0.2520
12 0.030390 3.102 0.4393 0.2912 VS ¼ (B þ 2Ht ) × Ww × Ts (5)
15 0.031600 3.270 0.4849 0.3349
20 0.033610 3.500 0.5536 0.3923
35 0.018550 4.904 0.6697 0.5062

The geometry of each structure determines the volume of

materials needed for its construction and also its construction
costs and time. Therefore, by optimizing the geometry of a
spillway, the lowest volume of concrete and cost may be
achieved. The total concrete volume of a trapezoidal
labyrinth spillway (VT) consists of three parts (Figure 3): Figure 3 | Illustration of different parts of a labyrinth spillway.

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 921 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

where N is the number of cycles, P is the crest height, Tw is the The labyrinth spillway loses its efficiency by increasing
wall thickness, Fb is the freeboard, H is the total head, and B is its head. A headwater ratio is defined as the total head
the labyrinth length (parallel to flow). (Ht) divided by the height of spillway crest (P). Generally,
labyrinth spillways are designed for a headwater ratio of
less than 0.9 (Crookston ). Crookston & Tullis ()
Labyrinth spillway optimum design
used the headwater ratio of 0.05–0.9. The following penalty
function is expressed to design an optimum labyrinth with a
To solve an optimization problem, it is needed to determine
headwater ratio between 0.05 and 0.9:
some terms and functions, including objective function, con-
stant parameters, design variables, and constraints (penalty 
Ht Ht

functions). The objective function in the optimization of fp2 ¼ 1 , 1 0 (8)
0:05P 0:9P
labyrinth spillway geometry includes three parts: volumes
of labyrinth wall, end walls, and floor slab (Equations
(3)–(5) and Figure 3). Hence, the objective function can be A relative thickness ratio is calculated by dividing the

formulated as follows: height of spillway crest (P) by spillway wall thickness (Tw).
Wall thickness depends on the hydraulic forces, ice loading,
and site conditions and is calculated from structural analy-
Minimize fCost ¼ VT ¼ VW þ Ve þ Vs (6)
sis. Tullis et al. () mentioned that a major decrease in
wall thickness causes separation and reduces the discharge
In optimization, some parameters are considered con-
coefficient. Their models are based on Tw ¼ P/6. Some
stant based on engineering judgment and previous studies.
studies have used models which are based on Tw ¼ P/8
These parameters, in addition to reduce the dimensions of
(Willmore ; Crookston & Tullis ). Equation (9) is
the problem, will also have a significant effect on saving a
written as a penalty function to control wall thickness and
computational time. These parameters in the present study
the spillway crest height of optimum models:
are free board (Fb ¼ 0.6 m), slab thickness (Ts ¼ 0.3 m),
number of cycles (N ¼ 14), and crest shape (HR). 
P P

On the other hand, design variables are considered fp3 ¼ 1 , 1 0 (9)
6Tw 8Tw
unknown and their values are calculated in the optimization
process. Design variables in this study are crest height (P),
total head (Ht), labyrinth length parallel to flow (B), wall thick- A magnification ratio is the length of a spillway

ness (Tw), angle of side legs (α), and apex center-line width (A). cycle (Lc) to the cycle width (W ). Previous optimization

The values of the design variables must be within a certain studies have recommended that this ratio must be limited

range. In other words, optimizing the value of the objective to a range of 3–9.5 for optimal performance (Kardan et al.

function is not possible at any cost. These limitations (or con- ). The penalty function of this ratio is considered

straints), which are the results of previously constructed as follows:

spillways, laboratory tests, numerical modeling, and codes 

LC LC

fp4 ¼ 1 , 1 0 (10)
(Chen ), improve labyrinth performance and its safety. 3W 9:5W
The penalty method is a common method to handle con-
straints into objective functions, which is used in the present One cycle width (W ) divided by crest height (P) is
study. The following section discusses the penalty functions. described as the cycle width ratio (or vertical aspect).
According to the first penalty (i.e., Equation (7)), the dis- Crookston & Tullis () suggested that this ratio should
charge of optimal labyrinth spillways will be equal to the be limited to 2–4. The following equation is used as the pen-
flood design (Qd) of Ute labyrinth: alty function of cycle width ratio:

W W

fp1 ¼ (Qd
pffiffiffiffiffi 3=2
2/3 2gCd LHt )  0 (7) fp5 ¼ 1 , 1 0 (11)
2P 4P

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 922 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

of the ith solution (frog) is represented as Xi ¼ (Xi1, Xi2,

The apex width (A) divided by the cycle width (W ) is …, XiD), where D is the variable number. Afterwards, sort-
described as the apex ratio. This ratio should be as low as ing of the frogs is performed in a descending order based
possible, usually, equal to 0.0765 or less (Hepler ). The on their fitness. In the next step, all members of the popu-
penalty function is written as follows: lation are divided into m subsets referred to as
 memeplexes, each containing frogs (i.e., P ¼ m × n). The
A

fp6 ¼ 1 0 (12) frog with the best performance (according to the objective
0:0765W
function) is placed in the first memeplex, the second best
A wall thickness ratio is obtained by dividing the apex frog is placed in the second memeplex, and the mth frog is
width (A) by the spillway wall thickness (Tw). Tullis et al. placed in the mth memeplex. In the next division process,
() limited this ratio to 1–2 for optimum performance. the (m þ 1)th frog is placed in the first memeplex. The pos-
Here, its penalty function is considered as follows: itions of the best and worst frogs, in each memeplex, are
identified as Xb and Xw, respectively. In addition, Xg is ident-

A A
 ified as the position of a frog with the global optimum. A
fp7 ¼ 1 , 1 0 (13)
Tw 2Tw similar process which exists in the PSO algorithm is applied
to enhance the frog with the worst fitness in each cycle,
The ratio of interference length (LD) to one side-leg within each memeplex. As a result, the position of the
length (LS) is important for limiting the effects of nappe frog, which has the worst fitness, goes toward the position
interference, which reduces spillway discharge efficiency. of the frog, which has the best fitness, as follows:
Indelkofer & Rouve () explored the idea of nappe inter-
ference and introduced a nappe effective disturbance length. Di ¼ rand × (Xb Xw ) (16)
Falvey (), by using data from previous studies and labyr-
Xnew
w ¼ Xold
w þ Di (Di min <Di <Di max ) (17)
inth spillway models, proposed this ratio as Equation (14)
and notified that this ratio should be equal to 0.30 or less.
where Di is the change of the frog position, rand is the
However, data presented by Falvey () indicated a limit
random number between 0 and 1, Xnew old
w and Xw are new
of 0.35 (Crookston & Tullis ). Therefore, its penalty func-
and old (i.e., current) positions, and Di_min and Di_max are
tion can be considered as Equation (15):
minimum and maximum step sizes for the position of a
frog, respectively.
LD
 
