Faculty of engineering

Mechanical Engineering Dept.

Measurements

Instructor:
Dr. Mohamed Abdallah Bhlol
Assistant Professor
Mechanical Engineering Department
Faculty of Engineering
Helwan University
 Faculty of engineering
Mechanical Engineering Dept.

Measurements

Chapter 2
Error and
uncertainty analysis

Dr. Mohamed Abdallah Bhlol

Topics
Errors in Measurements

Accuracy and precision

Sources of errors

Types of errors

Uncertainty analysis and Propagation of Error

Errors in Measurements
Introduction
The purpose of any measurement is to describe some
physical property of an object or a system quantitatively, ex:
length, temperature, pressure,…

All measured values are inaccurate to some degree. In fact,

it is impossible to find the true value of a physical quantity.

If the measured value is very close to the true value, we call
it to be a very accurate measuring system.

But before using the measured data for further use, one
must have some idea how accurate is the measured data. So
error analysis is an integral part of measurement.
Errors in Measurements
 Error:
Error is the difference between
the measured value and the true
value of the variable being
measured.
 True value:
Exact value of a variable (never known)

 Measured value:
Value of a variable as indicated by measurement system.
Errors in Measurements
 The difference between the measured value (Vm) and the
true value (Vt) of the quantity represents static error or
absolute error of measurement (Es), i.e.

E s V mV t

 Relative Error (Er):

Estimate of error based on a reference value used in place
of a known “true value” of a variable.

Es
Er 100
reference value
Errors in Measurements
The error may be either positive or negative.

For positive static errors the instrument reads high and for
negative static errors the instrument reads low.

From experimentalist's view point, static correction or

simply correction (Cs) is more important than the static error.

 static correction (Cs):

 The difference between the true value and the measured
value of a quantity.
C s V tV m  E s
Errors in Measurements
 Example 1:

 A voltmeter reads 112.68 V. If the true value of the voltage is

112.6 V, determine the following:
i) The static error
ii) The static correction for the voltmeter
Solution :
i) The static error

E s V mV t 112.68  112.6  0.08V

ii) The static correction for the voltmeter

C s V tV m  E s  0.08V
Errors in Measurements
Example 2:

 A thermometer reads 92.35 oC and the static correction

given in the correction curves is -0.07 oC. Determine the true
value of the temperature.

Solution :
The true value (Vt)

E s V mV t
 C s V tV m
V tV mC s  92.35  0.07  92.28C
Accuracy and precision
 The terms precision and accuracy are used in connection
with the performance of the instrument.

 Accuracy:
Closeness of agreement between the measured value and
the true value.

Accuracy is the ability of an instrument to show the exact

reading.

Less the error more accurate is the measurement.

Accuracy and precision
 Precision:

 Precision is the repeatability of the measuring process.

 Repeatability is the ability of a measurement system to

indicate the same value on repeated measurements for a
specific input value.

 Precision refers to the degree of agreement within a group

of measurements or instruments.
Accuracy and precision

 Distinction between Accuracy and Precision

Accuracy and precision
 Distinction between Accuracy and Precision
Accuracy and precision
Example 3:

 Two pressure gauges (pressure gauge A and B) have a full

scale accuracy of ± 5%. Sensor (A) has a range of 0-1 bar
and Sensor (B) 0-10 bar. Which gauge is more suitable to be
used if the reading is 0.9 bar?
Solution :
Sensor A :
Equipment max error = ± 5 x 1 bar = ± 0.05 bar
100
Equipment accuracy at 0.9 bar (%) = ± 0.05 bar x 100 = ± 5.6%
0.9 bar
Accuracy and precision
Solution :
Sensor B :
Equipment max error = ± 5 x 10 bar = ± 0.5 bar
100
Equipment accuracy at 0.9 bar (%) = ± 0.5 bar x 100 = ± 55%
0.9 bar
Conclusion :
 Sensor A is more suitable to use at a reading of 0.9 bar
because the error percentage (± 5.6%) is smaller
compared to the percentage error of Sensor B (± 55%).
Sources of errors
There are different sources of errors include:
1.Defect in instrument.

