EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
STATICS & PROBABILITY
The range within which the true value will lie with Number of bins (Histogram):
P% probability:
(Small N)
(Large N)
𝑥 ′ is the population mean, a.k.a. true value.
𝑥̅ is the sample mean.
𝑢𝑥̅ is the uncertainty interval in the estimation of the
true value at some probability level, 𝑃%.
N N
Population Mean xi Population Variance 2 x
1 1 2
N N i
i 1 i 1
n n
Sample Mean x x Sample Variance S 2 x x
1 1 2
n i n 1 i
i 1 i 1
Standard Deviation Variance
Probability Density Function Probability Density Function (cont’d)
(𝑓(𝑥) or 𝑝(𝑥))
P xi x xi dx f xi dx
P a x b f x dx
b
P x 1
xf x dx
2 x f x dx
2
Gaussian (Normal) Distribution; Standard Normal (z) Distribution
x ~ N ( , ) x
x ~ N ( , ) z ~ N (0,1)
1
1 x
2
2
f ( x) e
2
Statistics of Finite Data Sets Student’s t variable
x
t
Sx / N
xi x tv , P S x (P %)
n sample size, P % Probability
v degrees of freedom N 1
S x Sample standard deviation
t values are tabulated
Dr. Ernur Karadoğan Page 1 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
Standard deviation of the means: Estimate of the true mean value based on the
finite data set:
Sx
Sx
n x tv , P S x ( P %)
Pooled Statistics: Hypothesis Testing:
The pooled mean:
The pooled standard deviation:
The pooled standard deviation of the means
If the test statistic is the z-variable:
𝑃(𝑧) ≡ 1 − 𝛼
If the test statistic is the t-variable:
𝑃(𝑧) ≡ 1 − 𝛼
Dr. Ernur Karadoğan Page 2 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
Inference of Population Variance (Chi-square): Regression analysis:
N-number of data points
m- polynomial order
Confidence interval for regression:
S yx
yc t , P P% confidence interval
N
Least-Square Linear Fit:
a0
x x y x y
i i i
2
i i
x N x
2 2
i i
a1
x y Nx y
i i i i
x N x
2 2
i i
coefficient of determination:
N
( y i yci )2
r2 1 i 1
N
( y
i 1
i y )2
N
y i
y i 1
Dr. Ernur Karadoğan Page 3 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
Chauvenet’s Criterion: Three-sigma Test (assumes normal
distribution):
Data points that are outside the range of
Where:
99.73% probability of occurrence ( )
x x
z0 i
sx
Number of Measurements: Monte Carlo Simulation (basic idea):
Given the confidence interval (CI)
one-sided precision value d (desired):
Iteration process continues updating the data
the required number of measurements, N, is set for R until the predicted standard
estimated by: 1,2 deviation for R converges to an asymptotic
value.
1
The degrees of freedom in t depends on N,
therefore, N requires iteration to calculate.
2
If sx is not known a priori, a preliminary number of
measurements, N1, to obtain an estimate of the
sample variance, s1, to be expected:
Dr. Ernur Karadoğan Page 4 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
UNCERTAINTY ANALYSIS
Zero-order & Instrument Uncertainty: Combining Elemental Errors:
1
u0 resolution = 1 LSD (Least significant digit)
2
* Instrument uncertainty, uc, estimates systematic
uncertainty of the instrument (instrument calibration
errors)
Design stage uncertainty Systematic uncertainty (Systematic error is
(Minimum value for Uncertainty); constant in repeated measurements):
Systematic “standard” uncertainty ±b
(confidence level of one 𝜎; 68% probability
level for normal distribution)
Systematic uncertainty ±2b (confidence
level of one 2𝜎; 95% probability level for
normal distribution)
Random uncertainty: Multi-variable Error Propagation:
Random “standard” uncertainty: R R uR
1/2
L
uR (i uxi )2
i 1
The random uncertainty at a desired confidence level
R
i
xi x x
Dithering (Approximating a Sensitivity Index): Sequential Perturbation (Approximating a
Sensitivity Index):
For L independent variables:
Start with an initial guess of 𝛿𝑥 and reduce until the
𝜕𝑦
value of 𝜕𝑥 no longer changes (within a tolerance).
Dr. Ernur Karadoğan Page 5 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
Advanced stage uncertainty analysis: Multiple measurement uncertainty analysis
Random and systematic uncertainties are:
1/2
s x s x 1 s x 2 ... s x K
2 2 2
1/2
bx bx 1 bx 2 ... bx K
2 2 2
Where:
combined “standard” uncertainty in x:
1/2
ux bx 2 s x 2 ( P %)
expanded uncertainty in x:
1/2
u x t , P bx s x
2 2
( P %)
*total degrees of freedom in only “random
uc is the instrument uncertainty. uncertainties”:
1
u0 resolution
2
Where:
, K: total number of elemental
errors
Dr. Ernur Karadoğan Page 6 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
Multiple measurement uncertainty analysis (Single variable)
Propagation of uncertainty to a “result” (Consider R = f(xn), n=1 to L , L is # of independent variables)
When the degrees of freedom in each of the
variables, xi, are not the same, use the Welch-
Satterthwaite formulation* to calculate the degree of
freedom of the “result”:
Usually 𝜈𝑏𝑖 is very large; therefore, second term in the
denominator is zero.
Dr. Ernur Karadoğan Page 7 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
Correlated Errors
𝛿𝑖𝑗 = 1 correlated
𝛿𝑖𝑗 = 0 no correlation
Dr. Ernur Karadoğan Page 8 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
Dr. Ernur Karadoğan Page 9 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
Dr. Ernur Karadoğan Page 10 of 11
EGR 601 Advanced Engineering Experimentation
Selected Formulas & Tables
Dr. Ernur Karadoğan Page 11 of 11
