Applied Mathematical Modelling 37 (2013) 4430–4439

Contents lists available at SciVerse ScienceDirect

Applied Mathematical Modelling

journal homepage: www.elsevier.com/locate/apm

Optimal design of a sewer line using Linear Programming

Prabhata K. Swamee a,1, Ashok K. Sharma b,⇑
a
Dept. of Civil Eng., Indian Institute of Technology Kanpur, Kanpur 208016, Uttar Pradesh, India
b
CSIRO Land and Water, P.O. Box 56, Highett 3190, Victoria, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Wastewater collection systems greatly contribute to the cost of the overall municipal sew-
Received 14 March 2012 erage system; a cost-effective design of the collection system will provide significant sav-
Received in revised form 5 August 2012 ings towards the cost of wastewater services. It is impossible to evaluate the full impact
Accepted 11 September 2012
that each pipe size and slope would have on the overall cost of the collection system with
Available online 20 September 2012
intuitive designs. However, these solutions generally satisfy the design objectives within
the given constraints. A survey of the literature indicates that various optimisation tech-
Keywords:
niques are being applied for least-cost solutions. In general these approaches provide con-
Design
Optimisation
tinuous pipe sizes, which are converted to closest commercial sizes for adoption, which
Roughness height would heavily dilute the optimal outcome. Search methods are also adopted to obtain
Sewer cost-effective design solutions using directly commercial pipe sizes, which are computa-
Sewerage and wastewater collection tionally expensive. In the design of a sewerage system, a sewer line is a basic unit occurring
repeatedly in the design-process and finally the combinations of these basic units formu-
late the complete sewer system. However, the branch sewer lines, main sewers, trunk sew-
ers, pumping stations, treatment plant and outfall sewers are in general the main
components of an urban wastewater collection, treatment and disposal systems. A method
has been developed to optimise this basic unit using Linear Programming technique with-
out transforming nonlinear objective function or constraint equations into linear functions
and incorporating commercially available pipe sizes directly in the problem formulation.
The current research area of optimal sewer system design is focusing equally on economic
considerations and hydraulic feasibility and moving away from conventional design guide-
lines based on only self cleaning velocity concepts for node to node sewer link hydraulic
design. This paper is a step forward in developing optimal design approaches of sewer
systems.
Crown Copyright Ó 2012 Published by Elsevier Inc. All rights reserved.

1. Introduction

Wastewater from residential, commercial and industrial areas is collected and transported through the sewerage system
to a sewage treatment plant, where it is treated to the specified standards before it is reused or disposed to receiving water. A
sewerage system has a tree-like structure and is composed of various sewer lines which terminate at a junction that contains
a larger sewer line. This larger sewer line further terminates at the junction of a still larger sewer line – the main sewer line –
which eventually terminates at the wastewater treatment plant.
The hydraulic design of sewer system has not undergone any major change in the last 100 years; however a lot has been
done in the construction and management of these systems. A typical system involves laying out a sewer network along

⇑ Corresponding author.
E-mail addresses: swameefce@yahoo.com, swamee@iitk.ac.in (P.K. Swamee), ashok.sharma@csiro.au (A.K. Sharma).
1
Visiting Faculty.

0307-904X/$ - see front matter Crown Copyright Ó 2012 Published by Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.apm.2012.09.041
 P.K. Swamee, A.K. Sharma / Applied Mathematical Modelling 37 (2013) 4430–4439 4431

Notations

A flow area
ce unit excavation cost at ground level
cr increase in unit excavation cost per unit depth
cs sheeting and shoring cost coefficient
D link diameter
d nodal depth
d0 initial nodal depth
df outfall depth
dmax maximum nodal depth
dn terminal nodal depth
g gravitational acceleration
hmin minimum cover depth
kD diameter coefficient
ke earthwork cost coefficient
kh man hole cost coefficient
km pipe cost parameter
L link length
m pipe cost parameter
n number of links
P flow perimeter
Q link discharge
Qmax carrying capacity of a link
R hydraulic radius
So invert slope of a link
V average fluid velocity
Vmax maximum fluid velocity
Vs scouring velocity
Vsc self-cleaning velocity
w trench width
e average roughness height
c Lagrange multiplier
m kinematic viscosity
g relative depth

