Advanced Structural Analysis

Prof. Devdas Menon

Department of Civil Engineering
Indian Institute of Technology, Madras
Module No. # 2.2
Lecture No. # 08
Review of Basic Structural Analysis-2
Good morning to you. We are now on to lecture 8 in the second module on Review of
Basic Structural Analysis.
(Refer Slide Time: 00:18)

We are still on to force methods; this is the second lecture on solving force methods.
(Refer Slide Time: 00:26)

This is covered in part IV of this book on structural analysis.
(Refer Slide Time: 00:31)

In the last class, I broadly discussed the basic conceptual difference between force and
displacement methods. In this module, we will be focusing on two force methods,
namely, the method of consistent deformations and the theorem of least work. If you
recall, we had already demonstrated the application of these two methods to solving
statically indeterminate beams, continuous beams and fixed beams to frames; we also
applied it to trusses. In this session, we will extend that application to portal frames,
specifically two-hinged portal frames, two-hinged arches, fixed arches and also frames
with elastic supports. In the next section, tomorrow, we will cover approximate methods
of analysis under lateral loads.
(Refer Slide Time: 01:36)

I want you to look carefully at this simple problem. What you see in the picture is a two-
hinged portal frame, two-hinged because you have two hinge supports at the bottom, and
subjected to some arbitrary vertical gravity loading. The frame is symmetric; that brings
in some special features, which I want to you to understand. Let us be as generic as
possible; let us say that the beam has a second moment of area I
b
, which means it has a
flexural rigidity E I
b
and the two columns have the same flexural rigidity E I
c
. Is it clear?
How will this frame behave? First of all, can we calculate the reaction? You will notice
that yes, if the resultant vertical load is W, we can get the vertical reactions as you would
get in a simply supported beam. What about the horizontal reactions? What about the
bending moment diagram? What is the degree of static indeterminacy in this beam? 1. Is
it internally or externally determinate? Either way; you can treat it as externally
indeterminate by taking the horizontal reaction to be redundant or you could take maybe
the bending moment at the junction of the beam and column as a redundant.
(Refer Slide Time: 03:19)


Now, I want you to look at these pictures which I have drawn. Take look at this. This is
that same frame but it is simply supported; that means one of the supports is a roller
support. In what way is the behavior of this frame different from the original frame?
There will be no bending moments in the column and, in fact, this will behave exactly
like a simply supported beam. If you look at the bending moment diagram, it will be as
shown here (Refer Slide Time: 03:50) and these legs will remain straight.
(Refer Slide Time: 04:01)

You have to basically copy that diagram but it will sway; so, it is going to probably look
like this. If you were to draw the bending moment diagram of this frame, the vertical
elements will have no bending moment and the beam will have the same bending
moment as shown here (Refer Slide Time: 04:23). This is very straightforward; it is a
sagging moment diagram. If you arrest this movement, you are going to get a horizontal
reaction which means shear in these vertical columns. You will get this (Refer Slide
Time: 04:37) plus an additional effect. What is that additional effect?
(Refer Slide Time: 04:41)

It is as though you had the same frame but you also had this horizontal force H. This plus
this is that frame. Now, how do you analyze this frame? How do you understand its
behavior? As far as the bending moment diagram of this frame was concerned bending
moment or shear force or axial force (Refer Slide Time: 05:23), you will find that the
force response does not depend on I
b
and I
c
; it does not depend on the relative stiffness
of the members. It is statically determinate; the statics do not depend on the member
sizes, the material property and all that. It is a straightforward equilibrium calculation,
but the response of the portal frame, which is two-hinged, does depend on I
b
and I
c
.
Now, the ratio I
b
by I
c
has a big role to play and I want you to see that. Let us say the
beam is very slender compared to the column; that means the column is very wide
maybe it is a wall. Then, you could say I
c
tends to infinity or I
b
by I
c
tends to 0 (Refer
Slide Time: 06:15). So, the picture would look like this. Of course, you treat them still as
line elements. On the other hand, you can have a rigid beam and the column relatively
slender. So, I
b
by I
c
here tends to infinity (Refer Slide Time: 06:31).
(Refer Slide Time: 06:41)

Similarly, if you were to just isolate the beam part, one extreme is a simply supported
beam. When would you have this condition? When you have this condition, with an
important difference.
(Refer Slide Time: 06:59)

