Smart Structures

Professor Mohammed Rabius Sunny

Department of Aerospace Engineering
Indian Institute of Technology, Kharagpur
Week 07
Lecture No: 39
Analysis of composite laminate with piezoelectric patches(continued)
Part 04

Today, we will continue our discussion on Analysis of Composite Laminates which has
piezoelectric patches.

In the previous lecture, we saw a case where the deformation was purely in plane, it was
because our actuators were actuated in such a way that it generates only in plane force and
also the composite laminate was symmetric.

Now, today we will see a pure bending case. Now again the structure is same, we have a
composite laminate and the laminate is symmetric, dimensions are same, this is our x axis,
this is y. So, this dimension is L and this is c by 2 and this is c by 2 and it is fixed at one
end. And if we look at the view along the x z plane, it looks like this.

So, it has a piezoelectric patch and again we can assume that it is throughout, but that does
not matter even if we keep it at a small part of it, the analysis remains same. So, let us
assume that our thickness of the laminate is tb and thickness of the two patches are tc and
the laminate is symmetric. So, this is the differential equation that ordinary differential
equation, that we derived considering the Rayleigh-Ritz technique. So, we are solving it
using Rayleigh-Ritz method.

[𝑀𝐼𝐼 ]{𝑞̈ 𝐼 } + [𝑀𝐼𝑊 ]{𝑞̈ 𝑊 } + [𝐾𝐼𝐼 ]{𝑞𝐼 } + [𝐾𝐼𝑊 ]{𝑞𝑤 } = {𝐹𝐼 }

[𝑀𝑊𝐼 ]{𝑞̈ 𝐼 } + [𝑀𝑊𝑊 ]{𝑞̈ 𝑊 } + [𝐾𝑊𝐼 ]{𝑞𝐼 } + [𝐾𝑊𝑊 ]{𝑞𝑤 } = {𝐹𝑊 }

Now, within this differential equation ah these terms are 0, these terms do not come into
picture because this is 0 because it is a static problem. The actuation is static. So, there is
no acceleration and KI, KI, KIW equal to KIW equal to 0 as before since B matrix is 0,
symmetric laminate. So, we are left with only this equation. So, these two equations i.e.,
the first equation and the second equation can be solved separately.

Now, in this case the actuation is purely a bending actuation. So, we need to solve only
the only this equation. So, solve KWW qW is equal to Fw. So, now as before we need to
make some approximations. So, let us go to next slide.

(Refer Slide Time: 4:00)

Our approximation was a function of x y is the approximation that we make is as x square
by l square multiplied by qWI. So, this is we can call it phiWI and this is our this is qWI.

𝑥2
𝑤(𝑥, 𝑦) = 2 𝑞𝑊𝐼
𝑙
Now, this satisfies the essential boundary condition. So, at x equal to 0, w is 0 and at x
equal to 0 dw by dx is equal to 0. So, the cantilever boundary condition is satisfied slope
and displacements at x equal to 0 is 0.

Now, we need to find out our KWW. So, K we are just writing it as K because other
components of K are not there anymore and to do that, we have the expression as BW
transpose, multiplied by the D matrix, multiplied by BW and it is integral. Now, we need to
find out this BW transpose and to find out the BW transpose, we need our quantities like
kappa.

𝐾 = ∫[𝐵𝑊 ]𝑇 [𝐷][𝐵𝑊 ] 𝑑Ω
Ω

So, kappa as we know is del 2 w by del x 2 this is del 2 w by del y 2 and this is 2 del 2 w
del x del y and as per the approximation, this can be written as del 2 phiWI by del x 2. This
quantity is 0 because phi is a function of only x. So, again it is a very simplified
approximation. We are assuming that the bending i.e, the w is function of only x. So,
bending is with respect to only the y axis. And this quantity is 0 and this quantity is 0. And
this is equal to qWI. And also, del 2 phiWI by del x 2 is: if I just differentiate it twice, it
becomes 2 by l square. And this is our BW matrix we can say.
 𝜕 2𝑤
𝜕𝑥 2 𝜕 2 𝜙𝑊𝐼 2
2
𝜕 𝑤 2 2
𝐾= = { 𝜕𝑥 } = {𝑙 }
𝜕𝑦 2 0 0
𝜕 2𝑤 0 0
2
{ 𝜕𝑥𝜕𝑦}

And also, we need the MP's and MP's have only one component and which is this. Again,
that is an approximation. So, we have only one component of MP which is MXP and 0 0.
And we know how to find out MXP, we have done it before.