Subscripts
i link or node; and
j index

existing and proposed streets which terminates at the wastewater treatment plant  normally at the outskirts of urban
boundary. Each sewer link is then designed as a separate element using some relationships governing the hydraulics of flow
and a set of limiting constraints [1].
Camp [2] presented a method for the hydraulic design of sewer networks and highlighted the two main functions of the
sewer systems: to carry the maximum discharge for which it is designed and to transport suspended solids. Since then many
researchers [3–9] have contributed to the design of the sewer network and have applied various optimisation techniques.
They have described heuristic methodologies for sewer design that could be adapted on microcomputers. Gupta et al.
[10] used Powell’s method of conjugate directions for depth–diameter optimisation of wastewater collection systems. Arg-
aman et al. [11–14] applied dynamic programming for sewer systems design. Fisher et al. [15,1,16–18] used piecewise lin-
earization to apply Linear Programming (LP) for estimating the pipe sizes and slopes. Swamee [19] developed a sewer line
design method minimising nonlinear cost function and nonlinear constraints by iterative application of the Lagrange mul-
tiplier method. Genetic Algorithm (GA) is most popular and widely used search method. There are many examples of its
application in sewer system design [20–24]. Hanghighi and Bakhshipour [25] highlighted that GAs slowly progress in a ran-
dom-based framework and thus they are not computationally efficient compared to mathematical methods. As the number
of variables and constraints increase the GAs become slow. They developed an adaptive genetic algorithm so that only fea-
sible solutions are developed. The sewer pipe hydraulics estimates continuous pipe diameter which is rounded off to the first
larger size in the commercial list for subsequent analysis. To overcome the slow progress of GAs some researchers linked this
technique with other optimisation approaches. Cisty [26] hybridized the GA with LP and Haghighi et al. [27] hybridized GA
with Integer LP for optimisation of water distribution system for improving efficiency. The integration of GA and LP could be
4432 P.K. Swamee, A.K. Sharma / Applied Mathematical Modelling 37 (2013) 4430–4439

an attractive method; however the transformation of nonlinear functions into linear functions by piecewise linearization de-
stroys the originality of the function. Similarly, Pan and Kao [28] linked GA with quadratic programming (QP) model to im-
prove the solvability. In this approach, the nonlinear functions are transformed into quadric forms and solved using QP
instead of their linearization; however implementing such a combination would be very complex.
An ant colony optimisation algorithm for the storm sewer deign has also been applied [29,30]. In this approach nodal ele-
vations are used as decision variables and the pipe diameters are estimated satisfying hydraulic and other conditions for
each pipe link. Such a design cannot be termed as holistic design. Guo et al. [31] and Afshar and Rohni [32] have explored
the application of cellular automata based approaches for the optimal design of sewer systems, which is in its early stages of
development. Tabu search and simulated annealing techniques are also being applied for the sewer design, which require
experience in setting the parameters for their application [33].
In general, these approaches use the Manning equation or Hazen–Williams equation for resistance description; however,
ASCE Task Force [34] has disapproved the Manning equation and recommended the use of the Darcy–Weisbach equation for
open channel resistance. Liou [35] has strongly discouraged the use of the Hazen–Williams equation and pointed its limita-
tions. Similarly Brown [36] investigated the history of Darcy–Weisbech equation and indicated that due to the general accu-
racy and complete range of application only the Darcy–Weisbech equation should be applied as the standard and others
should be left for historians. Considering above conclusions only the Darcy–Weisbech equation has been adopted in the pres-
ent formulation of the design problem.
Huge amount of public money is invested around the world to provide sewerage services in the existing or upcoming
developments. The existing under capacity sewerage services are also upgraded in the growing areas. The design engineers
in general make decisions for the sewer systems based on calculations and their experience by analysing limited number of
options. Such solutions are seldom optimum. Thus, any small monitory efficiency in the provision of these services on such a
large sector would result in the substantial savings in the public funds.
Although the Linear Programming (LP) optimisation method has been applied to the optimal design of water distribution
networks, however its application in the design of sewer systems without linearization of objective functions is new. In this
approach the whole system is designed as single entity and not as individual pipe link. The algorithm terminates in limited
number of iterations depending upon the minimum and maximum sizes of commercial pipes used in the optimisation prob-
lem formulation. With the commercial pipe sizes used directly in the sewer line design methodology, the conversion of con-
tinuous estimated pipe diameters to nearest commercial pipe sizes could be avoided which otherwise would lose the
optimality of the whole deign.
The overall wastewater collection and treatment systems are comprised of sewer lines, trunk mains, sometimes com-
bined sewer systems, pumping stations, treatment plant and outfall sewer. Now wastewater reuse and recycling compo-
nents are also becoming integral part of the urban wastewater systems. However, in the design of a sewerage system the
sewer line is the basic unit occurring repeatedly in the design-process and finally the combinations of these basic units for-
mulate the complete sewer system. In this paper a method for optimal design of this basic unit is presented applying LP opti-
misation technique for the estimation of pipe diameters and sewer depths, using the Darcy–Weisbach equation as the
resistance equation and commercially available pipe diameters directly in the problem formulation.