Ht
¼ 0:244 ln þ (0:94 0:03α) (14) The SFLA has shown superior performance over
LS P
improved ant colony optimization, ant colony optimization,
LD
 
fp8 ¼ 1 0 (15) simulated annealing, Lagrange multiplier method, PSO, and
0:35LS GA in the previous studies (Orouji et al. , ).
Optimization tool: shuffled frog-leaping algorithm
Climate change
The SFLA has been introduced by Eusuff & Lansey (). It
was originally used to optimize the design of water distri- Scientific research has shown that climate change has
bution networks and is based on random and considerable impacts on rainfall, temperature, evapotran-
deterministic procedures. This algorithm employs a memetic spiration, streamflow, and water resources. To simulate the
metaheuristic approach based on the natural behavior of effects of future climate conditions, the output from global
frogs. High performance, simple concept, few parameters, coupled atmospheric-ocean general circulation models
and easy programming are some key advantages of SFLA (coupled GCMs) is used as the input to hydrologic. In the
(Li et al. ). Initially, the SFLA generates an initial present study, a low greenhouse gas emission scenario
random population of P frogs (i.e., solutions). The location (RCP2.6), a medium emission scenario (RCP4.5), and a

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 923 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

very high emission scenario (RCP8.5) are used. These scen- flood estimation is a prerequisite for planning, design, and
arios are introduced in the fifth phase of the coupled model management of hydraulic structures and has a crucial
intercomparison project (CMIP5) of the GCM. In recent impact on the investment and benefits of the projects, and
years, there has been considerable focus on the use of also on their safety (Duricic et al. ; Erdik et al. ;
these new RCP scenarios. Pektas & Erdik ; Bhagat ; Guo et al. ).

Statistical downscaling model

Flood frequency analysis

Statistical downscaling model (SDSM), developed by Wilby
et al. (), is a combination of regression and conditional
Generally, design flood is estimated from flood frequency
weather generator technique. SDSM relies on empirical
analysis when sufficiently long observed streamflow records
relationships between local scale predictand(s) and regional
are available. The main modeling problem in flood
scale predictor(s). The model performance has been accep-
frequency analysis is the selection of a probability distri-
table in different studies (Sarzaeim et al. ; Xing et al.
bution for flood magnitudes (Guo et al. ). Gumbel,
). Key functions are quality control and data transform-
generalized extreme value, log-normal, and log-Pearson
ation, selection of downscaling predictor variables, model
type III distributions are widely used in analyzing annual
calibration, weather generator, data analysis, graphical
maximum series (Wilks ). The general equation of fre-
analysis, and scenario generation.
quency analysis is given as follows:

Rainfall–runoff model
QTr ¼ Q þ K × s (18)

In the present study, a hydrologic model based on unit

where QTr is design flood with a return period of Tr, K is a
hydrograph theory (i.e., IHACRES) is used for streamflow
 and s are average and standard devi-
frequency factor, and Q
simulation based on observed data (1971–2000) and pre-
ation of maximum instantaneous flows, respectively.
dicted for the near-future period (2021–2050). Hydrologic
models are essential tools for assessing runoff changes in
the catchment area of interest in the evaluation of the
impacts of climate change (Najafi et al. ). The IHACRES Instantaneous peak flow
is a lumped and hybrid conceptual-metric model that can
simulate streamflow with minimum input data, i.e., rainfall Generally, the design of hydraulic structures for flood con-
and temperature (or potential evapotranspiration) at two trol is conducted with instantaneous peak flow, i.e. IPF
steps: (1) a nonlinear loss module and (2) a linear unit (Fill & Steiner ). However, the data relating to mean
hydrograph module. At the first step, rainfall is converted daily flow (MDF) are the most common recorded hydrologi-
to effective rainfall and then runoff is produced in the cal variable. The design of control structures using MDF
second step ( Jakeman et al. ; Croke et al. ). Despite data may cause underestimation, with a consequent risk of
the simplicity of IHACRES structure, the model has per- possible failure (Taguas et al. ). The following methods
formed well in many catchments worldwide (Sarzaeim are used to estimate IPF from MDF.
et al. ; Ehteram et al. b). Fuller () studied flood data of some watersheds in
the USA and suggested the following equation:
Design flood
0:3
IPF ¼ MDF × (1 þ 2:66 × A ) (19)
Engineers, hydrologists, and agriculturalists often need the
design flood for the design of hydraulic structures, such as Sangal (), based on a triangular hydrograph, pro-
dams, spillways, bridges, channels, or culverts. Design posed Equation (20), which is tested by some streamgauge

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 924 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

data in Ontario, Canada: considering the total length of the base period. In the pre-
sent study, the total length of the base (observation) period
4 × MDF Q1 Q2 was selected from January 1971 to December 2000 (30
IPF ¼ (20)
2 years). The calibration period was 1971–1991 (70% of the
data), and 30% of the data (1992–2000) was chosen for
Fill & Steiner () used data from watersheds in Brazil the validation phase. Results of the validation period are pre-
and proposed the following equation: sented in Table 4 using statistical indices (R, NSE, and Bias)
for both stations. These results are calculated based on the
0:80 × MDF þ 0:25 × (Q1 þ Q2 ) SDSM output for rainfall variables, maximum and minimum
IPF ¼ (21)
0:9123 × (Q1 þ Q2 )=2 × Q2 þ 0:36
air temperatures over the 9 years (1992–2000) compared
with the observed values. According to the statistical indi-
where IPF is in m3/s, MDF is in m3/s, A is the drainage cators of Table 4, the SDSM has a good performance in
area (km2), Q1 is the mean daily flow in the preceding day the studied area.
(m3/s), and Q2 is the mean daily flow in the following day A closer examination of the results of the correlation
(m3/s). indices shows that, in general, the SDSM is more capable
of modeling temperature than rainfall. In other words,
Performance assessment criteria based on Table 4, R and NSE indices for air temperatures
were greater than for rainfalls. Also, Bias values for tempera-
Various indices may be used to show the performance of tures are less than for rainfall, except for minimum
simulation models. In the current study, the performance temperature in the Conchas Dam station. A graphical rep-
of SDSM and IHACRES models was assessed with three resentation of the results of simulation of average monthly
statistics, namely correlation coefficient (R), Nash–Sutcliffe rainfall, and maximum and minimum temperatures in the
efficiency (NSE), and relative bias (Bias) listed in Table 3 validation period for the two stations are shown in Figure 4.
(Xing et al. ). The accepted values for streamflow predic-
tions suggested by Moriasi et al. () are 0.5 < NSE  1
and 25%  Bias  þ25%. Table 4 | Performance assessment of monthly observed and simulated results (SDSM
output) during the validation period (1992–2000)

Statistical criteria
RESULTS AND DISCUSSION
Station Variables R NSE Bias (%)

Conchas Dam Rain 0.930 0.807 4.991

SDSM calibration and validation
Tmax 0.996 0.975 4.134
Tmin 0.998 0.990 5.735
In the SDSM, calibration is done by the model itself, which Pasamonte Rain 0.926 0.838 8.598
only needs the length of the period to be determined. The Tmax 0.998 0.992 1.398
Tmin 0.999 0.991 6.982
length of the calibration and validation periods can vary

Table 3 | Equations of performance assessment criteria

Index Equation
h i0:5
Correlation coefficient R¼
Pn
(Oi O)  × (Si S)= Pn (Oi O)  2 × (Si 2
S)
i¼1 i¼1

Nash–Sutcliffe efficiency NSE ¼ 1

Pn P  2
Si )2 = ni¼1 (Oi O)
i¼1 (Oi
Pn Pn
Relative bias Bias ¼ i¼1 (Oi Si )= i¼1 (Oi ) × 100
 is the average of observed values, Si is the simulated value, S
Note: Oi is the observed value, O  is the average of simulated values, and n is the number of time-steps.

