THEORY of PHOTOELASTICITY

In this chapter in addition to the introduction of the concepts of temporary double refraction
and the stress-optic law the behavior of a stressed model in a polariscope will be described
shortly. A brief description of use of partial mirrors for fringe sharpening and multiplication is
also included.

1. Temporary Double Refraction

Many transparent non-crystalline materials that are optically isotropic when free from stress
become optically anisotropic and display characteristics similar to crystals when they are
stressed. These characteristics persist while loads on the materials are maintained but
disappear when the loads are removed. This behavior, known as temporary double refraction,
was first observed by Sir David Brewster in 1816. The method of photoelasticity is based on
this physical behavior of these transparent non-crystalline materials.
The optical anisotropy (temporary double refraction) which develops in a material as a result
of stress can be represented by an ellipsoid, known in this case as the index ellipsoid. The
semiaxes of the index ellipsoid represent the principal indices of refraction of material at the
point, as shown in Fig. 3.1. Any radius of the ellipsoid represents a direction of light
propagation through the point. A plane through the origin, which is perpendicular to the
radius, intersects the ellipsoid as an ellipse. The semiaxes of the ellipse represent the indices of
refraction associated with light waves having planes of vibration which contains the radius
vector and an axis of the ellipse. For a material which is optically isotropic, the three principal
indices of refraction are equal, and the index ellipsoid becomes a sphere. The index of
refraction is then the same for all directions through the material.

2. The Stress-Optic Law

The theory which relates changes in the indices of refraction of a material exhibiting temporary
double refraction to the state of stress in the material is due to Maxwell, who reported the
phenomenon in 1853. Maxwell noted that the changes in the indices of refraction were linearly
proportional to the loads and thus to stresses or strains for a linearly elastic material. The
relationships can be expressed in equation form as

n1 - n0 = c1 1 + c2 (2 + 3) (3.1a)

n2 - n0 = c1 2 + c2 (3 + 1) (3.1b)
n3 - n0 = c1 3 + c2 (1 + 2) (3.1c)

where 1, 2, 3 = principal stresses at point

Page 1 of 13
n0 = index of refraction of material in unstressed state
n1, n2, n3 = principal indices of refraction along the principal stress directions
c1, c2 = constants known as stress-optic coefficients

Equations (3.1) are the fundamental relationships between stress and optical effects and are
known as the stress-optic law. For plane stress situations (3 = 0) equations 3.1 reduce to

n1 - n0 = c11 + c22 (3.2a) n2 - n0 = c12 + c21 (3.2b)

3. The Stress-Optic Law in Terms of Relative Retardation

The method of photoelasticity makes use of relative changes in index of refraction which can
be written by eliminating n0 from equations 3.1 as

n2 - n1 = (c2 - c1) (1 - 2) = c(1 - 2) (3.3a)

n3 - n2 = (c2 - c1) (2 - 3) = c(2 - 3) (3.3b)
n1 - n3 = (c2 - c1) (3 - 1) = c(3 - 1) (3.3c)

where c = c2 - c1 is the relative stress-optic coefficient first expressed in terms of brewsters (1

brewster = 10-13 cm2 / dyn = 10-12 m2 / N = 6.895 x 10-9 in2 / lb).
Photoelastic materials are considered to exhibit positive birefringence when the velocity of
propagation of light associated with the principal stress 1 is greater than the velocity of the
wave associated with the principal stress 2. Since the principal stresses are ordered such that
123, the principal indices of refraction of a positive doubly refracting material can be
ordered such that n3 n2 n1. The form of equations 3.3 has been selected to make the relative
stress-optic coefficient c a positive constant.
Since a stressed photoelastic model behaves like a temporary wave plate, equation 2.29 can
be used to relate the relative angular phase shift  (or relative retardation) to changes in the
indices of refraction in the material resulting from the stresses. For example, consider a slice of
material (thickness h) oriented perpendicular to one of the principal-stress directions at the
point of interest in the model. If a beam of plane-polarized light is passed through the slice at
normal incidence, the relative retardation  accumulated along each of the principal-stress
directions can be obtained by substituting equation 2.29 into each of equations 3.3 to yield

12 = 2hc (1 - 2) /  (3.4a)