2. Resistance relationships

The ASCE Task Force [34] on Friction in Open Channels has recommended the use of Darcy–Weisbach equation. Thus, the
invert slope So is given by
2
fQ
So ¼ ; ð1Þ
8gRA2
where f = friction factor; Q = discharge; g = gravitational acceleration; and R = hydraulic radius defined as the ratio of the flow
area A to the flow perimeter P. For a circular section A and P are given by
h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii
A ¼ 0:25 D2 cos1 ð1  2gÞ  2ð1  2gÞ gð1  gÞ ; ð2Þ

P ¼ D cos1 ð1  2gÞ; ð3Þ

where D = sewer diameter; g = relative flow depth defined as the ratio of the normal depth y (depth of uniform flow as de-
picted in Fig. 1) to sewer diameter. For obtaining f ASCE Task Force [34] recommended the following variant of the Colebrook
equation:
" !#2
e 0:625m
f ¼ 1:325 ln þ pffiffiffi ; ð4Þ
12R VR f

where e = average roughness height of the surface; and m = kinematic viscosity of water. Swamee [37] converted Eq. (4) to the
following explicit form
 P.K. Swamee, A.K. Sharma / Applied Mathematical Modelling 37 (2013) 4430–4439 4433

Fig. 1. Definition sketch: Circular sewer section.

( "  0:9 #)2

4 e vP
f ¼ ln þ 1:63 : ð5Þ
3 12R Q

Using Eqs. (1) and (5) the bed slope So for open channel flow is obtained as
( "
 0:9 #)2
Q2 vP e
So ¼ 2
ln þ 1:63 : ð6Þ
6gA R 12R Q

Swamee [37] gave the following equation for the maximum average velocity Vmax and the maximum discharge Qmax
occurring at the flow depths 0.813D and 0.943D respectively as
!
pffiffiffiffiffiffiffiffiffiffiffi e 1:32m
V max ¼ 1:355 gDSo ln þ pffiffiffiffiffiffiffiffiffiffiffi ; ð7Þ
3:65D D gDSo

!
pffiffiffiffiffiffiffiffiffiffiffi
2 e 1:43m
Q max ¼ D gDSo ln þ pffiffiffiffiffiffiffiffiffiffiffi : ð8Þ
3:46D D gDSo

Further for circular sewer section, Swamee and Swamee [38] gave the following equation for the normal depth yn:
 " 0:665 #1
Q max
yn ¼ 0:943D 0:823 1 þ1 ; ð9Þ
Q

where Qmax is obtained from Eq. (8). Barring small depths g 6 0.3, the maximum error involved in Eq. (9) is 1%.