𝑀𝑃𝑋
𝑀𝑃 = { 0 }
0
So, finally, if we put everything in this expression this becomes an integral over x and y 0
0. And then, we have D matrix here and here we have 0 0. Let us write this as an integral
from x and y. So, this quantity reduces to del 2 phiWI by del x 2, D11, del 2 phiWI, del x 2
dx dy. And finally, on being integrated, this gives us 4 c. So, this is DXX generally as per
our convention. So, this D matrix many a times we write it as D11, D12, I mean all these A
B or D matrix, sometimes we write it from write it as A11 A12 A16 or we can write it as DXX
DXY and DXS. So, some books you will find out these subscripts to be 12 or 6 and in some
of the books you may find it to be x y or s and both are same.

𝑐/2 𝑙 𝜕 2 𝜙𝑊𝐼
𝜕 2 𝜙𝑊𝐼 2
𝐾11 = ∫ ∫ { 0 0} [𝐷] { 𝜕𝑥 } 𝑑𝑥𝑑𝑦
𝜕𝑥 2 0
−𝑐/2 0
0
𝑐/2 𝑙
𝜕 2 𝜙𝑊𝐼 𝜕 2 𝜙𝑊𝐼
𝐾11 = ∫ ∫ [𝐷] 𝑑𝑥𝑑𝑦
𝜕𝑥 2 𝜕𝑥 2
−𝑐/2 0

4𝐶𝐷𝑋𝑋
𝐾11 =
𝑙3
So, this is our expression for K and K has only one component. So, let's call it K11.

Now we have to find out the right-hand side the force term which comes due to the
piezoelectric actuation. Now, just to note one thing this we are we have been calling MPX
not MXP same thing.

(Refer Slide Time: 10:00)

Now, we have to find out the force term. So, we have FP is equal to minus c by 2, c by 2, 0
to l and then we have del 2 phiWI by del x 2 0 0. And this is multiplied with this vector MPX
0 0 and we have this.
𝑐/2 𝑙
𝑀𝑃𝑋
𝜕 2 𝜙𝑊𝐼
𝐹𝑃 = ∫ ∫ { 0 0} { 0 } 𝑑𝑥𝑑𝑦
𝜕𝑥 2 0
−𝑐/2 0

So, finally, this turns out to be this.

𝑐/2 𝑙
𝜕 2 𝜙𝑊𝐼 2𝑀𝑃𝑋 𝑐
𝐹𝑃 = ∫ ∫ 𝑀𝑃𝑋 2
𝑑𝑥𝑑𝑦 =
𝜕𝑥 𝑙
−𝑐/2 0

And then on being integrated the value comes to be 2 MPX c by L, and then we can solve
the equation and qw1 can come to. So, this is again it has only one component let us call it
FP1 by K11 and that gives me qw1 to be MPX, l square by 2 DXX. So, this is our qw1 and then
we can substitute qw1 to the original approximation and that gives us w as a function of x.

𝐹𝑃𝐼 𝑀𝑃𝑋 𝑙 2
𝑞𝑊𝐼 = =
𝐾11 2𝐷𝑋𝑋

Now, we will solve this similar problem using the Galerkin technique.

(Refer Slide Time: 11:47)

To do it using Galerkin technique, we need to again make a different approximation
because in Galerkin technique when we are having a purely bending problem, the highest
order derivative appearing is 4. So, accordingly the approximation has to be changed. So,
let us approximate w x as 6 x by l square, minus 4 x by l cube, plus x by l to the power 4
multiplied by qw1. So, this is our approximation now. So, this is our phiW1. Now this
satisfies both essential and natural boundary condition which means the geometric and
force boundary conditions.

𝑥 2 𝑥 3 𝑥 4
𝑤(𝑥, 𝑦) = [6 ( ) − 4 ( ) + ( ) ] 𝑞𝑊𝐼
𝑙 𝑙 𝑙
So, satisfies geometric and force boundary conditions. So, those conditions are w 0 y is 0,
now del w by del x at x equal to 0 is 0. And then we have del 2 w by del x 2, it is a cantilever
beam. So, at x equal to l is 0 and then we have del 3 w by del x 3 at x equal to l is 0. It has
the derivatives up to fourth order existing. So, we need to make sure that minimum fourth
order derivative exists because that is that is the highest order of derivative that would
appear in our equation.