23 = 2hc (2 - 3) /  (3.4b)
31 = 2hc (3 - 1) /  (3.4c)

where 12 is the magnitude of the relative angular phase shift (relative retardation) developed
between components of a light beam propagating in the 3 direction. Similar interpretations
can be applied to the retardations 23 and 31.
The relative retardation  is linearly proportional to the difference between two principal
stresses having directions perpendicular to the path of propagation of light beam. The third
principal stress, having a direction parallel to the path of propagation of the light beam, has no
effect on the relative retardation. Also, the relative retardation  is linearly proportional to the
model or slice thickness and inversely proportional to the wavelength  of the light being used.
The relative stress-optic coefficient c is usually assumed to be a material constant that is
independent of the wavelength  of light being used. However it has been shown that this
coefficient may become wavelength dependent in some cases as the model material passes

Page 2 of 13
from the elastic to the plastic state. The dependence of the relative stress-optic coefficient c on
the wavelength of light being used is referred to as photoelastic dispersion or dispersion of
birefringence.
It can be shown that equations 3.4 apply not only for principal stresses but also for secondary
principal stresses '1 and'2. Thus,
' = 2hc ('1 - '2) /  (3.5)

The secondary principal stresses ('1 '2) at the point of interest lie in the plane whose normal
vector is coincident with the path of the light beam. The stress-optic law in terms of secondary
principal stresses is widely used in applications of three-dimensional photoelasticity.
For two-dimensional plane-stress bodies where 3 = 0, the stress-optic law for light at normal
incidence to the plane of the model can be written without the subscripts on the retardation
simply as

 = 2hc (1 - 2) /  (3.6)

Here it is understood that 1 and 2 are in-plane principal stresses and that 1 is greater than 2
but not greater than 3 = 0 if both in-plane principal stresses are compressive. The form of
equation 3.6 is usually simplified to

1 - 2 = N f / h (3.7) where N =  / 2 (3.8) and f =  / c (3.9)

Here N represents the relative retardation in terms of cycles of retardation (and counted as
the fringe order) and h is model thickness. While f is known as material fringe value and is a
property of model material for a given wavelength .
NOTE. In some references equation 3.7 is written as  = (1 - 2) /2 = N f* /h, where f* is the
material fringe value in terms of shear and is equal to one-half the f value defined here.
If a photoelastic model exhibits a perfectly linear elastic behavior, the difference in the
principal strains 1 - 2 can also be measured by establishing the fringe order N (by using a
polariscope).
From the stress-strain relations for a two-dimensional or plane state of stress given by
equations 1.46 it can be shown that (1 - 2) = (1+) (1 - 2) / E. Substituting this result into
equation 3.7 yields

N f / h = E (1 - 2) / (1+) (3.10)

Which is rewritten as

N f / h = 1 - 2 (3.11)

It is clear that f = (1+) f / E (3.12)

Where f is the material fringe value in terms of strain.

NOTE: Calibration of model material is required for determining the value of f or f.

For a perfectly linear elastic photoelastic model, the determination of N is sufficient to

establish both 1 - 2 and 1 - 2 if any three of the material properties E, , f , or f are known.
However, many photoelastic materials exhibit viscoelastic properties, and equations 1.46 and
3.12 are not valid.

Page 3 of 13
4. Stressed Model in a Plane Polariscope

Fig. 3.2 illustrates a plane-stressed model inserted into the field of a plane polariscope with its
normal coincident with the axis of the polariscope. Note that the principal-stress direction 1 at
the point under consideration in the model makes an angle  with the axis of polarization of
the polarizer.

The intensity of the light is proportional to the square of the amplitude of the light wave. It
can be proved that the intensity of the light emerging from the analyzer of a plane polariscope
is given by

I = K Sin2 2 Sin2 /2 (3.13)

Where  = 2 - 1 = 2h (n2 - n1) /  = 2hc (1 - 2) / 

Examination of equation 3.13 indicates that extinction (I=0) occurs either when Sin2 2 = 0 or
when Sin2 /2 = 0. Clearly, one condition for extinction is related to principal-stress directions
and the other is related to principal-stress difference.

