Chapter 1

INTRODUCTION

1.1 GENERAL PRESENTATION

Designing structural concrete members for shear? (WHY SHEAR?! If

the method use the truss concept which is axial and without shear) is a

complicated problem, which has not been solved fully. Till now

concrete model codes use empirical equations as a shear design

procedure for structural elements that have no known models. Those

empirical equations are based on laboratory work, and need to be

changed when a new material is used, because of this reason we need

a rational approach to structural concrete.

Strut-and-tie modeling is a rational, unified approach that

considers all the loads simultaneously in one model (ACI 318-02).

Truss models were first introduced by Ritter in 1899 and were

verified experimentally by Mörsch in 1909 for the design of beams.

Researchers did not pursue the studies of Ritter and Mörsch until the

period between the 1960-1970s where valuable studies were presented

by Kupfer in 1964, Leonhardt in 1965, Lampert and Thurlimann in

1
1971, also by Collins and Mitchell in 1980 whom proposed the truss

model for the design of reinforced and prestressed concrete.

‫ هذه الطريقة تستعمل لسببين الول هو أنها بديلة للطرق الخآرى والثانية عند‬:‫ايناس‬

‫تعذر معرفة نموذج التحلبل وظأو التصمبم‬

The method is used for two reasons: a) as an alternative, b)

when we have no solution

:‫كما أرجو إدخآال النقاط التالية في اللمحة العامة‬

 The present work has the aim of developing a formulation and a

computational system, which should be used by the structural engineers as
an auxiliary tool in the definition of strut and tie models of bi-dimensional
structures. Or when we do not have complete knowledge about the design
model …. (as yu said)

 The strut models cover for concrete and ties will be steel ties.

 In the developed system the user can define initially a very complex
system with a great number of bars and many different directions. This
initial solution can be based on both the stress results of a linear elastic
finite element analysis (using SAP for example) and the experience of
the engineer.

 As the user may be in doubt about the best model to be used he may
superimpose in the initial model all the allowable solutions he may find.

 The aim of the system is to choose among all the allowable solution (these
ones witch satisfy equilibrium and strength) the more economical one. As
more economical solution is considered the one which correspond to the
least consumption of reinforcement.

 The problem formulation leads to a linear programming problem (LP)

which will be solved by using maple.

2
  The visualization of the model and results is also part of the system. It is
based on the data generated by the SAP program.

 For the validation: The problem formulation is shown and the last the
results of the program are compared with those old the technical literature
or solving by hand.

 The computational procedure presented here may be of great utility in the

definition of strut and tie models.

Truss models are applied only for the design of B-regions were the

Bernoulli hypothesis of plane strain distribution is valid, for D-regions

(disturbed regions) where the strain distribution in non-linear truss

models are not valid.

In 1987 Schlaich et al introduced the strut-and-tie model approach

which was extended from the truss theory, for the design of D-regions

of reinforced and prestressed concrete structures.

The contribution of concrete in a modified truss model with variable

angle of inclination which can be used in the design of reinforced and

prestressed concrete beams (Ramirez and Breen, 1991).Ramirez in

1994 indicated that strut-and-tie models could be used in the design of

pretensioned concrete members.

Finally the strut-and-tie model has been introduced in the ACI 318-02

for the design of structural concrete.

3
Trial and error procedure is used for the develop of strut-and-tie

models, based on the designer experience and insight. For the designer

it is a task to select an suitable strut-and-tie model that satisfies the

equilibrium conditions for a structure with complicated geometry and

loading conditions to reinforce it.

1.2 OBJECTIVE

The present work has for objective, the develop a system

denominated BSTM (Best Strut-and-Tie Models ), to choose the best

strut-and-tie models, based on an optimization procedure which will

find the more economical strut-and-tie model that satisfy the

equilibrium conditions, starting from initially known models that

allow a subsequent detachment of the structure.

1.3 ORGANIZATION OF THE THESIS

In chapter 2 basic concepts of strut-and-tie models is presented.

Chapter 3 introduces the fundamentals of linear programming and in

it an example of optimization using graphical, excel spreadsheets, and

simplex methods can be found, to estimate the solution of a linear

programming problem.

4
In chapter 4 the linear programming formulation of BSTM with the

constraints imposed is presented .

Chapter 5 illustrates several examples on the application of the LP

developed in chapter 4. The obtained models are compared with

those presented by previous literature.

In chapter 6, the conclusions of this work and some suggestions are

presented for subsequent researches.