If this is really true, it would degenerate to a beam with a hinge there. What is the
problem with this? If it is really infinity, if I
c
is really 0, then you get a hinge, an internal
hinge, here. This is not stable; this is unstable; this is not a good condition to have. On
the other hand, this is an extremely good condition to have; this is very stable (Refer
Slide Time: 07:43).
(Refer Slide Time: 07:49)

The reality is that I
b
by I
c
will lie between two extremes. They are limits; they are not
physically realizable; they are limits. If you really want to model it as a beam, how
would you model it? Tell me. This model is one limit (Refer Slide Time: 08:14). This
model is perhaps another limit where there is no rotation possible. There is no rotation
possible at the beam-column joint because the columns are very stiff. So, it is
approaching fixity conditions. If it is a fixed-fixed beam, you will find the fixed end
moment here and the fixed end moment here are going to be different.
Actually, this fixed end moment (Refer Slide Time: 08:44) is going to be more if b is
((.)), less than a; they are not going to be equal. Would this also simulate that behavior?
Or will there be a small difference? What is the difference? That is the crucial point to
know. What is the difference? ((.)) These two moments have to be equal in this situation
because that is nothing but H into h.
(Refer Slide Time: 09:22)

The bending moment you get from this diagram will look like this. Do you agree? This is
going to hog and the moment on this side and this side will be equal to H into h. We will
see later that if it is symmetric, this H into h turns out be an average of these two
moments; very interesting; it averages out, but that is the other extreme. If you had to
model isolate the beam, what is the boundary condition you will apply?
(Refer Slide Time: 10:04)

What kind of hinges? This cannot move. So, it is a simply supported beam with
rotational springs. Very good. With rotational springs. The rotational spring has a
stiffness k
theta
. If it is simply supported, then one limit is 0; if the columns are infinitely
rigid, then the other limit is infinity; broadly, you must have these feelings. Now, let us
look at the solution to this problem.
(Refer Slide Time: 10:51)

The energy method is particularly useful in solving such problems. You know that the
energy method says, the theorem of least work says that if you treat H as a redundant dou
U star or d U star by dH which is integral M into dM into dx by EI, which you see here,
is equal to 0. Now, what is n? You have to integrate over the full length of the frame; you
have to integrate separately for the two columns and for the beam. Will you try that and
give me an expression for capital H? A generalized expression which accounts for I
b
and
I
c
variations.
(Refer Slide Time: 11:48)

In this picture, I am showing you the free body diagrams where M
B
and M
C
are capital H
into small h. Does the diagram make sense? Your redundant is capital H; if you know
capital H, you know everything. Can you draw the bending moment diagram? What will
it look like? It will be nothing but this plus this (Refer Slide Time: 12:22). That is what
the diagram will look like because it is a superposition of two simply supported frames;
so, it is going to look like that. Of course, the picture that I have shown you there is a
distributed load and so you will have a curved bending moment diagram. The picture I
have shown you on this blackboard has a concentrated load; so, it has a linearly varying
free moment diagram (Refer Slide Time: 12:53). That is the only difference. Clear?
Now, you have to write an expression for M in the beams and in the columns. What will
be the expression in the column? It is a linearly varying bending moment. So, this is your
resultant bending moment diagram (Refer Slide Time: 13:13). Can we write this
expression for the column? That is, if you take the column and you say starting from the
base x pointing upward, is this expression correct? Now, whether you write plus or
minus depends on your convention. But, let us say, we are sitting inside that frame; so, it
is outside; we can treat it as hogging. For vertical elements, it is not right to use words
like hogging and sagging; for horizontal and inclined elements, it makes sense. Clear?
Do you agree that this is the expression and the limits are from 0 to small h? This is for
both column AB and column DC.
Now, can you write an expression for the beam? Can you write it in terms of M naught?
M naught is the free bending moment at any location x in a simply supported beam.
What will it be? ((.)) The sagging moment is M naught of x and minus capital H into h
because that is a constant moment throughout the length of the beam. Can we write that?
Got it? Here, the limits are from 0 to L.
Now, you invoke the theorem of least work. Can you substitute these values and derive a
formula for capital H? You have to integrate for two columns and one beam. If you
differentiate this, what do you get with respect to ((.))? You get minus x. If you
differentiate this, what will you get? What do you get? You are differentiating with
respect to capital H. Which is the function of capital H here? This is not a function; this
is the function, is it not? It is pretty easy to do.
(Refer Slide Time: 15:32)