4.1 Effect of Principal-Stress Directions

When 2 = n, where n = 0, 1, 2, ………., Sin2 2 = 0 and extinction occurs. This relation
indicates that, when one of the principal-stress directions coincide with the axis of polarizer (
= 0, /2, or any exact multiple of /2), the intensity of light is zero. When the entire model is
viewed in the polariscope, a fringe pattern is observed. The fringe pattern produced by the Sin 2

Page 4 of 13
2 term in equation 3.13 is the isoclinic fringe pattern. Isoclinic fringes are loci of points
where the principal-stress directions (either 1 or 2) coincide with the axis of plarizer.

4.2 Effect of Principal-Stress Difference

When /2 = n, where n = 0, 1, 2, ………., Sin2/2 = 0 and extinction occurs. In this case
the condition of extinction provides following relation if we substitute the value of  from
equation 3.6.

n = /(2) = hc (1 - 2) /  (3.14)

Now n = 0 implies that the principal stress difference is zero (irrespective of the wavelength of
the light used). While n = 1, 2, 3, …… (assuming monochromatic light) are termed as integral
order isochromatics which are going to be black due to the extinction caused by Sin2/2 term
of equation 3.13. The fringe pattern produced by Sin2/2 term of equation 3.14 is termed as
isochromatic fringe pattern. In the light of equation 3.14 isochromatic fringe may be defined as
the loci of points where the principal stress difference is constant.
The isochromatic fringe corresponding to principal stress difference equal to zero appears
black, even if a white light source is employed. However all other constant levels (except zero)
of principal stress differences appear as contours of various colors (not black because a given
level of principal-stress difference is unable to produce relative retardation equivalent to
integral number of wavelengths for all the colors of white light spectrum). The colors observed
in case of white light illumination result from the overall optical effect of different retardations
of various wavelengths for a constant level of principal stress difference.
As the level of principal stress difference goes on increasing the sequence of colors observed
in case of white light illumination is: Black, whitish, yellowish, reddish, bluish, yellowish,
reddish, greenish, yellowish, reddish, greenish,……… In some references reddish to bluish
and reddish to greenish transitions have been termed as tint of passage Additionally values of
n=1 and n = 2, 3, 4, … have been assigned to these color transitions.
At low level of principal stress difference only one color of white light spectrum may be
extinguished completely hence the complementary colors observed in case of white light
illumination are relatively sharp. At the higher levels of principal-stress difference, where
several wavelengths of light can be extinguished simultaneously, e.g., second-order red and
third-order violet, the isochromatic fringes become pale and very difficult to identify, and
should not be used for analysis. On the other hand with monochromatic light, the individual
fringes in an isochromatic fringe pattern remain sharp and clear to very high orders of
extinction.
Since wavelength of monochromatic light is fixed equation 3.14 can be written as

n = N = h (1 - 2) / f (3.15)

The number of fringes appearing in an isochromatic fringe pattern is controlled by the

magnitude of the principal stress difference 1 - 2, by the thickness h of the model, and by the
sensitivity of the photoelastic material, as denoted by the material fringe value f.
NOTE: Sometimes theoretically, the isoclinic and isochromatic fringes should be lines of zero
width; however, in the photographs fringes generally appear as bands with considerable width.
Also, direct visual examination of the fringe pattern in a polariscope will show again that
fringes are bands and not lines. In both instances, the width of the fringes is due to the
recording characteristics of the eye and the photographic film and not to inaccuracies in the
pervious development. If the intensity of light emerging from the analyzer is measured with a

Page 5 of 13
photodiode, a minimum intensity is recorded at some point near the center of the fringe that
coincides with the exact extinction line.

4.3 Frequency Response of a Polariscope

The frequency f for light in the visible spectrum is approximately 10 15 Hz. As a result of this
very high frequency, neither the eye nor any type of existing high-speed photographic
equipment can detect the periodic extinction associated with t term in the expression for the
light wave.

5. Effects of a Stressed Model in a Circular Polariscope (Standard Dark Field

Arrangement)

Fig. 3.5 illustrates a plane-stressed model inserted into the field of a circular polariscope (in a
standard dark field arrangement), with its normal coincident with the axis of the polariscope.

