Solutions to Exercise 1
Marvin Rausand
Email: [Link]@[Link]
2008-08-26
Problem 2.1
(a) The probability that the component survives two months is given by (1
month = 730 hours):
−5
Pr(T > 1460) = e −λ·1460 = e −2.5·10 ·1460
≈ 0.9642
(b) The mean time to failure of the component is:
1
MTTF = = 40 000 hours
λ
(c) The probability that the component survives its MTTF is:
Pr(T > MTTF) = e −λMTTF = e −1 ≈ 0.3679
Problem 2.2
(a) The failure rate can be determined by:
1
Pr(T > 100) = e −λ·100 =
2
that gives
1
ln e −λ·100 = ln = − ln 2
2
−λ · 100 = − ln 2
ln 2
λ = = 6.9 · 10−3
100
(b) The probability that the machine will survive 500 hours without failure
is:
ln 2
Pr(T > 500) = R(500) = e −λ·500 = e − 100 ·500
−5 1
= e ln2 = 2−5 = ≈ 0.031
32
1
(c) The probability that the machine will fail within 1000 hours, when you
know that the machine was functioning at 500 hours is:
Pr(T < 1000 ∩ T > 500)
Pr(T < 1000 | T > 500) =
Pr(T > 500)
R(500) − R(1000)
=
R(500)
2 − 2−10
−5
31
= = 1 − 2−5 = ≈ 0.97
2−5 32
Problem 2.3
(a) The total failure rate of the safety valve is:
1 1
λ= = [hours]−1 ≈ 1.70 · 10−5 [hours]−1
MTTF 2450 · 24
(b) The safety valve will survive three months (2190 hours) without failure
with probability:
−5
Pr(T > 2190) = e −1.70·10 ·2190
≈ 0.9634
(c) The failure rate with respect to critical failures is:
λcrit = 0.45 · λ = 7.65 · 10−6 [hours]−1
The mean time to a critical failure is hence:
1 1
MTTFcrit = = hours ≈ 130 719 hours ≈ 5447 days
λcrit 7.65 · 10−6
Problem 2.8
(a) The probability that the component survives t hours is:
t
Pr(T > t ) = R(t ) = e − = e −k t
2
0 ku du /2
Inserting k = 2.0 · 10−6 and t = 200 yield:
−6
Pr(T > 200) = e −2.0·10 ·200/2
= e −0.04 ≈ 0.96
(b) The mean time to failure, MTTF, is:
∞ ∞
−k t 2/2 1 2π
MTTF = R(t ) d t = e = ≈ 886 hours
0 0 2 k
The integral may be solved using mathematical tables or computer soft-
ware.
2
(c) The conditional probability is:
Pr(T > 400 ∩ T > 200) Pr(T > 400)
Pr(T > 400 | T > 200) = =
Pr(T > 200) Pr(T > 200)
1 −6
e − 2 2.0·10 ·(400)2
0.8521
= ≈ ≈ 0.8869
− 21 2.0·10−6·(200)2 0.9608
e
(d) Observe that the failure rate function z(t ) = k t is a linear function of t .
This is also the case for the Weibull distribution with shape parameter
α = 2. By comparing with the Weibull distribution we see that the above
distribution with shape parameter α = 2, and
distribution is a Weibull
scale parameter λ = k/2. Since α = 2, the distribution is also known as
the Raleigh distribution.
Problem 2.27
(a) The time to failure T is uniformly distributed over (a, b]. The survivor
function is easily found from the area under the probability density func-
tion.
⎧
⎪
⎨ 1 for t ≤ a
R(t ) = b − t for a < t ≤ b
⎩ b−a
⎪
0 for t > b
The failure rate function z(t ) is found from z(t ) = f (t )/R(t ) as
1/(b − a) 1
z(t ) = = for a < t ≤ b
(b − t )/(b − a) b − t
A sketch of z(t ) is given in Figure 1.
4
z(t)
0
0 1 2 3 4 5 6
Time t (a=2, b=5)
Figure 1: Failure rate function – Problem 2.27
3
Problem 2.28
(a) For f (t ) to be a valid probability density function, the area under the
curve f (t ) has to equal one. Therefore
1
2bc = 1
2
1
c =
b
This gives the probability density function:
⎧
⎨ 0 for t ∉ [a − b, a + b]
f (t ) = [t − (a − b)] /b 2 for a − b ≤ t ≤ a
⎩
− [t − (a + b)] /b 2 for a < t ≤ a + b
(b) The survivor function R(t ) is given by
⎧
⎪
⎪ 1 for 0 ≤ t < a − b
∞ ⎨ b 2 − 1 (t − (a − b))2 /b 2
⎪
for a − b ≤ t < a
R(t ) = f (t ) d t = 2
⎪
⎪ [t − (a + b)]2 /2b 2 for a ≤ t < a + b
t ⎪
⎩
0 for t ≥ a + b
(c) The failure rate function z(t) is given by
[t −(a−b)]
f (t ) for a − b ≤ t < a
z(t ) = = b 2 − 21 [t −(a−b)]2
2
R(t ) (a+b)−t for a ≤ t < a + b
A sketch of z(t ) is given in Figure 2 for a = 2 and b = 1.
6
z(t)
0
0,5 1,0 1,5 2,0 2,5 3,0
Time t (a=2, b=1)
Figure 2: Failure rate function – Problem 2.28