𝜕 2𝑤
𝑤(0, 𝑦) = 0 | =0
𝜕𝑥 2 𝑥=𝑙
𝜕𝑤 𝜕 3𝑤
| =0 | =0
𝜕𝑥 𝑥=0 𝜕𝑥 3 𝑥=𝑙

So, the equation is DXX del 4 w by del x 4 is equal to del 2 MPX by del x 2.
 𝜕 4 𝑤 𝜕 2 𝑀𝑃𝑋
𝐷𝑋𝑋 4 =
𝜕𝑥 𝜕𝑥 2
And then, we define an error as we have been doing and the error is DXX multiplied by the
fourth order derivative of w with respect to x minus this.

𝜕 4 𝑤 𝜕 2 𝑀𝑃𝑋
𝜖 = 𝐷𝑋𝑋 −
𝜕𝑥 4 𝜕𝑥 2
And then we say that we have to multiply this error with phiw1 and on being integrated over
the domain that should be 0. So, let us do that. Now we multiply phiw1 minus del 2 MPX by
del x 2 dx dy is equal to 0. So, this quantity would give us a term where I have q W1 and
this would give us a pure force term and by solving that we can find out our qW1.
𝑐/2 𝑙
𝜕 4 𝑤 𝜕 2 𝑀𝑃𝑋
∫ ∫ 𝜙𝑊𝐼 (𝐷𝑋𝑋 − ) 𝑑𝑥𝑑𝑦 = 0
𝜕𝑥 4 𝜕𝑥 2
−𝑐/2 0

So, if these entire expressions are evaluated finally, qW1 comes to be 5 by 36 multiplied by
MPX l square by DXX.

5 𝑀𝑃𝑋 𝑙 2
𝑞𝑊𝐼 = ( )
36 𝐷𝑋𝑋

And then it can be put in the original approximation and our solution is done.

(Refer Slide Time: 16:23)

Now, we will solve another problem where we have a anti symmetric laminate.

Now, we know that, in an anti symmetric laminate our B matrix is: so, with this anti
symmetric laminate is angle ply. It is not a cross ply. Because depending on whether the
anti laminate is angle ply or cross ply our B matrix changes. So, this is BXS this is 0 0, BYS,
this is BXS BYS 0.

0 0 𝐵𝑋𝑆
[𝐵] = [ 0 0 𝐵𝑌𝑆 ]
𝐵𝑋𝑆 𝐵𝑌𝑆 0

And then we have A matrix and we know that for any balance laminate, our A matrix has
AXS as 0, 0 0 0, ASS.

𝐴𝑋𝑋 𝐴𝑋𝑌 0
[𝐵] = [𝐴𝑋𝑌 𝐴𝑌𝑌 0 ]
0 0 𝐴𝑆𝑆

And then we have D matrix and because this laminate is anti symmetric. So, our DXS DYS
are also 0, 0 0 DSS.

𝐷𝑋𝑋 𝐷𝑋𝑌 0
[𝐷] = [𝐷𝑋𝑌 𝐷𝑌𝑌 0 ]
0 0 𝐷𝑆𝑆

And again, our structural details remain same. We have a plate laminate which has a
dimension of l along the x axis and a dimension of c along the y axis. And if we look at the
x z plane again, we have piezoelectric patch at the top and bottom and the patches are of
same property. Both material and geometric property are same.

So, this is tb this is tc and this is our x axis, this is z axis. Now, in this laminate, the
piezoelectric patches which are put here, if they are actuated even in the in purely in plane
mode, suppose I give same voltage at the top and bottom, then also it would induce some
out of plane deformation because our B matrix is non-zero. So, now because in the B
matrix, we have BXS and BYS present. So, it basically a coupling between the in plane
deformation and the twisting. Now, please understand that BXS, BXY and BYY are 0 and that
is why any in plane loading, any in plane actuation is not going to cause any bending, but
it would it is going to cause twisting. So, here twisting and in plane deformations are
coupled.

Now, there is no coupling between bending and twisting, but there is a coupling between
in plane normal strain components and torsional strain. So, we need to approximate both u
and w and solve it accordingly. So, now let us make some approximation. So, assume u 0
as a function of x and y is the approximation that we take is x by l, qu1.
So, this is our phiu1, and for w, let us make the approximation to be x square y by l square
c, qw1. And this is called phiw1. And we have v0 as 0.

𝑥 𝑥2𝑦
𝑢0 (𝑥, 𝑦) = 𝑞 𝑤(𝑥, 𝑦) = 𝑞
𝑙 𝑢1 𝑙 2 𝑐 𝑢1
Now in our approximation for w, we took a y component also because we expect a twisting
to be induced that is why we do not have a purely x component, I mean, w x y is not just a
function of x it is a function of y as well. And u0 is function of x only. Now all we need to
do is we need to solve that solve two equations. So, finally, in this case the equation would
look like this.