Chapter 2

Strut-and-Tie Models
This chapter will give an outline about strut-and-tie models, showing

when and why they are applied. The last paragraph of the chapter

includes an example of designing a deep beam using strut and tie

method using the ACI 318-02 method, which will indicate the

importance of knowing the appropriate strut-and-tie model in the

design of structural elements that have no defined models.

2.1 THE STRUCTURE’S B AND D REGIONS

Before the constructing a strut-and-tie model, it is important to

subdivide the structure into its B-and-D regions to know where strut-

and-tie models are applicable or not. Those regions of a structure in

5
which linear strain distribution (the Bernoulli hypothesis) is applied,

are referred to as B-regions, their internal forces or stresses can be

obtained from moment and shear diagrams, analyzed by means of the

statically system of beams, on B-regions truss models are applied.

Whereas, the regions of a structure where the strain distribution is

significantly nonlinear, e.g., near corners, near joints, corbels, wall

beams with or without openings, and other discontinuities and on

which the standard methods of codes fail to apply, are called D-

regions (D is for discontinuity” or “disturbed ). The internal flow of

forces in D-regions can be reasonably well described by strut-and-tie

models, examples of D regions are shown in Figure 2.1.

In B-regions, the stresses and stress trajectories are quite smooth as compared to

the turbulent pattern near D-regions, as shown in Figure 2.2. Stress intensities

decrease rapidly with the distance from the location of the concentrated load. This

behavior helps to identify the separation of B-and-D regions in a structure .

h1

h2

h2 h2

6
 h

h h

h h

Figure 2.1- Examples of D regions .

Trajectories D regions
Figure 2.2. Stress trajectories in B-and-D regions.

7
 2.2 GENERAL DESIGN PROCEDURE AND MODELING

Before starting the analysis of the structure it is recommended to

subdivide the structure into its B-and D-regions, because it isn’t

reasonable to model the whole structure with struts and ties. In Table

2.1 we can see when the strut-and-tie model is applied.

Table 2.1. Analysis leading to strut-and-tie forces

Structure consisting of :
Structure D-regions
B- and D-regions
only
e.g., linear structures, slabs
Analysis e.g., deep
and shells
beams
B-regions D-regions D-regions
Sectional Boundary forces
Overall structure effects
analysis Sectional Support
(M, N, V, T) effects reactions
Analysis of State 1 Via sectional Linear elastic analysis
inner (uncracked) values
forces or
Strut-and-Tie Models
stresses in State 2
individual and/or nonlinear stress analysis
regions (cracked)
Usually truss

8
Limit analysis helps to estimate the behavior of structural concrete

members under applied loads also the strength predictions. Lower-

bound and upper-bound methods are the divisions of limit analysis, for

the designing of new structures the lower bound method can be used,

and its based on strengthening its load transfer, while to check the

ultimate strength, dimensions, and details of existing structures the

upper-bound method is used. Strut-and-tie models match the lower-

bound method, for these models its required to determinate the

amount and location of steel reinforcement also the necessary

detailing.

The loads applied to a concrete member will be carried by the

concrete struts and the steel reinforcement (which form the load

transmit mechanism) when widespread cracking arise. Before

designing a structural concrete member, the locations of tensile ties

and the amounts of steel reinforcement are unknown. Actually, it is the

designer’s task to identify an appropriate strut-and-tie system in a

structural concrete member in order to reinforce it. As a result of this,

the nonlinear behavior of reinforced concrete cannot be taken into

account in the finite element model for developing strut-and-tie

systems. So the development of strut-and-tie model is based on the

9
linear elastic theory, while the design of the structural concrete

member is based on the theory of plasticity.

The development of a model for a D-region is simplified if the

elastic stresses and principal stresses directions are available, this can

be obtained easily by the wide variety of computer programs available

nowadays, however we can develop a strut-and-tie model even if no

elastic analysis is available and there is no time to prepare one, by

using “load paths”

Load path Method

The load path method is very important in perform of strut-and-tie

models, and can be used to find the path of compression and tension

forces from the point of applied loads to the supports, so that the

development of a strut-and-tie model will be easy by following the

flow of forces through the structure.

In this method we should first check that there is external load

equilibrium for the area to be modeled. The path of the loads inside

10
the structure happens through compression fields (struts) and tension

ones (ties).

The disturbed forces in the outline of the area should be substituted by

equivalent concentrated loads so that the loads on a side of the

structure, after they travel a certain load path, find on the other side

loads that balance them. These paths of loads should be continuous

and the shorter possible. After drawing all the load paths, they must be

substituted by straight lines which will represent the members of the

model, an example of forming a strut-and-tie model using load paths

is shown in Figure 2.3 .

q q q

Load
Path C
C C

T T T

(a) (b) (c)

Figure 2.3. The load path method (a) The structure and its loads. (b)
The loads path through the structure. (c) The corresponding strut-
and-tie model.

