Analysis and Optimisation of Fluid-Therm

Analysis and Optimisation of
Fluid-Thermal Systems using Excel
Mohamed M. El-Awad
Mohamed M. El-Awad 2
CONTENTS
Nomenclature 0
1. Introduction 1
1.1. A review of thermofluid subjects
1.1.1. Thermodynamics
1.1.2. Fluid dynamics
1.1.3. Heat transfer
1.2. Economic considerations
1.3. The Excel-based modelling platform used in this book
1.4. Closure
References
2. Excel formulae, built-in functions and iterative tools 25

2.1. Elements of Excel‟s user-interface
2.2. Excel formula and the use of cell labels in formulae
2.3. Excel built-in functions
2.3.1. Logical functions
2.3.2. Functions for matrix operations
2.3.3. Solution of linear systems of equations
2.5. Iterative solutions with Goal-Seek
2.6. Iterative solutions with circular calculations
2.7. Excel‟s graphical tools for data presentation and analyses
2.8. Closure
References
3. Solver 49
3.1. Activation of Solver
3.2. The GRG Nonlinear method
3.2.1. Solution of linear systems of equations
3.2.2. Optimisation analyses
3.4. The Simplex LP method
3.5. The Evolutionary method
3.6. The default settings of Solver options
3.7. Optimisation with the GRG Nonlinear and Evolutionary methods
3.8. Closure
References
4. VBA, user-defined functions, and the Thermax1 add-in 67

4.1. VBA and the development of user-defined functions
4.1.1. Development of fluid property functions
4.1.2. Development of general-purpose functions
4.2. Thermax1 property functions
3 Analaysis and Optimisation of Fluid-Thermal Systems using Excel
4.2.1. Property functions for saturated water and steam (Wat)

4.2.2. Property functions for ideal gases (Gas)
4.3. Activation and use of Thermax1
4.3.1. Accessing Thermax1 functions via the Function Wizard
4.3.2. Direct use of Thermax1 functions in Excel formulae
4.4. Using the general-purpose numerical tools of Thermax1
4.4.1. The interpolation functions
4.4.2. The Newton-Raphson solver
4.5. Closure
References
5. Analyses of multi-pipe and pump-pipe systems 93

5.1. Analyses of multi-pipe systems
5.1.1. Three pipes in series
5.1.2. Three pipes in parallel
5.1.3. Pipe-junctions
5.2. Determining the operating points for a pump-pipe system
5.2.1. A centrifugal pump connected to a single pipe
5.2.2. A centrifugal pump connected to two branching pipes
5.2.3. A centrifugal pump connected to a three-pipe branched network
5.3. Determining the operating point for two centrifugal pumps
5.3.1. The two pumps connected in parallel
5.3.2. The two pumps connected in series
5.4. Closure
References
6. Analyses of pipe networks 123

6.1. Mathematical analysis of looped pipe networks
6.2. The Hardy-Cross method
6.3. Analyses of looped pipe networks by Excel-Solver
6.4. Analyses of branched and mixed pipe networks by Excel-Solver
6.5. Closure
References
7. Determining the economic pipeline diameter 155

7.1. Development of the objective function for pipeline optimisation
7.2. The analytical method for pipeline optimisation
7.2.1. The cost-diameter formula
7.2.2. The pipe-friction formula
7.3. Analytical optimisation with the Hazen-Williams formula
7.4. Analytical optimisation with the Darcy-Weisbach formula
7.4.1. Pipeline optimisation for a laminar flow
7.4.2. Pipeline optimisation for a turbulent flow
7.5. An Excel-assisted analytical procedure for pipeline optimisation
7.6. Least-cost pipeline optimisation using Excel-Solver

7.7. Closure
References
8. Economic optimisation of pipe networks 181

8.1. A review of pipe network optimisation methods
8.2. Pipe-network optimisation using Excel
8.2.1 The analytical model
8.2.2 Optimisation analyses with alternative friction formulae
8.2.3 Comparison with optimised solutions of GA models
8.2.4 Effect of Solver options on the optimised solutions
8.3. Optimisation of a gravity-driven network with eight loops
8.3.1. Hydraulic analysis
8.4. Optimisation of a pump-driven two-loop pipe network
8.4.1. The analytical model
8.5. Closure
References
9. Thermodynamic and economic optimisation of the air-botomming cycle 221

9.1. The air-bottoming gas-turbine plant
9.1.1. The thermodynamic model
9.1.2. Modelling the gas-to-air heat-exchanger
9.2. Thermodynamic optimisation of the AB cycle
9.3. Economic optimisation of the AB cycle
9.4. Computational performance of the two models for the heat-exchanger
9.5. Closure
References
10. Optimisation of the gas-turbine cycle with cogeneration and inlet-air cooling 243
10.1. The cogeneration system
10.1.2. Otimisation analysis
10.2. The gas-turbine with inlet-air cooling
10.2.2. Optimisation analysis
10.2.3. Comparison with cogeneration and air-bottoming systems
10.3. Closure
References
Appendices 260
A.1. Properties of liquid water
A.2. Properties of air at atmospheric pressure
B.1. A list of the ideal gases supported by Thermax1
B.2. The linear and quadratic interpolation functions
B.3. The Newton-Raphson Solver
Subject index 267

Chapter 1
The term „fluid-thermal systems‟ refers to a wide range of residential or industrial
devices that involve transporting or utilising fluids. Fluid-thermal systems are found in
most energy systems such as refrigerators, air-conditioners, gas-turbine and steam-
turbine power plants, and internal-combustion engines. The components of fluid-
thermal systems include pumps, compressors, turbines, oilers, heaters, and heat
exchangers. Therefore, the designs of fluid-thermal systems are strongly based on
applying the principles of thermodynamics, fluid mechanics, and heat transfer.
Although fluid-thermal systems differ in their level of complexity and sophistication
from that of a thermo-flask or a hair-dryer to that of a new automobile engine or a
nuclear power plant, three considerations are common to the design of all these
systems, which are the efficiency of energy utilisation, the impact of any emissions on
the environment, and the economic feasibility. This chapter reviews the basic principles
of thermofluid and economic analyses and demonstrates their applications in
optimisation analyses of typical fluid-thermal systems.
Closure
This chapter gave a bird‟s eye view of the main concepts in the three thermofluid
subjects and illustrated the use of engineering economics principles in design analyses
of fluid-thermal systems. The chapter also described in general terms the Excel-based
modelling platform used for these analyses in the book. The following three chapters
describe the four components of the Excel-based platform in more details. Chapter 2
reviews the mostly needed Excel‟s functions and shows how they can be used in
Excel‟s formulae with the help of simple examples. Chapter 2 also illustrates the use of
Excel‟s iterative tools; Goal Seek and circular calculations. Chapter 3 introduces the
Solver add-in and shows how its three solution methods can be used for solving
different types of computer-based problems. Chapter 4 initially shows how VBA can be
used for developing user-defined functions. It then describes the functions provided by
Thermax1 and shows how Excel‟s user-interface can be used for selecting the
appropriate add-in function. Chapter 4 also illustrates the use of the interpolation
functions and the Newton-Raphson iterative solver provided by Thermax1.
Chapters 5 to 8 deal with the analyses and economic optimisation of fluid systems.
Chapter 5 uses Goal Seek and Solver to deal with the analyses of multi-pipe and pump-
pipe systems. Different arrangement are analysed so as to determine the system‟s
friction losses, power requirement, and operating point. Chapter 6, which deals with the
analyses of pipe networks, initially presents the conventional Hardy-Cross method and
illustrates its use by analysing simple cases of looped pipe-networks. The chapter then
shows how Solver provides a simpler method for pipe-network analyses that can be
used for the analyses of branched as well as looped pipe-networks. Chapter 7 focuses
on determining the economic diameter for a pipeline. It initially introduces the
traditional optimisation method before showing how Solver can be used for developing
Excel-based models for pipes with both laminar and turbulent flows. Chapter 8
combines the methods of Chapters 6 and 7 to demonstrate the adequacy of the Excel-
based platform for optimisation analyses of gravity-driven and pump-driven pipe-

networks. The chapter also studies the effects of the friction equation and Solver
options on the optimisation results.
Chapters 9 and 10 show how the Excel-based platform can be used for thermodynamic
and economic optimisation of thermal systems. Chapter 9 deals with the analyses the
air-bottoming cycle and studies the effect of modelling the gas-to-air heat-exchanger,
which is an important element of the three systems, on the optimised solutions. In this
respect, two models are tested which are the effectiveness method and the pinch-point
method. Thermax1 property functions for ideal gases enable the exact variable specific
heat method to be used in the thermodynamic models. Chapter 10 considers two options
for utilising the waste energy in the gas-turbine exhaust; which are a cogeneration
system and a vapour-absorption system for cooling the gas-turbine inlet-air. Thermax1
property functions for saturated water enable the Excel-based models to account for the
water-vaporisation process that takes place in the heat-recovery steam generator.
References
[1] Y.A. Cengel, and M.A. Boles, Thermodynamics an Engineering Approach,
McGraw-Hill, 7th Edition, 2007
[2] C. T. Crowe, D. F. Elger, B. C. Wiliams, and J. A. Roberson, Engineering Fluid
Mechanics, 9th edition, John Wiley & Sons, Inc., 2009.
[3] Y.A. Cengel and A.J. Ghajar, Heat and Mass Transfer: Fundamentals and
Applications. 4th edition, McGraw Hill, 2011.
[4] J. P. Holman, Heat Transfer, 10th edition, McGraw-Hill. 2010.
[5] W.G., Sullivan, E.M., Wicks, and C.P., Koelling, Engineering Economy 14th
Edition, Prentice- Hall International Series, 2009.
[6] T. D. Eastop and D. R. Croft, Energy Efficiency for Engineers and Technologists,
Longman Scientific & Technical, Copublished with Johan Wiley and Sons, 1990.
[7] A. Rivas, T. Gómez-Acebo, and J. C. Ramos. The application of spreadsheets to
the analysis and optimization of systems and processes in the teaching of
hydraulic and thermal engineering, Computer Applications in Engineering
Education, Vol. 14, Issue 4, 2006, pp. 256-268.
[8] D. Brkic, Spreadsheet-Based Pipe Networks Analysis for Teaching and Learning
Purpose, Spreadsheets in Education (eJSiE): Vol. 9: Iss. 2, Article 4, 2016.
Available at: http://epublications.bond.edu.au/ejsie/vol9/iss2/4
[9] A. Karimi, Using Excel for the thermodynamic analyses of air-standard cycles
and combustion processes, ASME 2009 Lake Buena Vista, Florida, USA.
[10] Z. Ahmadi-Brooghani, Using Spreadsheets as a Computational Tool in Teaching
Mechanical Engineering, Proceedings of the 10th WSEAS International
Conference on computers, Vouliagmeni, Athens, Greece, July 1315, 2006, 305-
310
[11] S.A. Oke, Spreadsheet Applications in Engineering Education: A Review, Int. J.
Engng Ed. Vol. 20, 2004, No. 6, 893-901
[12] L. Caretto, D. McDaniel, T. Mincer. Spreadsheet calculations of thermodynamic

