Measurement Errors
Measurement Errors: Gross error, systematic error,
absolute error and relative error, accuracy, precision,
resolution and significant figures, Measurement error
combination, basics of statistical analysis.
2/3/2017 1
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Measurement Errors
A measurable quantity is a property of phenomena, bodies, or
substances that can be defined qualitatively and expressed
quantitatively. Measurable quantities are also called physical
quantities
True value of a measurand is the value of the measured physical
quantity, which, would ideally reflect, both qualitatively and
quantitatively, the corresponding property of the object
Measurement Error is the deviation of the result of measurement
from the true value of the measurable quantity, expressed in
absolute or relative form
Error = Measured or Observed – True Value
2/3/2017 2
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Measurement Errors: Types
Gross Errors: Errors due to human carelessness. Ex. misreading of
Instrument or using wrong range. It can be avoided by two means
–Great care in reading and recording of data.
–More observations of measurement to avoid same error.
Systematic Error: Errors from measurement system/ instrument or
due to wrong use of instrument. Ex. Offset error, Zero setting error
–These are predictable, is typically proportional to true value,
–It can be generally measured/ eliminated if the cause is known.
2/3/2017 3
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Measurement Errors: Types
Systematic errors are of three types
•Instrumental Error : errors due to inherent shortcomings, loading effects
•Environmental Error: errors due to factors external to the instruments
•Observational Error : Errors due to observations; ex. parallax error,
Random Error : errors of unexplainable origin (unknown sources of
error) are referred as random errors. These error results due to
changes in environmental variables.
–Random Errors and generally of Gaussian nature
–It can be estimated by multiple measurements and its effects can be
reduced by averaging.
2/3/2017 4
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Absolute error and Relative error,
If Am is Observed or measured value of a physical quantity with At
is its true value, then
Error or Absolute Error A= Am- At
Relative Errors are Error expressed as fraction of true value.
Relative Error = A/ At
and can be expressed in % by multiplying with hundred.
2/3/2017 5
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Absolute error and Relative error,
Limiting Error (Guarantee Errors):
– Manufacturers guarantee component value to lie in certain % of its rated value.
– In Instruments, accuracy is provided and is generally, a certain % of Full Scale.
The manufacturers has to specify deviation from nominal value of
quantity. The limit of these deviations is called Limiting Errors.
Thus a nominal value A with limiting error A is referred as A±A.
Relative limiting error r = A/A = 0 /A
Or absolute limiting error 0 = A = r A
2/3/2017 6
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Absolute error and Relative error,
Ex. A resistance R of 600  is known to have possible absolute error as
± 60 . Express the value of resistor in relative error.
R= 600 ± 60 
Relative error= ± 60/600 = ± 0.1 = ± 10 %
Thus R = 600 ± 10% 
Percentages are usually employed to express errors in resistances and
electrical quantities. The terms Accuracy & Tolerance are also used. A
resistor with ± 10% error is said to be accurate to ± 10 % or having
tolerance of ± 10%.
2/3/2017 7
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Accuracy
A measure of how close a measurement is to the true value of the
quantity being measured.
2/3/2017 8
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
More accurate Less accurate
Ex. A voltmeter with 1% accuracy
indicates a value as 200 V.
Possible error = ± 1% of 100 V
= ± 2V
Thus the true value lies between
198 V to 202 V.
•Generally full scale accuracy is
referred by manufacturers
Precision
The term Precise means clearly and sharply defined.
Precision is measure of reproducibility of measurement, i.e. given a
fixed value of a quantity, precision is measure of degree of
agreement within a group of measurements.
2/3/2017 9
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
More Accurate,
More Precise
Less Accurate
More Precise
Less Accurate
Less Precise
More Accurate
Less Precise
Precision
Precision are composed of two characteristics
•Conformity
•Significant figures
High precision means conformity of repeated readings in a tight
cluster
While low precision means conformity of repeated readings in a
broad scattered cluster
The number of significant figure indicates the precision of
measurement.
2/3/2017 10
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Significant Figures
The number of significant figure indicates the precision of
measurement.
Let the true value of a register is 1.35786 K, but observer reads
from the scale to 1.4 K. This precision error is caused by limitation
of the instrument (less significant digits).
Let a measured voltage is 6.495 V.
–It indicates that instrument can read minimum value of 0.001 V (resolution).
–If the measurement is made with precision of 0.001 V, then the value read could
be 6.494V or 6.496 V, as the measurement has four significant digits.
2/3/2017 11
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Significant Figures
•Let a measured resistance (R) is close to 872.4 , than 872.3  or
872.5 . The four significant figure show that measurement
precision is 0.1 .
•Let R is close to 872.4 K, then precision is 0.1 K or 100 .
•If R is 52.0  then it implies that R is closer to 52  than 52.1  or
51.9 . Here ‘0’ is also a significant digit so the number of significant
figure of measurement is three.
•Let R is 5.00x106  (500000000 ). If the resistance is closer to the
value than 4.99x106  or 5.01x106 , then there are three
significant figures and not nine.
2/3/2017 12
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Significant Figures
•Let measurement of Resistor is done by measuring the voltage
across the resister (by Voltmeter) and current through resistor (by
Ammeter) respectively. Let the measured Voltage is 50.31 V and
2.33 A, then
Resistance R = V/I = 5.31/2.33 = 2.2789699571 
Clearly the answer in 10 fraction digits is irrelevant. Answer should
be in same number of significant digits as the original quantities i.e
R = 2.28  (three significant digits)
2/3/2017 13
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Resolution
If input is changed from some initial arbitrary value (may be zero),
then output will not change until a minimum increment of input is
observed.
Thus the smallest increment in input (quantity being measured) that
can be detected by the instrument is called resolution
(discrimination)
Smallest value of input which can be detected by instrument is
called Threshold.
2/3/2017 14
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Measurement Error Combination
Limiting Error (Guarantee Errors):
– Manufacturers guarantee component value to lie in certain % of its rated value.
– In Instruments, accuracy is provided and is generally, a certain % of Full Scale.
The manufacturers has to specify deviation from nominal value of
quantity. The limit of these deviations is called Limiting Errors.
Thus a nominal value A with limiting error A is referred as A ± A.
Relative limiting error r = A/A = 0 /A
Or absolute limiting error 0 = A = r A
2/3/2017 15
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Measurement Error Combination
2/3/2017 16
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
When two or more quantities, (each having limiting errors) is combined,
limiting error of result can be computed using algebraic relation .
Sum or difference of Quantities
Let y = ±u ±v ±w
Relative increment of the function
)(
errorlimitingresultingThus
then,and,,asdrepresenteareerrorslimitingif
)(
wvuwvu
w
w
w
v
v
v
u
u
uy
w
w
y
w
v
v
y
v
u
u
y
u
y
y
wvu
w
dw
y
w
v
dv
y
v
u
du
y
u
y
dw
y
dv
y
du
y
wvud
y
dy


