11
 Generally, designing structural elements requires the identify of

the load transmit mechanism, and reinforce the structural element in a

manner so that the load path transfer the applied loads safely to the

supports, actually not all the regions of the structural concrete member

are efficient in carrying loads, by the removal these regions the actual

load path could be found. An example of that is shown in

figure 2.4 where an elastic analysis using finite element was

performed. From the figure which represents a deep beam

supported to four points of loading with two openings, we can

observe the tension and compression regions and that the

concentration of stresses is near the openings, the direction of the

struts must be taken in the same direction of the compressive

stresses, the chief strut and ties can be localized in the center of

gravity of the compression and tension regions respectively.

Compression Tension

12
Figure 2.4 Load path based on an elastic analysis using finite element
method.

As a summary we can say that the strut-and-tie procedure is relatively

straightforward, and involves three key steps:

1. Develop the strut-and-tie model. The struts and ties serve to

condense or replace the real stress fields by resultant straight lines and

concentrate their curvatures in nodes.

2. Calculate the strut and tie forces, which must satisfy equilibrium.

These forces are the internal forces.

3. Dimension the struts, ties, and nodes for the internal forces of 2

with due consideration of crack widths.

2.3 DIMENSIONING THE STRUTS, TIES AND NODES

Dimensioning not only means sizing and reinforcing the individual

struts and ties for the forces they carry, but also ensuring the load

transfer between them by checking the node regions. There is a close

relation between the detailing of the nodes and the strength of the

struts bearing on these nodes and of the ties anchored in them because

the detail of the node chosen by the designer affects the flow of forces.

Therefore it’s necessary to check whether the strut and tie model

13
initially chosen is still valid after detailing or needs correction. Thus

modeling and dimensioning is in principle an iterative process. There

are basically three types of struts and ties to be dimensioned:

 CC: Concrete struts in compression.

 TC: Concrete ties in tension without reinforcement.

 TS: Ties in tension with reinforcement.

STRUTS and TIES

Whereas the TS are essentially linear or one dimensional elements

between two nodes, the CC and TC are two- (or three-) dimensional

stress fields tend to spread between two adjacent nodes. The struts in

the model are resultants of the compression stress fields. Since by

definition the curvatures or deviations of the forces are concentrated in

the nodes, the struts are straight. This is, of course an idealization of

reality. To cover all cases of compression fields including those of B-

regions three typical configurations are sufficient, see Figure 2.5:

1. The fan-shaped stress (a) field is an idealized stress field with no

(negligible) curvature and it does not develop transverse stresses.

2. The bottle-shaped (b) stress field with its bulging stress trajectories

develops considerable transverse stresses i.e., compression in the

14
 bottle neck and tension further away. The transverse tension can

initiate longitudinal cracks and cause an early failure. Therefore, its

essential to reinforce the stress fields in the transverse direction.

3. The prismatic (c) stress field is a frequent special case of the two

proceeding two stress fields in which transverse stress and

curvature is zero.

a b c

Figure 2.5 The basic compression fields: fan, bottle and prism.

NODES

15
The nodes of the model are derived as the intersection points of three

or more straight struts or ties, which themselves represent either

straight or curved stress fields or reinforcement bars. In a reinforced

concrete structure, a node is introduced to indicate an abrupt change

of direction of forces. In reality, the node usually occurs over a certain

length and width.

Types of nodes are illustrated in Figure 2.6 ;

 The smeared nodes (node B-consisting of three compressive

struts) represent the intersection point where wide concrete

stress fields join each other or with closely distributed

reinforcing bars. These types of nodes are normally not critical.

When sufficient anchorage of the reinforcing bars in the

smeared node is ensured, and sufficient reinforcement is

provided to catch the outermost fibers of the deviated

compressive stress fields, then the node is considered SAFE.

 On the other hand, the singular nodes (node A-consisting of two

struts and one tie) where concentrated forces are applied and the deviation

of the forces are locally concentrated. These nodes have to be carefully

detailed in order to prevent excessive deformations to the structures.

16
 Node B
CC C
C CC

Node A
C
T T C
C C

Figure 2.6 Types of nodes of strut-and-tie model and their stress

fields .

2.4 Reinforcement of different structural elements using

the strut-and-tie method

The table 2.2 below shows the affectivity of the strut-and-tie models

as a method to indicate where the main and additional reinforcement

should be placed for some structural elements.

Table 2.2 some strut-and-tie examples and the corresponding

reinforcement .