properties, Proceedings of the 2005 American Society for Engineering Education
Annual Conference & Exposition, American Society for Engineering Education ,
Copyright © 2005.
[13] The University of Alabama, Mechanical Engineering, Excel for Mechanical
Engineering project, Internet: http://www.me.ua.edu/excelinme/index.htm (Last
accessed November 23, 2015).
[14] J. Huguet, K. Woodbury, R. Taylor, Development of Excel add-in modules for
use in thermodynamics curriculum: steam and ideal gas properties, American
Society for Engineering Education, 2008, AC 2008-1751.
[15] K. Mahan, J. Huguet, K. Woodbury, R. Taylor, Excel in ME: Extending and
refining ubiquitous software tools, American Society for Engineering Education,
2009, AC 2009-2295.
[16] The University of Virginia, Free Excel/VBA Spreadsheets for Thermodynamics:
Rankine, Brayton, Otto and Diesel Cycles, Internet:
http://www.faculty.virginia.edu/ribando/ modules/xls/Thermodynamics/ (Last
[17] D. G. Goodwin, "TPX: thermodynamic properties for Excel",
http://www.tecnun.es/ asignaturas/termo/SOFTWARE/TPX/index.html (Last
[18] E.W. Lemmon, M.L. Huber, M.O. McLinden, NIST Reference Fluid
Thermodynamic and Transport Properties— REFPROP Version 8.0, User‟s
Guide, National Institute of Standards and Technology, Physical and Chemical
Properties Division, Boulder, Colorado 80305, 2007.
[19] T. K. Jack. Computerised calculations of thermo-physical steam and air
properties, Int. J. Pure Appl. Sci. Technol., 9(2) (2012), pp. 84-93 (Available
online at www.ijopaasat.in)
[20] C.O.C. Oko and E.O.Diemuodeke. MS Excel spreadsheet add-in for
thermodynamic properties and process simulation of R152a, Energy Science and
Technology, Vol. 5, No. 2, 2013, pp. 63-69.
Chapter 2
As a general-purpose application, Excel is equipped with numerous features and
functions that can be utilised by users with various backgrounds for the presentation
and analysis of their data. This chapter focuses on the features that are mostly needed
for its use in this book as a modelling platform for thermofluid analyses. These include
Excel‟s user-interface, formulae and built-in functions, and its graphical tools and
trendline feature. One of the important features highlighted in this chapter is the use of
cell-labelling instead of the commonly-used referencing by location. The chapter also
illustrates the use of Excel‟s matrix functions for the solution of linear systems of
equations and the use of its iterative tools, the Goal Seek command and circular
calculations, for the solution of nonlinear equations. Finally, the section on Excel‟s
graphical tools demonstrates the use of the trendline feature for data curve-fitting.
References
1. J. Walkenbach, Excel 2010 Formulas, Wiley Publishing Inc., 2010.
2. J. Walkenbach, Excel 2007 Charts, Wiley Publishing Inc., 2007.
3. Y. A. Cengel and M. A. Boles. Thermodynamics an Engineering Approach,
McGraw-Hill, 7th Edition, 2007.
Chapter 3
Solver is an Excel add-in developed by Frontline Systems [1] that enables iterative
solutions with multiple adjustable cells. It is a more useful tool for “What-if” analyses
than the Goal Seek command described in the previous chapter because it gives the user
more control over the solution process and offers three solution methods that use a
deterministic gradient method, a linear-programming method, and a stochastic
evolutionary method. The three solution options suit different types of analyses. This
chapter shows how Solver can be activated and its three solution methods used for
solving single nonlinear equations and systems of linear equations and optimisation
analyses. The Chapter also describes the settings of Solver‟s solution options and
illustrates the use of its GRG Nonlinear method and the Evolutionary methods for
thermofluid optimisation analyses.
References
1. Frontline Systems, internet: http://www.Solver.com/ (Last accessed November 23,
2015).
2. L.S. Lasdon, R.L. Fox, M.W. Ratner, Nonlinear optimization using the generalized
reduced gradient method, Revue Française d'Automatique, Informatique et
Recherche Opérationnelle, tome 8, V3 (1974), p. 73-103. Available at:
http://www.numdam.org/article/RO_1974__8_3_ 73_0.pdf
3. https://en.wikipedia.org/wiki/Evolutionary_algorithm
4. Wikkipedia, https://en.wikipedia.org/wiki/Gradient_method
5. Wikkipedia, https://en.wikipedia.org/wiki/Evolutionary_algorithm
6. W.S. Janna, Design of Fluid Thermal Systems, 3rd Edition, CENGAGE Learning,
2011.
Cahpter 4
Even with its numerous functions and the Solver add-in, Excel‟s capacity as a
modelling platform for thermofluid analyses is limited for two reasons. The first reason
is that thermofluid analyses require determination of the fluids‟ physical properties, but
Excel does not provide built-in functions for this purpose. The second reason is that
Excel‟s formulae are confined to individual cells and that makes them too restrictive for
model development in certain types of thermofluid analyses such as optimisation
analyses or iterative solutions that involve nonlinear equations. This chapter shows how
custom functions for thermofluid analyses can be developed with VBA and presents the
Thermax1 add-in that provides functions for the thermodynamic properties of saturated
water and saturated steam together with 29 ideal gases. The chapter describes the
procedures for activating Thermax1 and using its functions for thermofluid analyses
with Excel. In addition to the property functions, Thermax1 provides interpolation
functions and a Newton-Raphson solver for nonlinear equations. The chapter also
demonstrates the use of these general-purpose tools in thermofluid analyses.
References
[1] Y. A. Cengel and M. A. Boles. Thermodynamics an Engineering Approach,
McGraw-Hill, 7th Edition, 2007.
[2] J. Walkenbach, Excel 2013 Power Programming with VBA, Wiley Publishing, Inc.
2013.
[3] J.L. Latham, Programming in Microsoft Excel VBA, An Introduction, 2008.
Internet: http://ies.fsv.cuni.cz/default/file/download/id/21101 (Last accessed
November 29, 2015).
[4] http://www.fontstuff.com/vba/vbatut01.htm (Last accessed November 26, 2017).
[5] The University of Alabama, Mechanical Engineering, Excel for Mechanical
Engineering project, Internet: http://www.me.ua.edu/excelinme/index.htm (Last
accessed July 19, 2018).
[6] Huguet, K. Woodbury, R. Taylor, Development of Excel add-in modules for use in
thermodynamics curriculum: steam and ideal gas properties, American Society for
Engineering Education, 2008, AC 2008-1751.
[7] M. M. El-Awad, A multi-substance add-in for the analyses of thermo-fluid systems
using Microsoft Excel, International Journal of Engineering and Applied Sciences
(IJEAS), ISSN: 2394-3661, Volume-2, Issue-3, March 2015.
[8] M. Affandi, N. Mamata, S. M. Kanafiaha, N. Khalida, Simplified Equations for
Saturated Steam Properties for Simulation Purpose, Procedia Engineering, 53
(2013 ), 722 – 726
[9] C. R. Ferguson, Internal Combustion Engines Applied Thermosciences, John
Wiley, & Sons. Inc. 1986.
[10] Y. A. Cengel and A. J. Ghajar. Heat and Mass Transfer: Fundamentals and
Applications. McGraw Hill, 2011.
Analyses of multi-pipe and

pump-pipe systems
Practical pipe systems usually consist of multiple pipes connected in various ways. The
analyses of such systems are based on the same principles of mass and energy
conservation used for analysing the flow in a single pipe, but an iterative solution is
always needed. An iterative solution is also needed for determining the operating point
for a pump-pipe system that involves a single or multiple pumps. This chapter shows
how Excel‟s iterative tools, Goal Seek and Solver, can be used for analysing pipe
systems that involve parallel pipes, pipes in series, and pipe junctions and for
determining the operating point for a single pump with various pipe arrangements and
for multiple pumps arranged in parallel or in series.
5.1. Analyses of multi-pipe systems

Figure 5.1 shows three multi-pipe systems in which the pipes are connected in series, in
parallel, and at a joint. Analyses of multi-pipe systems are based on the same principles
of mass and energy conservation described in Chapter 1 with respect to a single pipe.
However, analyses of multiple pipe systems usually require iterative solutions. This
section shows how Excel and its two iterative tools can be used to determine the
magnitudes and directions of the flows in such mutiple pipe systems.
1 B 2 C 3
A D
(a)
1
2
A B
(b)
ZB
B
ZC
ZA
2
A 1 C
3
D
(c)
Figure 5.1. Three typical pipe arrangements in multiple pipe systems
The continuity equation

For steady flow of an incompressible fluid in an open system with multiple inlets and
outlets, the conservation of mass principle leads to the following continuity equation:
Q  Qin out (5.1)
Where Qin and Qout refer to the volume flow rate of the fluid entering and leaving the
system, respectively. Equation (5.1) applies to the whole pipe system as well as to any
part of the system that has defined inlets and outlets. Depending on the particular
configuration under consideration, the continuity equation can be expressed in more
specific forms as shown in the following sections.
The energy equation

Applied to the flow between any two points in a pipe system, the energy equation for
steady-flow of an incompressible fluid takes the following form [1]:
 h f  hc 
p1 V12 p2 V22
 z1   hP   z2  (5.2)
 2g  2g
Where p, γ, and V are the pressure, specific weight, and velocity of the fluid,
respectively, z the elevation, h p the head supplied by a booster pump (if any), h f the
major friction head loss in the pipe, and h c the minor friction head loss, which is the
summation of component losses (if any) in the pipe. The major friction head loss h f can
be determined by using the Darcy-Weisbach equation:
L V2
hf  f (1.21)
D 2g
Where, f is the friction factor, L and D are the length and diameter of the pipe,
respectively, and V is the fluid velocity through the pipe. The friction factor f depends
on whether the flow is laminar or turbulent. For laminar pipe flows, f is given by:
64
f  Re ≤ 2300 (1.22)
Re
Where Re is the Reynolds number defined as:
VD VD 4 Q
Re  = = (5.3)
  D
Where ρ, μ and ν are the density, dynamic viscosity, and kinematic viscosity of the
fluid, respectively, and Q is the volume flow rate. For turbulent pipe flows, the friction
factor can be obtained from the Moody diagram or calculated from the explicit
Swamee-Jain equation:
2
   5.74  
f  0.250 / log   0 .9   Re ≥ 2400 (1.25)
  3.7 D Re  
Where ε is the roughness of the pipe material. Table 5.1 gives values of ε for various
materials.
Table 5.1. Roughness of various pipe materials (adopted from Cengel and Cimbala [1]
Material Roughness ε (ft) Roughness ε (mm)
Glass, plastic 0 (smooth) 0
Concrete 0.003 – 0.03 0.9 – 9
Wood stave 0.0016 0.5
Rubber, smoothed 0.000033 0.01
Copper or brass tubing 0.000005 0.0015
Cast iron 0.00085 0.26
Galvanised iron 0.0005 0.15
Wrought iron 0.00015 0.046
Stainless steel 0.000007 0.002
Commercial steel 0.00015 0.045
The frictional head loss through the components h c is usually expressed as:
V2
hc   K L (1.28)
2g
Where KL is the component factor the value of which varies from one component to
another. Values of KL for the many components usually involved in pump-pipe systems
can be obtained from popular textbooks such as Cengel and Cimbala [1] and White [2].
5.1.1. Three pipes in series

Figure 5.1.a shows a pipe system with three pipes in series. In general, the pipes may
have different lengths, diameters, and roughness. The system may also have different
elevations at its entrance and exit points. If the flow rate is known, then the problem can
be solved directly by using the Darcy-Weisbach equation like a type-1 flow problem for
a single pipe. However, if the pipe diameters of flow rates are to be determined, then an
iterative solution is required like type-2 and types-3 problems for a single pipe. The
following analysis focuses on a type-2 flow problem involving thee pipes in series.
The analytical model

The analytical model is based on the assumptions that: (1) the flow is steady and
incompressible, (2) entrance effects are negligible and the flow is fully developed, (3)
and minor losses are negligible. Since the pipes are in series, the flow rate in the three
pipes is the same, i.e:
Q A  QB  Q1  Q2  Q3 (5.4)
Therefore, given the fluid velocity in pipe 1, the velocities in pipes 2 and 3 are given
by:
V1 A2
V2  (5.5.a)
A1
V1 A3
V3  (5.5.b)
A1
The head loss in each pipe is given by the Darcy-Weisbach equation:
L1 V12
h f 1  f1 (5.6.a)
D1 2 g
L2 V22
hf 2  f2 (5.6.b)
D2 2 g
L3 V32
h f 3  f3 (5.6.c)
D3 2 g
The total friction loss is the summation of the friction losses in the three pipes:
hf  hf1  hf 2  hf 3 (5.7)
The iterative solution proceeds as follows:
1. Assume the velocity in the first pipe (V1 ) and calculate the corresponding
velocities V2 and V3 from Equation (5.5.a) and Equation (5.5.b), respectively.
2. Based on the three velocities, calculate the friction head losses in the three
pipes
3. Compare the resulting total friction loss with the specified value. If they are
different go back to step 1 and repeat the procedure until the difference is ≈ 0.
The mass conservation principle makes V1 the only independent variable since V2 and
V3 can be calculated from Equations (5.5.a) and (5.5.b). In this case, Excel‟s Goal Seek
command can be used to perform the iteration. The following example, which is based
on Example 5.17 in White [2], illustrates the procedure.
Example 5.1. Three pipes in series

A pipe system that consists of three straight pipes in series transfers water (ρ = 1000
kg/m3 , ν = 1.02x10-6 m2 /s). The difference in elevation between the inlet and outlet
points of the system is 5 m and the pipes diameters and roughness are as follows:
Length (m) Diameter (m)  (m)

Pipe 1 100 0.08 0.00024
Pipe 2 150 0.06 0.00012
Pipe 3 80 0.04 0.0002
Determine the flow rate through the system that yierlds a pressure difference between
the inlet and outlet points of the system of 150 kPa.
Excel implementation
Figure 5.2 shows the Excel sheet developed for this example. The data part, positioned
on the left-side of the sheet, shows the given information about the system and its three
pipes. The density and kinematic viscosity of water are taken as ρ=1000 kg/m3 and ν
=1.02x10-6 m2 /s. The calculations part begins with an assumed value for the velocity in
the first pipe (V_1=1 m/s). Based on this guess, the velocities in the other two pipes
(V_2 and V_3) are calculated using Equation (5.5). The corresponding Reynolds
numbers Re_1, Re_2, and Re_3 are the calculated and, depending on whether the flow
is laminar or turbulent, the friction factors f 1 , f 2 and f 3 are determined. The pipes friction
losses are then calculated and the total friction loss (Δh_total) is determined.
Figure 5.2. Excel sheet with Goal Seek setup for Example 5.1
Figure 5.2 shows that at the guessed values of the velocities, the flows in the three pipes
are turbulent and the total head loss amounts to 62.8 m. Referring to Figure 5.1.a, the
required total head-loss across the system is given by:
PA  PD 150,000
hf  5  5  20.29052m
g 1000  9.81
The required velocity V_1 that makes the total head loss equal to the required value can
easily be determined by using Goal Seek. Figure 5.2 also shows the appropriate set-up
for Goal Seek. Figure 5.3 shows the solution obtained by Goal Seek, which is V_1 =
0.486 m/s. The corresponding flow rate, which is 10.17 m3 /s, is not significantly
different from the value of 10.22 m3 /s given by White [2].
Figure 5.3. Goal Seek solution for Example 5.1
5.1.2. Three pipes in parallel

Figure 5.1.b shows a pipe system that consists of three pipes connected in parallel. In
this case, the flow rates in the three pipes may not be equal, but the friction heal losses
are. As in the previous case, we assume that the head loss across the system is specified
and we want to compute the required total inlet flow rate QA .
Analytical model
For an incompressible steady flow, the conservation of mass leads to:
Q A  QB  Q1  Q2  Q3 (5.8)
The head losses in the three branches must be the same and equal to the specified head
loss h f :
hf1  hf 2  hf 3  hf (5.9)
The head loss in each pipe is given by the Darcy-Weisbach equation, but since the
friction factor depends on the Reynolds number and cannot be specified a priori, the
problem cannot be solved directly but has to be solved by an iterative process. In the
previous example, it was possible to perform the iteration with the Goal Seek command
by assuming the flow rate in the first pipe and then determining the flow rates in the
following two pipes from the mass conservation principle. However, in the present
situation Goal Seek cannot be used because all the three velocities are unknown. This
case demonstrates the usefulness of Solver that allows the three velocities to be
changing cells for dealing with iterative solutions with multiple variables.
Starting with assumed values for the three velocities (V1 , V2 and V3 ), values of the
friction losses in the three pipes, h f1 , h f2 and h f3 , can be calculated. For randomly chosen
initial guesses for the velocities, the total head loss will not be the same. Therefore, the
procedure has to be repeated with new values for V1 , V2 and V3 until Equation (5.9) is
approximately satisfied. The following example that illustrates the solution procedure is
based on Example 5.18 in White [2].
Example 5.2. Determining the flow rates through three parallel pipes
Assume that the same three pipes in Example 5.1 are now joined in parallel with the
same total head loss of 20.3 m. Compute the total flow rate Q, neglecting minor losses.
Figure 5.4 shows the Excel sheet developed for this example. The data part shows the
information provided in the question and the density and dynamic viscosity of water,
which are taken as ρ=1000 kg/m3 and ν = 1.02x10-6 m2 /s. The calculation part starts
with an assumed values for all three velocities in the pipes, which are V_1 = V_2 = V_3
= 1.0 m/s. Starting with these guessed velocities, the sheet proceeds to calculate the
corresponding Reynolds numbers, friction factors, and friction losses. The absolute
difference between the three friction losses and the specified head loss (h f1 -20.3), (h f2 -
20.3) and (h f3 -20.3) are then calculated and stored in the three cells named Error1,
Error2, and Error3 as shown in Figure 5.4. The formula bar reveals the formula used
for Error1. Since the initial values of the velocities are incorrect, values of these cells
are not zero, as they should be, but equal to 18.5, 16.9, and 16.9, respectively.
Similarly, the calculated value for the total flow rate (32.798 m3 /h) is not correct.
Figure 5.4. Excel sheet for Example 5.2
The correct velocities can be obtained by using Solver and seeking to reduce Error1,
Error2, and Error3 to small values by adjusting the values of the three velocities.
Figure 5.5 shows the set-up of Solver for this task. Note that the Set Target Cell has
been left blank and, therefore, the three options “Max”, “Min”, and “Value of” do not
matter. Also, note that the absolute values of the three errors are required to be less than
0.1. With this set-up, pressing the "Solve" button will prompt Solver to search for the
values of V_l, V_2 and V_3 that satisfy the three constraints. The solution obtained by
Solver is shown in Figure 5.6.
Figure 5.5. Solver set-up for Example 5.2
Figure 5.6. Excel sheet for Example 5.2 showing Solver solution
Figure 5.6 shows that the required values of the flow rates in the three pipes, are V1 =
3.439 m/s, V2 = 2.528 m/s and V3 = 2.513 m/s. The corresponding total flow rate is
99.340 m3 /h. White [2] used the Moody diagram to determine the friction factors. The
values he obtained for the three velocities were V1 = 3.46 m/s, V2 = 2.55 m/s and V3 =
2.52 m/s. His flow rate was 99.8 m3 /h. Taking into account the inaccuracy of
interpolating the Moody diagram, the present values can be considered to be in good
agreement with their corresponding values obtained by White [2].
Unlike pipes in series, parallel pipes require an iterative solution even when the total
flow rate is known instead of the total head loss. However, in this case the continuity
equation can be utilised so as to reduce the number of changing cells in Solver that
satisfy the energy equation by one.
5.1.3. Pipes junctions