We will have KIW multiplied by qI plane plus: we just wrote the equation in the, ok. So, we
have to solve two equations here. We have KII multiplied by qI which we call in plane plus
KIW multiplied by qw is equal to FI. And here we have KWI multiplied by qI plus KWW
multiplied by qW is equal to FW. So, we have to solve these two equations. So, we need to
evaluate these matrices.

[𝐾𝐼𝐼 ]{𝑞𝐼 } + [𝐾𝐼𝑊 ]{𝑞𝑊 } = {𝐹𝐼 }

[𝐾𝑊𝐼 ]{𝑞𝐼 } + [𝐾𝑊𝑊 ]{𝑞𝑊 } = {𝐹𝑊 }

(Refer Slide Time: 23:36)

So, finally, this entire thing looks like this if we write it in the, I mean, as per their
expressions. So, this is BI T multiplied by A matrix, multiplied by BI transpose dx dy and
we have qu1 plus minus c by 2, c by 2, 0 to L, then we have BI T multiplied by the B matrix,
multiplied by BW (transpose should not be there) and then we have dx dy. And finally, at
the right-hand side we have this is multiplied by qw1 and at the right-hand side we have
minus c by 2, c by 2, 0 to l. And then we have BI transpose into NP dx dy.
𝑐/2 𝑙 𝑐/2 𝑙

∫ ∫[𝐵𝐼 ]𝑇 [𝐴][𝐵𝐼 ]𝑑𝑥 𝑑𝑦 𝑞𝑢1 + ∫ ∫[𝐵𝐼 ]𝑇 [𝐵][𝐵𝑊 ]𝑑𝑥 𝑑𝑦 𝑞𝑤1

−𝑐/2 0 −𝑐/2 0
𝑐/2 𝑙

= ∫ ∫[𝐵𝐼 ]𝑇 {𝑁𝑃 }𝑑𝑥 𝑑𝑦

−𝑐/2 0

So, again if we follow the same procedure, this comes to be minus c by 2, c by 2, 0 to l.

And then, we have del phiu1 by del x, 0 0. And we have the A matrix here and this becomes
del phiu1 by del x, 0 0 dx dy, qu1 plus we have same thing del phiu1 by del x, 0 0 and here
we have the B matrix and here we have del 2 phiw1 del x 2, 0 twice of del 2 phiw1 del x del
y. Now please understand if we look at the ah approximations for w this phi w1, if we
differentiate it twice with respect to x there is a non zero derivative. If we differentiate it
twice with respect to y there is a zero derivative. And if we differentiate it with respect to
x once and y once again there is a non-zero derivative. So, that is why the middle term is 0
and the and the other two terms are non-zero. dx dy. And this is multiplied by qw1 minus c
by 2, c by 2, 0 to l. And this quantity is del phiu1 by del x, 0 0. Here we have NPX 0 0 and
dx dy.

𝑐/2 𝑙 𝜕𝜙𝑢1
𝜕𝜙
∫ ∫ { 𝑢1 0 0} [𝐴] { 𝜕𝑥 } 𝑑𝑥 𝑑𝑦 𝑞𝑢1
𝜕𝑥 0
−𝑐/2 0
0
𝜕 2 𝜙𝑤1
𝑐/2 𝑙
𝜕𝜙 𝜕𝑥 2
+ ∫ ∫ { 𝑢1 0 0} [𝐵] 0 𝑑𝑥 𝑑𝑦 𝑞𝑤1
𝜕𝑥 2
𝜕 𝜙𝑤1
−𝑐/2 0
2
{ 𝜕𝑥𝜕𝑦 }
𝑐/2 𝑙
𝜕𝜙𝑢1 𝑁𝑃𝑋
= ∫ ∫{ 0 0} [{ 0 } 𝑑𝑥 𝑑𝑦
𝜕𝑥 0
−𝑐/2 0