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 925 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Figure 4 | Comparison of average monthly rainfall, maximum air temperature (Tmax), and minimum air temperature (Tmin) in the validation period (January 1992–December 2000) simulated
by the SDSM.

Rainfall under climate change The forecast for rainfall at the Conchas Dam was similar.
In a study by Brauer et al. (), flow and rainfall trends
At first, the performance of the SDSM was assured. After- were compared during the baseline (1971–2000) and the
wards, using the SDSM and the downloaded datasets, the 2001–2010 period in the Canadian River watershed in the
daily rainfall, maximum temperature, and minimum tempera- Upper Lake Meredith in New Mexico and Texas. Among
ture for the near-future period (2021–2050) were downscaled. the stations in this study, rainfall at six stations had an increas-
Rainfall, maximum temperature, and minimum temperature ing trend in the 10-year observation period compared with
increased at both stations under all three RCPs (Figure 5). the base period. Among these six stations, there are three
For instance, the observed mean annual rainfall at the Pasa- stations (Eagle Nest, Ocata, and Villanueva) in New Mexico
monte station increased from 415.4 mm in the base period and near the studied stations in the present study. The Villa-
to 477.3 mm in the near-future period based on RCP4.5. nueva station had the highest rainfall. The mean annual

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 926 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Figure 5 | Comparison of mean monthly rainfall, maximum air temperature (Tmax), and minimum air temperature (Tmin) under RCP2.6, RCP4.5 and RCP8.5 scenarios, in 1971–2000 and
2021–2050.

rainfall of the Villanueva station during the observation period in the RCP2.6, RCP4.5, and RCP8.5 scenarios. At the Pasa-
was 231 mm, which, with 216 mm increment in the 2001– monte station, there is also a similar trend. The maximum
2010 period, rose to 447 mm. It can be seen that in the water- observed temperature at this station was 19.14  C, which
shed, the increase in rainfall has also occurred according to the will increase in the future period and, according to the
past research based on observed data. three scenarios, will reach 20.40, 20.50, and 20.75  C. The
minimum temperature, like the maximum temperature,
Air temperature under climate change shows an increasing trend at both stations. The minimum
temperature at the Conchas Dam was 6.75  C during the
The maximum temperature will increase at both stations observation period, and by the influence of climate
based on RCP scenarios. The maximum temperature of the change, it will reach 8.25, 8.36, and 8.62  C in the RCP2.6,
Conchas Dam station was 22.92  C in the observation RCP4.5, and RCP8.5 scenarios. At the Pasamonte station,
period, reaching 24.22, 24.31, and 24.57  C, respectively, the minimum observed temperature was 1.95  C, which in

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 927 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Table 5 | Performance assessment of the IHACRES model in the calibration and validation for the base period for the USGS 07226500 station with the
periods in both stations
simulated streamflow of the upcoming period, using the
Index RCP2.6 for each of the two stations, is shown. According to
Figure 6, the mean observed streamflow is less than the
Station Period Length (day) R NSE Bias (%)
flow in the future period in both stations. In the present
Conchas Dam Calibration 3,580–4,825 0.8 0.7 21.0
study, the peak flow is used to calculate design flood. Figure 7
Validation 7,903–8,225 0.8 0.6 7.5
shows the peak streamflow of 30 years in both past and future
Pasamonte Calibration 3,580–4,825 0.9 0.7 18.7
Validation 7,565–7,609 0.9 0.9 4.4 periods. In this figure, in contrast to Figure 6, the observed
streamflow values are higher than the simulated ones. With
rainfall and temperature increasing in the studied watershed,
the near future will be 3.44, 3.57, and 3.78  C, according to the amount of streamflow increased, and peak streamflow
the three RCP scenarios. values decreased. This is due to the proximity of predicted
rainfall values to normal values and the reduction of their
IHACRES calibration and validation standard deviations relative to the observation period. The
predicted rainfall, instead of concentrating in a few days,
Table 5 shows the results of calibration and validation of the will occur in a larger time period, which, in turn, produces
observed daily flow data of the two stations of USGS peak daily discharge. This can be an outcome of using the
07226500. Based on the three performance criteria, both SDSM downscaling model.
stations have an acceptable status in both calibration and vali-
dation periods. In other words, the value of R is close to 1, NSE
is more than 0.7, and the Bias index is in the ±25% range.
DESIGN FLOOD
Streamflow simulation
Instantaneous peak flow
After assuring the performance of the IHACRES model in
simulation of observed streamflow, it can be used to simulate The performance of three empirical equations (Fuller,
streamflow using downscaled data from the SDSM for the Sangal, and Fill-Steiner) in IPF estimation was investigated.
2021–2050 period. For this purpose, rainfall and temperature According to the performance criteria, the Fuller’s equation
data of the two stations were imported into IHACRES with was recognized as the best method (R ¼ 0.9, NSE ¼ 0.9, and
daily time-steps. In Figure 6, the mean annual streamflow Bias ¼ 16.9) for converting MDF to IPF.

Figure 6 | Comparison of observed mean annual streamflow (based on mean daily flows in 1971–2000) and simulated mean annual streamflow (in 2021–2050) in Conchas Dam and
Pasamonte stations using the RCP2.6 scenario.

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 928 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Figure 7 | Comparison of observed mean annual peak streamflow (based on mean daily in 1971–2000) and simulated mean annual peak streamflow (2021–2050) in Conchas Dam and
Pasamonte stations using the RCP2.6 scenario.