3. Cost function

The cost of a sewer line primarily consists of the link costs and nodal costs. Sewer pipes cost is a link cost whereas the cost
of manholes is a nodal cost. Although the cost of excavation is a link cost, for mathematical convenience, it would be lumped
at nodes. A sewer line and its longitudinal profile are shown in Figs. 2 and 3. Considering Fig. 2, the cost of ith link of sewer
pipe Cpi is written as
C pi ¼ ci Li ; ð10Þ
where ci = cost of pipe per unit length; and Li = length of ith link The coefficient ci depends on sewer diameter Di. As per Fig. 3,
the cost of ith manholes Cmi is
C mi ¼ khi di ; ð11Þ
where khi = cosy of manhole per unit depth; and di = ith nodal depth. The coefficient khi depends on di and the maximum sewer
pipe diameter D connecting the manhole as there will be two or more pipes connected at a manhole.
Local design guidelines should be consulted for manhole size based on manhole depths for estimating cost of manholes
and corresponding cost function developed. The following internal sizes of circular manholes for varying depths have been
adopted for costing [39]:
4434 P.K. Swamee, A.K. Sharma / Applied Mathematical Modelling 37 (2013) 4430–4439

Fig. 2. Definition sketch: Plan of a sewer line.

Fig. 3. Definition sketch: Longitudinal profile of sewer line.

Fig. 4. Definition sketch: Sewer trench.

 For depths above 0.90 m and up to 1.65 m – manhole 0.90 m diameter

 For depths above 1.65 m and up to 2.30 m – manhole 1.20 m diameter
 For depths above 2.30 m and up to 9.00 m – manhole 1.50 m diameter
 For depths above 9.00 m and up to 14.0 m – manhole 1.80 m diameter

The ith link of sewer trench has a trapezoidal cross-section having a side slope 0.1 horizontal to 1 vertical (Fig. 4). As an
approximation, assuming vertical sides, the earthwork cost of the ith link Cei is
 
1 1  2 2
C ei ¼ Li wi ce ðdi1 þ di Þ þ cr di1 þ di1 di þ di þ cs Li ðdi1 þ di Þ; ð12Þ
2 3
 P.K. Swamee, A.K. Sharma / Applied Mathematical Modelling 37 (2013) 4430–4439 4435

where ce = unit excavation cost at ground level (monetary unit per unit volume of earthwork, e.g., $m3); cr = increase in unit
excavation cost per unit depth (monetary unit per unit volume per unit depth, e.g., $m4) and cs = cost of sheeting and shor-
ing per unit area (monetary unit per unit area, e.g., $m2). Generally timbering or steel trench shields are used to protect
pipeline workers from trench collapse, including maintaining the ground slope and working space. Swamee et al. [40] have
studied excavation cost of various types of strata and found that ce/cr varied between 6.96 and 13.86 m. According to CPHEEO
[39] the bottom sewer trench width wi is given by the following criteria:

wi ¼ 1m; Di 6 0:6m; ð13Þ

wi ¼ Di þ 0:4m; Di > 0:6m: ð14Þ

Lumping half of ith link cost coefficient and half of the (i + 1)st link cost coefficient and adding the manhole cost coeffi-
cient as defined by Eq. (11), the ith nodal cost coefficient kni is written as

kni ¼ khi þ 14 ce ½Li wi ðmi1 þ 1Þ þ Liþ1 wiþ1 ð1 þ miþ1 Þ þ 12 cs ½Li ðmi1 þ 1Þ þ Liþ1 ð1 þ miþ1 Þ
1

ð15Þ
þ 12 cr di Li wi ðm2i1 þ mi1 þ 1Þ þ Liþ1 wiþ1 ð1 þ miþ1 þ m2iþ1 Þ ; i ¼ 1; 2; 3; . . . ; ðiL  1Þ;

where mi1 = di1/di;mi+1 = di+1/di; Li+1 and wi+1 = length and width respectively of (i + 1)st link; and iL = total number of links.
Lumping half of the 1st link cost to 0th node and adding the 0th manhole cost coefficient given by Eq. (11), the nodal cost
coefficient is written as:
 
1 1 1
kn0 ¼ kh0 þ L1 w1 ce ð1 þ m1 Þ þ cr d0 ð1 þ m1 þ m21 Þ þ cs L1 ð1 þ miþ1 Þ: ð16Þ
4 3 2

On the other hand, lumping half of the iLth link cost to iLth node and adding the iLth manhole cost coefficient given by Eq.
(11), the nodal cost coefficient kniL is written as
 