This is what you get. Agreed? This is the first expression for the two columns; so, 0 to h.
This is the bending moment; this is the derivative of that bending moment; this is for the
beam 0 to L; this is the moment which you can calculate minus h and the derivative of
this ((.)). Do you agree to this? It is a straightforward calculation. Will you expand this
and simplify?
Now, let us introduce a parameter which you can call as stiffness parameter where I take
the relative stiffness of the beam to the column; so, EI
b
by L to EI
c
by h. That gets rid of
all those variables and makes it generalized. What are the limits of gamma? gamma will
tend to 0 when the column is infinitely rigid; gamma will tend to infinity when the beam
is infinitely rigid. If you solve this equation in terms of gamma, do you get this
expression? It is a very straightforward thing to do. What does this remind you of? What
is this integral M naught into dx? It is an area. Area of what? Area of the free moment
diagram, right? Area of the free moment diagram. It is a useful relationship to remember
and we call that A naught.
Let us look at two extremes. If you have infinitely rigid columns, which is this case
(Refer Slide Time: 17:19), if you have infinitely rigid columns, gamma will tend to 0 and
that equation reduces to H equal to A naught by L A naught by L h. On the other hand, if
you have infinitely rigid beams, gamma tends to infinity and H tends to 0. That is
obvious because if you are approaching this condition, you do not get any horizontal
force because there is no moment transferred from here to here (Refer Slide Time:
17:49). There is no moment in the column; the horizontal force tends to 0.
Now, let us use a parameter eta to take care of the gamma. If we define eta as 1 divided
by 1 plus 2 into gamma by 3, which is essentially the denominator here, then capital H is
A naught divided by hL into eta. Let us take simple problems; let us take symmetric
loading problems. Is this clear? It is quite simple. If you take a symmetric. You will find
that this moment here, capital H into h, when you substitute in this equation will be A
naught divided by L. What does A naught divided by L remind you of? It is the average
value of the bending moment in the free moment diagram; so, it is interesting.
The value of the moments you get here, the maximum value of that moment is A naught
by L because the highest value of eta is 1. So, that bending moment will vary from 0 to A
naught by L. As a structural engineer, you must be sensitive to the bounds of your
results; you know that the answer is going to lie between this extreme and this extreme;
the deflected shape will be from one extreme to the other extreme. You get a clear
picture of the overall behavior in this procedure. Let us take a simple example. Will you
solve this problem?
(Refer Slide Time: 19:32)

Let us take a simple example. Will you solve this problem? Here, I have a symmetric
loading, a uniformly distributed load. Can you draw the bending moment diagram in
terms of eta? If you recall, this is the general bending moment diagram (Refer Slide
Time: 19:52). This is the general bending moment diagram. Here, you have a parabola
due to a UDL. What will it look like? These are your general expressions for any
symmetric loading (Refer Slide Time: 20:06).
Do you agree that this is the bending moment diagram? The bending moment diagram in
the beam will be a parabola which gets lifted up by capital H into small h; the area of that
free moment diagram is two-thirds into WL by 8; so, it is a parabola. Do you agree to
this expression? Two-thirds into WL by 8 into L will be the area of the free moment
diagram. That is A naught; that is all you have to calculate; that is WL squared by 12 and
that divided by L is WL by 12 into eta.
Now, WL by 12 reminds you of what? The fixed end moment; so, it is the other ((.)). So,
it is a fixed end moment into a parameter eta which can vary from 0 to 1. The highest
moment you can get is that of a fixed beam and the lowest is a simply supported beam. It
is a very simple calculation and your bending moment diagram at the mid span will be
WL by 8 minus whatever value you get here. Clear? It just gets lifted up.
Let us take a case of unsymmetric loading where W is at quarter span loading. You can
do any problem; we will just take these two and see the kind of results you get. It is
smart to convert the unsymmetric loading to a symmetric loading because there is a
relationship between these two. What is the relationship? In a horizontal reaction when
you have a single load, this h will get doubled. It will get doubled when you get this
because you invoke the principles of parity and superposition, is it not?
Solve for this first, which is very easy; we are just using these equations. Find 2 h, divide
by 2 and you get H; it is a clever way to solve this problem. Will you do that? Do you
agree the M naught diagram will look like that (Refer Slide Time: 22:41)? That is an
easy area to calculate. If you work it out, that is the area; these are very simple and so I
am going fast now. You can solve for 2 H, divide by 2 and you get H. These are very
quick ways of solving such problems but the real application of this procedure is for
arches which are very common in structural engineering.
Where do you come across two-hinged arches? In bridges, yes. You know the bow string
girder bridge; the deck is suspended through vertical elements into an arch; you get
roughly a uniformly distributed load on the arch but not truly uniformly distributed. The
arch is usually parabolic for that reason but because a loading is never always funicular,
you will always get some bending moments in the arch, which you want to minimize. Do
you do it for buildings as well? Yes, big gateways and so on but I will show you pictures
of a project which we are doing now where we have a series of arches.
(Refer Slide Time: 24:04)