Measurement Error Combination
2/3/2017 17
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Ex. Three resistances have following ratings
R1= 50 ± 10%, R2= 100 ± 5%, R3= 200 ± 10%
What is the result, when three resistances are combined in series
%57.835030350Thus
30)2055(error,limitingresulting
35020010050,resistancetheofMagnitude
20200200
100
10
200
5100100
100
5
100
55050
100
10
50
321
3
2
1
























R
wvuR
RRRR
R
R
R
Measurement Error Combination
2/3/2017 18
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Product of Quantities
Let y = uv
Taking logarithm





 








v
v
u
u
y
y
vu
v
dv
u
du
y
dy
dy
dv
vdy
du
uy
vuy eee
then,and,asdrepresenteareerrorslimitingif
or
111
yw.r.t.ationdifferentiTaking
logloglog
Measurement Error Combination
2/3/2017 19
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Quotient of Quantities
Let y = u/v





 











v
v
u
u
y
y
h
v
v
u
u
v
v
u
u
y
y
vu
v
dv
u
du
y
dy
dy
dv
vdy
du
uy
vuy eee
error,limitingrelativeso
signs)oppositeaveandwhenoccurserror(Max.
then,and,asdrepresenteareerrorslimitingif
or
111
y,w.r.t.ationdifferentiTaking
logloglog