1 1 1
kniL ¼ kLhiL þ LiL wiL ce ðmiL1 þ 1Þ þ cr diL ðm2iL1 þ miL1 þ 1Þ þ cs LiL ðmiL1 þ 1Þ; ð17Þ
4 3 2
Thus, the cost function of a sewerline Csewer is written as
XiL
C sewer ¼ kn0 d0 þ i¼1
ðkni di þ ci Li Þ; ð18Þ

where d0 = initial nodal depth. Swamee [37] recommended the following equation for d0 which provides a constant value of
the manhole cost.
d0 ¼ hmin þ 0:5m;
where hmin = minimum cover. As the sewer depth d0 is fixed, first term representing the manhole cost is also fixed.

4. Constraints description

The invert slope of ith link Soi is given by

di  di1 þ zi1  zi
Soi ¼ ; ð19Þ
Li
where zi1 and zi = elevations at (i  1)st and ith nodes. See Fig. 3, Eq. (6) for ith link is
( "  0:9 #)2
Q 2i ei mPi
Soi ¼ ln þ 1:63 : ð20Þ
6gA2i Ri 12Ri Qi

Using Eqs. (2) and (3) for circular sewer sections, Eq. (20) modifies to
( "  0:9 #)2
k1 Q 2i ei mDi
Soi ¼ ln þ k3 ; ð21Þ
gD5i k2 Di Qi

where
32 cos1 ð1  2gÞ
k1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3 ; ð22Þ
3 cos1 ð1  2gÞ  2ð1  2gÞ gð1  gÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

3 cos1 ð1  2gÞ  2ð1  2gÞ gð1  gÞ

k2 ¼ ; ð23Þ
cos1 ð1  2gÞ
4436 P.K. Swamee, A.K. Sharma / Applied Mathematical Modelling 37 (2013) 4430–4439

0:9
k3 ¼ 1:63 cos1 ð1  2gÞ : ð24Þ
The relative depth g depends on the sewer diameter. Table 1 provides g for various diameter ranges and the correspond-
ing values of slope coefficients k1, k2 and k3. The local design guidelines should be referred for relative depth g and corre-
sponding slope coefficients k1, k2 and k3 estimated to develop Table 1. Eliminating Soi between Eqs. (19) and (21)
  2
k1 Li Q 2i ei
di  di1 þ zi1  zi ¼ ln ; ð25Þ
gD5i k2 Di

where, the term containing m being small has been dropped.

5. Optimisation

Dropping the initial manhole cost as constant term, the cost function F of the sewer line having iL links is
XiL
F¼ i¼1
ðkni di þ ci Li Þ; ð26Þ

where d1, d2, d3,. . ., diL are decision variables. Splitting the sewer length Li into two parts x1i and x2i having diameters Di1 and
Di2 for LP application, Eq. (26) is rewritten as

X
iL
F¼ ðkni di þ ci1 xi1 þ ci2 xi2 Þ; ð27Þ
i¼1

where ci1 and ci2 = unit costs of sewer pipes of diameter Di1 and Di2 respectively. The decision variables xi1 and xi2 should
satisfy the following constrains

xi1 þ xi2 ¼ Li ; i ¼ 1; 2; 3 . . . iL : ð28Þ

Further, applying Eq. (25) to the segments of lengths xi1 and xi2, the following constraint is obtained:
  2   2
k1 xi1 Q 2i ei k1 xi2 Q 2i ei
di1 þ di  ln  ln ¼ zi1 þ zi ; i ¼ 1; 2; 3 . . . iL : ð29Þ
gD5i1 k2 Di1 gD5i2 k2 Di2