Let us take a generic case of a two-hinged symmetrical arch. You know that if you are
dealing with an unsymmetrical loading, it is easy to deal with it because you convert it to
symmetric, find the horizontal thrust, divide it by 2 and then you take a free body ((.)).
Can we use the same theory and do it here? Can you write an expression for capital H?
Do you agree that capital H?
Now, we have to integrate. There is no beam and column here; there is just one curved
length. Let us say that parameter is s, small s, and the full length is capital S. Do you
agree this is the expression for d U star by dH? There is only one redundant H and since
Mx at any section, the bending moment is M naught x minus Hy (y is the rise in the
arch at any location); if you take the derivative of this, you get y. Do you agree that this
reduces to this (Refer Slide Time: 25:10)?
Now, there is a problem with that integration. It is not easy to integrate because it will
look like this (Refer Slide Time: 25:19). This problem was tackled historically by
making some assumption because you do not get closed form integrals especially if EI is
constant along s. Is there a clever way of simplifying this equation, not making much
error? Any suggestions? I will give you a clue. Can we do something to integrate along
the horizontal length along x rather than s?
What is the relationship between s and x? ds is dx into square root of 1 plus y dash
squared; I hope you know that; that is a simple expression. If you substitute that, you will
get this expression; direct substitution. Now, you tell me what is the assumption that is
worth making. It has something to do with getting rid of EI from that entire equation.
Now, if you do that, you still have a square root. It is not easy to integrate this
expression. You do not want to see these square roots; you do not have a closed form
integral for such expressions.
What is a clever thing to do without making too much of a mistake? M naught can take
any shape depending on the loading; so, you cannot make an assumption on M naught.
You have to use your intelligence and do it. People have done this; today, that is the way
you analyze arches. It is simple mathematics a clever move. How do I get rid of square
root of 1 plus y dash square and just knock that off from that equation?
Polar coordinates is not suitable; that is suitable for segmental arches. Can we assume EI
to take some variation along the span? Where do you think the arch should be thicker
more at the crown or at the supports? Crown. Why? Where is the axial compression
more at the crown or at the supports? At the supports. It makes sense to thicken the
arch at the support and make it thinnest at the crown.
Let us say the EI at the crown is I
0
. Can I assume a variation of I of x in terms of I
0

which will help me crack this problem? Yes? I x is equal to I
0
into ((.)). That is what
they did. The formula for a parabolic arch is this; you can easily derive this to satisfy the
boundary conditions (Refer Slide Time: 28:32). This was the assumption made. Do not
worry; people do not really make arches to have exactly I equal to this. But they often
thicken it at the supports; it does not matter how you do it; you do it as per your
convenience. You can even make it uniform and still assume this and go ahead; you will
make a small error. We will see that the error is not significant but it is a tremendous
reduction in your work. If you make this assumption for analysis, how does that equation
reduce?
(Refer Slide Time: 29:14)

That equation now enormously simplifies; you are now integrating along the horizontal
length. You have the equation for the arch and it is symmetric; so, you need to integrate
only to L by 2. There is another beautiful thing that you can see which is independent of
the loading. What is that? One: the denominator is a constant. Can you integrate that and
tell me the value? It is very easy to do that integral y squared into dx from 0 to L by 2.
This denominator is a constant which you can remember because y is this. There you are;
4 into h squared into L by 15. That is an easy integration to do. Remember that and you
are ready to solve arches. The first question is let us say you have got an unsymmetric
loading. Do not worry; make it symmetric because then you need to integrate only your
half, find out the horizontal force and then divide that by 2 to get it for the original
problem. Secondly, in this integration, the denominator simplifies to 4 into h squared
into L by 15.
Let us say you have a case where you have a parabolic arch subjected to a uniformly
distributed load along the horizontal span. Do you know the answer for that? Do you
know the answer for this case? How can you forget? Yes? Do you know the answer for
capital H? What is it? ((.)) There is something nice about this arch. This arch is funicular
for this loading. That means there are no bending moments anywhere in the arch.
(Refer Slide Time: 31:37)

