Errors Propagation - Problem Set 2

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Don Bosco institute of Technology

Kurla (West), Mumbai -70

Department of Mechanical Engineering


Computational Methods AY: 2023-24 Sem – V

Module 1: Computation and Error Analysis


Error Propagation
What do you mean by error propagation?
Propagation of Error (or Propagation of Uncertainty) is defined as the effects on a function
by a variable's uncertainty.
 Propagation of Errors in Addition:
Suppose a result x is obtained by addition of two quantities say a and b
i.e. x = a + b
Let Δa and Δb are absolute errors in the measurement of a and b and Δ x be the corresponding
absolute error in x.
∴ x ± Δ x = (a ± Δa) + (b ± Δb)
∴ x ± Δ x = (a + b) ± (Δa + Δb)
∴ x ± Δ x = x ± (Δa + Δb)
∴ ± Δ x = ± (Δa + Δb)
∴ Δ x = Δa + Δb
Thus maximum absolute error in x = maximum absolute error in a + maximum absolute error
in b
Thus, when a result involves the difference of two observed quantities, the absolute error in
the result is equal to the sum of the absolute error in the observed quantities.
 Propagation of Errors in Subtraction:
Suppose a result x is obtained by subtraction of two quantities say a and b
i.e. x = a – b
Let Δ a and Δ b are absolute errors in the measurement of a and b and Δ x be the corresponding
absolute error in x.
∴ x ± Δ x = (a ± Δ a) – (b ± Δ b)
∴ x ± Δ x = (a – b) ± Δ a – ± Δ b
∴ x ± Δ x = x ± (Δ a + Δ b)

Page | 1
∴ ± Δ x = ± (Δ a + Δ b)
∴Δx =Δa+Δb
Thus the maximum absolute error in x = maximum absolute error in a + maximum absolute
error in b.
Thus, when a result involves the difference of two observed quantities, the absolute error in
the result is equal to the sum of the absolute error in the observed quantities.

 Propagation of Errors in Product:


Suppose a result x is obtained by the product of two quantities say a and b
i.e. x = a × b ……….. (1)
Let Δ a and Δ b are absolute errors in the measurement of a and b and Δ x be the corresponding
absolute error in x.
∴ x ± Δ x = (a ± Δa) x (b ± Δb)
∴ x ± x = ab ± a Δ ± b Δa ± Δa Δb
∴ x ± Δx = x ± a Δb ± b Δa ± Δ a Δb
∴ ± Δx = ± a Δb ± b Δa ± Δ Δb …… (2)
Dividing equation (2) by (1) we have

The quantities Δa/a, Δb/b and Δx/x are called relative errors in the values of a, b and x
respectively. The product of relative errors in a and b i.e. Δa × Δb is very small hence is
neglected.

Hence maximum relative error in x = maximum relative error in a + maximum relative error
in b.
Thus maximum % error in x = maximum % error in a + maximum % error in b
Thus, when a result involves the product of two observed quantities, the relative error in the
result is equal to the sum of the relative error in the observed quantities.
Page | 2
 Propagation of Errors in Quotient:
Suppose a result x is obtained by the quotient of two quantities say a and b.
i.e. x = a / b ……….. (1)
Let Δ a and Δ b are absolute errors in the measurement of a and b and Δ x be the corresponding
absolute error in x.

Binomial Expansion : (1 + x)n = 1 - x + x2 – x3 +…..(-1)r xr + …. + ∞


The values of higher power of Δ b/b are very small and hence can be neglected.

Now the quantity (Δa Δb / ab) is very small. hence can be neglected.

The quantities Δa/a, Δb/b and Δx/x are called relative errors in the values of a, b and x
respectively.

Page | 3
Hence maximum relative error in x = maximum relative error in a - maximum relative error
in b. Thus maximum % error in x = maximum % error in a - maximum % error in b.
Thus, when a result involves the quotient of two observed quantities, the relative error in the
result is equal to the difference of the relative error in the observed quantities.

 Propagation of Errors in Product of Powers of Observed Quantities:


Let us consider the simple case. Suppose a result x is obtained by following relation
x = an ……….. (1)
Let Δ a be an absolute error in the measurement of a and Δ x be the corresponding absolute
error in x.

Binomial Expansion:
𝒏 (𝒏 𝟏) 𝒏 (𝒏 𝟏)(𝒏 𝟐) 𝒏 (𝒏 𝟏)(𝒏 𝟐)….(𝒏 𝟒 𝟏)
(1 + x)n = 1 + n x + 𝟐!
x2+ 𝟑!
x3 + …….+ 𝒓!
……∞
The values of higher power of Δa/a are very small and hence can be neglected.

The quantities Δa/a and Δx/x are called relative errors in the values of a and x respectively.
Hence the maximum relative error in x = n times the relative error in a.
The greatest possible error (GPE) is one half of the measuring unit in use.