Figure 5.1.c shows three liquid reservoirs, A, B, and C, which are connected by pipes
that join at the junction D. Given the elevations of the three reservoirs relative to the
junction, zA , zB, and zC, it is required to determine the magnitudes and directions of the
flows in the three pipes. All flow rates in the three pipes are unknown, but using the
continuity equation the number of independent unknowns in the problem can be
reduced to two. Since the flow rates in the junction pipes depend on the size (length and
diameter) and roughness of the pipes, the energy equation can be used to provide the
additional equations required for solving the problem.
The analytical model

Assuming no variation in density, conservation of mass leads to:
Q3  Q1  Q2 (5.10)
Depending on the reservoirs elevations zA , zB, and zC, one or more of the pipe flows
may be opposite to that shown in Figure 5.1.c. Another requirement is that the variation
in pressure must be continuous. Applied at the junction, this leads to the following
relationships:.
z A  h f 1  zB  h f 2 (5.11.a)
z A  h f 1  zC  h f 3 (5.11.b)
z B  h f 2  zC  h f 3 (5.11.c)
Since there are only three unknowns, Equation (5.10) and any two of Equations
(5.11.a), (5.11.b), or (5.11.c) should be sufficient to solve the problem. The following
example, which has been adopted from Example 5.19 in White [2], illustrates the
method.
Example 5.3. Flow in a pipe -junction

Take the same three pipes in Example 5.2, and assume that they connect three
reservoirs at these surface elevations relative to the junction:
ZA = 20 m. zB = 100, m zC = 40 m
Neglecting minor losses, find the resulting flow rates in each pipe.
Figure 5.7 shows the Excel sheet developed for this example in which the given data are
shown at the left-side of the sheet. Two of the three pipe velocities are assumed, V_1
and V-2, while V_3 is calculated from Equation (5.14). Based on these velocities, the
sheet calculates the Reynolds numbers (Re_1, Re_2 and (Re_3), friction factors (f_1,
f_2 and f_3), and friction losses (hf_1, hf_2 and hf_3). Unlike the two previous
examples, Goal Seek cannot be used here since we need to start with guessed values of
the velocity in two pipes. This particular pipe arrangement also demonstrates the
usefulness of Solver for dealing with problems that require iterative solutions with
multiple variables since it allows the two velocities to be changing cells.
Figure 5.7. Excel sheet developed for Example 5.3
The following constraints are imposed by Equation (5.11):
 
z A  zB  h f 1  h f 2  0 (5.12.a)
 
z A  zC  h f 1  h f 3  0 (5.12.b)
Figure 5.8 shows Solver set-up for this example. This set-up requires Solver to find the
values of V_1and V-2 that make the two constraints required by Equation (5.12) close
to zero. Note that the formula bar in Figure 5.7 reveals the formula that determines the
first constraint. Figure 5.9 shows the solution obtained by Solver using the GRG
Nonlinear method. The velocity values determined by Solver are V_1 = - 2.9127, V_2 =
4.6032 and V_3 = 1.2938 m/s and the resultant flow rates are Q1 = - 52.7, Q2 = 46.9,
and Q3 = 5.9 m3 /h. The corresponding flow rates given by White [2] are 52.8, 47.0, and
5.8 m3 /h, respectively. The minus sign for Q1 indicates that the flow direction is away
from the junction.
Figure 5.9. Solver solution for Example 5.3
5.2. Determining the operating point for a pump-pipe system

Electrically-driven pumps are usually used in fluid-transporting systems in order to
overcome frictional losses in the pipes and the components. Different types and sizes of
pumps are manufactured for meeting various demands in terms of fluid flow rates and
pressure. Selection of a suitable type and size of the pump for a particular task depends
on the characteristics of the pump. For example, centrifugal pumps, which are the
mostly used type of pumps, can supply high heads at low flow rates and high flow rates
at low heads. The operating point for a pump-pipe system is the point at which the flow
rate and heat supplied by the pump match the required demand. This section deals with
the determination of operating point for a typical centrifugal pump with three common
pipe arrangements that include (a) a single pipe, (b) two pipes that branch at the pump
exit, and (c) a simple network consisting of three branching pipes.
5.2.1. A centrifugal pump connected to a single pipe

Figure 5.10 shows a centrifugal pump that is connected to a single pipe. The pump
supplies the pressure head needed to overcome the friction losses through the pipe so as
to maintain the fluid flow.
Figure 5.10. Pump-pipe system with a single pipe
The pump head h p (in m) is defined as the head of fluid delivered by the pump, i.e.:
h p  Pe  Pi  /  (5.13)
Where Pi and Pe are the inlet and exit pressures, respectively, and γ is the specific
weight of the fluid. The pump head is not constant but depends on its discharge Q (in
m3 /s). The curve that shows the relationship between the pump head and the pump
discharge at a given speed is known as the pump characteristic curve. Figure 5.11
shows a typical characteristic curve for a centrifugal pump at a certain speed. Pump
characteristic curves are supplied by pump manufacturers at different speeds or can be
prepared by direct measurement.
60
50 Operating point
40
30
h, m
20
Pump curve
10
System curve
0
0 0.001 0.002 0.003 0.004
Q, m3 /s
Figure 5.11. Determining the operating point for a centrifugal pump from its
characteristic curve and the system curve
Figure 5.11 shows that the maximum pump head is delivered at zero discharge and the
head decreases as the discharge is increased. How much pump head is actually needed
to overcome friction losses (h f ) and circulate the fluid through the pipe depends on:
1. The flow rate of the fluid Q

2. The viscosity of the fluid ν (which depends on the fluid‟s temperature)
3. The diameter of the pipe D
4. The length of the pipe L
5. The roughness (ε) of the pipe inside surface which depends on its material and
surface finish
The relationship between the frictional losses through the pipe and the flow rate of the
fluid being transported is called the system curve. The system curve is obtained by
calculating the friction loss h f at increasing fluid flow rates and plotting it versus Q as
shown in Figure 5.11. As the figure shows, the friction loss is zero at zero discharge and
increases as the discharge is increased. The pump curve and the system curve intersect
at a certain point as shown in Figure 5.11. The intersection point of the system and
pump curves is the operating point for the pump. At the operating point:
h p  h f ,total  h f  hc (5.14)
Where h f and h c refer to the friction head losses in the pipe and the components,
respectively. The power of the pump W at any given flow rate Q and pump head h p is
given by:
W p  Qh p /  (1.19)
Where is η is the pump efficiency. The pump efficiency reaches its maximum value at a
certain flow rate and a certain head. If we try to operate the pump at load requirements
(Q and h p ) which are different those values, then the pump efficiency will be reduced.
Therefore, it is important to have the operating point of the pump-pipe system as close
as possible to that point.
The operating point of the pump can be determined graphically by plotting the pump
curve and system curve on the same graph as shown in Figure 5.11. For the particular
pump-pipe system associated with Figure 5.11, the figure shows that the operating point
occurs about Q = 1.5x10-3 m3 /s and h p = 42 m. However, the operating point of the
system can be determined more accurately by representing the pump curve by a
mathematical relationship between the pump head and discharge. Using this
relationship for h p versus Q, together with Equation (5.14), the operation point can be
determined by iteration. The following example illustrates how Excel can be used to
determine the operating point for a pump-pipe system with a single pipe.
Example 5.4: Operating point for a pump-pipe system with a single pipe
A centrifugal pump is to supply water at (ρ = 1000 kg/m3 , ν = 1.02x10-6 ) through a
single pipe that has the following specifications:
Length, 100 m
Diameter, 2.5 cm
Roughness, 0.046 mm
According to the manufacturer, the pump has the following characteristics:
Flow rate Pump head

0.0 47.24
6.3080E-04 45.11
9.4620E-04 44.2
1.2316E-03 42.67
1.5770E-03 41.15
1.8924E-03 39.62
2.2078E-03 37.19
2.5232E-03 35.05
2.8386E-03 32.00
3.1540E-03 28.96
It is desired to determine the operating point and the required power for the pump.
Ignore component losses.
The solution method

Since there are no component losses, the friction head is obtained from Equation (1.21)
after we establish whether the flow is laminar or turbulent from the value of the
Reynolds number. In order to obtain a solution via Excel, the pump head has to be
expressed as a function of the flow rate. For the pump data shown above,
Suryanarayana and Arici [3] obtained the following polynomial:
h p  47.22  2.985  103 Q  1.549  105 Q 2  2.348  108 Q 3 (5.15)
The sought flow rate (Q) should be such that the calculated friction head loss in the pipe
is equal to pump head as obtained from Equation (5.15). The iterative solution is
obtained as follows:
1. Given an initial guess for the discharge, calculate the fluid velocity from V =
Q/A
2. Calculate the Reynolds number and the friction factor f
3. Calculate the corresponding friction loss h f from Equation (1.21)

4. Determine the pump head from Equation (5.15)
5. Calculate the difference between the pump head and the friction losses, which
is supposed to be zero at the operating point.
It is unlikely that our first guess will be the correct discharge at the operating point.
Therefore, the above procedure has to be repeated by an improved guess for the
discharge until the difference between h f and h p becomes sufficiently small.
Figure 5.12 shows the Excel sheet developed for solving this example. The data part of
the sheet shows the data provided for the pipe and the fluid, which is water. Figure 5.12
shows how the calculation part of the Excel sheet executes the iterative procedure
described above with an initial guess for the discharge of Q= 1.0x10-3 m3 /s. the figure
shows that there is a difference of 21.51 m between the system head (hf) and the pump
head (h_p). Goal Seek (or Solver) can be used to find the value of Q at which the
difference between hf and h_p becomes approximately zero. Figure 5.12 also shows
Goal Seek set-up for this example and Figure 5.13 shows the solution obtained by it. As
the figure shows, Goal Seek finds the operating point at Q = 1.395x10-3 m3 /s and h_p =
42.12 m. The corresponding power is 576.29 W. The flow rate obtained by
Suryanarayana and Arici [3] is 1.407x10-3 m3 /s.
Figure 5.12. Excel sheet and Goal Seek set-up for Example 5.4
Figure 5.13. Solution obtained by Goal Seek for Example 5.4

5.2.2. A centrifugal pump connected to two branching pipes

Figure 5.14 shows a pump that discharges into two branching pipes of different
diameters, lengths, and roughness. It is required to determine the operating point for the
pump and how the pump's discharge will be distributed between the two pipes. To
simplify the analysis, it is assumed that there are no differences in elevations between
the pipes inlets and outlets. Moreover, the two pipes are assumed to discharge at
atmospheric pressure.
P1 = Patm
Q1 , L1 , D1 , ε1
Q2 , L2 , D2 , ε2
P2 = Patm
Pi = Patm
Figure 5.14. Pump with two pipes in parallel
Continuity equation:
Given the values of the flow rate in pipe 1 (Q1 ) and in pipe 1 (Q2 ), the total flow rate Q
can be calculated from:
Q  Q1  Q2 (5.16)
The friction head loss in the each pipe can be obtained from Equation (1.21):
L1 V12
h f 1  f1 (5.17.a)
D1 2 g
L2 V22
hf 2  f2 (5.17.b)
D2 2 g
Since pressures at both ends of the two pipes are the same, the friction head losses in
the two pipes must be the same and equal to the pump head, i.e.:
h f 1  h f 2  hp (5.18)
Equation (5.18) will be satisfied at the correct values of Q1 and Q2 , but since both flow
rates are unknown an iterative solution is required which proceeds as follows:
1. Assume reasonable values for Q1 and Q2

2. Calculate the total flow rate Q from Equation (5.18)
3. Calculate V1 and V2 and determine the Reynolds number in the two pipes, Re 1
and Re2
4. Determine the friction factors in the two pipes, f 1 and f 2 , from the corresponding
Reynolds numbers
5. Determine the friction losses in the two popes, h f1 and h f2
6. Determine the pump head hp from Equation (5.15)
7. If h f1 ≠ h f2 , or h f1 ≠ h p , go to step 2 with improved values of Q1 and Q2
After the operating point is determined, the power required by the pump can be
obtained from the power equation as follows:
W p  gQh p (5.19)
Alternatively, the pump power can be calculated from the pipes flow rates and friction
losses as follows:

W p  g Q1 h f 1  Q2 h f 2  (5.20)
Suryanarayana and Arici [3] analysed the performance of the pump-pipe systems shown
in Figure 5.14 by following a slightly different iterative process. They also adopted a
different formula for the friction factor than Equation (1.25). The data of their example
will be used to demonstrate the procedure for using Excel-Solver in the analysis.
Example 5.5. Operating point for a pump with two parallel pipes
A pump-two-pipe system as shown in Figure 5.14 has the following dimensions.
D1 = 0.025m L1 = 500m ε1 =15x10-6 m
D2 = 0.05m L2 = 1,000m ε2 =46x10-6 m
The system carries water (ρ = 1000 kg/m3 , ν = 1.02x10-6 ). If the pump has the same
characteristic formula given in Example 5.4 as expressed by Equation (5.15), determine
the flow rates in the two branches and the power required by the pump.
The Excel sheet

The Excel sheet developed for this example is shown in Figure 5.15. The data part
contains the data for the two pipes and those for the transported fluid, which is water.
The sheet shows the calculations based on assumed values of Q1 = Q2 = 0.001 m3 /s. The
velocities, Reynolds number, friction factors and friction head losses are calculated on
the basis of these assumed flow rates. With the guessed flow rates, the friction losses in
the tw pipes, h f1 and h f2 , and the pump head, h p , have three different values. The formula
bar shows the formula in cell M5 that calculates the difference between the head losses
in pipe 1 and the head produced by the pump as 58.066 m. Also, note that Equations
(5.19) and (5.20) give two different values for the pump power in cells M7 and M9.
The required flow rates Q1 and Q2 that satisfy Equation (5.18) can be found by using
Solver. Figure 5.16 shows the set-up for Solver that requires it to find values of Q1 and
Q2 that reduces the difference between h f1 and h p to zero while satisfying the constraint
that the head losses in the two pipes h f1 and h f2 are equal according to Equation (5.18).
The solution found by Solver using the GRG Nonlinear method is shown in Figure
5.17. Solver determined the flow rates in the two pipes as Q1 = 0.000547 m3 /s and Q2 =
0.002279 m3 /s. At these flow rates, the friction head losses in the two pipes are equal to
32.243 m. Figure 5.17 also shows that the two values of the power required by the
pump at the operating point as determined by Equations (5.19) and (5.20) are the same
and equal to 894.04 W.