Normally the application of LP technique requires the linearization of objective function and constraints. Here it has been
applied in such a way that the originality of the objective function and constraints are maintained by selecting the pipe
lengths x1i or x2i of commercial pipe diameters Di1 and Di2 in the optimisation process (Eq. (28)). The formulation of con-
straint equations (Eq. (29)) for each pipe link at its downstream node J2(i) will ensure that the entire sewer line containing
iL links is designed as single entity.
Eqs. (27)–(29) constitute a LP problem which involves 3iL decision variables, consisting of 2iL equality constraints. By
including the lower and upper ranges of commercially available pipe sizes, such as Di1 and Di2, the problem is solved by using
simplex algorithm. Thus, the LP solution provides the minimum system cost according to the corresponding pipe diameters.
In order to begin the LP algorithm, the uniform sewer material is selected for all the links; and for this material the rough-
ness height is obtained. This roughness height is used for all the sewer links. For the assumed pipe diameters, Di1 and Di2,
relative flow depths gi1 and gi2 are obtained by using Table 1. Once gi1 and gi2 are known, the coefficients k1 and k2 are then
computed using Eqs. (22) and (23). Furthermore, for known diameters Di1 and Di2 the sewer cost coefficients ci1 and ci2 are
obtained. Similarly, as the sewer diameters are known, the bottom sewer trench widths wi and wi+1 are obtained by Eqs. (13)
and (14). The nodal cost coefficients kni are obtained by using Eqs. (15)–(17). For this purpose a tentative sewer invert profile
is assumed for obtaining the depth ratios mi and mi+1. To initiate the LP algorithm, the pipe invert levels can be estimated
based on the pipe gradients recommended in local manuals/design codes. Thus, the invert levels of all the pipes of the sew-
erline can be estimated. Using continuity conditions the sewer discharges Qi are computed, which is based on the upstream
pipe flow and the population load on the pipe i. Thus, the LP Eqs. (27)–(29) can be written. The LP solution indicates
preference for one diameter (lower or higher) in each pipe link. By knowing such preferences, the sewer pipe diameter
not preferred by LP is rejected and another diameter replacing it is introduced as Di1 or Di2. Once the new diameters and
the nodal depths are known, Eqs. (27) and (29) are then revised. After completing the replacement process for i = 1, 2,

Table 1
Relative Depth and Slope Coefficients.

Sewer diameter, D (m) Relative depth, g k1 k2 k3

0.15–0.25 0.50 4.323 3.00 2.447
0.30–0.50 0.60 2.471 3.332 2.728
0.55–1.20 0.70 1.632 3.555 3.018
Above 1.20 0.75 1.384 3.620 3.171
 P.K. Swamee, A.K. Sharma / Applied Mathematical Modelling 37 (2013) 4430–4439 4437

Fig. 5. Plan of the sewer line.

Table 2
Sewer line data.

Pipe link/node i/j 1st node of i J1(i) 2nd node of i J2(i) Length of link i L(i) (m) Elevation of node j z(j) (m) Discharge at node j q(j) (m3/s)
1 1 2 300 120.00 0.01
2 2 3 350 120.50 0.02
3 3 4 300 121.00 0.02
4 4 5 350 120.00 0.03
5 5 6 400 119.50 0.05
6 120.00 0

Table 3
Cost of various commercial pipe sizes including the cost of laying.

S. No. Pipe diameter (mm) Cost per meter (IRs) S. No. Pipe diameter (mm) Cost per meter (IRs)
1 150 345 10 700 2580
2 200 452 11 800 3485
3 250 542 12 900 3981
4 300 710 13 1000 4450
5 350 1181 14 1100 5103
6 400 1289 15 1200 5774
7 450 1450 16 1400 6992
8 500 1640 17 1600 8906
9 600 2126 18 1800 11,272

3,. . .iL, another LP solution is carried out to obtain the new preferred diameters and nodal depths. The process of LP and pipe
size replacement is continued until Di1 and Di2 are two consecutive commercial sewer pipe sizes. One more LP cycle now
obtains the diameters and depths to be adopted. The algorithm is further explained with the following example.

Example

It is proposed to design a sewerline containing five sewer links and six nodes as depicted in Fig. 5. The nodal elevations
and nodal discharges are listed in Table 2.
The commercial pipe sizes and costs are listed in Table 3. Soil excavation cost ce = 100 IRs/m3 and additional cost of exca-
vation cr = 30 IRs/m3/m depth were also considered. The cost of timbering the trenches has been considered as 300 IRs/m.
The algorithm terminated in LP 17 iterations. The variation of system cost with LP iterations is shown in Fig. 6. Thus the sys-
tem cost is minimum at the terminal iteration, which is the optimal solution. The final pipe sizes and invert depths are de-
picted in Fig. 7.
4438 P.K. Swamee, A.K. Sharma / Applied Mathematical Modelling 37 (2013) 4430–4439

Fig. 6. Variation of sewer line cost with LP cycles.

