Metrology

L. Motra & P. R. Shakya

ME 503

Lecture 2

1/24/2025
PRShakya 1
 Metrology and Measurement
Lecture 2
Errors in Measurement
➢ Types and Sources of Errors
➢ Error Propagation
➢ Effect of Averaging Results
➢ Method of Least Squares
➢ Calculations of Error
➢ Assignment

1/24/2025 1
 Errors in Measurements
➢ It is never possible to measure the true value of a dimension,
there is always some error. So, measurement is merely an estimated
value of dimension.
➢The error in the measurement is the difference between the
measured value and the true value of measured dimensions.
Error in Measurement = Measured value-True value
➢The error in measurement may be expressed or evaluated either as
an absolute error or as a relative error.

1/24/2025 3
 Errors in Measurements
Absolute Error

➢ True absolute error: It is the algebraic difference between the result of

measurement and the conventional true value of the quantity measured.

True Absolute Error = | Measured Value – Conventional True Value |

➢ Apparent absolute error: If the series of measurement are made then the
algebraic difference between one of the result of measurement and the
arithmetical mean is known as apparent absolute error.

Apparent Absolute Error = | Measured Value – Arithmetic Mean Value |

1/24/2025 4
 Errors in Measurements
Relative Error

➢ It is the quotient of the absolute error and the value of comparison (may
be true value or the arithmetic mean of a series of measurements) used for
calculation of the absolute error.

➢ It is an error with respect to the actual value.

| 𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑑 𝑉𝑎𝑙𝑢𝑒 − 𝑇𝑟𝑢𝑒/𝑀𝑒𝑎𝑛 𝑉𝑎𝑙𝑢𝑒 | Xi − X0
Relative Error = , erelative =
| 𝑇𝑟𝑢𝑒/𝑀𝑒𝑎𝑛 𝑉𝑎𝑙𝑢𝑒 | X0
Xi − X
=
X
1/24/2025 5
 Errors in Measurements
Uncertainty
Uncertainty of measurement means doubt about the validity of the result of a measurement. A
measure of range of measurements from the average value. Also called standard deviation.
Uncertainty characterize the dispersion of the measured quantity values.
Average Deviation: Estimation of Uncertainty by Repeated Measurements
The statistical method for finding a value with its uncertainty is to repeat the measurement
several times, find the average value, and find the either the average deviation or the standard
deviation.
Standard Deviation
The statistical measure of uncertainty. Average deviation
1/24/2025 6
 Errors in Measurements
Factors Affecting the Amount of Error

➢ The accuracy and design of the measuring instruments

➢ The skill of the operator

➢ Method of measurement

➢ Temperature variations

➢ Elastic deformation of the part or instrument etc.

1/24/2025 7
 Errors in Measurements
Sources of Errors
i. Calibration Error
Each measures instrument should be calibrated with a standard one at certain time
interval (may be once in a year or once in every 6 months.)
If the above procedure is not followed the instrument may give erroneous result, it
is called calibration errors.

ii. Environmental Error

These errors are due to surrounding pressure, temperature and humidity.
Internationally agreed standard value of temperature pressure are:
0
Temperature = 20 C.
Pressure = 760 mm of Hg + 10 mm of Hg vapour pressure
If the ambient condition varies from the above standard values the measured value
will be erroneous.
1/24/2025 8
 Errors in Measurements
Sources of Errors
iii. Contact Pressure/Stylus Pressure
Errors are also introduced due to pressure
exerted at stylus. It is more prominent in case
of soft work piece. Ideally the stylus should
touch the top surface of the w/p, due to stylus
pressure both deformation and deflection of
w/p take place. This type of errors are also
induced when the force applied on the anvils
of micrometer varies.
To minimize this error:
Skill inspector
Fit with ratchet mechanism, the ratchet slips
when the applied pressure exceeds the
minimum required operating pressure.
1/24/2025 9
 Errors in Measurements
Sources of Errors

iv. Error due to Supports

The elastic deformation/deflection of a For Minimum Deflection

long measuring bar due to position of
support cause error in measurement.
So, Sir G.B. Airy found out the position
of supports to give minimum error.
Two support conditions are shown in
figures. For zero slope at ends

1/24/2025 10
 Errors in Measurements
Sources of Errors
v. Error due to Alignment

➢ Abbe’s alignment principle: It states that “ the axis or line of measurement

should coincide with the axis of measuring instrument or line of the measuring
scale.”