What is the bending moment at mid span in a simply supported frame? q
0
into L squared
by 8. That will get eliminated at the crown. You have to divide it by h; so, it is q
0
into L
squared by 8 h; you know the answer. But let us say we pretend we do not know the
answer and we do this integration making that assumption of I
x
equal to ((.)). Do you
think we will get exactly that answer or will there be a small error? You would expect a
small error; let us check it out.
This is an easy integration to do. M naught. You can see that the support reaction is q
0

into L by 2. So, if you cut a section anywhere in a simply supported arch, it will be q
0

into L by 2 into x minus q
0
into x squared by 2. You can really work this out; it will
reduce to this form; please check it out at your convenience. When you complete this
integration, you get in the numerator q
0
into hL cubed by 30; you can check it out.
Now, you substitute here and you get exactly, without any error, the expression q
0
into L
squared by 8 h. When people saw this result, they said the error, if at all, in a generic
case will be negligible. This gave confidence in using this technique; whoever suggested
this was a genius because it simplified the work so much. Otherwise, you will have to
resort to numerical integration, which is not worth ((.)). This is a very simple way to
solve any arch. Is it clear? This is a funicular arch, statically determinate.
(Refer Slide Time: 33:20)

Now, let us do something more interesting. Let us say one half of the arch is loaded.
Without any calculation, can you get capital H? It will be q
0
into L squared by? There
you are. You should write applying the principles of parity and superposition. Let us say
you have got it. Can you now draw the bending moment diagram, axial force diagram
and shear force diagram? Only then is the analysis complete. Getting H makes it
statically determinate but then you have to finish it.
You have done this exercise for three-hinged arches, remember? We have done this
before. This is now like a three-hinged arch. In fact, in this particular case, it is a three-
hinged arch because the bending moment at C will be 0. You can do this; I am not going
to do this all over again; I am giving you an assignment where you have to do it; we have
done this earlier.
What about this? Let us say you had a triangular loading like this (Refer Slide Time:
34:35). What do you think capital H is for this second case, load case 1B, when you have
a triangular load? It is the same. It looks tricky but then if you put a triangle the other
way, it will add up to a UDL. Here also, you get the same answer (Refer Slide Time:
34:52).
(Refer Slide Time: 34:55)

Let us take the case of concentrated loads. Can you derive an expression? In fact, these
are the two standard loading cases. You will have a uniformly distributed load and you
have a concentrated load. You could have the concentrated load at the crown but that is a
special case of this. How will you solve this problem? Let us do this together. What is
the expression for M naught? Very easy; you get pure bending in the simply supported
arch between the two concentrated loads and that is W into a; in the region up to the
concentrated load, it is W into x. It is enough to integrate to one half; do not bother about
the other.
Do you agree to this expression? You have got an expression for M naught. Please check
this out. You just write the equation in the final form. Do you not think it is worth doing
it on your own later? Check it out; it is a standard integration. You are good at
integration; it is a simple integration. You know the equation for y; it is a parabola. We
wrote the equation here for M naught; you have to integrate it over two parts: 0 to a and a
to L by 2. Is it clear?
What is the denominator? It is a fixed quantity, is it not? 4 h squared by L; so, 4 h
squared into L by 15. That is an equation which practicing engineers like to see because
all you need to do is to substitute the value of a by L, a by L is a ratio, and you have got
the answer. This is a standard solution. You can use this; you do not need to reinvent the
wheel but we have worked the background for doing this. Is it clear?
Now, if you had just one load, the vertical reactions are easy to calculate; the horizontal
reaction is what you get in that expression divided by 2. Instead of 5 by 4, it becomes 5
by 8 times that same expression (Refer Slide Time: 37:13). What happens if you had one
concentrated load at the crown? All you need to do is to substitute a by L equal to half, 1
by 2, and this is worth remembering. This is a standard formula for a single concentrated
load on a symmetric arch acting at the crown. All these are special cases of the eccentric
case. It is 25 by 128 into WL by h.
(Refer Slide Time: 37:47)