Fig. 7. Estimated pipe sizes and invert depths after 17 LP cycles.

6. Conclusions

An algorithm for the optimal design of a sewerline using the Linear Programming approach has been developed. In this
algorithm, the Darcy–Weisbach resistance equation is used in the formulation as the preferred resistance equation. The com-
mercial sewer pipe sizes are directly used in the design of sewer system. This eliminates the problem of rounding off the
estimated pipe sizes to the nearest commercial sizes as required in some optimisation techniques, which forfeits the purpose
of system optimisation to large extent. At this stage the methodology has been developed for a sewer line having any number
of links, which will be extended to a typical sewer network in future.

References

[1] J.S. Dajani, R.S. Gemmel, E.K. Morlok, Optimum design of urban wastewater collection networks, J. San. Eng. Div., ASCE 98 (6) (1972) 853–867.
[2] T.R. Camp, Design of sewers to facilitate flow, Sewage Works J. 18 (1) (1946) 3–16.
[3] J.C. Liebman, A heuristic aid for the design of sewer networks, J. San. Eng. Div., ASCE 93 (SA4) (1967) 81–90.
[4] L.A. Cook, B. Lockwood, The investigation of sewer network by computer, J. Inst. of Civ. Eng., Part 2 (1977) 481–494.
[5] L.W. Mays, Sewer network scheme for digital computations, J. Envir. Eng., ASCE 104 (3) (1978) 535–539.
[6] J.M. Lorine, Sanitary sewer design, Civ. Eng., ASCE 52 (6) (1982) 66–68.
[7] D.P. Dasher, P.K. Davis, Designing sanitary sewers with microcomputers, J. Envir. Eng., ASCE 112 (6) (1986) 993–1007.
[8] S.W. Miles, J.P. Heaney, Better than ‘optimal’ method for designing drainage systems, J. Wat. Resour. Plng. Mangt., ASCE 114 (5) (1988) 477–499.
[9] C. Charalombus, A.A. Elimam, Heuristic design of sewer networks, J. Envir. Eng., ASCE 116 (6) (1990) 1181–1199.
[10] J.M. Gupta, S.K. Agarwal, P. Khanna, Optimal design of wastewater collection system, J. Envir. Eng., ASCE 102 (5) (1976) 1029–1041.
[11] Y. Argaman, U. Shamir, E. Spivak, Design of optimal sewerage system, J. Envir. Eng., ASCE 99 (5) (1973) 705–716.
[12] L.B. Merritt, R.H. Bogan, Computer-based optimal design of sewer systems, J. Envir. Eng. Div., ASCE 99 (1) (1973) 35–53.
[13] J.B. Walsh, L.C. Brown, Least cost method for sewer design, J. Envir. Eng. Div., ASCE 101 (4) (1975) 537–553.
[14] S.L. Gupta, P. Mehndiratta, P. Khanna, Gravity wastewater collection system optimisation, J. Envir. Eng., ASCE 109 (5) (1983) 1195–1209.
[15] J.M. Fisher, G.M. Karadi, W.W. McVinnie, Design of sewer systems, Water Resour Bull, AWRA 7 (2) (1971) 294–302.
[16] J.S. Dajani, Y. Hasit, Capital cost minimization of drainage networks, J. Envir. Eng. Div., ASCE 100 (2) (1974) 325–337.
[17] A.A. Elimam, C. Charalombus, F.H. Ghobrial, Optimal design of large sewer networks, J. Envir. Eng., ASCE 115 (6) (1989) 1171–1189.
[18] A.A. Elimam, C. Charalambous, F.H. Ghobrial, Optimum design of large sewer networks, J. Environ. Eng., ASCE 115 (6) (1989) 1171–1189.
[19] P.K. Swamee, Design of sewer line, J. Envir. Eng., ASCE 127 (9) (2001) 776–781.
[20] L.Y. Liang, R.G. Thompson, D.M. Young, Optimising the design of sewer networks using genetic algorithms and tabu search, Eng., Construct. Architect.
Mangt 11 (2) (2004) 101–112.
[21] H.T. Weng, S. Liaw, W Huang, Establishing an optimization model for sewer system layout with applied genetic algorithm, Environ. Inform. Archives 2
(2004) 781–790.
[22] Y. Guo, G.A. Walters, S.-T. Khu, E. Keedwell, Optimal design of sewer networks using hybrid cellular automata and genetic algorithm, in: Proceedings of
IWA World Water Congress 2006, Beijing, China.
 P.K. Swamee, A.K. Sharma / Applied Mathematical Modelling 37 (2013) 4430–4439 4439