The corresponding values found by Suryanarayana and Arici [3] after nine iteration are
Q1 = 0.0005738 m3 /s, Q2 = 0.002325 m3 /s, and hp = 32.85 m. The slight differences
could be attributed to the fact that Suryanarayana and Arici [3] used a different formula
for calculating the friction factor. It should be mentioned that it was necessary in this
example to use the automatic-scaling option of Solver so as to find the solution.
Apparently, this is due to the use of Equation (5.15).
5.2.3. A centrifugal pump connected to a three-pipe branched network

Figure 5.18 shows a pump connected to a simple pipe network that consists of three
pipes that joint at point J. The three pipes can have different lengths, diameters, and
roughness. Another complication with this system is that some fluid is discharged at the
junction, which is shown in Figure 5.18 as QJ. This flow rate can either be pre-specified
as a given value or left to be determined by the iterative solution.
P1 = Patm
Q2 , L2 , D2 , ε2
Q1 , L1 , D1 , ε1
J
QJ
Q3 , L3 , D3 , ε3 P3 = Patm
Pi = Patm
Figure 5.18. Pump with a three-pipe network
The principles of mass and energy conservation can be applied in order to determine the
operating point for the pump.
Mass conservation
Q1  Q2  Q3  Q J (5.21)
Therefore, if three of the flow rates are known, the fourth can be calculated from the
above equation.
Conservation of energy:
Since the pressures at the inlet of pipe 2 and pipe 3 are the same, and the pressures at
their discharge points are both atmospheric, the energy equation leads to:
hf 2  hf 3 (5.22)
At the operating point, the power required by the pump to balance the friction losses in
the network can be calculated from.
W p  gQ1 h p (5.23)
Or from:

W p  g Q1 h f 1  Q2 h f 2  Q3 h f 3  (5.24)
The requirement that the two values of the pump power determined by Equations (5.23)
and (5.24) should be equal provides another constraint on the iterative process besides
Equation (5.22).
Example 5.6. Pump with three branched pipes

A pump-system shown in Figure 5.18 has the following dimensions.
D1 = 0.05m L1 = 500m ε1 =46x10-6 m
D2 = 0.025m L2 = 500m ε2 =15x10-6 m
D3 = 0.03m L3 = 500m ε3 =46x10-6 m
The system carries water (ρ = 1000 kg/m3 , ν = 1x10-6 ). The network supplies a demand
of 0.0002 m3 /s at the junction J. If the pump has the same characteristic formula given
in Example 5.4 and expressed by Equation (5.15), determine the operating point and the
power required by the pump.
The Excel sheet

The Excel sheet developed for this example extends that developed for Example 5.5 by
adding the data and the calculations relevant to the third pipe and the flow rate at the
junction (Q_J). The calculations start by guessed values for Q1 = 0.001 m3 /s and Q2 =
0.0005 m3 /s. As the formula bar shows, Q3 is determined from Equation (5.20). The
sheet then calculates the velocities, Reynolds numbers, friction factors and friction
losses in the three pipes from the Darcy-Weisbach equation using the relevant data. The
sheet also calculates the pump head (h_p) from Equation (5.15) at the guessed value of
the flow rate (Q = Q1 ). The pump power is then calculated on the basis of Equation
(5.23) as W_pump and on the basis of Equation (5.24) as W_pipe.
Values of h f1 , h f2 , h f3 , and h p shown in Figure 5.19 indicate that Equation (5.22) is not
satisfied with the guessed flow rates. The two values of the pump power determined by
Equations (5.23) and (5.24), i.e. W_pump and W_pipe, are also different. Solver can be
used in order to find the flow rates that satisfy these two conditions. Figure 5.20 shows
the set-up for this example that requires Solver to find the values of Q1 and Q2 that
makes the difference between h f2 and h f3 diminishes (Equation (5.21)) while making
pump power determined by Equation (5.23), which is W_pump, equal to that
determined by Equation (5.24), which is W_pipe. A third constraint has been added that
requires Q2 ≤ Q1 so as to ensure a physically acceptable solution. By leaving the Set
Objective blank, Solver will iterate so as to find the solution that satisfies the three
specified constraints.
Figure 5.21 shows the solution found by Solver with the GRG Nonlinear method.
According to this solution, Q1 = 0.001671, Q2 = 0.00058 and Q3 = 0.000892 m3 /s.
Judging from the Reynolds numbers in the three pipes, all pipe flows are turbulent. At
the flow rates obtained for the three pipes, the friction losses in pipe 2 and pipe 3 are
both equal to 35.9247 m, while the total friction head to be provided by the pump is
40.70285 m. Both Equations (5.23) and (5.24) determine the pump power as 667.31 W.
Figure 5.21. The solution determined by Solver for Example 5.6
The automatic-scaling option is needed for this example also. In this example the
discharge at the pipes junction has been specified, but it is also possible to allow it to be
determined by Solver. In this case, there will be three adjustable cells for the three flow
rates, Q2 , Q2 , and Q3 or QJ. This is left as an exercise to the reader.
5.3. Operating point for centrifugal pumps in serial and parallel arrangements
Practical considerations may prefer the use of two or more smaller pumps instead of
one large pump. Deciding the number, size, and arrangement of the pumps needed for a
particular pipe system requires careful analysis of the combined pump-pipe system.
This section deals with the determination of the operating point for two centrifugal
pumps arranged in parallel and in series.
5.3.1. Two centrifugal pumps connected in parallel

Figure 5.22 shows the pump curves for a single pump and for two identical pumps
connected in parallel as shown in Figure 1.7.a. As Figure 5.22 shows, the operating
point shifts from point A to point B, with the two pumps giving a larger flow rate and
head compared to the single pump. It is desired to determine the operating point and
power requirement of the pump-pipe system in this case.
hp
Two pumps
One pump B
A
Figure 5.22. Characteristic curve for two centrifugal pumps connected in parallel
For a steady flow of an incompressible fluid in the common pipe, mass conservation
leads to:
Q  Q1  Q2 (5.25)
Parallel arrangement of pumps is referred to as "flow-rate additive" [4] because the

combined flow rate in the system is the addition of those of the individuals pump as
indicated by Equation (5.25). For two identical pumps:
Q1  Q2  Q / 2 (5.26)
Neglecting the small loss in the pipe joining the two pumps and assuming no losses
within the two pumps, the energy equation applied between points 1 and 2 leads to:
h p ,1  h p , 2  h p (5.27)
At the operating point, the friction head loss in the pipe equals the head delivered by
one pump, i.e.:
h f  h p ,1  h p , 2 (5.28)
Example 5.7. Operating point for two pumps in parallel

Reconsider the pump-pipe system considered in Example 5.4. A second pump, which is
identical to the existing pump, has now been connected in parallel as shown in Figure
1.7.a. Determine the operating point and total power requirement of the two pumps for
the same system that carries water at 25o C.
The Excel sheet

The data part of the Excel sheet shown in Figure 5.23 is identical to that in the sheet
developed for Example 5.4. Calculations start at an assumed flow rate in the main pipe
Q = 0.001 m3 /s. Based on this flow rate, the water velocity in the pipe and the
corresponding Reynolds number are determined. Since the two pumps are identical, the
total flow rate is divided equally between the two pumps as stated by Equation (5.26).
Therefore, the heads delivered by the two pumps h p,1 and h p,2 are the same as required
by Equation (5.27). However, the friction head loss in the main pipe (hf) shown in
Figure 5.23 is not equal to the head supported by the pumps, h_p1 and h_p2, as required
by Equation (5.28). Therefore, the flow rate Q has to be adjusted to a value that satisfies
both equations (5.27) and (5.28). This can be done by using Goal Seek to eliminate the
difference between hf and h_p1. Figure 5.24 shows the solution found by Goal seek.
Figure 5.23. Excel sheet developed for Example 5.7 with Goal Seek set-up
Figure 5.24. Goal Seek solution to Example 5.7

The solution determined by Goal Seek matches the pipe friction loss with the pumps'
head. The obtained flow rate is 0.001442 m3 /s, which leads to a friction loss of 44.9 m
and a power requirement of 635.177 W. Compared to Example 5.4, in which a single
pump was used, the addition of the second pump in parallel has increased the pipe flow
rate from 0.0014 m3 /s to 0.00144 m3 /s, i.e. by 2.86%. However, the friction loss in the
pipe has also been increased from 42.12 m to 44.9 m, which is a 6.6% increment, and
the power requirement has increased from 576.29 W to 635.18 W, i.e., an increment of
about 10.22%. This example problem can also be solved by using Solver. Figure 5.25
shows the relevant set-up for Solver. The solution found by Solver is shown in Figure
5.26. Comparison with Figure 5.24 confirms that the two solutions are identical.
Figure 5.26. Solver solution to Example 5.7

5.3.2. Two centrifugal pumps connected in series

Two pumps are connected in series when the outlet of the first pump leads directly to
the inlet of the second as shown in Figure 1.7.b. The corresponding head-versus
discharge curve is shown in Figure 5.27. Figure 5.27 shows that the operating point for
a single pump is at point A. When another pump is added in series the operating point
shifts to point B where both the head and flow rate are increased. For simplicity, the
following analysis assumes that the two pumps have the same capacities.
hp
Two pumps
One pump B
Figure 5.27. Characteristic curves for two centrifugal pumps connected in series
Assuming steady flow of an incompressible fluid in the common pipe, mass

conservation leads to:
Q  Q1  Q2 (5.29)
Neglecting the small loss in the pipe joining the two pumps and assuming no losses
within the two pumps, the energy equation leads to
h p  h p ,1  h p , 2 (5.30)
Pumps in series are called "pressure-additive" [4] because, as Equation (5.30) indicates,
the total pressure in the pipe system is the addition of those of the individual pumps. At
the operating point, the friction head loss in the pipe equals the head delivered by the
two pumps, i.e.:
h f  h p  h p,1  h p, 2 (5.31)
Example 5.8. Operating point for two pumps in series

Two pumps were connected in series as shown in Figure 1.7.b. Both pumps have the
same characteristic curve expressed by Equation (5.15). The length, diameter, and
roughness are the same as those given in Example 5.4. Determine the operating point
and power requirement of the pump-pipe system.
Excel sheet
The sheet developed for this example is shown in Figure 5.28. The data and
calculations parts are similar to those in Example 5.7 but, instead of two flow rates in
the pipes, the present sheet assumes the same flow rate Q passes through both pipes but
calculates two pump heads hp1 and hp2. Given an initial value for Q, the sheet
calculates the velocity (V), Reynolds number (Re), friction factor and head losses in the
two pipes (hf).
Using the given value for Q, the sheet also determines the pumps heads h_p1 and h_p2,
which are identical in this case. Figure 5.28 shows that there is a difference between the
pumps' heads and the friction loss in the pipe which indicates that the guessed value for
Q is incorrect. Figure 5.28 also shows the set-up for Goal Seek for finding the value of
Q that makes the friction loss (hf) equals the pump head (hp). Figure 5.29 shows the
solution obtained by Goal Seek, which is Q = 0.0019 m3 /s.
Figure 5.29. Solution for Example 5.8 found by Goal Seek

Compared to Example 5.4, in which a single pump was used, the addition of the second
pump in series has increased the pipe flow rate from 0.0014 m3 /s to 0.001927 m3 /s, i.e.
by 37.86%. However, the friction loss in the pipe also increased from 42.12 m to
78.429 m, which is a 86.21% increment, and the power requirement increased from
576.29 W to 1482.38 W, i.e., an increment of 157.23%. The solution can also be found
by using Solver and Figure 5.30 shows Solver set-up for this example. Figure 5.31
shows that the solution found by Solver is identical to that obtained by Goal Seek.
Table 5.2 compares the calculated values for pipe flow rate, friction loss, and pump
power for the three arrangements considered in Example 5.4 (a single pump), Example
5.7 (two pumps in parallel), and Example 5.8 (two pumps in series).
Figure 5.31. Solution for Example 5.8 found by Solver