Now, let us put together all that we have learned and apply it to this problem. J ust go
through the motions of solving this problem. You have two concentrated loads: one is at
the crown and the other is eccentric. You have a UDL but not all the way till one half.
Can you write an expression for capital H with whatever we have derived? You do not
need to invoke the theorem of least work and solve for M naught; you can very well use
the formulas we have derived; that is what engineers do.
As students, you should know how to derive but you do not you need to do that; today,
you have handbooks and so on; you need to apply as well; so, let us apply here. If you
apply that, first you have to find the support reactions which are pretty straightforward.
You should also have the ability to check the reactions. Let us take the UDL. The total
load of the UDL is 30 into 6, that is, 180 kilonewton. How much of it will go to the left
support? Three-fourths of it; one-fourth will go to the right support. Take the
concentrated load at the crown 200 kilonewton (half, half 100, 100) and take that 120;
2 by 12 will go to the left support and 10 by 12 of that will go to the right. It is simple
stuff; you can find it; you can take moment equilibrium and solve it. Clear?
For the expression for horizontal force, you can just use what we have derived. Due to
the UDL Do you remember the formula? q
0
into L squared by 8 h into half; that is
easy to remember. You just have to plug in those values. q
0
is 30 kilonewton per meter,
L is 12 and small h is 6 meters. Then, take the eccentric load first. We have derived this:
5 by 8 into W I am calling it W
1
here. Here, a is how much? 2 meters. a is the smaller
value, please note. Then, due to the symmetric concentrated load at the crown, 25 by
128. Can I move ahead? This is easy to do; it is solved in the book. You get some
answer. You should not take too much time over this. The assignment problem is
actually a simpler problem. Clear?
Now, what do you do? You have found the support reactions. What do you do? You
draw the free body; we have to go through this. From a practicing engineer's point of
view, you need to have a feel for the bending moment diagram at least get the shape
right. Then, you need to know only the maximum and minimum values; that is all that
we expect from you. You do not need to get all the values; if you want to, you can.
Similarly, if you want to get the shear force and axial force, you can get them from the
vertical component of the force at any section and the horizontal; the horizontal is
constant throughout the arch.
(Refer Slide Time: 41:02)

Do you remember this equation? We have derived this. You have an expression for y; if
you differentiate this, you get an expression for y dash; you have an expression for
moment. What you can do is just tabulate over a few points on the arch; wherever you
have a concentrated load acting, do not miss out that point. We have done that 0
meters, 2 meters, 4, 6 minus and 6 plus show that there could be a singularity there
because of that load (6 is the middle of the arch), then 8 and 10. Every 2 meters I have
done something. It is easy to calculate all this; it is straightforward; you just have to
substitute this in the equation; we need to plug in.
We have done this for three-hinged arches. This is not something worth doing in the
class here but definitely worth doing for an assignment so that you are familiar with it.
Please go through this problem; see that you are getting the results and then you plot.
Look for the peak values. We have got the maximum moment as 108.84 and that is
exactly at the crown. It makes sense; that is where you will get it.
What about the maximum negative moment? This was checked at 8 meters but where do
you think it will be maximum? Will it be this value? Will it be here? You need to check
it out. How do you find out where to get the maximum moment, especially from the
shear force ((.)) where the shear force is changing sign? You will find that it is
changing sign at 8.92; when you work that out, you get a slightly higher value than 42.76
46.3. These are things you can do and we have done this earlier; that is all.
(Refer Slide Time: 42:55)

You can plot the bending moment diagram, the shear force diagram and normal thrust
diagram. The rough shape is all that we want; we want a rough shape and we want to
know the maximum hogging moment, the maximum sagging moment and the location.
Is it clear? I am not going through this exercise in detail because you have already done
this for a three-hinged arch and it is similar. Is it clear? We are done with arches.
(Refer Slide Time: 43:20)