[23] Y. Guo, G.A. Walters, S.-T. Khu, E. Keedwell, Efficient multiobjective storm sewer design using cellular automata and genetic algorithm hybrid, J. Water
Resour. Plng. Mangt. 134 (6) (2008) 511–515.
[24] M.H. Afshar, Rebirthing genetic algorithm for storm sewer network design, Scientia Iranica 19 (1) (2012) 11–19.
[25] A. Haghighi, A.E. Bakhshipour, Optimization of sewer networks using an adaptive genetic algorithm, Water Resour. Manag. (2012), http://dx.doi.org/
10.1007/s11269-012-0084-3.
[26] M. Cisty, Hybrid genetic algorithm and linear programming method for least-cost design of water distribution systems, Water Resour. Manag. 24 (1)
(2010) 1–24.
[27] H. Haghighi, M.V. Samani, Z. Samani, GA-ILP method for optimization of water distribution networks, Water Resour. Manage 25 (7) (2011) 1791–1808.
[28] T. Pan, J. Kao, GA-QP model to optimize sewer system design, J. Environ. Eng. 135 (1) (2009) 17–24.
[29] M.H. Afshar, A parameter free continuous ant colony optimization algorithm for the optimal design of storm sewer networks: constrained and
unconstrained approach, Adv. Eng. Softw. 41 (2) (2010) 188–195.
[30] R. Moeini, M.H. Afshar, Layout and size optimization of sanitary sewer network using intelligent ants, Adv. Eng. Softw. 51 (2012) 49–62.
[31] Y. Guo, G.A. Walters, S.T. Khu, E. Keedwell, A novel cellular automata based approach to storm sewer design, Eng. Optim. 39 (3) (2007) 345–364.
[32] M.H. Afshar, M. Rohani, Optimal design of sewer networks using cellular automata-based hybrid methods: discrete and continuous approaches, Eng.
Optim. 44 (1) (2012) 1–22.
[33] S.F. Yeh, C.W. Chu, et al, Applying tabu search and simulated annealing to the optimal design of sewer networks, Eng. Optim. 43 (2) (2011) 159–174.
[34] ASCE Task Force, Friction factors in open channels: Progress report of task force on friction factors in open channels, J. Hydraul. Eng. Div., ASCE 89 (2)
(1963) 97–143.
[35] C.P. Liou, Limitation and proper use of the Hazen–Williams equation, J. Hydraul. Eng. ASCE 124 (9) (1998) 951–954.
[36] G.O. Brown, The history of the Darcy–Weisbach equation for pipe flow resistance, in: J.R. Rogers, A.J. Fredrich (Eds.), Environmental and Water
Resources History, Proceedings and Invited Papers for the ASCE 150th Anniversary presented at the ASCE Civil Engineering Conference and Exposition,
Washington, 2002, pp. 34–43.
[37] P.K. Swamee, Critical slope equations for open channels, ISH J. Hydraul. Eng. 8 (2) (2002) 44–49.
[38] P.K. Swamee, N. Swamee, Design of noncircular sewers sections, J. Hydraul. Res. IAHR 46 (2) (2008) 277–281.
[39] CPHEEO, Manual on Sewerage and Sewage Treatment, second ed., Central Public Health and Environmental Engineering Organisation, Ministry of
Urban Development, New Delhi, India, 1993.
[40] P.K. Swamee, G.C. Mishra, B.R. Chahar, Minimum cost design of lined canal sections, J. Wat. Resour. Manag. 14 (1) (2000) 1–12.