Table 5.2. Comparison of the operating conditions for a single pump, two pumps in
parallel, and two pumps in series
Case Arrangement Q (m3 /s) h p (m) Power (W)
Example 5.4 A single pump 1.40 42.12 576.29
Example 5.7 Two pumps in parallel 1.44 44.89 635.18
Example 5.8 Two pumps in series 1.93 78.43 1482.38
5.4. Closure
This chapter dealt with the analyses of various arrangements of multi-pipe and pump-
pipe systems by using the iterative tools provided by Excel. In the analyses of multi-
pipe systems, the iterative solution is required for simultaneously satisfying the
continuity equation and the energy equation. For pump-pipe systems, it is required for
determining the operating point of the system. The chapter shows how the iterative
solutions in these analyses can be handled by using Goal Seek and Solver. Some of the
analyses involved a single variable and, therefore, could be solved by using Goal Seek,
while other analyses required the use of Solver for dealing with multiple changing cells
or allowing constraints to be applied on the iterative process.
All the analyses in this chapter that used Solver applied the GRG Nonlinear method and
required less than 5 iterations. Determining the operating point for a pump-pipe system
by using the third-order pump curve formula in Equation (5.15) required the use of the
automatic-scaling feature of Solver. It should be noted that the analyses only utilised
Excel‟s Goal Seek and Solver and did not need the Thermax1 property add-in. Most
fluid-flow and heat-transfer analyses do not involve variations of fluid-properties and,
therefore, the Excel-Solver combination is adequate for them. The following three
chapters show how this combination can be used for the analyses and optimisation of
pipe networks and pump-pipe systems.
References
[1] Y.A. Cengel and J.M. Cimbala, Fluid Mechanics Fundamentals and Applications,
McGraw- Hill, 2006.
[2] F. M., White, Fluid Mechanics, Seventh Edition, McGraw-Hill, 2011.
[3] N. N. Suryanarayana, and O. A. Arici, Design and Simulation of Thermal Systems,
McGraw-Hill, 2004.
[4] W. S. Janna, Design of Fluid Thermal Systems, Third Edition, CENGAGE
Learning, 2011.
Analyses of pipe networks

Pipe-network flow analyses are based on the two principles of mass and energy
conservation coupled with an auxiliary formula for determining the pipe friction. The
same principles were applied in the previous chapter for analysing simple multi-pipe
systems, but pipe-network analyses are more challenging because they involve more
continuity equations to be solved with more energy constraints to be satisfied.
Moreover, practical pipe-network analyses require computer-based methods because
closed-form analytical solutions are only possible for very simple cases. Before
introducing the computer-based methods for pipe-network analyses, this chapter
initially highlights the limitation of mathematical analytical methods by considering a
simple three-pipe network. The chapter then introduces the Hardy-Cross method, which
is the earliest computer-based method for the analyses of looped networks. An Excel-
based methodology is then presented that uses Solver to deal with the analyses of both
looped and branched pipe networks.
6.1. Mathematical analysis of looped pipe-networks

Figure 6.1 shows two types of pipe network arrangements: (a) a looped network and (b)
a branched network. As shown in Figure 6.1, the flow in a pipe network can be driven
by gravity effect or by a motor-pump system. The connection points, called nodes, can
be sources (supply points), sinks (demand or load points), or just junctions. The looped
network in Figure 6.1.a has two sources, which are the reservoirs A and B, and six load
points. The branched network in Figure 6.1.b has a single source, which is the pump
(A), and six load points. According to the notation scheme adopted in Figure 6.1, a
single letter refers to a location or a pipe junction, e.g. A, B, etc., two letters refer to a
pipe, e.g. AB, BC, etc., whereas numbers refer to pipe loops, e.g. loop 1, loop 2, etc.
I
A J
B
H
F G
C
D
Loop 1 E
E C D
Loop 2 Loop 3 B
F
H
G M A
(a) (b)
Figure 6.1. Pipe network arrangements: (a) looped and (b) branches networks
In a typical analysis, the flow rates at the sources and outlets are given together with the
lengths, diameters, and roughness of the pipes and it is required to determine the flows
and friction losses in the different network pipes. In order to analyse the looped pipe
network shown in Figure 6.1 that has eight pipes, eight independent equations have to
be formed from mass and energy balances.
The continuity equations

The continuity equation can be applied at two levels; (i) to the whole pipe network, and
(ii) to any junction (node) in the network. For a steady flow of an incompressible fluid,
the continuity equation applied to the whole network shown in Figure 6.1.a leads to:
QA + QB = QC + QD + QE + QF + QG + QH (6.1)
Applied to any junction in the network, the continuity equation requires the algebraic
sum of all the flow rates meeting at the junction to be zero. For example, the flow at
junction C of the network shown in Figure 6.1.a is given by:
QC = QAC - QCD - QCE - QCF (6.2)
Application of the continuity equation at the six nodes in the network shown in Figure
6.1.a leads to five (number of nodes minus 1) independent linear equations. The
remaining three equations have to be obtained from the energy equation.
The energy equations

In general, the energy equation should not only account for the friction losses in the
pipes, but also for any differences in elevations between the end points. However, for
simplicity let us assume that all pipes of the network under consideration lie in the same
horizontal plane. A special form of the energy equation (Equation (6.2)) can be written
for a closed loop in which point 2 in the equation coincides with point 1. In this case,
the algebraic sum of the head losses in all the pipes forming a closed loop is balanced
by any heads generated by inline booster pumps in the loop.
M N
1
hf  h
1
P (6.3)
Where M is the number of pipes forming the closed loop, N is the number of booster
pumps in the loop, h f is the head loss in each pipe (including the minor losses), and h P is
the head produced by a booster pump. When there is no booster pump within the loop,
Equation (6.3) reduces to:
h
1
f 0 (6.4)
Friction loss in a pipe (h f ) can be determined by using the Darcy-Weisbach equation.

For pipes carrying water, the Hazen-Williams equation can be used instead of the
Darcy-Weisbach equation.
Since we cannot be certain about the flow directions in the network pipes in advance,
we have to assume them. The flow directions in pipes CD, CE and CF in Figure 6.1.a
are assumed flow directions. The true directions can only be determined after the
analysis is done. Applying Equation (6.4) to the three loops in the network provides the
three (number of loops) additional equations needed to solve the problem. The solution
of the eight equations should give the unknown values of the flow rates in the eight
pipes of the network. However, the energy equations are nonlinear because of the
dependency of friction losses on the unknown pipe flow rates. The following example
demonstrates the mathematical solution methodology by considering a simple network
that consists of only three pipes forming a single loop.
Example 6.1. Mathematical analysis of a simple looped pipe network

Figure 6.2 shows a pipe network that distributes water from a single source to two
consumption points as shown in the figure. The friction factor for all three pipes can be
taken as 0.025. What are the flow rates in the three pipes?
A B
100 l/s 60 l/s
L=200m
D=0.2m
L=400m
D=0.15m L=600m
D=0.1m
40 l/s
Figure 6.2. Pipe network in Example 6.1
Solution
Since there are only three flow rates in this case that have to be determined, three
independent equations are needed to solve the problem. There are two possible flow
arrangements as shown in Figures 6.3,a and 6.3.b that show the flow in pipe BC in
opposite directions. For this simple network, let us refer to pipe AB by pipe 1, pipe BC
by pipe 2, and pipe AC by pipe 3. Starting with the arrangement shown in Figure 6.3.a,
mass conservation leads to the following two equations:
100 l/s A Q1 B Q1
100 l/s A B
60 l/s 60 l/s
Q3 Q2 Q3 Q2
C C
40 l/s 40 l/s
(a) (b)
Figure 6.3. Possible flow directions in the network
Q2  Q1  60 (6.5)
Q3  100  Q1 (6.6)
The third equation can be obtained by applying the energy equation (Equation (6.4)).
h f ,1  h f , 2  h f ,3  0 (6.7)
Using Equation (1.21) and substituting for V by Q/A, Equation (6.8) becomes:
L1 Q12 L2 Q22 L3 Q32

f1  f2  f3 0 (6.8)
D1 2 gA12 D2 2 gA22 D3 2 gA32
This equation can be put in the following simpler form:
K1Q12  K 2 Q22  K 3Q32  0 (6.9)
Where,
L1 1
K1  f1 (6.10)
D1 2 gA12
L2 1
K2  f2 (6.11)
D2 2 gA22
L3 1
K3  f3 (6.12)
D3 2 gA32
Now, substituting for Q2 and Q3 from Equations (6.5) and (6.6), Equation (6.9)
becomes:
K1Q12  K 2 Q1  60  K 3 100  Q1   0

2 2
(6.13)
Expanding and rearranging the equation leads to:
aQ12  bQ1  c  0 (6.14)
Where:
a  K 1  K 2  K 3  (6.15)
b  120K1  200K 2  (6.16)
c  60 2 K 2  1002 K 3 (6.17)
Equation (6.14) is a quadratic equation that can be solved for Q1 once the numeric
values of K1 , K2 , and K3 are determined. There are two solutions for the quadratic
equation, which are given by:
 b  b 2  4ac
Q1  (6.18)
2a
Where a = 114350.4, b = -1.3x107 , and c = 3.37x108 . Substituting for a, b, and c in

Equation (6.18) gives the following solutions:
First solution: Q 1 = 66.99 l/s, Q2 = 6.99 l/s, and Q3 = 33.0 l/s
Second solution: Q 1 = 44.04 l/s, Q2 = -15.96 l/s and Q3 = 55.96 l/s
The second solution, which leads to a negative value for Q2 , cannot be accepted because
it either means that the magnitude of Q2 is negative, which is physically meaningless, or
its direction is reversed, which goes against our initial assumption embedded in the
formation of Equation (6.7). Trying to solve the problem assuming that the flow rate in
pipe BC goes in the opposite direction as shown in Figure 6.3.b leads to values of the
flow rates in the pipes as complex numbers, which is also a physically unacceptable
solution. Therefore, the only acceptable solution is the first solution shown above.
6.2. The Hardy-Cross method

The pipe-network considered in Example 6.1 could be analysed mathematically because
of its simplicity, but the analyses of complex industrial pipe networks are usually
performed by using computer-based methods [1]. This section introduces the Hardy
Cross (H-C) method which is the earliest and mostly used method for the analysis of
looped networks such as those shown in Figure 6.1.a. To apply the method, the
magnitudes and directions of the pipe flows are assumed such that the continuity
equation is satisfied. The flow rates are then corrected on the basis of the energy
equation. Further corrections of the newly obtained flow rates are performed until the
corrections become practically zero and all the pipe flows converge to constant values.
Suppose that the assumed flow at a particular pipe is Q. Then the corrected flow rate is
given by:
Qnew  Q  Q (6.19)
Where ΔQ is the flow correction for the particular pipe being considered. The question
now is how ΔQ is determined by using the energy equation. Equation (1.21) can be put
in the following form:
h f  KQ 2 (6.20)
Where K is given by:
Lf 2 Lf
K  2 3 (6.21)
2 gDA 2
g D
Applying Equation (6.20) to the corrected flow rate given by Equation (6.19), we get:
h f  K Q  Q
2
(6.22)
Expanding the right-hand side of Equation (6.22):

h f  K Q 2  2Q.Q  Q 2  (6.23)
Neglecting the term (ΔQ2 ), the equation becomes:

h f  K Q 2  2Q.Q  (6.24)
Applied to all pipes in the same loop, the energy equation, Equation (6.4), leads to:
 K Q 2
 2Q.Q  0 (6.25)
Rearranging the equation, leads to the following:
 KQ 2
 Q  2KQ  0 (6.26)
Or,
 KQ   h 2
 hf 
Q    /  2  Q  (6.27)
 2 KQ
f

To apply the H-C method, a sign convention must be adopted that takes into
consideration the flow direction in the particular pipe. In the following discussion, the
friction losses in the clockwise direction are taken as positive while those in the
counter-clockwise direction are taken as negative. For pipe CE in Figure 6.1.a which is
common to the two loops 1 and 2 this sign convention means that the friction loss in
this pipe will be given a negative sign with loop 1, but a positive sign with loop 2. The
procedure for applying the H-C method is summarised as follows:
1. Make a guess for the flow rate (magnitude and direction) in one pipe of each loop
and calculate the flow rates in the other pipes in the network by applying the
continuity equation at the respective nodes.
2. Calculate the friction loss h f in each pipe of the network and determine the net
friction loss in each loop according to the flow direction in the relevant pipes
(clockwise positive and counter-clockwise negative). Most likely, these will not
be zeros at the initially guessed pipe flows.
3. Calculate a correction increment ΔQi for the flows in loop i from Equation
(6.27). The summations in Equation (6.27) are performed over the number of
pipes in each loop.
4. Obtain new pipe flows in each loop by modifying the old ones according to
Equation (6.19).
5. Repeat steps 2, 3, and 4 until the correction increments become small and the
variations in the flow rates become negligible.
A number of articles and videos in the internet explain how the H-C method can be
implemented in Excel [2,3]. The following example demonstrates its application to
analyse a simple network.
Example 6.2. Analysis of a simple pipe -network by the H-C method

Figure 6.4 shows a simple pipe network that distributes water to three consumption
points as shown in the figure. Determine the flow rates in the four pipes if a friction
factor for all pipes can be taken as 0.025.
L=200m
D=0.20m
100 60
l/s l/s
L=300m
D=0.15m
L=400m
D=0.20m
L=500m
D=0.15m
30
l/s
10 l/s
Figure 6.4. Pipe network in Example 6.2
Solution
Step 1: The first step is to assume flow rates in the four pipes that satisfy continuity. It
is enough to assume the value of the flow rate in one pipe since the other flow rates
must follow the continuity equation. Let us assume that the flow in pipe AB is 50 l/s in
the direction shown in Figure 6.5. The magnitudes and directions of other flow rates are
automatically determined by applying the continuity equation at the four nodes. Figure
6.5 shows the magnitudes and directions of the flow rates in the network that satisfy the
continuity equation at all the nodes based on the assumed flow in pipe AB.
A B
100 60
l/s 50 l/s
l/s
10
50 l/s
l/s C
40
l/s 30
D l/s
10
l/s
Figure 6.5. Assumed flow rates in the network
Step 2: The second step is to determine the resulting velocities and friction head losses
in the different pipes. Figure 6.6 shows the sheet developed for the example. The given
information about the network pipes, i.e. their lengths and diameters, are entered on the
left side of the sheet. Values of the gravity acceleration (gc) and friction factor (f) are
entered at the top left-side of the sheet.
The cross-sectional areas of the different pipes are calculated in column E so as to

determine the fluid velocities in column G. Based on these velocities the friction losses
are determined in column H. Table 6.1 reveals the formulae in columns G and H that
calculate the pipe velocities and friction head losses. To reveal the formulae in the
Excel sheet, press the Ctrl key and the tilde (~) key together. To go back to the usual
form, press the two keys together again. Note that in the formula for the friction head
loss (hf), the head loss is multiplied by SIGN(Q) so as to account for the direction of
the flow in the pipe – which can change during the iteration process.
Table 6.1. Formulae in columns G to K