Measurement Error Combination
2/3/2017 20
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Power of a factor/ Composite factor
Let y = un vm





 







u
v
m
u
u
n
y
y
vu
u
dv
m
u
du
n
y
dy
dy
dv
v
m
dy
du
u
n
y
vmuny eee
then,andasdrepresenteareerrorslimitingif
or
1
y,w.r.t.ationdifferentiTaking
logloglog
Basics of Statistical Analysis
A number of measurements of a quantity have data scattered
around a central value. Statistical tools can be employed to reach
best approximation to the true value of the quantity.
Arithmetic mean
Let x1, x2,--------, xn are n different measurements of same quantity
(samples). Most probable value from various observations of value
can be obtained using arithmetic mean.
2/3/2017 21
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad




n
i
i
n
x
nn
xxx
X
1
21 1
or 
Basics of Statistical Analysis
Dispersion (Spread or Scatter)
The property which denotes the extent to which samples are
dispersed around a central value (mean) .
2/3/2017 22
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Curve 1: Data spreads in the
range x1 to x2. The observations
are less disperse (more precise)
Curve 2: Data spreads in the
range x3 to x4. The observations
are more disperse (less precise)
Curve 1: More precision
Curve 2: Less precision
x3 x1  x2 x4 x
Curves showing different ranges and precision
Probability or
frequency of
occurrence
Basics of Statistical Analysis
Range: Range is difference of largest and lowest value of the
samples. It is simplest measure of dispersion.
Deviation: deviation is departure from arithmetic mean of the
sample.
2/3/2017 23
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
  


n
i
i
n
i
i
ii
x
n
d
n
D
xd
11
0
11


As average deviation will be zero, therefore deviation is not suitable
measure of dispersion
Variance: Variance is average of squared deviation of samples from
mean
Basics of Statistical Analysis
Variance: Variance is average of squared deviation of samples from
mean
2/3/2017 24
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
  

n
i
i
n
i
i x
n
d
n
V
1
2
1
2 11

Variance is a finite value and is occasionally used for representing
dispersion
Standard deviation: is square root of averaged squared deviation of
samples from mean.
  Vx
n
d
n
n
i
i
n
i
i    1
2
1
2 11

Normal or Gaussian Curve
2/3/2017 25
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
-3 -2 -1 1 2 3 x
y
Probability or frequency
of occurrence
•Gaussian curve is symmetric about arithmetic mean and area under
the curve is zero. Therefore, data is normalized to be zero mean
•Most of natural events, measurements having some amount of
randomness follows Gaussian curve
For Normal curve, probability y is given as
0i.e)(deviationvaluenormalizedx
indexprecisionhwhere
22


 


xh
e
h
y
Normal or Gaussian Curve
2/3/2017 26
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Normal or Gaussian Curve:
Specification
2/3/2017 27
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Parameter Data range probability Explanation
Standard Deviation
1 range ± 0.6828 68.28 % of data lie in the range
2 range ±2 0.9546 95.46 % of data lie in the range
3 range ±3 0.9974 99.74 % of data lie in the range
Probable Error
r range ±3r 0.5 50.00 % of data lie in the range
Combination of Components
2/3/2017 28
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
Let X is combination of several independent variables each of which
is subject to random effects
       
nx
n
xxX
n
n
n
n
n
n
V
x
X
V
x
X
V
x
X
V
x
x
X
x
x
X
x
x
X
X
x
x
X
x
x
X
x
x
X
X
xxx
xxxfX
22
2
2
1
2
2
2
2
2
2
2
1
2
1
2
2
2
1
1
21
21
21
Thuszero.totendllproduct wicrossofsumt variableindependenforas
X,invariationtheninput,inchangesmallare,,,if
),,,(

































































Uncertainty & Confidence Level
For single sample data, statistical analysis can not be applied as
its scatter can’t be observed.
Thus sample value (X) can be expressed in terms of mean value
(value for single samples) and uncertainty interval based upon
odds (confidence level)
2/3/2017
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
29
levelconfidenceasexpressedalsomaychance,orOdds1tob
intervalyuncertaintw
1)to(b


 wXX
Treatment of uncertainty is done in same way as to the erro
analysis
Thanks
2/3/2017
NEC 403 Unit I by Dr Naim R Kidwai,
Professor & Dean, JIT Jahangirabad
30

