Sensitivity Analysis

Often in engineering analysis, we are not only interested in predicting the
performance of a vehicle, product, etc, but we are also concerned with the
sensitivity of the predicted performance to changes in the design and/or errors in
the analysis. The quantification of the sensitivity to these sources of variability is
called sensitivity analysis.
Let’s make this more concrete using an example. Suppose we are interested in
predicting the take off distance of an aircraft. From Anderson’s, Intro. To Flight,
an estimate for take-off distance is given by Eqn (6.104):
T
=
·
max
2
44 . 1
L
LO
SC g
W
S
µ
To make the example concrete, let’s consider the jet aircraft CJ-1 described by
Anderson. In that case, the conditions were:
0 . 1
19815
7300
/ 2 . 32
@ / 002377 . 0
318
max
2
3
2
=
=
=
=
=
=
·
L
C
lbs W
lbs T
s ft g
level sea ft slugs
ft S
µ
Thus, under these “nominal” conditions:
( )
( )( )( )( )( ) 7300 0 . 1 318 002377 . 0 2 . 32
19815 44 . 1
2
=
LO
S
ft S
LO
3182 = @ nominal conditions
Suppose that we were not confident of the value and suspected that it
might be ±0.1 from nominal.
max
L
C
That is,
0.9 ” ”1.1
max
L
C
Sensitivity Analysis
Also, let’s suppose that the weight of the aircraft may need to be increased for a
higher load. Specifically, let’s consider a 10% weight increase:
1) Linear sensitivity analysis
2) Nonlinear sensitivity analysis (i.e. re-evaluation)
They both have their own advantage and disadvantages. The choice is often
made based on the problem and the tools available. We’ll look at both options.
Linear Sensitivity Analysis
Linear sensitivity based on Taylor series approximations. Suppose we were
interested in the variation of with w &
LO
S
max
L
C
Then:
W
W
S
C
C
S
C S
W W C C S
LO
L
L
LO
L LO
L L LO
A
c
c
+ A
c
c
+
~ A + A
max
max
max
max max
) (
) , (
That is, the change in is:
LO
S
W
W
S
C
C
S
S
LO
L
L
LO
LO
A
c
c
+ A
c
c
÷ A
max
max
The derivatives
W
S
C
S
LO
L
LO
c
c
c
c
&
max
are the linear sensitivities of to changes in
& , respectively.
LO
S
max
L
C j W
Returning to the example:
max
max
max
2
2
44 . 1
L
LO
L
L
LO
C
S
T SC g
W
C
S
÷ =
÷
=
c
c
·
µ
W
S
T SC g
W
W
S
LO
L
LO
2
44 . 1 2
max
=
÷
=
c
c
·
µ
These can often be more information by looking at percent or fractional changes:
W
W
W
S
S
W
C
C
C
S
S
C
W
W
S
S
C
C
S
S S
S
LO
LO L
L
L
LO
LO
L
LO
LO
L
L
LO
LO LO
LO
A
c
c
+
A
c
c
=
A
c
c
A
c
c
~
A
+
max
max
max
max
max
max
1 1
Fractional sensitivities
16.100 2
Sensitivity Analysis
For this problem:
W
W
S
S
W
S
S
W
C
C
S
S
C
S
S
C
LO
LO LO
LO
L
L
LO
LO
L
L
LO
L
A
~
A
¬ =
c
c
A
÷ ~
A
¬ =
c
c
0 . 2 0 . 2
1 0 . 1
max
max
max
max max
C
C
S
S
L
L
LO
LO
A
÷ ~
A
1
max
max
W
W
S
S
LO
LO
A
~
A
0 . 2
Thus, a small fraction change in will have an equal but opposite effect on
the take-off distance.
max
L
C
The weight change will result in a charge of which is twice as large and in
the same direction.
LO
S
Thus, is more sensitive to W than changes at least according to linear
analysis.
LO
S
max
L
C
Example
We were interested in varying ± 0.1 which according to linear analysis
would produce variation in take-off distance:
max
L
C
LO
S 1 . 0 µ
ft S S C
ft S S C
LO LO L
LO LO L
318 1 . 0 1 . 1
318 1 . 0 9 . 0
max
max
÷ = + ~ A ÷ =
+ = + ~ A ÷ = ¬
For a weight increase of 10% we find
( )( ) ( )
ft
S S lb W
LO LO
636
1 . 0 2 19815 1 . 1
+ ~
+ ~ A ÷ =
Nonlinear Sensitivity Analysis
For a nonlinear analysis, we simply re-evaluate the take off distance at the
desired condition (including the perturbations). So, to assess the impact of the
variations we find:
max
L
C
ft C S
C S
L LO
L LO
3535 ) 9 . 0 (
) 7300 )( 9 . 0 )( 318 )( 002377 . 0 )( 2 . 32 (
) 19815 ( 44 . 1
) 9 . 0 (
max
max
2
= =
= =
¬ ft C S
L LO
6 . 353 ) 1 . 0 (
max
+ = ÷ = A A
which agrees well with linear result
ft S S C
ft S S C
LO LO L
LO LO L
318 1 . 0 1 . 1
318 1 . 0 9 . 0
max
max
÷ = + ~ A ÷ =
+ = + ~ A ÷ =
16.100 3
Sensitivity Analysis
Similarly,
( ) ft C S
L LO
2892 1 . 1
max
= =
¬ ( ) ft C S
L LO
290 1 . 0
max
÷ = + = A
Finally, a 10% W increase to 21796lb’s gives:
( )
( ) ft W W S
ft lb W S
LO
LO
668 1 . 0
3850 21796
+ = + = A A
= =
16.100 4

Sign up to vote on this title
UsefulNot useful