V hf hf/Q ΔQ Qnew
=F5/E5 =f*(B5/C5)* G5^2/(2*gc)*SIGN(F5) =H5/F5 =-SUM(H5:H8)/SUM(I5:I8)/2 =F5+$J$5
=F6/E6 =f*(B6/C6)* G6^2/(2*gc)*SIGN(F6) =H6/F6 =F6+$J$5
Step 3: The third step is to determine ΔQ according to Equation (6.27) which requires
the determination of hf/Q. Table 6.1 also reveals the formulae that calculate hf/Q in
column I and ΔQ in column J. Since ΔQ is the same for all pipes, it is calculated for
pipe AB only but used for the other pipes as well as shown in the last column of Table
6.1. As shown in Figure 6.6, the value obtained for this first iteration is ΔQ =0.0157.
Step 4: New values for the flow rates in the four pipes according to Equation (6.19).
Column K in the sheet shown in Figure 6.6 shows the new values of the flow rates.
Step 5: The adjusted flow rates obtained in step 4 can now be used as initial values for
the second iteration. Table 6.2 shows the calculations in the second iteration which are
based on the new values of the flow rates following the same steps described for the
first iteration. Note that the ΔQ has dropped to 0.0047. Table 6.2 also shows the new
values of the flow rates after the second iteration. These new values of the flow rates
can now be used as initial values for the third iteration. As shown in Table 6.3,
calculations of the third iteration yield a very small value for ΔQ of 0.0001.
Table 6.2. Results of the second iteration

Pipe Q V hf hf/Q ΔQ Qnew
AB 0.066 2.092834 5.5809832 84.88403 0.004707 0.070
BC 0.006 0.325289 0.2696557 46.91027 0.010
CD -0.024 -1.37236 -7.999415 329.8501 -0.020
AD -0.034 -1.09026 -3.029247 88.44087 -0.030
Table 6.3. Results of the third iteration

Pipe Q V hf hf/Q ΔQ Qnew
AB 0.070 2.242649 6.4086065 90.96041 0.000144 0.071
BC 0.010 0.591626 0.8920015 85.31902 0.011
CD -0.020 -1.10603 -5.195783 265.8355 -0.019
AD -0.030 -0.94045 -2.25394 76.2881 -0.029
Figure 6.7 shows the variation of ΔQ with the iteration number. The figure indicates
that the change in the flow rates after the third iteration is small and, therefore, the flow
rates obtained after the third iteration can be taken as the final solution. The plus sign
with the flows in AB and BC means that their direction is clockwise, while the minus
sign with the flows in CD and AD means that their direction is counter-clockwise. Note
that the flow in pipe BC was initially assumed to be in the counter-clockwise direction,
but the final solution showed that it is clockwise.
0.02
0.016
0.012
ΔQ
0.008
0.004
0
0 2 4 6
Iteration number
Figure 6.7. Variation of ΔQ with iteration number
This example shows the basic steps of the H-C method that also apply for more
complex networks. However, in a network with multiple loops, the nodes that are
common to two or more loops will have different ΔQ‟s. For example, pipe CE in Figure
6.1.a is common to loop 1 and loop 2. Suppose, the two values of ΔQ are ΔQ1 for loop
1 and ΔQ2 for loop 2. The effective ΔQ for pipe CE is calculated from [2]:
QCE ,1  Q1  Q2 Loop 1 (6.28)
QCE , 2  Q2  Q1 Loop 2 (6.29)
The two values of ΔQ have the same magnitude but opposite signs. The following
example, which is based on Example 5.2 in Nalluri and Feather [4], illustrates the
procedure.
Example 6.3. Analysis of a two-loop network by the Hardy-Cross method

Figure 6.8 shows a network that consists of two pipe loops with one source and five
discharge (load) junctions. The valve in pipe BC is partially closed, which produces a
local head loss of 10.0 V2 /2g.
A B C
Valve
1 2
D
E
Figure 6.8. The two-loop pipe network of Example 6.3
The following table gives the length and diameter of each pipe in the network. The
roughness of all pipes is 0.06 mm. It is required to determine the flows of water in the
network pipes.
Pipe AB BC CD DE EF AF BE
Length (m) 500 400 200 400 600 300 200
Diameter (mm) 250 150 100 150 200 250 150
Analytical model
Figure 6.9 shows the initially assumed flows in each pipe of the two loops forming the
network. These flows may change in both magnitude and directions in the final answer.
We need to guess the flow rate in only one pipe in each loop since the flow rates in the
other pipes can be determined from the continuity equation.
60
200 A 120 B 50 C
40
Valve
1 10 2 10
80
D
E 20 30
F 40 30
40
Figure 6.9. Guessed flows and sign convention for the pipe network of Example 6.3
Assume that the flow rates in pipe AB (QAB) and pipe BC (QBC) are guessed. Then the
flow rates in the other pipes are determined as follows:
Loop 1:
QBE  Q AB  QB  QBC  (6.30)
Q AF  Q A  Q AB  (6.31)
QEF  Q AF  QF  (6.32)
Loop 2:
QCD  QBC  QC  (6.33)
QBE  Q AB  QB  QBC  (6.34)
QDE  QD  QCD  (6.35)
In determining the value of ∆Q from Equation (6.27), the friction losses in all the pipes
are calculated from Equation (1.21) except pipe BC for which the equation takes the
following form to allow for the loss in the valve:
L V2 V2
h f , BC  f  10 (6.36)
D 2g 2g
The Excel sheet

Figure 6.10 shows the Excel sheet developed for this problem. As usual, the left side of
the sheet shows the problem data that consist of the pipes lengths (L) and diameters
(D), the roughness factor of the pipe material (ε), and the viscosity of the fluid (visc).
Column E shows the flow rates in the pipes; two of which are guessed (QAB and QBC)
and the rest are calculated by applying the continuity equation at the different nodes.
Based on these initial flow rates, the sheet determines the Reynolds numbers, friction
factors, and friction losses. The friction factors are determined by using the Swamee-
Jain formula (Equation (1.25)). The formula bar reveals the formula that applies
Equation (6.36). The flow rate adjustment (ΔQ) in each loop is determined according to
Equation (6.27). For loop 1, ΔQ =-0.0057251 and for loop 2, ΔQ =-0.011222. For pipe
BE, which is common to both loops, the effective values of ΔQ have the same
magnitude for both loops, which is 0.0054965. The new flow rates in the different pipes
are obtained according to Equation (6.19).
Figure 6.10. Excel sheet developed for solving Example 6.3 by the Hardy Cross method
To avoid repetition of the whole procedure for the following iterations, the new Q
values in column (N) can be copied and pasted in column (E) that stores the old values.
(Make sure that you use the "Paste Values" option). This procedure is to be repeated
until ΔQ became sufficiently small so that there was no significant change in Q in all
the network pipes. Figure 6.11 shows the result after 5 iterations when ΔQ dropped to
small values in both loops. The minus sign before the flow rates in pipes EF, AF, CD,
DE, and BE means that their directions are counter-clockwise relative to the respective
loop. Note that, the flow in pipe BE, which is common to the two loops, has the same
magnitude (0.01645 m3 /s) but opposite directions. While its direction is clockwise (+)
in loop 1, it becomes counter-clockwise (-) in loop 2. The present pipe flows, in m3 /s,
are compared with those given by Nalluri and Feather [4] in Table 6.4, which shows a
good agreement between the two solutions.
Figure 6.11. The Excel sheet for Example 6.3 with the adjusted flows after five
iterations
Table 6.4. Comparison of the present pipe flow rates with those obtained by Nalluri and
Feather [4]
Pipe Nalluri and Present
Feather [4] solution
AB 0.11152 0.11150
BC 0.03505 0.03505
CD 0.00495 0.00495
DE 0.03495 0.03495
EF 0.04848 0.04850
AF 0.08848 0.08850
BE 0.01648 0.01645
6.3. Analyses of looped pipe -networks with Excel-Solver

Solver offers an alternative method for pipe-network analyses with Excel that is easier
to apply, yet more general than the H-C method discussed in the previous section [5,6].
The method utilises the multi-variable iterative capability of Solver that can take into
consideration multiple constraints on the iterative process. Like the H-C method, the
procedure starts with guessed pipe flows that satisfy the continuity equation. Since
these initially guessed flows are unlikely to satisfy the energy equation, Solver is used
to adjust the pipe flow rates so that the energy equation (Equation (6.4)) is satisfied in
all loops of the network. To illustrate this method, consider the two-looped network in
Example 6.3. The first two steps in the solution procedure are the same as those of the
H-C method, viz.:
1. Guess the magnitude and direction of flow rate in pipe AB (loop 1) and pipe BC
(loop 2) and calculate the flow rates in the other pipes in the network by applying
mass conservation at the nodes (junctions) according to Equations (6.31) to
(6.36).
2. Calculate the friction loss in each pipe of the network from the Equation (6.5)
and Equation (6.37) and calculate the net friction loss in each loop according to
the assumed flow direction in the relevant pipes. More likely, Equation (6.4) will
not be satisfied with the initially guessed flows.
The last three steps of the Hardy Cross method are replaced by using Solver to adjust
the flow rates in pipes AB and BC so as to satisfy the energy equation (Equation (6.4))
by imposing the following constraints on the iterative process:
h f , AB  h f , BE  h f , EF  h f , AF   Loop 1 (6.37)
h f , BC  h f ,CD  h f , DE  h f , BE   Loop 2 (6.38)
Where ││ refers to the absolute value and ε is a prescribed positive tolerance that is
acceptably small. Note that only the flow rates in pipes AB and BC need to be adjusted
by Solver since the other pipe flows are adjusted automatically by applying Equations
(6.30) to (6.35) in the main Excel sheet. Therefore, this method ensures that continuity
is satisfied in a better way than the H-C method that updates each flow rate separately.
Since the absolute value of the error is considered, the positive direction for friction
doesn‟t have to be the same for all the loops, i.e., it can be clockwise or anti-clockwise
for different loops.
Figure 6.12 show the Excel sheet developed for solving Example 6.3 with this method.
The problems data (pipe lengths and diameters, roughness factor, and water viscosity)
are shown at the left part of the sheet. The two values for the flow rates in pipe AB
(Q_AB) and pipe BC (Q_BC) are guessed values. The flow rates in the other pipes
(Q_CD – Q_BE) are calculated according to Equations (6.30) to (6.35).

The sheet then calculates the velocity (Q/A) and Reynolds number (Re) for each pipe.
Based on these values, the sheet determines the friction factor (f) and friction loss (hf)
in the pipes. Finally, the sheet calculates the balance of friction losses in the two loops
as required by Equations (6.37) and (6.38). The formula bar shows how the balance of
friction losses in loop 1 is calculated. Note that the formula takes into consideration that
the flow may not be in the suggested direction or may change its direction during the
solution process. Figure 6.12 shows that the imbalances in friction losses in the two
loops with the initially guessed flows are not close to zero. Solver can now be used to
find the values of Q_AB and Q_BC that make the friction imbalances in the two loops
become less than a specified value. Figure 6.13 shows the required set-up for this task.
Note that the Set Target Cell has been left blank. With this set-up, Solver will search for
the values of Q_AB and Q_BC that reduce the absolute value of the friction balances in
the two loops below 0.1. The sheet automatically calculates values of the flow rates in
the other pipes that satisfy continuity. Figure 6.14 shows the solution found by Solver
on the Excel sheet. Note that the imbalances in friction losses in the two loops have
been reduced to 0.1 or less. Solver‟s solution shown in Figure 6.14 agrees closely with
that given by Nalluri and Feather [4] and the solution obtained by the H-C method
shown in Table 6.4.
Example 6.4. Analysis of a three -loop network by Excel-Solver

Determine the flow rates in all the pipes in the network shown in Figure 6.15. If the
pressure head at point A is 40 m, find the pressure head at D (which might represent a
fire-demand, for example).
The case in this example was used by Benjamin [7] to illustrate the Hardy-Cross
method and his solution will be used to verify the present solution with Excel-Solver.
60 L/s
B
A
30 L/s
C E
D
F G H
15 L/s 15 L/s
Figure 6.15. Pipe-network for Example 6.4
Pipe Length (m) Diameter (m)

AB 250 0.3
BE 100 0.2
DE 125 0.2
CD 125 0.2
AC 100 0.15
DG 100 0.25
FG 125 0.15
CF 100 0.1
EH 100 0.25
GH 125 0.1
Analytical model
Following the procedure described above, the flow rate in one of the pipes in each loop
is guessed and the flow rates in the other pipes are calculated from relevant mass
balance equations. In what follows, it is assumed that the flows in pipes AB, CD, and
DE are guessed. Then flows in the other seven pipes are determined as follows:
QBE  Q AB (6.39)
Q AC  Q A  Q AB (6.40)
QDG  QDE  QCD  QD (6.41)
QFG  QCF  QF (6.42)
QCF  Q AC  QCD (6.43)
QEH  QBE  QDE (6.44)
QGH  QH  QEH (6.45)
Figure 6.16 shows the initial values and assumed directions of the pipe flows.
60 L/s
10 L/s B
A
50 L/s 1 30 L/s 10 L/s

30 L/s 10 L/s
C E
D
20 L/s 2 10 L/s 3 0 L/s
F
5 L/s G
15 L/s H
15 L/s
15 L/s
Figure 6.16. Pipe-network for Example 6.4 with initial flows

Energy constraints:
h f , AB  h f , BE  h f , DE  h f , AC  h f ,CD  0 Loop 1 (6.46)
h f ,CD  h f , DGg  h f ,CF  h f , FG  0 Loop 2 (6.47)
 h f , DE  h f , EH  h f , DG  h f ,GH  0 Loop 3 (6.48)
Figure 6.17 shows the Excel sheet developed for this example. The first flow rates are
guessed flow rates for pipes AB, DE, and CD. The flow rates for the other seven pipes
are determined according to Equation (6.39) to (6.45). Based on the guessed flow rates,
the sheet calculates the Reynolds number (Re), friction factor (f), and friction loss (hf)
in each pipe. Figure 6.17 shows that the guessed flow rates do not satisfy the
requirements of the energy equation. Solver can now be used to adjust the three guessed
flow rates in pipes AB, DE, and CD until Equations (6.46) to (6.48) are satisfied and
Figure 6.18 shows Solver set-up for this problem. Note that Solver is only required to
adjust the flow in the three pipes with the guessed flows and that the constraints used
for Solver solution represent Equations (6.46), (6.47) and (6.48). Figure 6.19 shows the
final results obtained by Solver. Table 6.5, which compares these results with that given
by Benjamin [7], proves their proximity to the reference solution.
Figure 6.17. Excel sheet for Example 6.4