Flood frequency analysis test. According to Table 6, the observed and predicted
data fitted to the four statistical distributions such as
To determine the design flood of the labyrinth spillway extreme value distribution, log Pearson type III, Gumbel,
using the flood frequency analysis method, an appropriate and log-normal. The normal distribution is the only distri-
distribution was selected first based on the maximum bution that does not fit the three IPF series.
annual peak streamflow in the base period. Subsequently, There were other reasons for choosing Gumbel distri-
using this distribution and the amount of actual design bution. Commonly, Gumbel distribution is used for predicting
flood of Ute spillway (16,143 m3/s), the return period of extreme events (i.e., floods) in hydrology (Haan ; Cunnane
the Ute spillway was determined (as described by Equation ). Onen & Bagatur () chose Gumbel distribution for
(8)). The calculated return period was used to determine the several reasons such as: (1) the relatively long streamflow
simulated design flood for the future period with the data (>10 years) and (2) the river is less regulated; hence is
IHACRES model. Since in this study, only the annual not significantly affected by reservoir operations, diversions
peak streamflow (with constant return period) was required or urbanization, which is true in the present study. The return
to determine the design flood of the future period, using period of the design flood of the Ute spillway (16,143 m3/s)
three-parameter distributions such as Pearson type III and would be 100,000 years by the Gumbel distribution. Calcu-
log Pearson type III, was practically impossible to deter- lation of design flood for future period was also based on the
mine the coefficient K in Equation (18). Therefore, among same calculated return period so that by equating all the con-
the two-parameter distributions, the Gumbel distribution ditions, the effect of climate change phenomenon (i.e.,
was selected after a Kolmogorov–Smirnov (K–S) fitting reduction of peak discharge) on the design flood is taken into

Table 6 | Results of the Kolmogorov–Smirnov test for each probability distribution

Base period (USGS 07226500) Predicted period (Conchas Dam) Predicted period (Pasamonte)
Significance level (α) Significance level (α) Significance level (α)

Distribution 0.1 0.05 0.02 0.01 0.1 0.05 0.02 0.01 0.1 0.05 0.02 0.01

Generalized extreme value ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓

Log-Pearson type III ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Gumbel ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Normal × ✓ ✓ ✓ × × ✓ ✓ × × ✓ ✓
Log-normal ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 929 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

account. According to Table 7, the observed IPF was 170 m3/s, Optimization
which has been changed by simulation in both Conchas
Dam and Pasamonte stations. According to this table, at Labyrinth spillway optimization was performed considering
both stations, and under climate change conditions, the the following six models: SFLA-Base, IWO-Base, and CS-
maximum and minimum predicted streamflow are less Base models used SFLA, IWO, and CS algorithms, respect-
than the observed values, reducing the design flood from ively, which are based on the actual design flood of Ute
3
16,000 to 12,250 m /s (average of the two stations). spillway (16,143 m3/s) and SFLA-RCP2.6, IWO-RCP2.6,
and CS-RCP2.6 models are based on design flood under cli-
mate change (12,250 m3/s).
Canadian River watershed: past (1941–2016) and future
(2021–2050)
Sensitivity analysis of the algorithms
Figure 8 is based on the peak observed streamflow of the
USGS 07226500 station during the 1941–2016 period. A sensitivity analysis on the SFLA parameters was per-
This information was extracted from the USGS website. A formed to obtain the best performance. The number of
linear trendline of these data is also plotted in Figure 8. frogs, memeplexes, and the maximum number of iterations
The observed data from 1941 to 2016 in this figure confirms were determined as 50, 5, and 1,000, respectively. In order
the reduction of peak streamflow (present research findings) to implement the IWO sensitivity analysis, the initial popu-
at this station. Several studies (e.g., Wilson & O’Brien ; lation of 10, 20, and 30; the maximum number of
King et al. ; Spencer & Salazar ) have been con- population of 50, 75, 100, 150, and 200; the minimum
ducted on the causes of reduced water storage in dams on number of seeds of 0, 1, and 2; the maximum number of
the Canadian River. Different assumptions have been seeds of 2, 5, and 10; and the nonlinear modulation index
made according to these studies, such as: (1) changes in of 2, 3, and 5 were considered. In CS sensitivity analysis,
baseflow due to reductions in seepage from the neighboring Pa parameter of 0.15, 0.20, and 0.25; best number of host
Ogallala aquifer, (2) brush invasion, and (3) changes in rain- birds’ nests (or population size) of 15, 20, and 25; and differ-
fall patterns. According to Spencer & Salazar (), there is ent numbers of iterations were assumed. The best values for
no significant change in annual rainfall in the watershed these parameters are shown in Table 8. According to this
between Ute and Meredith reservoir. These results are differ- table, the best value for parameter Pa is 0.15. For the
ent from those of Brauer et al. (). According to the results number of host nests, the minimum value of the objective
of this study, the main cause of the flood-pattern disruption in function is 20 nests. Also, the CS algorithm converged
the Canadian River watershed is the change in the rainfall after 1,000 runs, and the objective function did not change
pattern (in other words, the lesser frequency of huge rainfall by increasing the number of iterations. The values of mini-
events). According to Brauer et al. (), the peak rainfall mum, average, maximum, and coefficient of variation for
(i.e., occurrence of 50.9–139.7 mm rainfalls) in the 1960– the 10 runs are listed in Table 9. According to this table,
1979 and 1990–2009 periods was 0.9% and in 2000–2009 the best values of objective function were 9,351.170 and
was 0.3% of total rainfall, which means a decrease in peak. 8,440.865 m3 for the SFLA-Base and SFLA-RCP2.6, respect-
This is almost the same as the results of the present study. ively. The models based on the SFLA, IWO, and CS

Table 7 | Description of design flood based on observed and predicted data using the Gumbel distribution

Design flood (m3/s) Minimum (m3/s) Maximum (m3/s) Year Feature Station

16,143 0.54 170.28 1971–2000 Observation USGS 07226500

12,493 8.17 156.16 2021–2050 RCP2.6 Conchas Dam
12,007 5.09 137.53 2021–2050 RCP2.6 Pasamonte

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 930 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Figure 8 | Annual observed peak streamflow at USGS 07226500 Ute Creek near Logan, New Mexico, for the observed period (1941–2016) and the simulated period (2021–2050).

Table 8 | Values of parameters for labyrinth spillway optimization by SFLA, IWO, and CS are shown in Table 10. The spillway discharge, calculated
according to Equation (1), depends on discharge coefficient
Parameter Symbol Value
(Cd), total length of the spillway (L), and total upstream
SFLA Number of frogs N 50 head (Ht). All six models have the same discharge capacity
Number of memeplexes M 5
of their design flood (16,143 and 12,250 m3/s) or more.
Maximum number of iterations itmax 1,000
The total length of SFLA-Base, IWO-Base, and CS-Base
IWO Number of initial population N0 20
Maximum number of population pmax 150 models decreased by 30%, i.e., from 1,024.529 m (Ute
Minimum number of seeds Smin 2 labyrinth) to 720.469, 719.321, and 722.089, respectively.
Maximum number of seeds Smax 5
The SFLA-RCP2.6, IWO-RCP2.6, and CS-RCP2.6 models’
Nonlinear modulation index n 3
Initial value of standard deviation σ initial 1 total length also decreased by 35% to 671.334, 671.383,
Final value of standard deviation σ final 0.001 and 684.991 m, respectively. According to Table 10, all
Maximum number of iterations itmax 1,000
six models have a smaller concrete volume than the Ute
CS Detection probability of the eggs pa 0.15
Host nests (population size) n 20
spillway volume (14,789.391 m3). The total concrete
Maximum number of iterations itmax 1,000 volume of SFLA-Base, IWO-Base, and CS-Base spillway
models was calculated as 9,342.012, 9,368.336, and
1,0038.766 m3, respectively (36.83, 36.66, and 32.12%