Case Strut-And-Tie Model Reinforcement

Deep beam with

uniform load

17
 Point support

Point load and point

support

Deep beam with

central point load

2.5 Illustrative Design Example of a Deep Beam Using Strut-

and-Tie Method per ACI 318-02 Appendix A

This example shows the importance on strut-and-tie models in the

design of structural elements of D-regions and which have no defined

models. Design of such elements was based on empirical equations

in the past, but recently the ACI code in 2002 recommended the use of

18
strut-and-tie models in the design of reinforced concrete as indicated

in chapter 1 .

Consider the 2 meter deep beam described in the figure 2.8 below. Use

the strut and tie model to determine the required amount of

 f y  420 MPA
reinforcement. Additional details: f c  25 MPA , ,

PDL  800 kN , and PLL  400 kN .

2000 mm
P 2000 mm
P 2000 mm
All Bearing Plates are
450mm*500mm
2000 mm

400 mm 6000 mm 400 mm Cross Section

500 mm wide
P P

Figure 2.8 Deep beam subjected to two points of loading

Step 1: Evaluate the Total Factored Load, Pu

19
Pu  1.2 PDL  1.6 PLL  1.2(800)  1.6( 400)  1600kN

Step 2: Check Bearing Capacity at Loading and Support

Locations

bearing strength at points of loading:


  0.85 f c  n Ac  0.75(0.85)( 25)(1.0)( 450)(500) / 1000

 3586  1600 kN  OK

bearing strength at supports :


  0.85 f c  n Ac  0.75(0.85)( 25)(0.80)(450)(500) / 1000

 2868  1600 kN  OK

Step 3: Select the Strut-and-Tie Model to Use in Design

From table 2.2 it could be see that the appropriate strut-and-tie model for the deep

beam shown in figure 2.8 is the model illustrated in figure 2.9 .

Pu=1600 kN Pu=1600 kN
2000 mm 2000 mm 2000 mm

B FBC C
2000 mm
F CD
B
A
F

A
 FAD 
D

400 mm 6000 mm 400 mm

Pu Pu

Figure 2.9 Strut-and-tie model for the deep beam .

20
Step 4: Isolate Disturbed Region and Estimate Member Forces

and Dimensions

The entire deep beam is a disturbed region, but it is only necessary to

consider the left third of the structure to complete the design. The

horizontal position of nodes A and B are easy to define, but the

vertical position of these nodes must somehow be estimated or

determined. What we do know is that the design strength of strut BC

must be greater than or equal to the factored load in strut BC. That is:


Strut BC:   Fnc   f cu Ac   (0.85 s f c ) bwc  FBC

Where  s  1 .0 ( prismatic strut )

Similarly, the design strength of tie AD must be greater than the

factored load in tie AD. In addition, this tie must be anchored over a

large enough area (wt b) such that the factored load is less than  Fnn

Tie AD :  Fnn   Ay f y  FAD and


Tie AD :  Fnn   f cu Ac  (0.85 n f c )bwt  FAD ,

Where  n  0.8 ( on tie anchored in Node A )

By setting the design strength equal to the required capacity, jd will be

a maximum and wt=1.25wc. The flexural lever arm will be :

21
jd=2000 - wc / 2 - wt / 2 = 2000 - 1.125wc

P u =1600 kN
CL 
 f cu   0 . 85  s f c
F BC
wc
B

jd  2000  1.125wc
AB
F

A

wt
FAD

2000 mm  f cu   0 . 85  n f c
Pu

By taking summation of moments about point A :

M A 1600(2000)(1000)  FBC (2000  1.125wc ) .

By substituting  (0.85 f c  )bwc for FBC , wc=231 mm, and therefore

wt=288 mm .

If these values are used for the dimensions of the struts and ties, the

stress in strut FBC will be at its limit, and the force in tie FAD will be

anchored in just sufficient area. It is often wise to increase these

values a little to leave some margin. wc will be selected to be 240 mm,

and wt will be selected to be 300 mm.

 jd  2000  240 / 2  300 / 2  1730 mm and ;

22
FBC = FAD =1600 (2000) / 1730=1850 kN

Check capacity of strut BC :

 Fnc  (0.85 s f c  )bwt

 0.75(0.85)(1.0)(25)(500)(240) / 1000  1912 kN OK

Step 5: Select Reinforcement

Tie AD :  Fnt   As f y  FAD  1850 kN

 As  1850(1000) / 0.75 / 420  5872 mm2.

 Consider 1 layer of 6#36(11) bars = 6036 mm2 @ 150 mm from

bottom.

 Consider 2 layers of 5#29 (9) bars = 6450 mm2 @ 80 mm and

220 mm from bottom.