➢ If while measuring the length of a work-piece the measuring scale is inclined to

the true line of the dimension being measured there will be an error in the
measurement.

➢ The length recorded will be more than the true length. This error is called “Cosine
error”. In many case the angle 𝜃 is very small and the error will be negligible.
1/24/2025 11
 Errors in Measurements

Both sine and cosine error

Cosine Error If D = True diameter
The errors of above nature can be L = apparent length
avoided by using gauges with spherical d = micrometer anvil diameter
ends. Then, D = L cosθ – d sinθ
Error = L – D = L – (L cosθ – d sinθ)
= L(1 – cosθ) + d sinθ
1/24/2025 12
 Errors in Measurements
Sources of Errors
vi. Parallax Error
This error is occur when the line of vision is not
directly in line with measuring scale.
In Figure, ∆𝑃𝐴𝑆 𝑎𝑛𝑑 ∆𝑁𝐸𝑆 𝑎𝑟𝑒 𝑠𝑖𝑚𝑖𝑙𝑎𝑟 𝑇𝑟𝑖𝑎𝑛𝑔𝑙𝑒
Then,
𝑃𝐴 𝑃𝑆 𝑑 𝑑
= ,𝑃𝐴 = ∗ 𝑁𝐸 = ∗ 𝑑 + 𝐷 𝑡𝑎𝑛𝜃
𝑁𝐸 𝑁𝑆 (𝑑+𝐷) (𝑑+𝐷)
Error = d*tan𝜃

Now generally, 𝜃 is small so tan𝜃= 𝜃 and error e = d𝜃

For least error d should be minimum possible, value of θ can be reduced to zero by
placing mirror behind the pointer which ensures normal reading scale.
1/24/2025 13
 Errors in Measurements
Sources of Errors
vii. Error due to Dust
Presence of dust in the atmosphere change reading in the order of fraction of
micron. When high accuracy in measurement is required dust should be
cleaned by clean chamois cloth or by soft brush and gauges should never be
touched by moist fingers.

viii. Error due to Vibrations

The instrument anvil will not give consistent and repetitive reading if it is
subjected to vibration. So the measurement should be taken away from the
source of vibration, or use rubber padding for damping.
1/24/2025 14
 Errors in Measurements
Sources of Errors
ix. Error due to Location
If the datum surface is not perfectly flat or if any foreign matter such as dirt, chip
etc are present between the datum and w/p, error occurs in measurement.
x. Error due to Poor Contact
The measured dimension will be greater than the actual
dimension due to poor contact as shown in figure.
xi. Error due to Wear in Gauges
The anvil of micrometer is subjected to wear due to
repeated use and lead to error in measurement. The lack
of parallelism due to wear of anvil can be checked by
optical flat.
1/24/2025 15
 Types of Errors
Basically errors are of two types:
i. Controllable or Systematic Error
ii. Uncontrollable or Random Error
Controllable or Systematic Error
These type of errors are regularly repetitive in nature and are of similar form.
If properly analyzed, they can be determined and reduced effectively.
Systematic errors includes:
➢ Calibration errors,
➢ Ambient or atmospheric conditions,
➢ Stylus pressure error, etc.
1/24/2025 16
 Types of Errors
Uncontrollable or Random Error
Random errors are accidental, non-consistent in nature and as they occur
randomly, they cannot be eliminated since no definite cause can be located.

The possible source of random errors are:

➢ Small variations in the position of setting standard and work-piece.

➢ Slight displacement of lever joints of measuring instruments.

➢ Operator error in scale reading.

➢ Fluctuations in the friction of measuring instrument etc.

