Mechanical Measurements

Prof. S. P. Venkateshan

temperature. The data points indicated by the full symbols appear also to hug the trend line. However the data points do not lie on it. This is due to random errors that are always present in any measurement. Actually the standard

thermocouple would also have the random errors that are not indicated in the figure. We have deliberately shown only the trend line for the standard

thermocouple.

Sub Module 1.3 3. Statistical analysis of experimental data Statistical analysis and best estimate from replicate data:
Let a certain quantity X be measured repeatedly to get Xi , i=1,n Because of random errors these are all different. How do we find the best estimate Xb for the true value of X? It is reasonable to assume that the best value be such that the measurements are as precise as they can be! In other words, the experimenter is confident that he has conducted the measurements with the best care and he is like the skilled shooter in the target practice example presented earlier! Thus, we minimize the variance with respect to the best estimate Xb of X. Thus we minimize:
S = [ Xi X b ]
i =1 n 2

(1)

(2)

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

This requires that:

n S = 2 [ Xi Xb ] (-1) =0 X b i=1

or Xb =

Xi
i =1

(3)

The best estimate is thus nothing but the mean of all the individual measurements!

Error distribution: When a quantity is measured repeatedly it is expected that it will be distributed around the best value according to some distribution. Many times the random errors may be distributed as a normal distribution. If and are, respectively, the mean and the standard deviation, then, the probability density is given by
f(x) = 1 2
1 x 2 e
2

(4)

The probability that the error around the mean is (x-) is the area under the probability density function between (x-)+dx and (x-) represented by the product of the probability density and dx. The probability that the error is

anywhere between - and x is thus given by the following integral:

1 F(x) = 2

1 v 2 2 dv e

(5)

This is referred to as the cumulative probability. It is noted that if x the integral tends to 1. Thus the probability that the error is of all possible

magnitudes (between - and +) is unity! The integral is symmetrical with

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

respect to x= as may be easily verified. The above integral is in fact the error integral that is a tabulated function. A plot of f(x) and F(x) is given in Figure 5.

Cumulative
1 0.8 0.6 0.4 0.2 0 -3 -2 -1 0

Probability density

(x- )/
Figure 5 Normal distribution and its integral

Many times we are interested in finding out the chances of error lying between two values in the form p. This is referred to as the confidence interval and the corresponding cumulative probability specifies the chances of the error occurring within the confidence interval. Table 1 gives the confidence intervals that are useful in practice:

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

Table 1 Confidence intervals according to normal distribution

Cumulative Probability Interval p

0 0

0.95 +1.96

0.99 +2.58

0.999 +3.29

The table indicates that error of magnitude greater than 3.29 is very unlikely to occur. In most applications we specify +1.96 as the error bounds based on 95% confidence.

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

Example 1
Resistance of a certain resistor is measured repeatedly to obtain the following data. No. 1 2 3 4 5 6 7 8 9

R, k 1.22 1.23 1.26 1.21 1.22 1.22 1.22 1.24 1.19

What is the best estimate for the resistance? What is the error with 95% confidence? Best estimate is the mean of the data.

R=

1.22 4 + 1.23 + 1.26 + 1.21 + 1.24 + 1.19 9 = 1.223 1.22 k

Standard deviation of the error :

Variance =
2 1 9 Ri -R 9 1

=3.33 10-4 Hence : = 3.33 10-4 = 0.183 0.02 k

Error with 95% confidence :

Error 95% = 1.96 = 1.96 0.0183 = 0.036 0.04 k

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

Example 2

Thickness of a metal sheet (in mm) is measured repeatedly to obtain the following replicate data. What is the best estimate for the sheet thickness? What is the variance of the distribution of errors with respect to the best value? Specify an error estimate to the mean value based on 99% confidence.

Experiment No. t, mm Experiment No. t, mm

0.202 0.198 0.197 0.215 0.199 0.194 7 8 9 10 11 12

0.204 0.198 0.194 0.195 0.201 0.202

The best estimate for the metal sheet thickness is the mean of the 12 measured values. This is given by

0.202 + 0.198 + 0.197 + 0.215 + 0.199 + 0.194 + 0.204 +0.198 + 0.194 + 0.195 + 0.201 + 0.202 = 0.2 mm 1 = tb = t = 12 12

ti

12

The variance with respect to the mean or the best value is given by (on substituting t for t b ) as

2b =

t i t
1

12

ti
=
1

12

12

12

t2

0.2022 + 0.1982 + 0.197 2 + 0.2152 + 0.1992 + 0.1942 + 0.2042 +0.1982 + 0.1942 + 0.1952 + 0.2012 + 0.2022 0.2 mm = 12 = 3.04 10-5 mm 2

The corresponding standard deviation is given by

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

b = 3.04 105 =0.0055 0.006 mm

The corresponding error estimate based on 99% confidence is

Error = 2.58 b = 2.58 0.0055 0.014 mm

Principle of Least Squares Earlier we have dealt with the method of obtaining the best estimate from replicate data based on minimization of variance. No mathematical proof was given as a basis for this. We shall now look at the above afresh, in the light of the error distribution that has been presented above.