So, after all these evaluations this quantity comes to be minus c by 2, c by 2, 0 l. So, here
we have del phiu1 by del x, AXX multiplied by del phiu1 by del x, dx dy, qw1. We are just
evaluating this term. And then we have minus c by 2, c by 2, 0 to l. If we evaluate this term,
we will get del phiu1 by del x multiplied by B11. So, it is it is not B11, it will be BXS BXS
multiplied by twice of del phiw1 x y by del x del y, and dx dy qw1. Finally, at the right-hand
side we have minus c by 2, c by 2, 0 l del phiu1 by del x NPX dx dy.
 𝑐/2 𝑙 𝑐/2 𝑙
𝜕𝜙𝑢1 𝜕𝜙𝑢1 𝜕𝜙𝑢1 𝜕 2 𝜙𝑤1
∫ ∫( ) 𝐴𝑋𝑋 ( ) 𝑑𝑥 𝑑𝑦 𝑞𝑢1 + ∫ ∫ ( ) 𝐵𝑋𝑆 (2 ) 𝑑𝑥 𝑑𝑦 𝑞𝑤1
𝜕𝑥 𝜕𝑥 𝜕𝑥 𝜕𝑥𝜕𝑦
−𝑐/2 0 −𝑐/2 0
𝑐/2 𝑙
𝜕𝜙𝑢1
= ∫ ∫( ) 𝑁𝑃𝑋 𝑑𝑥 𝑑𝑦
𝜕𝑥
−𝑐/2 0

So, there is one equation where qu1 and qw1 are the unknowns.

(Refer Slide Time: 29:26)

Next, we have to form another equation and, in that equation, would be again with qu1 and
qw1 as the unknowns and that equation would look like this. 0 to L and here we have BI
transpose multiplied by BW transpose multiplied by the B matrix, dx dy. And this is
multiplied with qu1 plus minus c by 2, minus c by 2, 0 to l BW transpose multiplied by D
multiplied by BW dx dy. And then at the right-hand side we have del 2 w1 by del x 2, 0, 2
phiw1 del x del y, but there is no actuation because the actuation is purely in plane actuation.
Our piezo's are actuated symmetrically with same voltage. So, we can write it here. Pure
in plane actuation.
 𝑐/2 𝑙 𝑐/2 𝑙

∫ ∫[𝐵𝑊 ]𝑇 [𝐵][𝐵𝐼 ]𝑑𝑥 𝑑𝑦 𝑞𝑢1 + ∫ ∫[𝐵𝑊 ]𝑇 [𝐷][𝐵𝑊 ]𝑑𝑥 𝑑𝑦 𝑞𝑤1

−𝑐/2 0 −𝑐/2 0
𝑐/2 𝑙
𝜕𝜙𝑤1 𝜕 2 𝜙𝑤1 0
= ∫ ∫{ 0 2 } {0} 𝑑𝑥 𝑑𝑦
𝜕𝑥 𝜕𝑥𝜕𝑦 0
−𝑐/2 0

Now, if these expressions are simplified, this comes to be this 2 phiw1 x y. Then we have
BXS phiu1 x dx dy qu1 and then we have minus c by 2, c by 2, 0 to L and here we have
contributions from D matrix and that is phiw1 x x DXS phiw1 x x plus phiw1 x y and then we
have DSS again twice phiw1 comma x y dx dy. And then at the right-hand side we have 0.
𝑐/2 𝑙

∫ ∫ 2𝜙𝑤1,𝑥𝑦 𝐵𝑋𝑆 𝜙𝑢1,𝑥 𝑑𝑥 𝑑𝑦 𝑞𝑢1

−𝑐/2 0
𝑐/2 𝑙

+ ∫ ∫(𝜙𝑤1,𝑥𝑥 𝐷𝑥𝑥 𝜙𝑤1,𝑥𝑥 + 2𝜙𝑤1,𝑥𝑦 𝐷𝑠𝑠 2𝜙𝑤1,𝑥𝑦 )𝑑𝑥 𝑑𝑦 𝑞𝑤1 = 0

−𝑐/2 0

So, here the derivatives are written as comma with subscripts and here it was written as the
derivative itself, but again they are same. So, we have got 2 equations, the previous
equation is this. Here I have unknowns qu1, qw1. Next equation is this. Here I have
unknowns qu1, qw1. And so, 2 equations 2 unknowns we can solve these 2 equations. After
solving these equations, we can find out qu1 and qw1. So, let us give some number to it
maybe we call it 1 and maybe we call it 2. So, solving 1 and 2, qu1, qw1 can be found out.
So, in this problem ah it was an antisymmetric laminate the actuation was a pure in plane
actuation. Now although the actuation was actuation was in plane actuation because of the
B matrix being non zero we had to solve the 2 coupled equations. So, we got qu1 and qw1
each of these u0 and w had 1 term in the approximation.

So, after this solution we can find out the constants associated with each of these terms
unknown constants and after getting these 2 constants we can again back substitute and
find out your u0 and w and that solves the problem.

(Refer Slide Time: 34:20)

So, with this I would conclude this lecture.

Thank you.