convergence graphs are shown in Figure 9 for 1,000 lower than the Ute labyrinth spillway). The SFLA-

iterations. RCP2.6, IWO-RCP2.6, and CS-RCP2.6 have minor dis-

charge design and hence smaller concrete volume (42.96,
42.89, and 40.53% lower than the Ute labyrinth spillway,
Optimal labyrinth spillways respectively). According to Table 10, the labyrinth
spillway using SFLA has less concrete volume than the
The properties of Ute labyrinth spillway and proposed opti- optimal spillways using IWO and CS algorithms. The
mized labyrinth spillways, based on actual design flood volume of spillway walls has the greatest effect on
(base model) and under climate change conditions the total concrete volume of labyrinth spillway. This
(RCP2.6 model) using SFLA, IWO, and CS algorithms, volume has been reduced from 9,972.916 to 5321.531 m3

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 931 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Table 9 | Results of 10 runs of SFLA, IWO, and CS

Objective function (m3)

SFLA IWO CS

Run no. Base RCP2.6 Base RCP2.6 Base RCP2.6

1 9,351.638 8,438.257 9,369.668 8,446.162 10,040.286 8,795.063

2 9,343.536 8,440.865 9,374.560 8,446.911 10,038.857 8,794.675
3 9,349.027 8,436.49 9,376.578 8,445.586 10,042.973 8,794.974
4 9,347.029 8,440.031 9,370.338 8,446.927 10,039.304 8,795.805
5 9,348.046 8,439.222 9,371.694 8,447.760 10,039.180 8,794.649
6 9,342.012 8,436.664 9,376.718 8,447.280 10,038.766 8,794.888
7 9,346.753 8,437.791 9,373.106 8,448.333 10,038.918 8,794.881
8 9,345.483 8,439.898 9,368.606 8,448.155 10,042.396 8,795.249
9 9,351.170 8,436.333 9,372.481 8,446.112 10,039.045 8,795.100
10 9,344.193 8,436.724 9,368.336 8,456.228 10,039.830 8,794.989
Minimum 9,342.012 8,436.333 9,368.336 8,445.586 10,038.766 8,794.649
Average 9,346.889 8,438.227 9,372.209 8,447.945 10,039.956 8,795.027
Maximum 9,351.170 8,440.865 9,376.718 8,456.228 10,042.973 8,795.805
Coefficient of variation 0.00034 0.00020 0.00033 0.00036 0.00015 0.00004

Figure 9 | Convergence of SFLA, IWO, and CS.

in the SFLA-RCP2.6 model, which means almost 50% lower to Table 11), this height reduction does not result in loss
than the built Ute spillway walls. The volume of end walls of performance and safety of the spillway. The B parameter
has been reduced by more than 37% and the volume of in the SFLA-RCP2.6 and IWO-RCP2.6 models is about 12 m
spillway floor slab has decreased by about 35%. Spillway less than the Ute spillway. The B parameter does not
crest height (P) has decreased in all six models. P and L directly interfere in any constraint, but it is involved in cal-
play important roles in reducing the concrete volume of culating the wall length. In Figure 10, one cycle of all
spillway walls. Crest height is involved directly in the calcu- models is shown to compare B and W. The Ws in the Ute
lation of the three constraints (headwater ratio, relative spillway was about 256 m, which was decreased to 223 m
thickness ratio, and cycle width ratio), and since the value in the SFLA-RCP2.6 and IWO-RCP2.6 models. The Ws of
of this constraint is within the acceptable range (according each spillway is shown in Figure 11 for a better comparison.

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 932 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Table 10 | Parameters of Ute and optimal models

SFLA IWO CS

Parameter Ute spillway Base RCP2.6 Base RCP2.6 Base RCP2.6

3
Q (m /s) 16,143 16,145 12,250 16,144 12,251 16,144 12,250
N (–) 14.000 14.000 14.000 14.000 14.000 14.000 14.000
P (m) 9.140 8.022 7.963 8.050 7.967 8.395 8.063
Tw (m) 1.065 1.003 0.995 1.006 0.996 1.049 1.008
Ht (m) 5.790 6.511 6.202 6.513 6.201 6.751 6.279
B (m) 33.990 23.540 21.839 23.496 21.830 23.560 22.366

α( ) 12.162 17.836 17.500 17.812 17.500 17.761 17.500
A (m) 1.820 1.003 1.078 1.012 1.088 1.049 1.013
W (m) 18.290 17.154 15.927 17.121 15.943 17.192 16.130
Ws (m) 256.060 240.157 222.974 239.695 223.197 240.692 225.815
Cd (–) 0.383 0.457 0.400 0.457 0.400 0.432 0.385
Ls (m) 34.770 24.728 22.899 24.678 22.890 24.740 23.451
Lc (m) 73.181 51.462 47.953 51.380 47.956 51.578 48.928
L (m) 1,024.529 720.469 671.334 719.321 671.383 722.089 684.991
Vw (m3) 9,972.916 5,795.709 5,321.531 5,826.764 5,328.667 6,360.948 5,566.052
Ve (m3) 1,315.879 912.080 824.251 915.694 824.780 1,001.650 862.704
Vs (m3) 3,500.596 2,634.222 2,290.550 2,626.148 2,292.138 2,676.167 2,365.893
VT (m3) 14,789.391 9,342.012 8,436.333 9,368.336 8,445.586 10,038.766 8,794.650

This parameter is important in locating spillways, especially Increasing the angle of side legs (α) can increase the dis-
in earth dams that their spillways are situated outside the charge coefficient. In optimal models, this parameter was
main body of the dam. In fact, the limitation of this width increased by more than 30%. Totally, in the optimal
is expressed as one of the main reasons for the choice of models, the discharge coefficient, angle of side legs, and
labyrinth spillways. The Tw is another parameter that con- total head of the spillway were increased, and the height,
tributes to the calculation of volume of spillway walls, thickness, length and width (spillway and apex) of the labyr-
which has decreased in the optimal models. inth spillway were decreased.

Table 11 | Results of models’ constraints

SFLA IWO CS

Constraint Ute spillway Base RCP2.6 Base RCP2.6 Base RCP2.6

0:05  Ht =P  0:90 0.634 0.812 0.779 0.809 0.778 0.804 0.779

1:00  A=Tw  2:00 1.709 1.000 1.083 1.005 1.093 1.000 1.005
6:00  P=Tw  8:00 8.582 8.000 8.000 7.999 7.998 7.999 7.999
2:00  W=P  4:00 2.001 2.138 2.000 2.127 2.001 2.048 2.001
3:00  Lc =W < 9:50 4.001 3.000 3.011 3.001 3.008 3.000 3.033
A=W  0:0765 0.099 0.059 0.068 0.059 0.068 0.061 0.063
LD =LS  0:35 0.460 0.35 0.35 0.35 0.35 0.35 0.35

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 933 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

As is obvious in Table 10, discharge constraint is more

than the desired value. According to Table 11, other con-
straints in the design of Ute labyrinth spillway are in the
desired range. In the Ute spillway, the value of the interfer-
ence length ratio was 0.46, which was reduced to 0.35 in the
optimal models. To increase the spillway capacity, the apex
ratio (A/W ) should be limited to 0.0765, which, according
to Table 11, is not considered in the Ute design, but it is
within the desired range in the proposed models.