Measurement errors, Statistical Analysis, Uncertainty

  • 1.
    Measurement Errors Measurement Errors:Gross error, systematic error, absolute error and relative error, accuracy, precision, resolution and significant figures, Measurement error combination, basics of statistical analysis. 2/3/2017 1 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 2.
    Measurement Errors A measurablequantity is a property of phenomena, bodies, or substances that can be defined qualitatively and expressed quantitatively. Measurable quantities are also called physical quantities True value of a measurand is the value of the measured physical quantity, which, would ideally reflect, both qualitatively and quantitatively, the corresponding property of the object Measurement Error is the deviation of the result of measurement from the true value of the measurable quantity, expressed in absolute or relative form Error = Measured or Observed – True Value 2/3/2017 2 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 3.
    Measurement Errors: Types GrossErrors: Errors due to human carelessness. Ex. misreading of Instrument or using wrong range. It can be avoided by two means –Great care in reading and recording of data. –More observations of measurement to avoid same error. Systematic Error: Errors from measurement system/ instrument or due to wrong use of instrument. Ex. Offset error, Zero setting error –These are predictable, is typically proportional to true value, –It can be generally measured/ eliminated if the cause is known. 2/3/2017 3 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 4.
    Measurement Errors: Types Systematicerrors are of three types •Instrumental Error : errors due to inherent shortcomings, loading effects •Environmental Error: errors due to factors external to the instruments •Observational Error : Errors due to observations; ex. parallax error, Random Error : errors of unexplainable origin (unknown sources of error) are referred as random errors. These error results due to changes in environmental variables. –Random Errors and generally of Gaussian nature –It can be estimated by multiple measurements and its effects can be reduced by averaging. 2/3/2017 4 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 5.
    Absolute error andRelative error, If Am is Observed or measured value of a physical quantity with At is its true value, then Error or Absolute Error A= Am- At Relative Errors are Error expressed as fraction of true value. Relative Error = A/ At and can be expressed in % by multiplying with hundred. 2/3/2017 5 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 6.
    Absolute error andRelative error, Limiting Error (Guarantee Errors): – Manufacturers guarantee component value to lie in certain % of its rated value. – In Instruments, accuracy is provided and is generally, a certain % of Full Scale. The manufacturers has to specify deviation from nominal value of quantity. The limit of these deviations is called Limiting Errors. Thus a nominal value A with limiting error A is referred as A±A. Relative limiting error r = A/A = 0 /A Or absolute limiting error 0 = A = r A 2/3/2017 6 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 7.
    Absolute error andRelative error, Ex. A resistance R of 600  is known to have possible absolute error as ± 60 . Express the value of resistor in relative error. R= 600 ± 60  Relative error= ± 60/600 = ± 0.1 = ± 10 % Thus R = 600 ± 10%  Percentages are usually employed to express errors in resistances and electrical quantities. The terms Accuracy & Tolerance are also used. A resistor with ± 10% error is said to be accurate to ± 10 % or having tolerance of ± 10%. 2/3/2017 7 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 8.
    Accuracy A measure ofhow close a measurement is to the true value of the quantity being measured. 2/3/2017 8 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad More accurate Less accurate Ex. A voltmeter with 1% accuracy indicates a value as 200 V. Possible error = ± 1% of 100 V = ± 2V Thus the true value lies between 198 V to 202 V. •Generally full scale accuracy is referred by manufacturers
  • 9.
    Precision The term Precisemeans clearly and sharply defined. Precision is measure of reproducibility of measurement, i.e. given a fixed value of a quantity, precision is measure of degree of agreement within a group of measurements. 2/3/2017 9 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad More Accurate, More Precise Less Accurate More Precise Less Accurate Less Precise More Accurate Less Precise
  • 10.