 Consider 3 layers of 6#22 (7) bars = 6699mm2 @ 60, 150, and

240 mm from bottom.

Check capacity of tie AD :

 Fnt   As f y  0.75(6450)(420)  2032 kN  1850 kN OK

23
Step 6: Calculate Force in Diagonal Compressive Strut FAB and

Check Capacity

Pu=1600 kN
450 mm CL
B

240 mm
FBC
wc
B

wct
jd  1730 mm

B
wcb

A
F
A

300 mm

FAD
A
2000 mm
450 mm
Pu

tan   1730 / 2000 and   40.9 

Therefore, the force in the diagonal compressive strut,

FAB=1600 / sin 40.9 o = 2444 kN.

Width at top of strut

= wct = lb sin θ + ha cos θ = 450 sin 40.9o + 240 cos 40.9o = 476 mm.

Width at top of strut

= wcb = lb sin θ + ha cos θ = 450 sin 40.9o + 300 cos 40.9o = 521 mm.

Assuming that sufficient crack control reinforcement is used,

then : βs = 0.75.

24
check capacity of strut AB:  Fnc  (0.85  s f c  )bwct

 0.75(0.85)(0.75)(25)(500)(476) / 1000  2885 kN  2444 kN OK

Step 7: Minimum Distributed Reinforcement and Reinforcement

for Bottle Shaped Struts

Horizontal Web Reinforcement :

Use one # 13 (4) on each face at sh= 300 mm over entire length,

Ah / (b sh) = 2 (129) / 500 / 300 = 0.0017 > 0.0015 OK

Vertical Web Reinforcement :

Use one # 16 (5) on each face at sv = 300 mm over entire length,

Av / (b sv) = 2 (199) / 500 / 300 = 0.00265 > 0.0025 OK

Check of Reinforcement to Resist Bursting Forces in Bottle Shaped

Struts:

 vi sin i  0.0017 sin 40.9   0.00265 sin 49.1

 0.00312  0.003 OK

Chapter 3

25
 Fundamentals of Linear
Programming (LP)

In this chapter an idea of what LP is will be given, mentioning the

methods used, additional attention will be given for the method that

will be used to determine the BSTM, and an example demonstrating a

LP problem using various methods is included in the end of chapter.

3.1 INTRODUCTION

In 1947 G. Danzing invented the simplex method for solving linear

programming problems, this method together with the revolution of

computers paved the way for other methods of constrained

optimization presented by a number of investigators, specially

mentioned Charnes.

Old techniques for determining the optimum solution depending on

differential calculus are still used, by students, in their calculus

courses.

What does Optimization means ?

26
Optimization is looking for a point on a function, this point is either

the minimum or the maximum. The optimum is the point where the

curve is flat. In mathematical terms this refers to the x value where the

derivative f ( x)  0 , and the second derivative, indicates whether the

optimum is a minimum (if f ( x)  0 ) or a maximum (if f ( x)  0 ).

Optimization always deals with finding the optimum solution, in

civil engineering, optimization means design structures for minimum

cost, constrained by the limitations of safety.

3.2 LINEAR PROGRAMMING (LP)

Linear programming (LP) is an optimization approach that achieves a

desired objective, such as maximizing profit or minimizing cost

(which is the case of this thesis) in the presence of constraints such as

limited resources. The term linear refers that the mathematical

functions representing both the objective and the constraints are linear.

The term programming does not mean “computer programming”, but

rather scheduling or setting an agenda.

We can say that LP is one of the most widely applied techniques and

owes its popularity principally to George Danzig's simplex method

and the revolution in computing. It is a very powerful technique for

27
solving allocation problems and has become a standard tool for many

businesses and organizations. Although Danzig's simplex method

allows solutions to be generated by hand, the iterative nature of

producing solutions is so tedious, assuming that computers had never

been invented, then linear programming using the simplex method

would have remained an interesting academic idea, relegated to the

mathematics classroom. Fortunately, computers were invented and as

they have become so powerful for so little cost, linear programming

has become possibly one of the most widespread uses for a personal

PC.

3.3 METHODS OF LP

There are numerous software packages which deal with solving linear

programs ; LINDO, GAMS and XPRESS-MP are the most popular.

All these packages tend to be DOS based and are intended for a

specialist market which requires tools dedicated to solving LPs. In

recent years, however, several standard business packages, such as

spreadsheets, have started to include an LP solving option, and

Microsoft Excel is one of them. The insertion of an LP solving

capability into applications such as Excel is attractive for at least two

reasons:

28
  Firstly, Excel is perhaps the most popular spreadsheet used both

in business and in universities and as such is very accessible.