CONCLUSION

In this paper, a novel approach was presented to investigate

the effect of climate change uncertainty on the optimization
of Ute Dam labyrinth spillway design. The concrete volume
of the labyrinth spillway was selected as the objective
function. The concrete volume may represent cost and
environmental impacts of spillway construction. According
to the present study results and historical streamflow data
from 1941, the peak flow at the USGS 07226500 streamgage
is decreasing. Hence, the design flood should be decreased.
Results revealed that all three algorithms (SFLA, IWO, and
CS) are able to optimize the problem. But, the SFLA was a
Figure 10 | One cycle of Ute and optimal labyrinth spillways.
little better in finding optimal solutions.

Figure 11 | Plan view of the Ute and optimal labyrinth spillways.

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 934 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

It is worth mentioning that because of different uncer- Bozorg-Haddad, O. (ed.)  Advanced Optimization by Nature-
tainties that exist in methodology of this study such as the Inspired Algorithms. Springer, Singapore. doi:10.1007/978-
981-10-5221-7.
downscaling method and climate change scenarios, the pre- Bozorg-Haddad, O., Mirmomeni, M. & Mariño, M. A. 
sent result is one of the possible solutions. For example, in Optimal design of stepped spillways using the HBMO
the IHACRES model, the snow and vegetation cover were algorithm. Civil Engineering and Environmental Systems
27 (1), 81–94. doi:10.1080/10286600802542465.
not considered, which might affect the streamflow and
Brauer, D., Baumhardt, R. L., Gitz, D., Gowda, P. & Mahan, J. 
design flood. As was stated, IHACRES is a lumped model. Characterization of trends in reservoir storage, streamflow,
In the future studies, other rainfall–runoff models (i.e., dis- and precipitation in the Canadian River watershed in New
tributed or semi-distributed) can be used. In addition, in Mexico and Texas. Lake and Reservoir Management 31 (1),
64–79. doi:10.1080/10402381.2015.1006348.
future studies, discharge volumes can be used in design
Chen, S. H.  Hydraulic Structures. Springer. doi:10.1007/978-
flood estimation. 3-662-47331-3.
In brief, climate change, as a hydrologic uncertainty, has Croke, B. F. W., Andrews, F., Spate, J. & Cuddy, S. M. 
great impact on rainfall, temperature, streamflow, and design IHACRES User Guide. Technical Report 2005/19, 2nd edn.
ICAM, School of Resources, Environment and Society, The
flood. Spillway geometry is strongly affected by design flood, Australian National University, Canberra.
and optimization algorithms can be very effective in reducing Crookston, B. M.  Labyrinth Weirs. PhD Dissertation, Utah
the spillway construction costs. By optimizing hydraulic struc- State University, Logan, UT.
Crookston, B. M. & Tullis, B. P.  Labyrinth weirs: nappe
tures geometry (e.g., reducing the concrete volume), such as
interference and local submergence. Journal of Irrigation and
spillways, construction time and other costs associated with Drainage Engineering 138 (8), 757–765. doi:10.1061/(ASCE)
extraction, production, transport, storage, destruction and IR.1943-4774.0000466.
recycling of materials may be reduced. These will also be effec- Crookston, B. M. & Tullis, B. P.  Hydraulic design and analysis
of labyrinth weirs. I: discharge relationships. Journal of
tive in ameliorating environmental impacts.
Irrigation and Drainage Engineering 139 (5), 363–370. doi:10.
1061/(ASCE)IR.1943-4774.0000558.
Cunnane, C.  Statistical Distributions for Flood Frequency
CONFLICT OF INTEREST Analysis Operational. Hydrology Report No. 33, World
Meteorological Organization, No. 718, Geneva, Switzerland.
Dong, N. D., Jayakumar, K. V. & Agilan, V.  Impact of climate
We declare that we have no conflict of interest with any change on flood frequency of the Trian Reservoir in Vietnam
person or institution. using RCMs. Journal of Hydrologic Engineering 23 (2),
05017032. doi:10.1061/(ASCE) HE.1943-5584.0001609.
Duricic, J., Erdik, T., Pektaş, A. O. & Van Gelder, P. H. A. J. M.
 Mean normalized force computation for different types
REFERENCES of obstacles due to dam break using statistical techniques.
Water 5 (2), 560–577. doi:10.3390/w5020560.
Ashofteh, P. S., Bozorg-Haddad, O., Akbari-Alashti, H. & Marino, Ehteram, M., Mousavi, S. F., Karami, H., Farzin, S., Singh, V. P., Chau,
M. A.  Determination of irrigation allocation policy K. W. & El-Shafie, A. a Reservoir operation based on
under climate change by genetic programming. Journal of evolutionary algorithms and multi-criteria decision-making
Irrigation and Drainage Engineering 141 (4), 04014059. under climate change and uncertainty. Journal of
doi:10.1061/(ASCE)IR.1943-4774.0000807. Hydroinformatics 20 (2), 332–355. doi:10.2166/hydro.2018.094.
Azimi, A. H. & Seyed Hakim, S.  Hydraulics of flow over Ehteram, M., Singh, V. P., Karami, H., Hosseini, K., Dianatikhah,
rectangular labyrinth weirs. Irrigation Science 37 (2), M., Hossain, M., Ming Fai, C. & El-Shafie, A. b Irrigation
183–193. doi:10.1007/s00271-018-0616-6. management based on reservoir operation with an improved
Bates, B. C., Kundzewicz, Z. W., Wu, S. & Palutikof, J. P. (eds) weed algorithm. Water 10 (9), 1267. doi:10.3390/w10091267.
 Climate Change and Water: Technical Paper of the Ehteram, M., Singh, V. P., Ferdowsi, A., Mousavi, S. F., Farzin, S.,
Intergovernmental Panel on Climate Change. IPCC Karami, H., Mohd, N. S., Afan, H. A., Lai, S. H., Kisi, O.,
Secretariat, Geneva, Switzerland. Malek, M. A., Ahmed, A. N. & El-Shafie, A.  An
Bhagat, N.  Flood frequency analysis using Gumbel’s improved model based on the support vector machine and
distribution method: a case study of Lower Mahi Basin, cuckoo algorithm for simulating reference
India. Journal of Water Resources and Ocean Science 6 (4), evapotranspiration. PLoS One 14 (5), e0217499. doi: 10.
51–54. doi:10.11648/j.wros.20170604.11. 1371/journal.pone.0217499.