It can be proved that the light emerging from the analyzer of a circular polariscope (standard
dark field arrangement) is given by

I = K Sin2 /2 (3.16)

This result indicates that the intensity of light beam emerging from the analyzer of the circular
polariscope is a function only of the principal-stress difference since the angle  does not
appear in the expression for the amplitude of the wave. This fact indicates that the isoclinics
have been eliminated from the fringe pattern observed with the circular polariscope.

5.1 Effects of a Stressed Model in a Circular Polariscope (Standard Light Field

Arrangement)

Page 6 of 13
The circular polariscope can be converted from standard dark field arrangement into standard
light field arrangement simply by rotating the analyzer through 90 o. The advantage of
employing both light and dark field arrangements is that twice as many data are obtained for
the whole-field determination of 1 - 2. Recall that the order of the fringes N coincides with n
for the plane polariscope and for the dark-field circular polariscope; therefore, the fringes are
counted in the sequence 0, 1, 2, 3, ……However, with the light-field arrangement of the
circular polariscope (illustrated in Fig. 3.9), N and n do not coincide, for the reasons given
below.

For the light field arrangement it can be proved that

I = K Cos2 /2 (3.17)

Above equation shows that extinction (I=0) will occur when

/2 = [(1+2n)/2]  for n = 0, 1, 2, 3, …….

And N =  / 2 = (1/2) + n (3.18)

Which implies that the order of the first fringe observed in a light-field polariscope is 1/2,
which corresponds to n = 0. Isochromatic fringe patterns for the dark- and light-field
arrangements are illustrated in Figs. 3.10a and 3.10b respectively.

Page 7 of 13
Page 8 of 13
6. Effects of a Stressed Model in a Circular Polariscope (Arbitrary Analyzer Position,
TARDY COMPENSATION)

Tardy Compensation method is used for determining fractional fringe orders. This method
involves the rotation of the analyzer.
It can be proved that the intensity of the light emerging from the analyzer of a circular
polariscope with the analyzer oriented at an arbitrary angle  with respect to its dark field
position is given by the following relation.

I = K (1- Cos 2 Cos  - Cos 2 Sin 2 sin ) (3.19)

It can be shown that for minimum intensity following relation holds

 = n/2 and  = 2  2n , n = 0, 1, 2, 3, ……… (3.20)

The above conditions for extinction (I=0) indicates that a principal-stress direction must be
parallel to the axis of polarizer i.e.  = 0, /2, …. (NOTE: This conditioning is practically
attained by passing an isoclinic fringe through the point at which the value of isochromatic
fringe order is to be determined). The fringe order at the point is then

N =  / 2 = n   /  (3.21)

To illustrate the procedure of Tardy compensation, consider the hypothetical dark-field fringe
pattern and points of interest shown in Fig. 3.12. At point P 1, which lies between fringe orders
2 and 3, the value assigned to n is 2. As the analyzer is rotated through an angle , the second-
order fringe will move towards point P 1 until extinction is obtained. The fringe order at P 1 is
then given by N = 2 +  / . For point P2 the value of n is also taken as 2, and the analyzer is
rotated through an angle 1 until the second-order fringe produces extinction, giving a value
for the fringe order of N= 2 + 1 / . In this instance n could also be taken as 3, and the
analyzer rotated in the opposite direction through an angle -2 until the third-order fringe
produced extinction at point P2. In this instance the fringe order would be given by
N = 3 - 2 / .
The accuracy of Tardy Compensation method depends upon the quality of quarter-wave plates
employed in the polariscope; however, accuracies of  0.02 fringes can usually be achieved.

Page 9 of 13
6. Fringe Sharpening with Partial Mirrors

The bandwidth of the isochromatic fringe can be reduced by a novel technique due to Post,
which employs partial mirrors in a circular-lens polariscope. The partial mirrors are inserted
into the field of the polariscope on both sides of the model and parallel to it, as illustrated in
Fig. 3.15.

The effect of partial mirrors is to cause the light to propagate back and forth through the
model in the manner illustrated in Fig. 3.16. As the light is reflected back and forth between
the two mirrors, a portion of it is transmitted at each reflection point. Hence, the intensity of
the ray passing back and forth through the model is progressively decreasing. For instance, ray
1 is the most intense, ray 3 less intense, etc. The rays shown in Figure 3.16 are drawn
obliquely only to present the effects of the partial mirrors; in practice, normal incidence is
employed and all the rays coincide and enter and emerge from the same point in the model.
The reflection and transmission coefficients of the partial mirror are normally indicated by R
and T. The expression for the intensity of kth ray is given below.