    Precision Precision are composedof two characteristics •Conformity •Significant figures High precision means conformity of repeated readings in a tight cluster While low precision means conformity of repeated readings in a broad scattered cluster The number of significant figure indicates the precision of measurement. 2/3/2017 10 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 11.
    Significant Figures The numberof significant figure indicates the precision of measurement. Let the true value of a register is 1.35786 K, but observer reads from the scale to 1.4 K. This precision error is caused by limitation of the instrument (less significant digits). Let a measured voltage is 6.495 V. –It indicates that instrument can read minimum value of 0.001 V (resolution). –If the measurement is made with precision of 0.001 V, then the value read could be 6.494V or 6.496 V, as the measurement has four significant digits. 2/3/2017 11 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 12.
    Significant Figures •Let ameasured resistance (R) is close to 872.4 , than 872.3  or 872.5 . The four significant figure show that measurement precision is 0.1 . •Let R is close to 872.4 K, then precision is 0.1 K or 100 . •If R is 52.0  then it implies that R is closer to 52  than 52.1  or 51.9 . Here ‘0’ is also a significant digit so the number of significant figure of measurement is three. •Let R is 5.00x106  (500000000 ). If the resistance is closer to the value than 4.99x106  or 5.01x106 , then there are three significant figures and not nine. 2/3/2017 12 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 13.
    Significant Figures •Let measurementof Resistor is done by measuring the voltage across the resister (by Voltmeter) and current through resistor (by Ammeter) respectively. Let the measured Voltage is 50.31 V and 2.33 A, then Resistance R = V/I = 5.31/2.33 = 2.2789699571  Clearly the answer in 10 fraction digits is irrelevant. Answer should be in same number of significant digits as the original quantities i.e R = 2.28  (three significant digits) 2/3/2017 13 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 14.
    Resolution If input ischanged from some initial arbitrary value (may be zero), then output will not change until a minimum increment of input is observed. Thus the smallest increment in input (quantity being measured) that can be detected by the instrument is called resolution (discrimination) Smallest value of input which can be detected by instrument is called Threshold. 2/3/2017 14 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 15.
    Measurement Error Combination LimitingError (Guarantee Errors): – Manufacturers guarantee component value to lie in certain % of its rated value. – In Instruments, accuracy is provided and is generally, a certain % of Full Scale. The manufacturers has to specify deviation from nominal value of quantity. The limit of these deviations is called Limiting Errors. Thus a nominal value A with limiting error A is referred as A ± A. Relative limiting error r = A/A = 0 /A Or absolute limiting error 0 = A = r A 2/3/2017 15 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 16.
    Measurement Error Combination 2/3/201716 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad When two or more quantities, (each having limiting errors) is combined, limiting error of result can be computed using algebraic relation . Sum or difference of Quantities Let y = ±u ±v ±w Relative increment of the function )( errorlimitingresultingThus then,and,,asdrepresenteareerrorslimitingif )( wvuwvu w w w v v v u u uy w w y w v v y v u u y u y y wvu w dw y w v dv y v u du y u y dw y dv y du y wvud y dy                  
  • 17.
    Measurement Error Combination 2/3/201717 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad Ex. Three resistances have following ratings R1= 50 ± 10%, R2= 100 ± 5%, R3= 200 ± 10% What is the result, when three resistances are combined in series %57.835030350Thus 30)2055(error,limitingresulting 35020010050,resistancetheofMagnitude 20200200 100 10 200 5100100 100 5 100 55050 100 10 50 321 3 2 1                         R wvuR RRRR R R R
  • 18.
    Measurement Error Combination 2/3/201718 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad Product of Quantities Let y = uv Taking logarithm                v v u u y y vu v dv u du y dy dy dv vdy du uy vuy eee then,and,asdrepresenteareerrorslimitingif or 111 yw.r.t.ationdifferentiTaking logloglog
  • 19.
    Measurement Error Combination 2/3/201719 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad Quotient of Quantities Let y = u/v                   v v u u y y h v v u u v v u u y y vu v dv u du y dy dy dv vdy du uy vuy eee error,limitingrelativeso signs)oppositeaveandwhenoccurserror(Max. then,and,asdrepresenteareerrorslimitingif or 111 y,w.r.t.ationdifferentiTaking logloglog 