 The second reason is that, the spreadsheet offers very

convenient data entry and editing features which allows to gain

a greater understanding of how to construct linear programs.

For these two reasons and many others the excel spreadsheet will be

used in this thesis.

STANDARD FORM

The basic linear programming problem consists of two major parts:

the objective function and a set of constraints. For a maximization

problem, the objective function is generally expressed by:

Maximize Z  c1 x1  c2 x2  .....  cn xn (3.1)

Where ci j = payoff of each unit of the jth activity that is undertaken

and xj = magnitude of the jth activity. Thus the value of the objective

function, Z, is the total payoff due to the total number of activities, n.

The constraints can be expressed generally as :

ai1 x1  a i 2 x 2  ....  ai n x n  bi (3.2)

Where ai j = amount of the ith resource that is consumed for each unit

of the jth activity and bi = amount of the ith resource that is available. That

29
is, the resources are limited. The second general type of constraints specifies that

all activities must have a positive value,

x1  0 (3.3)
This expresses the realistic notion that, for some problems, negative

activity is physically impossible (for example, we cannot produce

negative goods).

The objective function and the constraints specify the linear

programming problem.

GRAPHICAL SOLUTION

When the LP problem involves only two variables, the solution can be

obtained graphically and because of that its limited practically.

However, graphical solutions are useful for demonstrating some basic

concepts that underlie the general algebraic techniques used to solve

higher-dimensional problems with the computer.

For a two dimensional problem, such as the problem shown below, the

solution space is defined as a plane with x1 measured along the

abscissa and x2 measured along the ordinate. Because the linearity of

constraints, they can be plotted on this plane as straight lines. If the LP

problem has a solution (was formulated properly), these constraint

lines will define a region, called the feasible solution space,

30
surrounding all possible combinations of x1 and x2 that obey the

constraints and hence represent the feasible (possible) solutions.

The objective function for a particular value of Z can then be plotted

as another straight line and superimposed on this space. The value of

Z can then be adjusted until it is at the maximum value while still

touching the feasible space. This value of Z represents the optimal

solution.

EXAMPLE

Its required to develop a graphical solution for this problem ;

Maximize Z  5 x1  3 x2 (3.4)

Subjected to the following constraints

3 x1  5 x2  15 (3.5.a)
5 x1  2 x2  10 (3.5.b)
x1  0 (3.5.c)
x2  0 (3.5.d)

GRAPHICAL SOLUTION

First, the constrains can be plotted on the solution space, and that

can be achieved by replacing the inequality by an equal sign and

solving for a variable, see Figure (3.1).

31
 x2

4 5 x1  2 x 2  10

2 3x1  5x2  15
x1  0
1

x2  0
1 2 3 4 5 x1

Figure 3.1 x1 – x2 plane showing the constraints of 3.5

The shaded area represents the feasible solution space, in order to

represent the objective function, a value of Z must be chosen, and

since we are interested in maximizing Z, we can increased it step by

step, this will make it moves from the origin. We look for the

maximum value of Z, inside the feasible solution, Figure (3.2).

As shown in the figure below the maximum value of Z equals to

(12.37) at this value x1 and x2 are respectively (1.05, 2.37).

32
 x2

5 5 x1  2 x 2  10

4
Z3
3
A
2 3x1  5x2  15
x1  0
1
Z1
x2  0
1 2 Z2 3 4 5 x1

Figure 3.2 Representation of the objective function with the restraints.

SOLUTION USING THE SIMPLEX METHOD

The simplex method is one of the well known mathematical

techniques of solving linear programming problems. The simplex

method is a search method, that is simply searches for the optimal

solution by going from one vertex to the next one of the area of

feasible solutions until optimum solution is found . the simplex

method can be found in several mathematical software like LINDO,

MathCAD and Maple.

Maple was used to solve the same problem indicated above as

33
follows :

> restart;
> with(simplex):
> maximize( 5*x1+3*x2, {3*x1+5*x2 <= 15,
5*x1+2*x2 <= 10}, NONNEGATIVE );
{ x0, y0 }
Warning, the protected names maximize and
minimize have been redefined and unprotected

54 02
}2x , 
1x {
91 91

we can notice that the values obtained from the simplex method using

Maple are the same obtained from the graphical solution.

SOLUTION USING EXCEL SPREADSHEETS

Solving the same example above mentioned using the excel

spreadsheet, step by step will give a good picture of how

does Solver work.

The best approach to entering the problem into Excel is first to list in a

column the names of the objective function, decision variables and

constraints. We can then enter some arbitrary starting values in the

cells for the decision variables, usually zero, shown in Figure (3.3).