1/24/2025 17
 Types of Errors
Systematic Error Random Error

• These errors are repetitive in nature and are of • These are non-consistent. The sources giving
constant and similar form. rise to such errors are random.
• These errors result from improper conditions • Such errors are inherent in the measuring
or procedures that are consistent in action. system or measuring instruments.
• Except personal errors, all other systematic • Specific causes, magnitude and sense of these
errors can be reduced or eliminated. errors cannot be determined from the
knowledge of measuring system or condition.
• If properly analyzed these can be determined • These errors cannot be eliminated, but the
and reduced or eliminated. results obtained can be corrected.
• These include calibration errors, variation in • These include errors caused due to variation
contact pressure, variation in atmospheric in position of setting standard and work-
conditions, parallax errors, mis-alignment piece, errors due to displacement of lever
errors etc. joints of instruments, errors resulting from
backlash, friction etc.
1/24/2025 18
 Types of Errors
During measurement several types of error may arise, these are:

➢ Static errors which includes:

❖ Reading errors.
❖ Characteristic errors.
❖ Environment errors.

➢ Instrument Loading errors:

➢ Dynamic errors:

1/24/2025 19
 Types of Errors
Static errors
These errors result from the physical nature of the various components of
measuring system. Static errors result from the intrinsic imperfections or
limitations in the hardware and apparatus compared to ideal instruments. There
are three basic sources of static errors.
❖Reading Errors : These types of errors apply exclusively to instruments
These errors may be the result of parallax, readability, and interpolation.
Parallax error creeps in when the line of sight is not perpendicular to
the measuring scale. The magnitude of parallax error increases if the measuring
scale is not made flush to the component.
Interpolation error is the error resulting from the inexact evaluation of
the position of index with regards to two adjacent graduation marks between
which the index is located.
1/24/2025 20
 Types of Errors
❖Characteristic Errors : It is defined as the deviation of the output of the
measuring system from the theoretical predicted performance or from
nominal performance specifications.
linearity errors, repeatability, hysteresis and resolution errors are the examples
of characteristic errors. Calibration error is also included in characteristic error.
❖Environmental Errors : These errors results from the effect of surrounding
such as temperature, pressure, humidity etc. on measuring system.
Environmental errors of each component of the measuring system make a
separate contribution to the static error. It can be reduced by controlling the
atmosphere according to the specific requirement.

1/24/2025 21
 Types of Errors
Instrument loading errors
Instrument loading error is the difference between the value of the measurand
before and after the measuring system is connected/contacted for measurement.
Example: a soft or delicate component is subjected to deformation during
measurement due to the contact pressure of the instrument.

Dynamic errors
Dynamic errors are caused by time variations in the measurand and results
from the inability of a measuring system to respond faithfully to a time-varying
measurand.
Usually the dynamic response is limited by inertia, damping, friction or other
physical constraints in the sensing or readout or display system.

1/24/2025 22
 Propagation of Errors
Propagation of Errors
The method of determining an uncertainty in a function of the given
independent variables each with an uncertainty is propagation of errors.

Propagation of Errors during Summation

Let; z=x+y
Let the error in measurement of x and y be δx and δy respectively.
Then, measured value of z; z +  z = ( x   x) + (y   y )
z +  z = ( x + y)   x   y

Hence, error in z; z = x + y
1/24/2025 23
 Propagation of Errors
Propagation of Errors during Subtraction
Let; z=x–y
Let the error in measurement of x and y be δx and δy respectively.
Then, measured value of z; z +  z = ( x   x) − (y   y )
z +  z = ( x − y)   x   y
 z =  x   y
Hence, error in z; z = x + y For example, z = 3 x + 4 y
  z =  (3 x + 4 y)
For, z = 3 x − 4 y
  z =  (3 x + 4 y)
1/24/2025 24
 Then, measured value of z;
Propagation of Errors z   z = ( x   x).(y   y )
during Multiplication  z x y
 z 1   = x(1  ).y(1  )
Let;  z  x y
z = x.y  z x y x y
 1   = 1    .
Let the error in  z  x y x y
measurement of x, y and z x y
But, .  0.
be δx, δy and δz x y
respectively. z x y
Then,  = 
z x y
Hence, maximum probable relative error in z is;
z x y
= +
z x y
1/24/2025 25
 Then, measured value of z is;
Propagation of Errors z   z = ( x   x).(y   y ) −1
during Division  z  x −1  y −1
 z 1   = x(1  ).y (1  )
Let; z = x = xy −1  z  x y
y  z  x −1 y
 z 1   = x(1  ).y (1  + .........)
Let the error in  z  x y
measurement of x, y and z  z x y x y
 1   = 1    . + ......
be δx, δy and δz  z  x y x y
respectively. x y z x y
But, .  0., Then,  = 
x y z x y
Hence, maximum probable relative error in z is;
z x y
 = +
z x y
1/24/2025 26
 Propagation of Errors during powers and exponents
Let z be a quantity dependent on three other variables w, x, y as,
z = kwl x m y n where k is a constant.
Then, measured value of z is;
z   z = k ( w   w)l ( x   x) m ( y   y ) n
z w x y
1 = (1  ) (1 
l
) (1 
m
)n
z w x y
If  w,  x,  y are very small, then binomial expansion results;
z w x
 y
w x x  y 
1 = 1 l m n  As,     0
z w x y  w x x y 
Hence, maximum probable relative error in z is;
z w x y
 =l +m +n
1/24/2025 z w x y 27
 Propagation of Errors
The general method of getting formulas for propagating errors involves the
total partial differential of a function.
Suppose, z = f ( w, x, y,...) , where w, x, y,….etc. are the independent variables.
Then, total differential is;