  • 20.
    Measurement Error Combination 2/3/201720 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad Power of a factor/ Composite factor Let y = un vm               u v m u u n y y vu u dv m u du n y dy dy dv v m dy du u n y vmuny eee then,andasdrepresenteareerrorslimitingif or 1 y,w.r.t.ationdifferentiTaking logloglog
  • 21.
    Basics of StatisticalAnalysis A number of measurements of a quantity have data scattered around a central value. Statistical tools can be employed to reach best approximation to the true value of the quantity. Arithmetic mean Let x1, x2,--------, xn are n different measurements of same quantity (samples). Most probable value from various observations of value can be obtained using arithmetic mean. 2/3/2017 21 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad     n i i n x nn xxx X 1 21 1 or 
  • 22.
    Basics of StatisticalAnalysis Dispersion (Spread or Scatter) The property which denotes the extent to which samples are dispersed around a central value (mean) . 2/3/2017 22 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad Curve 1: Data spreads in the range x1 to x2. The observations are less disperse (more precise) Curve 2: Data spreads in the range x3 to x4. The observations are more disperse (less precise) Curve 1: More precision Curve 2: Less precision x3 x1  x2 x4 x Curves showing different ranges and precision Probability or frequency of occurrence
  • 23.
    Basics of StatisticalAnalysis Range: Range is difference of largest and lowest value of the samples. It is simplest measure of dispersion. Deviation: deviation is departure from arithmetic mean of the sample. 2/3/2017 23 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad      n i i n i i ii x n d n D xd 11 0 11   As average deviation will be zero, therefore deviation is not suitable measure of dispersion Variance: Variance is average of squared deviation of samples from mean
  • 24.
    Basics of StatisticalAnalysis Variance: Variance is average of squared deviation of samples from mean 2/3/2017 24 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad     n i i n i i x n d n V 1 2 1 2 11  Variance is a finite value and is occasionally used for representing dispersion Standard deviation: is square root of averaged squared deviation of samples from mean.   Vx n d n n i i n i i    1 2 1 2 11 
  • 25.
    Normal or GaussianCurve 2/3/2017 25 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad -3 -2 -1 1 2 3 x y Probability or frequency of occurrence •Gaussian curve is symmetric about arithmetic mean and area under the curve is zero. Therefore, data is normalized to be zero mean •Most of natural events, measurements having some amount of randomness follows Gaussian curve For Normal curve, probability y is given as 0i.e)(deviationvaluenormalizedx indexprecisionhwhere 22       xh e h y
  • 26.
    Normal or GaussianCurve 2/3/2017 26 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad
  • 27.
    Normal or GaussianCurve: Specification 2/3/2017 27 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad Parameter Data range probability Explanation Standard Deviation 1 range ± 0.6828 68.28 % of data lie in the range 2 range ±2 0.9546 95.46 % of data lie in the range 3 range ±3 0.9974 99.74 % of data lie in the range Probable Error r range ±3r 0.5 50.00 % of data lie in the range
  • 28.
    Combination of Components 2/3/201728 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad Let X is combination of several independent variables each of which is subject to random effects         nx n xxX n n n n n n V x X V x X V x X V x x X x x X x x X X x x X x x X x x X X xxx xxxfX 22 2 2 1 2 2 2 2 2 2 2 1 2 1 2 2 2 1 1 21 21 21 Thuszero.totendllproduct wicrossofsumt variableindependenforas X,invariationtheninput,inchangesmallare,,,if ),,,(                                                                 
  • 29.
    Uncertainty & ConfidenceLevel For single sample data, statistical analysis can not be applied as its scatter can’t be observed. Thus sample value (X) can be expressed in terms of mean value (value for single samples) and uncertainty interval based upon odds (confidence level) 2/3/2017 NEC 403 Unit I by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad 29 levelconfidenceasexpressedalsomaychance,orOdds1tob intervalyuncertaintw 1)to(b    wXX Treatment of uncertainty is done in same way as to the erro analysis
  • 30.
    Thanks 2/3/2017 NEC 403 UnitI by Dr Naim R Kidwai, Professor & Dean, JIT Jahangirabad 30