Excel will vary the values of the cells as it determines the optimal

solutions. Having assigned the decision variables with some arbitrary

starting values we can then use these cell references explicitly in

34
writing the formulae for the objective function and constraints,

remembering to start each formula with an '='

Starting entering the formulae for the objective and constraints, the

objective function in B3 will be given by :

=5*B4+3*B5

The constraints will be given by (putting the right hand side {RHS}

values in the adjacent cells):

#1 (B8) =3*B4+5*B5

#2 (B9) =5*B4+2*B5

#3 (Non-neg 1) (B10) =B4

#4 (Non neg 2) (B11) =B5

A B C
1 Example
2
3 Objective 0
4 x1 0
5 X2 0
6
7 Constraints
8 #1 0 15
9 #2 0 10
10 #3 0 0
11 #4 0 0
Figure 3.3 : Introduction of the problem in Excel

We are now ready to use Solver.

If we select from the menu of Tools the option Solver

35
the dialogue box shown in Figure (3.4) is opened, and if we select the

objective cell before call up Solver the correct Target Cell will

be identified. This is the value Solver will attempt

either to maximize or minimize.

Figure 3.4 : The Solver Dialogue Box

Then we select max, or min , in this case we want to set the target cell

(the objective) to a Max. By typing in the white box the cells B4 &

B5, solver will know the cells which we want to vary. Then we can

enter the constraints by first clicking the 'Add ..' button. This reveals

the dialogue box shown in Figure (3.5).

36
 Figure 3.5 : Entering Constraints

The reference cell is to the cell containing our constraint formula, so for

constraint #1 we enter B8. By default <= is selected but we can change this by

clicking on the drop down arrow to open a list of other constraint types. In the

right hand white box we enter the cell reference to the cell containing the

constraint value, which for the #1 constraint is cell C8. Then by clicking the 'Add'

bottom we can add the rest of the constraints. Having added all the constraints,

click 'OK' and the Solver dialogue box should look like that shown in Figure

(3.6).

Figure 3.6 : The Completed Solver Dialogue Box

If we Click 'Solve' now will see Solver find the optimal values for x1

and x2. On doing this, at the bottom of the screen Excel will inform us of Solver

progress, then on finding an optimal solution the dialogue box shown in Figure

(3.7) will appear. We will also observe that Solver has changed all the values in

37
our spreadsheet, replacing them with the optimal results. We can use the Solver

Results dialogue box to generate three reports.

Figure 3.7 : Solver Results

It's better to get Solver to restore our original values in the spreadsheet

so that we can return to the original problem formulation and make

adjustments to the model. The three reports are generated in new

sheets in the current workbook of Excel.

If we choose the Keep Solver Solution the table shown in Figure (3.3)

will change to the to the one shown in Figure (3.8).

A B C
1 Example
2
3 Objective 12.36842
4 x1 1.052632
5 X2 2.368421
6
7 Constraints

38
 8 #1 15 15
9 #2 10 10
10 #3 1.052632 0
11 #4 2.368421 0
Figure 3.8: Output of the problem in Excel

The Answer Report gives details of the solutions and information

concerning the status of each constraint is provided. The Sensitivity

Report for the problem, which provides information about how

sensitive the solution is to changes in the constraints, the answer

report is shown in Figure (3.9).

Cell Name Original Value Final Value

$B$3 objective 0 12.36842105

Cell Name Original Value Final Value

$B$4 x1 0 1.052631579
$B$5 x2 0 2.368421053

Figure 3.9 : Answer Report for the problem

All the reports can simply be copied and pasted into Word and this is

perhaps one of the big advantages of using Excel over a DOS based

LP solver. Although the reports paste into Word as tables, they are

easily converted into text and can then be manipulated if one is

producing a written report on our finding.

39
The same result obtained from the graphical solution was reached

faster and easier.

As a conclusion, Excel Solver provides a simple, yet effective,

medium for allowing to explore linear programs. It can be used for

large problems containing hundreds of variables and constraints, and

does these relatively quickly.

On the downside, we can't view the Tableau as it is generated at each

iteration and so those who want to find the results of each iteration

will find Solver less superior to, other DOS based programs as

LINDO, which does allow this. It does, however, produce a superior

set of results and sensitivity reports when compared to LINDO, and,

due to the spreadsheet nature, it allows the user to observe very

quickly the effects of any changes made to constraints or the objective

function.

Chapter 4

Problem Formulation

40
This chapter includes the BSTM problem formulation using LP, but

firstly the objective function and the constraints will be defined.