f f f
f = w+ x+  y + .........
w x y

Here, δz, δw, δx, δy……. etc are the error in z, w, x, y……….etc.

1/24/2025 28
 Propagation of Errors
Let z be a quantity dependent on three other variables w, x, y as,
z = kwl x m y n where k is a constant.
Error in z by method of partial differential is given by,
z z z
z = w+ x+  y
w x y
  z = lkwl −1 x m y n w + mkwl x m−1 y n x + nkwl x m y n−1 y
z w x y
 =l +m +n
z w x y
Hence, maximum probable relative error in z is;
z w x y
 =l +m +n
z w x y
1/24/2025 29
 Propagation of Errors
Let z be a quantity dependent on three other variables w, x, y as,
z = kwl x m y n where k is a constant.
Taking ln on both sides,
 ln z = ln k + l ln w + m ln x + n ln y
Error in z by method of partial differential is given by,
z z z
 z = w+  x+  y
w x y
Differentiating,
z w x y
 =l +m +n
z w x y
Hence, maximum probable relative error in z is;
z w x y
 =l +m +n
1/24/2025 z w x y 30
 Propagation of Errors
Trigonometric function, For example, z = 3 x 2 y 4
z = sin x z z
z = x+ y
  z = sin( x +  x) − sin x x y
z x y
1 1 1  =2 +4
In case, = + z x y
z x y
Then, the error is calculated using For example, z = 3 x 2 y −4
formula; z x y
 =2 + −4
z x y z x y
= +
z 2
x 2
y2 z x y
 =2 +4
z x y
1/24/2025 31
 Propagation of Errors
Example Example
1 What is the error in
4a 3b 2 −
Let, p = = 4a 3b 2c −1d 2 perimeter of a room with
c d lengths;
Then, maximum probable error in
L1 = 5.0  0.5cm, L2 = 6.0  0.3cm
p is given by;
The perimeter is;
p a b
1 d c
=3 +2 + −1 +− P = L1 + L2 + L1 + L2 = 22cm
p a b c 2 d
The error is;
p a1 d b c
 =3 +2 + +  P =  L1 +  L2 +  L1 +  L2 = 1.6cm
p a b c 2 d

1/24/2025 32
 Propagation of Errors
Examples What is the error in circumference of
a circle with radius R = 2.15  0.08 cm?
What is the error in area  C = 2 R = 2  0.08 = 0.50cm
of a room? What is the error in volume
L = 5.0  0.5cm, W = 6.0  0.3cm of sphere with radius;
The area is; A = L  W = 5  6 = 30cm 2 R = 2.15  0.08 cm?
Relative Error is; 4 3
V =  R = 41.6cm3
 A  L W 0.5 0.3 3
= + = + = 0.15 OR 15%  V R 0.08
A L W 5 6 =3 =3
Absolute error is; V R 2.15
=0.11 OR 11%
 A = A  0.15 = 30  0.15 =4.5cm 2