Table 6.5. Comparison of Solver solution for pipe flow rates in m3 /s with the results of
Benjamin [7]
Pipe Benjamin [7] Present Pipe Benjamin [7] Present
AB 0.0098 0.00978 DG 0.0082 0.00808
BE 0.0098 0.00978 FG 0.0035 0.00363
ED 0.0065 0.00649 CF 0.0185 0.01863
DC 0.0317 0.03159 EH 0.0033 0.00329
AC 0.0502 0.05022 HG 0.0117 0.01171
6.4. Analyses of branched and mixed pipe -networks with Excel-Solver

The procedure for analysing branched and mixed pipe-network with Excel and Solver is
basically the same as that applied for the analysis of looped-networks. Accordingly, the
basic equations for the analysis of branched and mixed networks are the continuity and
energy equations supplemented by an equation for determining the frictional losses.
The continuity equation can also be applied to the whole network as well as to any of its
pipe junctions. The energy equation requires that the friction loss between any two
points (nodes) A and B in the network to be balanced by the difference in potential and
kinetic heads between these two points in addition to the head produced by any booster
pump along the line. Therefore, the energy equation takes the following form:
M
V A2  V B2
1
h f  hP  z A  z B  
2g
(6.49)
Where M is the total number of pipes between the two points A and B. If there is no
booster pump along the line, then the equation simplifies to:
M
V A2  V B2
1
h f  z A  z B  
2g
(6.50)
Equation (6.49) or Equation (6.50) can be used to generate the required number of
constraints such that the problem can be solved with Solver. The following example,
which is based on Example 5.4 in Nalluri and Feather [4], illustrates the procedure.
Example 6.5. Branched network connecting multi-reservoirs

Determine the gravity-driven discharges in the pipes of the branched network shown in
Figure 6.20 neglecting minor losses. The network carries water and the data of the pipes
are given below. The elevations at nodes A, B, C, and D are shown in the figure. The
following table shows the pipe data.
200 m
A 120 m
B
100 m
C
J
75 m
D
Figure 6.20. Branched network for Example 6.5

Pipe Length (m) Diameter (mm)

AJ 10 000 450
BJ 2 000 350
CJ 3 000 300
DJ 3 000 250
Solution
Nalluri and Feather [4] solved the problem by using the quantity balance method
('nodal' method). Their solution started with an assumed value for the head at joint J
based on which the flow rate and head loss in each pipe were calculated. The guessed
heat at joint J was then adjusted successively in order to satisfy the energy equation. It
is more convenient with Excel-Solver to follow the same procedure used in the previous
section by assuming the flow rates in each pipe, determining the friction losses, and
then using Solver to adjust the flow rates until the energy equation is satisfied within a
specified tolerance. Figure 6.21 shows the magnitudes (litre/s) and directions of the
flow rates initially assumed in each pipe.
200 m
A 100 120 m
30 B
100 m
20 C
J
50 75 m
D
Figure 6.21. Assumed flows in the network
Three of the four flows at junction J have been assumed (AJ, BJ, and CJ), while the
fourth (DJ) is left to satisfy continuity:
QDJ  Q AJ  QBJ  QCJ (6.51)
Since there are no booster pumps in the network, the energy equation (6.50) applies.
For pipes that have constant cross-sections, the velocities at both ends of the pipe are
the same and, therefore, Equation (6.50) can be utilized to produce the following
equations:
h f , AJ  h f , BJ  z A  z B   0 (6.52)
h f , AJ  h f ,CJ  z A  z C   0 (6.53)
h f , AJ  h f , DJ  z A  z D   0 (6.54)
Equations (6.52) – (6.54) will be used as constraints for Solver solution. Once the
friction losses in the different pipes are known, the elevation at the junction can be
determined from any of the following equations:
z J  z A  h f , AJ (6.55)
z J  z B  h f , BJ (6.56)
z J  z C  h f ,CJ (6.57)
z J  z D  h f , DJ (6.58)
The Excel sheet

Figure 6.22 shows the Excel sheet developed for this example. The flow rates in the
three coloured or shaded cells are assumed, but the flow rate in pipe DJ is calculated
from the continuity equation. Note that the formula bar shows the formula in cell E19.
Based on the assumed pipe flow rates, the sheet calculates the relevant Reynolds
numbers, friction factors, and friction losses.
The top four cells in column J (i.e. J15, J16, J17, and J18) show the height at the
junction J as calculated from Equations (6.55)-(6.58). These cells show different values
because the assumed flow rates in the pipes are not correct. This is also indicated by the
bottom three cells in column J that show the imbalances in the energy equations,
Equations (6.52)-(6.54). These imbalances should be zeros but, since the assumed flow
rates do not satisfy the energy equations, the values shown in the sheet are relatively
large. Solver can now be used to find the correct flow rates (their magnitudes and
directions) in the three pipes AJ, BJ, and CJ that satisfy the energy equation. Solver set-
up is shown in Figure 6.23, which requires Solver to adjust the flow rates in the top
three pipes such that the last three entries in column J, i.e. J20, J21, and J22, are close to
zero. Note that the flow in pipe DJ is automatically determined by the continuity
equation.
Figure 6.24 shows the solution found by Solver. This solution satisfies the energy
equation as shown by the small values of the three entries in cell J20, J21, and J22. Also
note that the resulting head losses in all four pipes now yield the same value for the
elevation at joint J, which is 125.5 m. Table 6.6, which compares the present solution
with that given by Nalluri and Feather [4], shows a close agreement between the two
solutions.

Table 6.6. Comparison of the pipe flow rates in m3 /s obtained by Solver with those
given by Nalluri and Feather [4]
Pipe Nalluri and
Present analysis
Feather. [4]
AJ 0.344 0.344780
BJ 0.105 0.104696
CJ 0.127 0.127590
DJ 0.112 0.112606
To illustrate the application of the Excel-Solver method for the analyses of general pipe
networks that consist of both pipe branches and loops, consider the pipe network shown
in Figure 6.25. This network, which is formed by four loops and two branched pipes
and has two sources and six discharge points, was analysed by Rivas et al [6] by using
Excel and Solver. The network has sixteen pipes and, therefore, there are sixteen
unknown flow rates. Applying the continuity equation at the eleven nodes in the
network provides eleven linear equations. Four non-linear equations can be formed by
applying the energy equations over the four loops in the network. The fifth non-linear
equation can be formed by applying the energy equation between the two tanks A and
M, for which the elevations are known. The following example, which is based on data
provided by Rivas et al [6], demonstrates the procedure.
123 m
E
B C D
2 3
G F
1
H
4
J I K
L
112 m
Figure 6.25. Pipe network (adapted from Rivas et al [6])

Example 6.6. Analysis of a pipe network with loops and branches

Determine the flow rate in each pipe in the network shown in Figure 6.25 using the
following data:
Pipe size Pipe size

Pipe Diameter Length Pipe Diameter Length
(mm) (m) (mm) (m)
AB 250 200 CD 200 180
BJ 150 400 LM 250 140
IJ 200 300 KL 200 360
HI 300 190 FK 150 200
CH 80 210 FG 150 340
BC 200 300 GL 80 180
GH 150 160 EF 150 180
DG 300 200 DE 200 345
All pipes are made of steel with a roughness of 0.3 mm. The network distributes water
and the demand in litres per second at the different nodes is shown below:
Node Demand (l/s)

C 100
D 30
E 30
I 20
F 10
K 10
Analytical model
As in the previous two sections, the solution procedure starts by assuming the flows in
some pipes and calculating those in the other pipes by satisfying the continuity equation
around each node. In this case, it is enough to assume the flow in five pipes, AB, BC,
GH, KL and DE (i.e. QAB, QBC, QGH , QKL , and QDE ). The remaining flows are calculated
from the continuity equation as follows:
QBJ  Q AB  QBC (6.59)
QIJ  QBJ (6.60)
QHI  QIJ  Q I (6.61)
QCH  QHI  QGH (6.62)

QDG  QCD  QDE  QD (6.63)
QCD  QBC  QC  QCH (6.64)
QLM  QKL  QGL (6.65)
QFK  QKL  QK (6.66)
QFG  QEF  QFK  QF (6.67)
QGL  QGH  QFG  QDG (6.68)
QEF  QDE  QE (6.69)
Four additional equations can be formed by applying the energy equation to the four
loops in the network shown in Figure 6.25. Since there are no pumps in the system, the
closed clockwise (or counter-clockwise) summation of the friction losses in each loop
must be zero:
 h f , HI   
 h f ,CH  h f , BC  h f , BJ  h f , IJ  0 Loop 1 (6.70)
 h f , DG  
 h f ,CD   h f ,CH  h f ,GH  0 Loop 2 (6.71)
h f ,15   
 h f ,16   h f ,8  h f ,13  0 Loop 3 (6.72)
 h f , FK   
 h f , FG  h f ,GL  h f , KL  0 Loop 4 (6.73)
The fifth needed friction equation can be formed by applying the energy equation along
any route that connects points A and M. Selecting the route given by A-B-J-I-H-G-L-
M:
h f , AB 
 h f , BJ  h f , IJ  h f , HI  h f ,GH  h f ,GL  h f , LM  z A  z M   0 (6.74)
Since the guessed flows are unlikely to satisfy the energy equations (6.70) to (6.74),
they must be adjusted by using Solver such that the above conditions are approximately
satisfied.
The Excel sheet developed for this example is shown in Figure 6.26. The first columns
on the left side of the sheet show the given network data that include the pipe roughness
(ε), the kinematic viscosity of water (visc), the discharges at the six demand nodes, and
the pipes diameters and lengths. Initial guesses are made for the flow rates of five pipes,
one pipe in each loop. These are shown in the figure by the coloured or shaded cells.
The flow rates in the other eleven pipes are then calculated from the continuity
equations (6.59 – 6.69).
The calculations in the sheet determine the velocity in each pipe followed by the
Reynolds number, friction factor, and friction loss. Based on the estimated friction
losses, the sheet then determines the imbalance in the energy equation in the four loops
(cells R2:R6) and along the route between points A and M. The last column shows the
head obtained at each node (U2:U12). The specified and calculated pipe flows
automatically satisfy continuity but, as the magnitudes of the friction imbalances show,
they do not satisfy the energy constraint (Equations (6.70) – (6.74)). Also, note that
some nodes acquired negative elevations based on the assumed flow rates. Solver can
now be used to search for the flows in the five pipes that satisfy the specified energy
constraints, which will also give correct elevations.
Figure 6.27 shows Solver set-up for this example which requires Solver to perform the
iteration with Equations (6.70) to (6.74) as constraints. Note that the "Target Cell" has
been left blank while the objective of the iterative solution has been set to a "Value of
0". The "Changing Cells" correspond to the five pipe flow rates that Solver has to
adjust. Figure 6.28 shows the solution determined by Solver that satisfied the five
constraints with a specified tolerance of 0.01. The solutions obtained for the flow rates
in the sixteen pipes and for the heads at the eleven nodes are compared with those given
by Rivas et al [6] in Figure 6.29 and Figure 6.30. The figures show excellent
agreements with their results.
0.2
Rivas [6]
0.15
Present
0.1
Flow rate (m3/s)
0.05
0
0 2 4 6 8 10 12 14 16 18
-0.05
-0.1
-0.15
Figure 6.29. Comparison of the present results with those given by Rivas et al [6] for
the pipe flow rates
120
110
Elevation (m)
100
90
Rivas [6]
Present
80
0 2 4 6 8 10 12
Figure 6.30. Comparison of the present results with those given by Rivas et al [6] for
the nodal elevations
6.5. Closure
This chapter demonstrates the use Excel and Solver as a platform for the analyses of
pipe networks. The solution procedure allows the students to focus more on applying
the principles of mass and energy conservation than on the complications of numerical
solution methods such as the Hardy-Cross method. An important advantage of the
Excel-Solver method for pipe-network analyses is that it can be applied to looped-
networks as well branched and mixed networks. The analyses can also be for pipe
networks with specified heads as well as specified discharges. Apart from its generality,
the method ensures that continuity is automatically satisfied as the iterative process
approaches the correct values.
All the analyses were performed with the GRG Nonlinear method of Solver without
using the automatic-scaling option. The number of iterations increased from 5 for the
two-loop network to 55 for the four-loop network.
References
[1] W.S. Janna, Design of Fluid Thermal Systems, third edition, CENGAGE Learning,
2011
[2] A. N. El–Bahrawy, A spreadsheet teaching tool for analysis of pipe networks,
Engineering Journal of University of Qatar, vol. 10, 1997, p. 33 – 50, the internet:
https://www.researchgate.net/profile/Aly_El-Bahrawy/contributions, last accessed
24/1/2018.
[3] TM‟sChannel, Pipe network analysis in Excel using Hardy Cross method, Video,
available at: https://www.youtube.com/watch?v=M8f1FNgeq7o, last accessed
20/6/2017.
[4] C. Nalluri and R.E. Feather, Civil Engineering Hydraulics, Fourth Edition,
Blackwell Publishing, 2008.
[5] A. M. Elfeki, Pipe network analysis example with Excel Solver, Researchgate,
available at: https://www.researchgate.net/publication/258051650_Pipe_Network_
Analysis_Example_with_Excel_Solver, last accessed 20/6/2017.
[6] A. Rivas, T. Gomez-Acebo, J.C. Ramos, The application of spreadsheets to the