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 935 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Erdik, T.  Fuzzy logic approach to conventional rubble Hydraulic Engineering: Saving A Threatened Resource-In
mound structures design. Expert Systems with Applications Search of Solutions, August 2–6, Baltimore, Maryland.
36 (3), 4162–4170. doi:10.1016/j.eswa.2008.06.012. Hosseini, K., Nodoushan, E. J., Barati, R. & Shahheydari, H. 
Erdik, T. & Savci, M. E.  TAKAGI-SUGENO fuzzy approach Optimal design of labyrinth spillways using meta-heuristic
in rock armored slopes for 2% wave runup estimation. algorithms. KSCE Journal of Civil Engineering 20 (1),
Coastal Engineering Journal 50 (2), 161–177. doi:10.1142/ 468–477. doi:10.1007/s12205-015-0462-5.
S0578563408001776. Houston, K. L.  Hydraulic Model Study of Ute Dam Labyrinth
Erdik, T., Savci, M. E. & Ş en, Z.  Artificial neural networks Spillway. Report GR-82-7.
for predicting maximum wave runup on rubble mound Indelkofer, H. & Rouve, G.  Discharge over polygonal weirs.
structures. Expert Systems with Applications 36 (3), Journal of the Hydraulics Division 110 (HY3), 385–401.
6403–6408. doi:10.1016/j.eswa.2008.07.049. Jakeman, A. J., Littlewood, I. G. & Whitehead, P. G. 
Erdik, T., Duricic, J. & Van Gelder, P. H. A. J. M.  The Computation of the instantaneous unit hydrograph and
probabilistic assessment of overtopping reliability on Akyayik identifiable component flows with application to two small
dam. KSCE Journal of Civil Engineering 17 (7), 1810–1819. upland catchments. Journal of Hydrology 117 (1–4), 275–300.
doi:10.1007/s12205-013-1355-0. doi:10.1016/0022-1694(90)90097-H.
Eusuff, M. M. & Lansey, K. E.  Optimization of water Karamouz, M., Doroudi, S., Ahmadi, A. & Moridi, A. 
distribution network design using the shuffled frog leaping Optimal design of water diversion system: a case study.
algorithm. Journal of Water Resources Planning and In: World Environmental and Water Resources Congress
Management 129 (3), 210–225. doi:10.1061/(ASCE)0733- 2009: Great Rivers 1–10. doi:10.1061/41036(342)342.
9496(2003)129:3(21 0). Karamouz, M., Nazif, S. & Zahmatkesh, Z.  Self-organizing
Falvey, H. T.  Hydraulic Design of Labyrinth Weirs. ASCE Gaussian-based downscaling of climate data for simulation of
Press, Reston, VA. doi:10.1061/9780784406311. urban drainage systems. Journal of Irrigation and Drainage
Ferdowsi, A., Farzin, S., Mousavi, S. F. & Karami, H.  Hybrid Engineering 139 (2), 98–112. doi:10.1061/(ASCE) IR.1943-
bat and particle swarm algorithm for optimization of 4774.0000500.
labyrinth spillway based on half and quarter round crest Kardan, N., Hassanzadeh, Y. & Bonab, B. S.  Shape
shapes. Flow Measurement and Instrumentation 66, optimization of trapezoidal labyrinth weirs using genetic
209–217. doi:10.1016/j.flowmeasinst.2019.03.003. algorithm. Arabian Journal for Science and Engineering
Fill, H. D. & Steiner, A. A.  Estimating IPF from mean daily 42 (3), 1219–1229. doi:10.1007/s13369-016-2355-4.
flow data. Journal of Hydrologic Engineering 8 (6), 365–369. Kaveh, A.  Advances in Metaheuristic Algorithms for Optimal
doi:10.1061/(ASCE)1084-0699(2003)8:6(365). Design of Structures, 2nd edn. Springer, Switzerland.
Fuller, W. E.  Flood flows. Transations of the ASCE 77, doi:10.1007/978-3-319-05549-7.
564–617. Khatsuria, R. M.  Hydraulics of Spillways and Energy
Ghaderi, A., Abbasi, S., Abraham, J. & Azamathulla, H. M.  Dissipators. CRC Press, New York.
Efficiency of trapezoidal labyrinth shaped stepped spillways. Khode, B. V. & Tembhurkar, A. R.  Evaluation and analysis of
Flow Measurement and Instrumentation 72, 101711. doi:10. crest coefficient for labyrinth weir. World Applied Sciences
1016/j.flowmeasinst.2020.101711. Journal 11 (7), 835–839.
Ghare, A. D., Mhaisalkar, V. A. & Porey, P. D.  An approach King, J. P., Hawley, J. W., Hernandez, J., Kennedy, J. F. &
to optimal design of trapezoidal labyrinth weirs. World Martinez, E. L.  Study of Potential Water Salvage on the
Applied Sciences Journal 3 (6), 934–938. Tucumcari Project Arch Hurley Conservancy District. New
Guo, S., Muhammad, R., Liu, Z., Xiong, F. & Yin, J.  Design Mexico Water Resources Institute, New Mexico State
flood estimation methods for cascade reservoirs based on University, Report TR335.
copulas. Water 10 (5), 560. doi:10.3390/w 10050560. Li, Y., Zhou, J., Zhang, Y., Qin, H. & Liu, L.  Novel
Haan, C. T.  Statistical Methods in Hydrology. Iowa State multiobjective shuffled frog leaping algorithm with
University Press, Ames, IA. application to reservoir flood control operation. Journal of
Hager, W. H., Pfister, M. & Tullis, B. P.  Labyrinth weirs: Water Resources Planning and Management 136 (2),
developments until 1985. In E-Proceedings of the 36th IAHR 217–226. doi:10.1061/_ASCE_ WR.1943-5452.0000027.
World Congress, June 28–July 3, Hague, Netherlands. Mirnaseri, M. & Emadi, A.  Hydraulic performance of
Hassanvand, M. R., Karami, H. & Mousavi, S. F.  Use of multi- combined flow labyrinth weir-gate. Advance in Agriculture
criteria decision-making for selecting spillway type and and Biology 2, 54–60. doi:10.15192/PSCP.AAB.2014.1.2.
optimizing dimensions by applying the harmony search 5460.
algorithm: Qeshlagh dam case study. Lakes & Reservoirs: Monsef, H., Naghashzadegan, M., Jamali, A. & Farmani, R. 
Research and Management 24 (1), 66–75. doi:10.1111/lre. Comparison of evolutionary multi objective optimization
12250. algorithms in optimum design of water distribution network.
Hepler, T.  Innovative spillway designs. In Proceedings of the Ain Shams Engineering Journal 10 (1), 103–111. doi:10.
Hydraulic Engineering Sessions at Water Forum 92, 1016/j.asej.2018.04.003.