1/24/2025  V = 0.11 V = 4.6cm3 33

 Propagation of Errors
Example: The radius of curvature R of a concave surface is determined by noting
two dimensions such that,
d2 h
R = + . If error in measurement of d is  1% and in measuring
8h 2
h is  0.5%, determine the total error in measuring R if d = 50 mm and
h = 5 mm. R 1 d
= ( 2d ) + 0 =
d2 h d 8h 4h
Given, R = + R d 2 1 d 2
1
8h 2 = ( −1) h + = − 2 +
−2

Total error in R is given by; h 8 2 8h 2

R R d  d2 1 
R = d + h R = d +  − 8h 2 + 2   h
d h 4h  
1/24/2025 34
 Propagation of Errors
d  d2 1 
R = d +  − 8h 2 + 2   h
4h  
d  1   d 2
1 1 
R =  d + − 2 +   h
4h  100   8h 2   200 
d2 1 d2  1 
R =  + − 2  h
400h 2 8h  200  Relative Error;
50 2
1 50 2
 5  d 2 h 502 5
R =  + − R= + = + = 65mm
  8h 2 8  5 2
400  5 2 8  52  200 
 R 1.55
5 3  = = 2.3846%
 R =   +  = 1.55mm R 65
 4 10 
1/24/2025 35
 Propagation of Errors
Example: Determine the compound error in angle θ determined by a sine bar by
measuring height of a slip gauge and the length between rollers of sine bar such that,
h
sin  = .
l  dh dl 
 d =  −  tan 
h  h l 
Given, sin =
l  dh dl   h   dl   1 
Taking log;  d =  −    =  dh − h   
 h l  l − h  
2 2 l  l − h 
2 2

 ln sin  = ln h − ln l For small finite errors  h and  l, the error in  is;

Differentiating;
  l  1 
cos dh dl  =   h + h   2 2 
 d = −  l  l − h 
sin  h l
1/24/2025 36
 The Effect of Averaging Results
➢ The accuracy of determination is the amount by which it is estimated that a
measurement could deviate from its true value, but it is not necessarily true that it
does so.
➢ By chance the measurement could possibly be exactly right or it could deviate
from the true size by any fraction of the accuracy of determination. There is no
way of knowing other than using a method of measurement with a better accuracy
of determination.
➢ The statistical parameters of arithmetic mean and standard deviation may be used
to assess random errors by taking a series of repeated measurements. Thus, the
complete measurements are repeated a great many times, then frequency
distribution is obtained by plotting a tally chart of these values.

1/24/2025 37
 The Effect of Averaging Results
➢ Let a series of measurement be denoted by, x1 , x2 , x3 , x4 ,............., xn
Then,
x1 + x2 + x3 + x4 + .......... + xn
Arithematic Mean, x =
n

Standard deviation,  =
 i
( x −x ) 2

n
99.27% of observations lie within  3 .
95% of observations lie within  2 . So, for all practical purposes the
65% of observations lie within   . estimated accuracy of
determination is equal to δ = ± 3σ
1/24/2025 . 38
 The Effect of Averaging Results
If now we take the observations and divide them into random sub-groups of n
and for each sub-group calculate its mean size, we can produce a frequency
distribution for the values of mean. It can be shown that

m =
n
It follows that the accuracy of
where,  m = standard deviation
determination of the mean size
of mean x .
of a sample of n observations is,
 = standard deviation of the
3
individual observations. 3 m =
n
n = sub-group size or sample size.
1/24/2025 39
 The Effect of Averaging Results
If we apply this to the mean size of n observations, we see that,
3 
99.27% confidence limits=  =
n n To do this with accuracy we should
2 have to take a large no. of

2 observations and from them
95% confidence limits=  =3
n n calculate the true value of the
standard deviation, but by
1
 estimating the accuracy of

65% confidence limits=  =3 determination we can not only give
n n its value but also the approximate
 = the estimated accuracy of degree of confidence we can assign
to that value.
determination of a single observation.
1/24/2025 40
 The Effect of Averaging Results