4.1 INTRODUCTION

In reinforced concrete structural elements subjected to rapture we can

observe the development of tension zones where the reinforcement is

placed, and zones of compression where the struts are placed. If we

ignore the existing material between the struts and ties it can be

considered that the frame structure is reduced to an articulate

structure named a strut-and-tie model.

For the type of structures above mentioned two problems can be

identified, in one side , it is necessary to determine the position of the

reinforcement in a manner that assures the equilibrium of forces in

each node, on the other hand, it is needed to determine the required

amount of reinforcement and concrete section so that the stresses in

them don’t exceed the maximum allowable stresses.

In the problem formulation proceeding, a linear programming problem

will be developed with the affair of identifying the position of

reinforcement through the representation of ties in the strut-and-tie

model, without calculating the cross sectional area of reinforcement

for the obtained tie forces.

41
 4.2 FORMATION OF THE STRUT-AND-TIE MODEL

The formulation developed has for objective to choose the best

solution among all the solutions or combination of solutions that are

introduced. At this phase it’s necessary to insert on the projected area

several possible strut-and-tie models, considering reasonable

placement positions of reinforcement. The release of the initial model

was based on an elastic analysis for the structure using SAP2000 for

finite element analysis. It must be in account that the strut axes should

not exceed the outline limits of the structure, since the struts have

dimensions that should be kept inside the structure.

4.3 BASIC CONDITIONS THAT MUST BE IMPOSED

The basic conditions imposed to obtain the (Strut-and-tie model) are:

 Equilibrium conditions

 Reinforcement of ties

 Minimum forces of struts

As indicated before, struts and ties will carry the compression and

tension axial forces respectively, the reinforcement is strictly

necessary in ties while is not in struts, these variables can be

composed for :

N  f y d * As (4.1)
where

42
N axial force
f yd yield strength of steel
As cross-sectional area of bar

4.3.1 EQUILIBRIUM CONDITION

The equilibrium conditions shown in figure 4.1 , for each

node in the strut-and-tie model are:

 In the x-direction:

NM NF
 N i * cos ( i )   F * cos (  )  0 (4.2)
i 1 j j
j 1

 In the y-direction

NM NF
 N i * sin ( i )   F * sin (  )  0 (4.3)
i 1 j j
j 1

Fi Ni

i
i
N3

N1
N2
x

Figure 4.1 .Equilibrium condition

Where

NM number of convergent members in each node

NF number of external forces subjected to each node
Ni axial force of member i
Fj external concentrated force j

43
i the angle between the axial force N i and the abscissa
i the angle between the external force F j and the
abscissa

4.3.2 REINFORCEMENT OF TIES

The reinforcement of reinforced concrete structural elements, is

limited to the maximum value (as a function of the element type and

the cross sectional area of the concrete ) and as the reinforcement

cannot be negative, we get :

N
i
0
(4.4)

4.3.3 MINIMUM FORCES OF STRUTS

The axial force in a possible strut can’t match tension

so :
N
j
0
(4.5)
In terms of absolute values :
N
j
0
(4.6)

4.3.4 CRITERION OF OPTIMIZATION

It is required to choose the most appropriate model among all the

possible articulate strut-and-tie models which satisfy the conditions

(4.2),(4.3),(4.4),(4.5),(4.6) above mentioned.

44
The optimization criterion used is strictly economical, in other words,

it finds the minimum cost. When the reinforced concrete section is

kept constant, the minimum cost corresponds to minimum

reinforcement to be used.

So for the optimization criterion :

NT
Minimize Z   As ,i * li (4.7)
i 1

where

Z cost “objective function ”

NT number of ties
As ,i reinforcement in tie i
li length of tie i

As the variables used are the strut and ties axial forces it results that;

NT l
i
Minimize Z   N *
i 1 i f
yd
(4.8)

Where fyd the yield strength of the steel .

45
 4.4 LINEAR PROGRAMMING PROBLEM AND ITS

SOLUTION

It results from the combination of the basic conditions that must be

imposed with the criterion of optimization that:

l
Minimize
NT
i
Z   N *
i 1 i f
yd

Subjected to :

 For each node in the articulated strut-and-tie model :

NM NF
 N i * cos ( i )   F * cos (  )  0
i 1 j j
j 1

NM NF
 N i * sin ( i )   F * sin (  )  0
i 1 j j
j 1

 For each tie :

N 0
i

 For some ties, depending on the model, the minimum force is given by :

N N
i i , min

 For each strut :

N 0
i

All the relationships indicated in the paragraph 4.4 are linear in

variables of Ni, that is why the problem is of LP. The solution of this

problem using the solver of excel worksheets outcome the best strut-

and-tie models.

46
47