2. Instrument calibration
-Due to frequent use of a measuring instrument and also of
aging, the instrument may go out of calibration. Which leads
to systematic error. Therefore the instrument should be sent
for calibration at frequent intervals.

3. Instrument reproducibility
- Even if the instrument has been calibrated under a set of
conditions, there may still be error in measurement due to
the difference in the calibration conditions.
Sources of errors
4. Imperfection in design of instrument.

5.Environmental effects.

6.Error due to properties of object.

7. Error due to surface finish of object.

8.Observational error.
Types of errors
The errors in instrument readings may be classified in to
three categories as:

1. Gross errors
2. Systematic errors
3. Random errors.

1) Gross errors
These are basically human errors caused by the operator
or person using the instrument.

The instrument may be good and may not give any error
but still the measurement may go wrong due to the
operator.
Types of errors
1) Gross errors
The different types of gross errors are:

1. Observational errors.
2. Reading with parallax error.
3. Incorrect adjustments of zero
and full-scale adjustments.
4. Improper applications of
instruments (i.e. using a 0–
100 V voltmeter to measure
0.1 V, etc)
Types of errors
2) Systematic errors (Bias errors)

The portion of the absolute error that remains constant

during repeated measurements (i.e. measured values have
similar deviation from correct value).

May result from shortcomings of the instruments, such as

defective or worn parts, and effects of the environment on the
equipment or the user.

They are sometimes called bias due to error in one

direction- high or low.

Can be result from mis-calibrated device and loading errors.

 This kind of error can be eliminated by calibration.

 Types of errors
3) Random errors (precision error)

Random errors are distinguishable by their lack of

consistency (i.e. measured values deviate randomly around
mean value).

The variations in temperature, vibrations of external medium,

etc. cause errors in the instrument.

Errors of this type are normally of limited duration and are

inherent to specific environment.

Random errors can never be corrected, they can only be

reduce.
Types of errors

Systematic and random error

Types of errors
 Types of errors
Noise and Interference
Noise
Random variation of a value of a measured signal.
Due to variation of the extraneous variables.
Increases data scatter.
Statistical methods can reduce the effect of noise
Interference
Undesired deterministic trend in measured value.
Example: AC power source superimposed on electrical
signal.
Interference is often obvious, but if period is longer than
measured signal, it will appear as a false trend and this harder
to detect. More than one test required.
Types of errors
Noise and Interference
Types of errors
Overall error (u)

Overall error represents the combination of all known errors.

u c  [e e e e ]
2
1
2
2
2
3 M
2 1/ 2

Example: an instrument with known hysteresis, linearity, and

sensitivity errors has the following total instrument error.

u c  [e  e  e ]
2
h
2
L
2 1/ 2
K
Types of errors
Example 4:
 A speedometer with resolution 5 mph (8 kph) and accuracy
within ±4% reading. Find: the design-stage uncertainty (ud) at
60 mph (90 kph) .
Solution :

u c  [e e e e ]
2
1
2
2
2
3 M
2 1/ 2

u c u d  [e  e ]
2
R
2 1/ 2
A

1
u R  [  Re solution]
2
Types of errors
Solution :
1
u R  [  5]  2.5 mph
2
4
u A  4% reading    60  2.4mph
100

u c u d  [( 2.5) 2 (2.4) 2 ]1/ 2  3.5mph

Uncertainty analysis and propagation
of error
Any system that relies on a measurement system will
involve some amount of uncertainty (doubt).
The uncertainty may be caused by individual inaccuracy of
sensors, limitations of the display devices, random variations
in measurands, or environmental conditions.
What is uncertainty?
Uncertainty of measurement is a parameter that describes
the distribution of the (thinkable) measured values.

The word ‘uncertainty’ expresses the doubt to the

exactness of the result of the measurement.

Measurement result is the measurement value with its

uncertainty
 Uncertainty analysis and propagation
of error
Propagation of Error
If there is a series of measured variables (x1, x2, x3,--- xk)
which are commonly used with functional relationship to
determine some resultant quantity (say y).