It is also important to note that you get some secondary effects which are normally
ignored in a first-order analysis. What do you think are the secondary effects that a
designer should be concerned about when you do two-hinged arches? In cables also we
looked at those secondary effects. First, change in the shape. What do you think will
happen to the arch? Will it increase in length or reduce in length?
The cable will increase in length and the arch will reduce in length because of the axial
compression. That is sometimes called rib shortening, rib because the arch looks like the
rib in your skeleton. That is the first effect that might be worth considering when you
have slender arches and when the axial force in the arch is very high. The axial force in
an arch will be high when the rise of the arch is small because you know that the
beneficial effect of the horizontal thrust is not there.
Let us look at it conceptually. If the arch was a straight member with a constant axial
force, the elastic shortening will be NL by EA; you know that; EA is the axial rigidity. In
this case, you have a varying axial force and so you will have to integrate along the
curved length of the arch. Is it clear? That is the rib shortening. How will this affect the
horizontal thrust in the arch? Will it increase compared to the situation where you
ignored it? Or will it reduce? How to find out?
Let us say I want to do an exact analysis including rib shortening. What should I do? Go
back to the first principles. How did we solve for capital H? Which method did we use?
We used the energy method; we did not explicitly calculate any deflection. In that
method, how do you include rib shortening? We did not use any work done in that; we
used complementary strain energy. We have to add an additional term. A term due to
what? Complementary strain energy is equal to strain energy assuming linear elastic
behavior. What is that term we need to add?
In a frame, you can have many components to strain energy. The main component is
flexural strain energy (bending strain energy) but you also have two components from
axial strain energy and shear strain energy, which we usually treat as negligible. Here,
which is the strain energy you have to include? Axial. That is all you do. Do not forget
your fundamentals; you just have to include that second component of axial strain
energy; then, you invoke this theorem (Refer Slide Time: 46:34).
When you invoke this theorem, you get an additional term, where you account for the
actual truss; when you substitute in your capital H you can prove this, the final form
takes this shape. Look at this carefully and tell me if you were to include the axial force,
will your capital H increase or decrease? The numerator has decreased and the
denominator has ((.)). What does it mean? It will decrease. I am not asking you to
calculate but least you should know that it can be done.
(Refer Slide Time: 47:16)

The second effect that we should worry about, because it is an indeterminate structure, is
what would happen if you had a temperature rise or reduction in temperature. That can
also be worked out. Here, you can use the consistent deformation principle. Treat it as a
roller support, a simply supported beam, and let the heating take place. What do you
think will happen to the arch? It will expand in length and so the roller will move to
some distance. Then, you have to push it back and that force which you need to push it
back is capital H.
You can work this out; you can easily prove that capital H takes this form (Refer Slide
Time: 48:00). It is related to the stiffness of the arch. H increases and the last thing that
you have to worry about is important. Supposing your support slips; we have
conveniently assume that the two supports remain in place but when the horizontal thrust
is very large, it will move.
In bow string girder arches, what prevents the movement is the tie member at the base,
but the tie is not infinitely stiff; it can move. Is it clear? That movement also you can call
it a slip. If it moves, will capital H increase or decrease? It will decrease. If it moves all
the way, it will go to 0. It is a function; you can prove it; it is a function of the axial
stiffness that you get.
(Refer Slide Time: 48:54)

Finally, if you want to deal with fixed arches, you have three unknowns now. You
cannot take advantage of symmetry in loading now; you have three unknowns and so it is
convenient to take the two bending moments M
A
and M
B
and the horizontal reaction.
Earlier, we had only the horizontal reaction; you need three equations now. What are the
three equations? Vertical reactions can be expressed in terms of those unknowns. The
three equations are, remember, dou u star by dou x
1
equals 0 (in this case x
1
is H), x
2

equal to 0, M
A
and M
B
. Is it clear?
You can write an expression for the bending moment at any section in the arch in terms
of the free bending moment and the effect caused by all these values which you can write
an expression for (Refer Slide Time: 49:49). You take the derivative of this with respect
to these. You can simplify and they are easy to work out. I want you to go through this
carefully; you can write these.
(Refer Slide Time: 50:03)

If you do this and if you take the case of a symmetric arch, then your two moments will
be equal and opposite; you have only two unknowns; it simplifies to this form (Refer
Slide Time: 50:16). Remember earlier when you had only the unknown horizontal force,
you had only integral y squared by EI; now, you have this. This turns out to be a
flexibility matrix. Your two unknowns are H and M f. There are some solved problems
in the book. These are properties of the arch which you can calculate by integration. We
will stop with this; you can go through this later. Thank you very much.
(Refer Slide Time: 50:45)