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest
 936 A. Ferdowsi et al. | Optimization of dam’s spillway design under climate change Journal of Hydroinformatics | 22.4 | 2020

Moriasi, D. N., Arnold, J. G., Van Liew, M. W., Bingner, R. L., Harmel, assessment under climate change conditions. Environmental
R. D. & Veith, T. L.  Model evaluation guidelines for Monitoring and Assessment 189 (7), 359. doi:10.1007
systematic quantification of accuracy in watershed simulations. /s10661-017-6067-3.
Transactions of the ASABE 50 (3), 885–900. Sarzaeim, P., Bozorg-Haddad, O., Zolghadr-Asli, B., Fallah-
Najafi, M. R., Moradkhani, H. & Jung, I. W.  Assessing the Mehdipour, E. & Loáiciga, H. A.  Optimization of run-of-
uncertainties of hydrologic model selection in climate change river hydropower plant design under climate change
impact studies. Hydrological Processes 25 (18), 2814–2826. conditions. Water Resources Management 32 (12),
doi:10.1002/hyp.8043. 3919–3934. doi:10.1007/s11269-018-2027-0.
Nazif, S. & Karamouz, M.  Evaluation of climate change Spencer, S. & Salazar, A.  Historical Trends in key
impacts on streamflow to a multiple reservoir system using a Components of the Hydrologic Cycle for the Lake Meredith
data-based mechanistic model. Journal of Water and Climate Watershed. Available from: http://www.panhandlewater.org/
Change 5 (4), 610–624. doi:10.2166/wcc.2014.012. 2011 adopted plan.html.
Novak, P., Moffat, A. I. B., Nalluri, C. & Narayanan, R.  Tabari, M. M. R. & Hashempour, M.  Development of GWO–
Hydraulic Structures. CRC Press, New York. DSO and PSO–DSO hybrid models to redesign the optimal
Onen, F. & Bagatur, T.  Prediction of flood frequency factor dimensions of labyrinth spillway. Soft Computing 23,
for Gumbel distribution using regression and GEP model. 6391–6406. doi:10.1007/s00500-018-3292-9.
Arabian Journal for Science and Engineering 42 (9), Taguas, E. V., Ayuso, J. L., Pena, A., Yuan, Y., Sanchez, M. C.,
3895–3906. doi:10.1007/s13369-017-2507-1. Giraldez, J. V. & Pérez, R.  Testing the relationship
Orouji, H., Bozorg Haddad, O., Fallah-Mehdipour, E. & Marino, between instantaneous peak flow and mean daily flow in a
M. A.  Estimation of Muskingum parameter by meta- Mediterranean Area Southeast Spain. Catena 75 (2),
heuristic algorithms. Proceedings of the Institution of Civil 129–137. doi:10.1016/j .catena.2008.04.015.
Engineers-Water Management 165 (6), 1–10. Tullis, J. P., Amanian, N. & Waldron, D.  Design of labyrinth
Orouji, H., Mahmoudi, N., Fallah-Mehdipour, E., Pazoki, M. & spillways. Journal of Hydraulic Engineering 121 (3), 247–255.
Biswas, A.  Shuffled frog-leaping algorithm for optimal doi:10.1061/(ASCE)0733-9429 (1995)121:3(247).
design of open channels. Journal of Irrigation and Drainage Wilby, R. L., Dawson, C. W. & Barrow, E. M.  SDSM – a
Engineering 142 (10), 06016008. doi:10.1061/(ASCE)IR. decision support tool for the assessment of regional climate
1943-4774.0001059. change impacts. Environmental Modelling & Software 17 (2),
Pektas, A. O. & Erdik, T.  Peak discharge prediction due to 145–157. doi:10.1016/S1364-8152(01)00060-3.
embankment dam break by using sensitivity analysis based Wilks, D. S.  Comparison of three-parameter probability
ANN. KSCE Journal of Civil Engineering 18 (6), 1868–1876. distributions for representing annual extreme and partial
doi:10.1007/s12205-014-0047-8. duration precipitation series. Water Resources Research
Raju, K. S. & Kumar, D. N.  Impact of Climate Change on 29 (10), 3543–3549. doi:10.1029/93WR01710.
Water Resources with Modeling Techniques and Case Willmore, C.  Hydraulic Characteristics of Labyrinth Weirs.
Studies. Springer, Singapore. doi:10.1007/978-981-10-6110-3. MSc Thesis, Utah State University, Logan, UT.
Safarrazavi Zadeh, M., Esmaeili Varaki, M. & Biabani, R.  Wilson, L. & O’Brien, V.  Canadian River Watershed: Brush
Experimental study on flow over sinusoidal and semicircular Control Planning, Assessment and Feasibility Study.
labyrinth weirs. ISH Journal of Hydraulic Engineering 1–10. Available from: http://www.tsswcb.texas.gov/reports#
doi:10.1080/09715010.2019.1644679. feasibilitystudy.
Safavi, H. R., Sajjadi, S. M. & Raghibi, V.  Assessment of Xing, W., Wang, W., Shao, Q. & Ding, Y.  Estimating net irrigation
climate change impacts on climate variables using requirements of winter wheat across central-eastern China
probabilistic ensemble modeling and trend analysis. under present and future climate scenarios. Journal of Irrigation
Theoretical and Applied Climatology 130 (1–2), 635–653. and Drainage Engineering 144 (7), 05018005. doi:10.1061/
doi:10.1007/s00704-016-18983. (ASCE)IR.1943-4774.0001320.
Sangal, B. P.  Practical method of estimating peak flow. Yang, X. S.  Nature-Inspired Metaheuristic Algorithms, 2nd
Journal of Hydraulic Engineering 109 (4), 549–563. doi:10. edn. Luniver Press, Frome, UK.
1061/(ASCE)0733-9429(1983)109:4(549). Yang, X. S. (ed.)  Nature-inspired Algorithms and Applied
Sangsefidi, Y., Mehraein, M., Ghodsian, M. & Motalebizadeh, Optimization. Springer. doi:10.1007/978-3-319-67669-2.
M. R.  Evaluation and analysis of flow over arced weirs Zahmatkesh, Z., Karamouz, M., Goharian, E. & Burian, S. J. 
using traditional and response surface methodologies. Analysis of the effects of climate change on urban storm
Journal of Hydraulic Engineering 143 (11), 04017048. doi:10. water runoff using statistically downscaled precipitation data
1061/(ASCE)HY.1943-7900.0001377. and a change factor approach. Journal of Hydrologic
Sarzaeim, P., Bozorg-Haddad, O., Fallah-Mehdipour, E. & Engineering 20 (7), 05014022. doi:10.1061/(ASCE)HE.1943-
Loáiciga, H. A.  Environmental water demand 5584.0001064.

First received 24 January 2020; accepted in revised form 22 April 2020. Available online 5 June 2020

Downloaded from http://iwaponline.com/jh/article-pdf/22/4/916/715609/jh0220916.pdf

by guest