Ik = KT2 Rk-1 Sin2 (k/2) k = 1, 3, 5, 7,……. (3.22)

Page 10 of 13
These intensities I1, I3, I5,…….Ik add arithmetically; hence the resultant intensity of the
superimposed rays is given by the series expansion

I = KT2  Rk-1 Sin2 (k/2) (3.23)

If this relationship is expanded and plotted as a function of /2 = N, the intensity-versus-

fringe-order plot shown in Fig. 3.17 is obtained. When this plot is compared with the
conventional intensity-versus-/2 plot, also shown in Fig. 3.17, it is clear why the fringes are
sharpened. The eye will begin to record a fringe at some minimum intensity I 0; hence the
sharpened intensity function produces a much narrower fringe than the conventional intensity
function.

7. Fringe Multiplication with Partial Mirrors

Post has shown that partial mirrors can be usefully employed to multiply the number of fringes
which can be observed in a photoelastic model. Fringe multiplication is quite important since
the standard methods of compensation used to evaluate fractional fringe orders are time-

Page 11 of 13
consuming. Fringe multiplication is, in a sense, a whole field compensation technique where
the fractional orders of the fringe can be determined simultaneously over the entire field of the
model.

When partial mirrors are used in fringe multiplication, they are inserted into a lens polariscope
on both sides of the model with one of the mirrors inclined slightly, as illustrated in Fig 3.18.
The effect of the inclined mirror is to reflect the light back and forth through the model, as
illustrated in Fig. 3.19. From this figure it is clear that each ray of light emerges from the
mirror system at an angle which depends on the number of times the light ray has traversed the
model. For instance, rays 1, 3, 5, and 7, which have traversed the model the same number of
times as their ray number, emerge at angles 0, 2, 4, and 6. Although the rays do not pass
through the same point, the inclination angle  used in the illustration was greatly exaggerated.
The length of the path over which the photoelastic effect is averaged depends upon the angle
of inclination , the ray number, and the separation distance between the mirrors. In practice,
multiplication by a factor 5 to 7 can be achieved without introducing objectionable errors due
to the averaging process, which is inherent in this method.

The fact that different rays of light are inclined at different angles with respect to the axis of
polariscope permits each ray to be isolated. The rays are all collected by the field lens but
focussed at different points in the focal plane of the field lens (see Fig. 3.18). Any one of these
rays can be observed by placing the eye or a camera lens at the proper image point. A
diaphragm stop is used to eliminate all images except the one under observation.
Suppose that the light- and dark-field photographs are obtained for rays 1, 3, and 5. The fringe
patterns recorded from ray 1 may be interpreted in the conventional sense where the orders of

Page 12 of 13
the fringes are sequenced as 0, 1/2, 1, 3/2, 2, 5/2,……. However, for ray 3, the light has
passed through the model 3 times and the order of the fringes are sequenced as 0, 1/6. 1/3,
1/2, 2/3, 5/6, 1, ……… Finally, for ray 5, the orders of the fringes are sequenced as 0, 1/10,
1/5, 3/10, 2/5, 1/2, ….. The superposition of the results obtained from these 3 rays is sufficient
to determine the fringe order to the nearest one-tenth of an order over the entire model. The
fringe-multiplication technique can be interpreted as a whole field compensation method where
fractional orders of the fringes can be determined with a high degree of accuracy.
The intensity relationship for the mth ray, where m = 1, 2, 3, etc. as shown in Fig. 3.19 is given
below.

Im = KT2 Rm-1 Sin2 (m/2) (3.24)

For minimum intensity loss it can be proved that

R = (m-1) / (m+1) (3.25)

Since R is a function of m (the number of times light traverses through the model), it is not
possible to optimize the mirrors for all rays simultaneously. A multiplication factor of 5 is
usually sufficient for most applications and, for this case, the mirrors should have R = 0.667
and T = 0.333 . It has been assumed that R + T = 1.

Page 13 of 13