Page | 4
What is the GPE for 9.76 cm?.
The measurement is in hundredths of a centimetre (in numbers, that’s 1/100 ths).
Multiply the measurement by one half:
1/2 x 1/100th cm = 0.005 cm .
The GPE for 9.76 cm is 0.005 cm.

Error in polynomial functions:


Suppose in a function f(x) where f is differentiable and the approximate value xa of x is
given. In such cases, mean-value theorem of calculus is used to estimate error bound in f(x).
According to this theorem,
f(x) – f(xa) = (x – xa) f’(θ)
where θ is some value between x and xa and f’ is the first derivative of the function f. Then
the error in f(x) is
ef = |f(x) − f′(x )| = |e f′(θ)|

Problem set 2
1) Find the approximate maximum error in 5.43 x 27.2.
.
2) The quotient gives the result 2.053. To what extent can this be relied upon?
.

3) If y = 7 x7 + 3 x3, find the percentage error in y at x = 1, when error of x = 0.05.


4) If u = x y z2 + x2 y2 z and the error for each of x, y, z is 0.001 at x = 2, y = 4 and z = 5,
then compute maximum relative error in u.
5) Volume of a cylinder is calculated after measuring its diameter (2.5 ± 0.02) m and its

height as (4.8 ± 0.05) m respectively. Estimate the absolute error in calculation of


volume.

Page | 5
1) Find the approximate maximum error in 5.43 x 27.2.
Solution:
Greatest possible error (GPE) in 5.43 is 0.005 and in 27.2 is 0.05 i.e Δa = 0.005 and Δb
= 0.05
Hence the approximate maximum error in 5.43 x 27.2 = 0.005 x 27.2 + 0.05 x 5.43
= 0.4075
𝟐𝟓.𝟒
2) The quotient gives the result. To what extent can this be relied upon?
𝟏𝟐.𝟑𝟕

Solution:
Greatest possible error (GPE) in 25.4 is 0.05 and in 12.37 is 0.005. i.e Δa = 0.05 and Δb
= 0.005
We know that maximum relative error in x = maximum relative error in a + maximum relative
error in b. 2.053
∆ ∆ ∆ ∆ ∆ ∆
So = − ===== = −

𝑎 ∆𝑎 ∆b . 0.05 0.005
∴ ∆x =
b 𝑎
− b
= .
x
25.4
− 12.37
= 0.003212
Hence, true quotient can lie between 2.053 ± 0.003212 i.e. between 2.05621 and 2.0497.

3) If y = 7 x7 + 3 x3, find the percentage error in y at x = 1, when error of x = 0.05.


Solution:
Given y = 7 x7 + 3 x3 and Δx = 0.05

We know that, ef = |e f′(θ)|. Here ef = Δy, ex = Δx and f′(θ) = f’(y)


So f’(y) = 49 x6 +9 x2
yat x =1 = 7 + 3 = 10
f’(y)at x = 1 = 49 + 9 = 58
so Δy = 0.05 x 58 = 2.9
∆ .
% error in y = x 100 = x 100 = 29

Page | 6
4) If u = x y z2 + x2 y2 z and the error for each of x, y, z is 0.001 at x = 2, y = 4 and z =
5, then compute maximum relative error in u.
Solution:
Given u = x y z2 + x2 y2 z and Δx = Δy = Δz = 0.001

u at given values of x,y,z = (2 x 4 x 52) + (22 + 42 + 5) = 200 + 320 = 520


f’(u) = (y z2 + x z2 + 2 x y z) + (2x y2 z + x2 2y z + x2 y2)
so Δu = (y z2 Δx + x z2 Δy + 2 x y z Δz) + (2x y2 z Δx + x2 2y z Δy + x2 y2 Δz)
= [(4 x 52 x 0.001) + (2 x 52 x 0.001) + (2 x 2 x 4 x 5 x 0.001)] +
[(2 x 2 x 42 x 5 x 0.001) + (22 x 2 x 4 x 5 x 0.001) + (22 x 42 x 0.001)]
= 0.774
∆ .
% error in y = x 100 = x 100 = 0.149

5) Volume of a cylinder is calculated after measuring its diameter (2.5 ± 0.02) m and
its height as (4.8 ± 0.05) m respectively. Estimate the absolute error in calculation
of volume.
Solution: V = d2 h, dt = 2.5 m, Δd = 0.02 m, ht = 4.8 m and Δh = 0.05 m

f’(V) = d2 + 2 x d h

Absolute error in volume, ΔV = ( d2 x Δh) + (2 x d h x Δd)

= ( x 2.52 x 0.05) + (2 x x 2.5 x 4.8 x 0.02)

= 0.6224 m3

Page | 7

You might also like