1/24/2025 41
 Graphical Methods
If an experiment is carried out to find the law relating two measured variables x
and y, it is usual to plot a graph of the readings and determine the law of graph
by plotting a mean line, i.e. we are averaging out the errors in the individual
observations.
This is obviously a tedious process and a better method is known as the method
of least squares.
Method of Least Squares
The least square principle states that the most probable value of observed
quantities is that which renders the sum of the squares of residual errors to
minimum.
(a) A series of Observed Values
Deviation of any particular observed value
x1 , x2 , x3 , x4 ,............., xn
x from the most probable value X is,
X − the most probable value
x− X
1/24/2025 42
 Method of Least Squares
From least square principle;

 ( x − X ) should be minimum.
2

d ( x − X )
2

i.e., =2 ( x − X ) = 0.
dx
  ( x ) - nX = 0

X=
 ( x)
= Arithmetic mean
n

1/24/2025 43
 Method of Least Squares
(b) A Series of Observed Values of Two Dependent Variables
Let the observed values be given by ( xi , yi ) where i varies from 1 to n.
The equation of straight line is given by,
y = ax + b, a is the slope or gradient of the line and b is the the intercept on y axis.
 x ( y )
( x − x )( y − y )  xy −
i i

a=
 i i
=
i i
n
( x − x ) (  x)
2 2

x − n
i 2 i
i

y = ax + b, where y =
 yi
, x=
 x i

1/24/2025
n n 44
 Method of Least Squares
Errors by using Standard Deviations
z = x+ y or z = x− y z = xm y n
or z = x + y − w z  2

 x  y
2

=  m  +  n  + ....
z = ( x 2
+ y + w
2 2
) z  x   y  
z = xy  z 
2
 z   z 
2 2

z=x y  z =    x +    y +    w + ...
2 2 2 2

 x   y   w 
z  2
x   y  w
2 2

=   +   +   + ....
z  x   y   w  

1/24/2025 45
 Assignment
❖A cylinder of 80 mm diameter was placed between the micrometer anvils. Due
to inaccurate placement, the angle between the micrometer and cylinder axis
was found to be 1 minute. Calculate the amount of error in the measured
diameter of the above cylinder if the micrometer anvil diameter is 6 mm. Use
suitable approximations.
❖A test indicator is used to check the concentricity of a shaft but its stylus is so
set that its movement makes an angle of 350 with the normal to the shaft. If
the total indicator reading is 0.02 mm, calculate the true eccentricity.

1/24/2025 46
 Assignment
❖An error in measuring outside diameter D, inside diameter d and length of a
hollow cylinder was found to be 2%, 3% and 5% respectively. Determine the
compound error in measurement of volume of a cylinder, if D = 50mm, d = 30
mm and L = 80 mm.
❖For particular measurement the relation is;
w2
D = L + , where L = 400mm and  L = 0.025mm
8L
. w = 50mm and  w = 1.00mm
Determine the error in measurement of D.
Ans:  (0.024 + 0.031) mm
1/24/2025 47
 Assignment
w2
Given, D = L +
8L
1 1
D D   D = 1− ( 0.025 ) + ( 1)
Then,  D= L+ w 512 32
L w
511 1
2 D = ( 0.025) + ( 1)
  D= 1 + ( −1L )  L +
w w
−2
w 512 32
8 4L
  D =  ( 0.02495 + 0.03125 ) mm
w2 w
  D= 1 − 2  L + w
8L 4L

502 50
  D= 1 − ( 0.025 ) + (1)
8 ( 400 ) 4  400
2

1/24/2025 48
 Assignment
❖ The diameter of a steel ball is measured five times with a micrometer giving the following
results: 8.011, 8.005, 8.009, 8.014 and 8.011 mm.
(a) Calculate the most probable value of the diameter and its standard deviation and
parameters of the infinite population of similar measurements from which these are supposed
to have come.
(b) Another measurement yields 8.021 mm. Is it wrong? If wrong, comment the reason
of measurement being wrong.

❖ The resistance value at a temperature t of a metal wire, Rt is given by the expression;

Rt=Ro(1+αt) where, Ro is the resistance at) 0°C and α is the resistance temperature
coefficient. The resistance values of metal wire at different temperature have been tabulated
as given below. Obtain the values of Ro and α using least square straight-line fitting.
Temp. (°C) 20 40 60 80 100
Resistance (Ω) 107.5 117.0 117.0 128.0 142.5
1/24/2025 49
.