Let, y is a given function of the independent measured

variables x1, x2, x3,--- xk
y = f (x1, x2, x3,--- xk)

 how would the uncertainties in each of the individual

measured variable (x1, x2, x3,--- xk), contributes to the
uncertainty in the resultant quantity y?
 Uncertainty analysis and propagation
of error
Propagation of Error
Let uy be the uncertainty in the result y, and u1, u2,….,uk be
the uncertainties in each of the independent variables.

Thus, according to Taylor series expansion, and Root-Sum-

Square (RSS) method, the overall uncertainty can be
computed as:
y 2 y 2 y 2 1/ 2
u y  [( u1. )  (u 2 . )    (u k . ) ]
x1 x 2 x k
Where:
u1, u2,….,uk are the uncertainties in each of the independent
variables and can be calculated as follow:
Uncertainty analysis and propagation
of error
1 i
n
( x  x ) 2

ui   
n 1


n
xi
x  1
n
The Root-Sum-Square (RSS) method of combining
uncertainties is based on the following assumptions:

1. Individual uncertainties are following the Normal distribution

2. Each individual uncertainty is estimated at the same
probability level
3. All individual uncertainties are consistent in units
 Uncertainty analysis and propagation
of error
Example 5:

 Calculate the uncertainty (uσ) in stress (σ) for the shown

steel rod having a length of L = 50 mm, and diameter (d mm),
if the indicated reading of the load cell (0 : 2500 N capacity), F
= 1500 N ± 0.2 % fsd

Measurements for the rod diameters (d mm) were as follows:

d: 5.21 5.18 5.25 5.26 5.22 5.19 5.23 5.24
5.18 5.26
Uncertainty analysis and propagation
of error
Solution :
F F 4F
   
A  2 d 2
d
4

 
n 10
di di 52.22
d  1
 1
  5.222mm
n 10 10

1 i
n
( d  d ) 2

ud   0.0311mm
n 1

F = 1500 N ± 0.2 % fsd

uF = ± (0.2/100)x2500 = ±5 N
 Uncertainty analysis and propagation
of error
To find overall uncertainty:
 2  2 1/ 2
u   [(u F . )  (u d . ) ]
F d
 4 4 2
   0.047mm
F d 2  (5.222) 2

 4F 3  8 x1500
 x  2d    26. 823N / mm 3

d   (5.222) 3

u  [(5x0.047)2 (0.0311x  26.823)2 ]1/ 2  70.03N / mm2

Uncertainty analysis and propagation
of error
Example 6:
For mass flow through a pipe find fractional uncertainty
in (m˚), where:

P  2 P L  2
  vA 
m v D  D
RT 4 RT t 4

Then Assume the following uncertainties for the variables:

Uncertainty analysis and propagation
of error

Find the total uncertainty in single measurement of (m˚).

 Uncertainty analysis and propagation
of error
Solution :
m˚ = f (P, T, L,Δt, D)
m 1 L  2
 x x D
P RT t 4
1 L  2 P 
m
 x x D x 
RT t 4 P P
Similarly,
m
 
m m
 
m m m m
 
m
   2
L L T T t t D D
 Uncertainty analysis and propagation
of error
Solution :
m 2 m 2 m 2 m 2 m 2 1/ 2
u m  [(u P . ) (u T . ) (u L . ) (u t . ) (u D . ) ]
P T L t D

m 2  m 2 m 2  m 2 2m 2 1/ 2
u m  [(u P . ) (u T . ) (u L . ) (u t . ) (u D . ) ]
P T L t D

u m u P 2 u T 2 u L 2 u t 2 2u D 2 1/ 2
  [( ) ( ) ( ) ( ) ( )]
m P T L t D
Uncertainty analysis and propagation
of error
Solution :

Variable up/p ut/t uT/T uL/L uD/D Sum

(∑u2) 1/2
Bias (ux/x) 0.55% 0.55% 0.01% 0.1% 1% 1.27%

Precision (ux/x) 0.1% 0.4% 2% 0% 0% 2%

 
1/ 2
u m
 u 2precision u bias2
m

 
1/ 2
u m
 (1.27) 2 (2) 2  2.36%
m