analysis and optimization of systems and processes in the teaching of hydraulic
and thermal engineering, DOI%2010.1002_cae.20085
[7] M.M. Benjamin, Analysis of complex pipe networks with multiple loops and inlets
and outlets, the internet: http://faculty.washington.edu/markbenj/CEE342/
Abbreviated%20Hardy-Cross.pdf, last accessed 20/6/2017.
Chapter 7
Piping systems can cost more than 20% of the total cost of a chemical plant and even
more than that in the case of district heating or cooling systems and oil or gas pipelines
[1,2]. Since piping systems are not only costly to install but also to operate, economic
optimisation analyses form an important step in the design process of these systems. A
number of analytical formulae were developed in the past for determining the economic
pipeline diameter [3,4]. The development of these formulae involves arduous calculus
manipulations even if the fluid flow is known to be laminar. For turbulent flows, the
analytical formulae may also require the use of dimensionless charts in order to avoid
iterative solutions. This chapter shows how Excel and its iterative tool, Goal Seek, can
be used to make the analytical method more convenient. The chapter then presents an
Excel-based approach for pipeline diameter optimisation by using Solver that can be
used with both laminar and turbulent flows.
References
[1] G.K. Roy, Prediction of optimum economic pipe diameter by nomograph, the
Journal of the Institute of Engineers (India), Vol 68, part CH 3, June 1988, 83 - 85.
[2] S.B. Genic, B.M. Jacimovic, and V.B. Benic, Economic optimization of pipe
diameter for complete turbulence. Energy and Buildings, 45 (2012), 335-338.
[3] R.P. Genereaux, Fluid-flow design methods, Industrial and Engineering
Chemsitry, 29 (1937) 385-388.
[4] T.A. Akintola and S.O. Giwa, Optimum pipe size selection for turbulent flow,
Leonardo Journal of Sciences, Issue 14, January-June 2009, 112- 123.
[5] W.S. Janna, Design of Fluid Thermal Systems, third edition, CENGAGE Learning,
2011
[6] USP United States Plastic Corporation, http://www.usplastic.com/ Last accessed
January 31, 2018.
[7] British Metals LLC. Brismet, http://brismet.com/ Last accessed January 31, 2018.
[8] Copper tube:www.camlee.com/wp-content/uploads/2013/06/CPRICE.pdf, Last
accessed January 31, 2018.
[9] Wikipedia contributors, https://en.wikipedia.org/wiki/Hazen%E2%80%93
Williamsequation, Last accessed January 31, 2018.
[10] D. Alciator and W.S. Janna, Modified pipe friction diagrams that eliminate trial-
and-error solutions, Proc. Of the National Fluid Dynamics Congress, Part 2, July,
1988, available at: https://www.engr.colostate.edu/~dga/dga/papers/pipefriction.
pdf. Last accessed January 31, 2018.
Chapter 8
Considering the large initial and running costs of water and gas pipe-networks, the
designer may not simply seek to minimise their total costs, but to maximise their
economic benefits. Moreover, the design of large pipe networks is not based on
economic considerations alone, but also involves other important factors such as the
network‟s reliability and environmental impact. Even the network layout itself may
have to be determined by the design optimisation process. Since the scope of pipe-
network optimisation is both broad and complex, most researchers focussed on certain
aspects of the process. The majority of the published work dealt with “least-cost”
optimisation analyses of water pipe-networks with pre-determined layouts. This chapter
aims to demonstrate the usefulness of Excel and Solver for such analyses for both
gravity-fed and pump-fed pipe-networks. The effect of the friction formula on the
optimised solution is investigated by using three models to calculate the major pipe
friction losses, which are the Darcy-Weisbach equation with the Swamee-Jain equation,
the Darcy-Weisbach equation with the Colebrook-White equation, and the Hazen-
Williams equation. The analyses also investigate the effect of Solver‟s options on the
optimisation results. Compared to hydraulic analyses of pipe-network, for which a
number of computer models are available, pipe-network optimisation is still an active
area of research. A brief review of the research efforts in this field is given first.
References
[1] D. F. Yates, A. B. Templeman and T. B. Boffey, The computational complexity of
the problem of determining least capital cost designs for water supply networks,
Engineering optimization, 1984, vol. 7, no. 2, pp. 142– 155.
[2] E. Alperovits and U. Shamir, Design of optimal water distribution systems. Water
Res. Research, 13(6), (1977), pp 885-900.
[3] G.A. Walters, A Review of Pipe Network Optimization Techniques. In: Coulbeck
B., Evans E.P. (eds) Pipeline Systems. Fluid Mechanics and Its Applications, vol
7. Springer, Dordrecht, (1992), pp 3-13
[4] M. Shrivastava, V. Prasad, and R. Khare, Optimization techniques for water
supply network: a critical review, International Journal of Mechanical
Engineering and Technology (IJMET), Volume 5, Issue 9, September (2014), pp.
417-426
[5] G. C. Dandy, A. R. Simpson, and L.J. Murphy, A review of pipe network
optimisation techniques, Presented at Watercomp '93, The 2nd Australasian
Conference on Technical Computing in the Water Industry, 30 March - 1 April
1993, Melbourne, Australia
[6] I.C. Goulter, B.M Lussier, and D.R. Morgan, Implications of head loss path choice
in the optimization of water distribution network, Water Res. Research, 1986,
22(5), pp. 819-822.
[7] A. Kessler, and U. Shamir 1989. Analysis of the linear programming gradient
method for optimal design of water supply networks. Water Res. Research, 25(7),
pp. 1469-1480.
[8] G. Eiger, U. Shamir, and A. Ben-Tal, Optimal design of water distribution

networks. 1994, Water Res. Research, 30(9), pp. 26637-2646
[9] O. Fujiwara, and D.B. Khang, A Two-Phase Decomposition Method for Optimal
Design of Looped Water Distribution Networks, Water Res. Research, Vol. 26,
No. 4, 1990, pp. 539-549.
[10] I. Sârbu and F. Kalmár, Optimization of looped water supply networks, Periodica
Polytechnica ser. Mech. Eng. Vol. 46, No. 1, pp. 75–90 (2002)
[11] P. Bales, B. Geißler, O. Kolb, J. Lang, A. Martin, and A. Morsi. Comparison of
Nonlinear and Linear Optimization of Transient Gas Net-works. Technical
Report Preprint Nr. 2552, Darmstadt University of Technology, 2008.
[12] D. Dobersek, D. Goricanec, Optimisation of tree path pipe network with nonlinear
optimisation method, Applied Thermal Engineering, (2009), 29(8):1584-1591
[13] B. Djebedjian, A. Yaseen, and M. Abou Rayan, Optimization of large water
distribution network design using genetic algorithms, Tenth International Water
Technology Conference, IWTC10, 2006, Alexandria, Egypt
[14] L.A. Rossman, EPANET 2: users manual, US Environmental Protection Agency,
Ofﬁce of Research and Development, National Risk Management Research
Laboratory, 1994.
[15] X. Zhou, D.Y. Gao, and A.R. Simpson, Optimal Design of Water Distribution
Networks by Discrete State Transition Algorithm, Engineering Optimization,
48(4):1-26 · April 2015
[16] D.A. Savic and G.A. Walters, Genetic algorithms for least-cost design of water
distribution networks. Water Res. Planning & Management, 1997, 123(2), 67-77.
[17] M. Čistý, Z. Bajtek, Optimal design of water distribution systems by a
combination of stochastic algorithms and mathematical programming, Slovak
Journal of Civil Engineering, 2008/4, 1-7
[18] A.J. Abebe and D.P. Solomatine, Application of global optimization to the design
of pipe networks, Proc. 3rd International Conference on Hydroinformatics,
Copenhagen, August 1998. Balkema, Rotterdam.
[19] B.S. Sil, A. Kumar, P. Saikia, P.J. Bui, P. Banerjee, Computation of Least Cost
Pipe Network –An Alternate Method, International Journal Of Computational
Engineering Research (ijceronline.com), 2013, Vol. 3 Issue. 2
[20] D. Brkic, Spreadsheet-Based Pipe Networks Analysis for Teaching and Learning
Purpose, Spreadsheets in Education (eJSiE): Vol. 9: Iss. 2, Article 4, 2016.
Available at: http://epublications.bond.edu.au/ejsie/vol9/iss2/4
[21] A. Rivas, T. Gomez-Acebo, J.C. Ramos, The application of spreadsheets to the
analysis and optimization of systems and processes in the teaching of hydraulic
and thermal engineering, DOI%2010.1002_cae.20085
Chapter 9
The energy of the gas-turbine‟s exhaust can be recovered and used for improving the
efficiency and economic benefit of the plant in a number of ways. One such option is
the air-bottoming system in which the exhaust-gas energy is used as the high-
temperature source for another gas turbine. Although the quantity of the exhaust-gas
energy is significant, its temperature and quality are relatively low. Therefore, the
economic feasibility of the air-bottoming system depends on the effectiveness of the
heat exchanger used between the top and bottom cycles. The efficiency of the system
also depends on the pressure ratio and air flow rate in the bottom cycle. This chapter
presents thermodynamic and economic optimisation analyses of the air-bottoming
cycle. For thermodynamic optimisation, Solver is used to maximise the thermal
efficiency while for economic optimisation, it is used to minimise the total annualised
cost of the system. The analyses apply the exact variable specific-heat method by using
the Thermax1 add-in for determining the thermodynamic properties of the working
fluid. The analyses compare two methods for modelling the heat-exchanger which is a
key component in the air-bottoming system.
References
[1] M.A Korobitsyn,. New and advanced energy conversion technologies. Analysis
of cogeneration, combined and integrated cycles, ISBN 90 365 1107 0, 1998.
[2] M. Saghaﬁfar, A. Poullikkas, Thermo–economic optimization of air bottoming
cycles, Journal of Power Technologies, 95(3) (2015) 211–220
[3] Y.A. Cengel, and M.A. Boles, Thermodynamics an Engineering Approach,
McGraw-Hill, 7th Edition, 2007
[4] L. Pierobon & F. Haglind, Design and optimization of air bottoming cycles for
waste heat recovery in off-shore platforms. Applied Energy, Vol. 118, 2014, 156–
165.
[5] A. Bejan, G. Tsatsaronis, M. Moran, Thermal Design & Optimisation, John
Wiley & Sons, 1996.
[6] Y.A. Cengel and A.J. Ghajar. Heat and Mass Transfer: Fundamentals and
Applications. McGraw Hill, 2011
[7] D. Czaja, T. Chmielniak, S. Lepszy, Selection of Gas Turbine Air Bottoming
Cycle for Polish compressor Stations, Journal of Power Technologies , Vol. 93,
2013, No. 2, 67–77
[8] F.C. Knopf, Modeling, analysis and optimization of process and energy systems,
John Wiley & Sons, 2012.
Chapter 10
Gas turbines are the preferred option for providing small and medium-size industries
with power and for meeting the peak-load demands on electric utilities than steam-
turbines and combined-cycle systems. Gas turbines installed for these purposes are
usually of the open-cycle type in which the exhaust-gas energy is lost. With the
increasing cost of energy, the utilisation of this energy becomes more attractive but the
required time and initial investement may not allow adding an air-bottoming (AB)
cycle. This chapter compares two other options for utilising the lost energy in
cogeneration and inlet air cooling systems. Both options involve a heat-recovery steam
generator (HRSG) to be installed. The chapter presents least-cost optimisation analyses
of the two options and compares them to the AB system. Like the previous chapter, the
performance of the HRSG is estimated by using two heat-exchanger models, which are
the effectiveness model and the pinch-analysis model. The exact analysis method is also
adpoted by using the Thermax1 add-in to determine the air and gas properties.
References
[1] A. Bejan, G. Tsatsaronis, M. Moran, Thermal Design & Optimisation, John
Wiley & Sons, 1996.
[2] J. E. Kraft, Turbine Inlet Cooling System Comparisons, Energy-Tech, 36-37,
August 2006
[3] L.C. Burmeister, Elements of Thermal-Fluid System Design, Prentice Hall, 1989.

Analysis and Optimisation of Fluid-Therm

Cargado por

Información del documento

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Analysis and Optimisation of Fluid-Therm

Cargado por

Copyright:

Formatos disponibles

Analysis and Optimisation of

Fluid-Thermal Systems using Excel

2. Excel formulae, built-in functions and iterative tools 25

4. VBA, user-defined functions, and the Thermax1 add-in 67

4.2.1. Property functions for saturated water and steam (Wat)

5. Analyses of multi-pipe and pump-pipe systems 93

6. Analyses of pipe networks 123

7. Determining the economic pipeline diameter 155

7.6. Least-cost pipeline optimisation using Excel-Solver

8. Economic optimisation of pipe networks 181

9. Thermodynamic and economic optimisation of the air-botomming cycle 221

Subject index 267

based platform for optimisation analyses of gravity-driven and pump-driven pipe-

[12] L. Caretto, D. McDaniel, T. Mincer. Spreadsheet calculations of thermodynamic

Analyses of multi-pipe and

5.1. Analyses of multi-pipe systems

The continuity equation

Q  Qin out (5.1)

The energy equation

Where Re is the Reynolds number defined as:

5.1.1. Three pipes in series

The analytical model

The head loss in each pipe is given by the Darcy-Weisbach equation:

The iterative solution proceeds as follows:

Example 5.1. Three pipes in series

Length (m) Diameter (m)  (m)

Figure 5.3. Goal Seek solution for Example 5.1

5.1.2. Three pipes in parallel

Figure 5.4. Excel sheet for Example 5.2

Figure 5.5. Solver set-up for Example 5.2

5.1.3. Pipes junctions

The analytical model

Example 5.3. Flow in a pipe -junction

Figure 5.7. Excel sheet developed for Example 5.3

The following constraints are imposed by Equation (5.11):

Figure 5.8. Solver set-up for Example 5.3

Figure 5.9. Solver solution for Example 5.3

5.2. Determining the operating point for a pump-pipe system

5.2.1. A centrifugal pump connected to a single pipe

Figure 5.10. Pump-pipe system with a single pipe

1. The flow rate of the fluid Q

According to the manufacturer, the pump has the following characteristics:

Flow rate Pump head

The solution method

h p  47.22  2.985  103 Q  1.549  105 Q 2  2.348  108 Q 3 (5.15)

3. Calculate the corresponding friction loss h f from Equation (1.21)

Figure 5.13. Solution obtained by Goal Seek for Example 5.4

5.2.2. A centrifugal pump connected to two branching pipes

1. Assume reasonable values for Q1 and Q2

D1 = 0.025m L1 = 500m ε1 =15x10-6 m

D2 = 0.05m L2 = 1,000m ε2 =46x10-6 m

The Excel sheet

Figure 5.15. Excel sheet developed for Example 5.5

Figure 5.16. Solver set-up for Example 5.5

Figure 5.17. Solver solution for Example 5.5

5.2.3. A centrifugal pump connected to a three-pipe branched network

Figure 5.18. Pump with a three-pipe network

Example 5.6. Pump with three branched pipes

D1 = 0.05m L1 = 500m ε1 =46x10-6 m

D2 = 0.025m L2 = 500m ε2 =15x10-6 m

D3 = 0.03m L3 = 500m ε3 =46x10-6 m

The Excel sheet

Figure 5.19. Excel sheet developed for Example 5.6

Figure 5.20. Solver set-up for Example 5.6