Consider a set of replicate data xi. Let the best estimate for the measured quantity be xb. The probability for a certain value xi within the interval

x i , x i + dx i to occur in the measured data is given by the relation

1 p(x i ) = e 2

( x b x i )2
22

dx i

(6)

The probability that the particular values of measured data are obtained in replicate measurements must be given by the compound probability given by
1

p=

e
i =1

( x x )2 b i
22

dx i =

( x b x i )2
22

i =1

dx i
i =1

(7)

The reason the set of data was obtained as replicate data is that it was the most probable! Since the intervals dx i are arbitrary, the above will have to be maximized by the proper choice of x b and such that the exponential factor is a maximum. Thus we have to choose x b and such that

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

p' =

1 n

( x b x i )2
22

i =1

(8)

has the largest possible value. As usual we set the derivatives get the values of the two parameters xb and . We have:

p ' p ' = = 0 to x b

1 p ' = n + 2 ei =1 x b 2
Or

( x i x b )2
22

2 ( x i x b ) (1)
i =1 This part should go to zero

=0

(9)

( x i x b ) =0 or x b = x i = x
i =1 i=n

(10)

It is clear thus that the best value is nothing but the mean of the values! We also have:
1 n p ' n 2 = - n+1 + n+3 ( x i x b ) ei=1 i =0
This part should go to Zero
n

( x i -x b )2
2 2

=0

(11)

Or (12) n This last expression indicates that the parameter 2 is nothing but the variance of the data with respect to the mean! Thus the best values of the measured quantity and its spread is based on the minimization of the squares of errors with respect to the mean. Principle of Least Squares. This embodies what is referred to as the

2 =

( xi x b )
i =1

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

Propagation of errors: Replicate data collected by measuring a single quantity repeatedly enables us to calculate the best value and characterize the spread by the variance with respect to the best value, using the principle of least squares. Now we look at the case of a derived quantity that is estimated from the measurement of several primary quantities. answered is the following: A derived quantity Q is estimated using a formula that involves the primary quantities. a1 , a 2 ,.....a n Each one of these is available in terms of the The question that needs to be

respective best values a1 , a 2 ,.....a n and the respective standard deviations

1 , 2 .... n .

What is the best estimate for Q and what is the corresponding

standard deviation Q ? We have, by definition

Q =Q(a1 ,a 2 ,.......a n )

(13)

It is obvious that the best value of Q should correspond to that obtained by using the best values for the as. Thus, the best estimate for Q given by Q as

Q =Q(a1 ,a 2 ,.......a n )
Again, by definition, we should have:

(14)

2 1 N (15) ( Qi Q ) N i =1 The subscript i indicates the experiment number and the ith estimate of Q is given 2 Q =

by
Qi = Q ( a1i , a 2i ,....a ni )

(16)

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

If we assume that the spread in values are small compared to the mean or the best values (this is what one would expect from a well conducted experiment), the difference between the ith estimate and the best value may be written using a Taylor expansion around the best value as
2 Q

Q Q 1 N Q = a1i + a 2i + ...... + a ni a 2 a 2 N i =2 a1

(17)

where the partial derivatives are all evaluated at the best values for the as. If the as are all independent of one another then the errors in these are unrelated to one another and hence the cross terms. (17) may be rewritten as
2 Q 2 2 2 Q Q 1 N Q = a1i + a 2i + ....... + a ni N i =1 a1 a 2 a n N

a mi a ki = 0 for m k
i =1

Thus equation

(18)

Noting that a ji
i =1

=N 2 we may recast the above equation in the form j

2 2 2

2 Q

Q 2 Q 2 Q 2 = 1 + 2 +.......+ n a1 a 2 a n

(19)

Equation (19) is the error propagation formula. It may also be recast in the form

Q 2 Q 2 Q 2 Q = 1 + 2 +.......+ n a1 a 2 a n

(20)

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

Example 3
The volume of a sphere is estimated by measuring its diameter by vernier calipers. In a certain case the diameter has been measured as D = 0.0502 0.00005 m. Determine the volume and specify a suitable uncertainty for the same. Nominal volume of sphere:

V=

D3 0.05023 = 6.624 105 m3 =3.14159 6 6

The error in the measured diameter is specified as:

D = 0.00005m

The influence coefficient is defined as

ID =

V D2 0.05022 = = 3.14159 =3.958 10-3 m 2 D 2 2

Using the error propagation formula, we have

V=ID D=3.958 10-3 0.00005 = 1.979 107 m3

Thus

V = 6.624 105 1.979 107 m3

Alternate solution to the problem By logarithmic differentiation we have

dV dD =3 V D

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

This may be recast as

V = 3V

D 0.00005 = 3 6.624 10-5 = 0.0198 10-5 m3 D 0.0502

This is the same as the result obtained earlier.

Example 4
Two resistances R1 and R2 are given as 1000 25 and 500 10 . Determine the equivalent resistance when these two are connected in a) series and b) parallel. Also determine the uncertainties in these two cases. Given Data:

R1 = 1000, 1 = 25; R 2 = 500 2 = 10 All Values are in

Case a) Resistances connected in series:

Equivalent resistance is

R s =R1 + R 2 = 1000+500=1500
Influence coefficients are:

I1 =

R s R s = 1 ; I2 = =1 R1 R 2

Hence the uncertainty in the equivalent resistance is

s =

( I11 )2 + ( I22 )2

( 25 )2 + (10 )2

= 26.93

Case b) Resistances connected in parallel: Equivalent resistance is given by

Indian Institute of Technology Madras

Mechanical Measurements

Prof. S. P. Venkateshan

Rp =

R1R 2 1000 500 = 333.3 = R1 + R 2 1000 + 500 Rp = R1R 2 ( R1 + R 2 )

Influence coefficients are:

I1 = I1 = R p R1 R p R 2 = R2 R1R 2 500 1000 500 = = 0.111 2 1500 ( R1 + R 2 ) ( R1 + R 2 ) 1500 R1 R1R 2 1000 1000 500 = = 0.444 2 ( R1 + R 2 ) ( R1 + R 2 ) 1500 15002

Hence the uncertainty in the equivalent resistance is

s =

( I11 )2 + ( I22 )2

( 0.111 25)2 + ( 0.444 10 )2

= 5.24

Thus the equivalent resistance is 1500 26.9 in the series arrangement and 333.6 5.24 in the parallel arrangement.

Indian Institute of Technology Madras