SOLUTION: A commercial jet aircraft has four engines. For an aircraft in flight to land safely, at least two engines should be in working condition. Each engine has an independent reliabilit

Algebra ->  Probability-and-statistics -> SOLUTION: A commercial jet aircraft has four engines. For an aircraft in flight to land safely, at least two engines should be in working condition. Each engine has an independent reliabilit      Log On


   



Question 1121297: A commercial jet aircraft has four engines. For an aircraft in flight to land safely, at least two engines should be in working condition. Each engine has an independent reliability of p=92%.
a) if the probability of landing safely must be at least 99.5% what is the minimum value for p ?
b)...
c)...
The question is quoted from the book "Complete Business Statistics" 7th Edition by Sounderpandian (Asian edition)
Chapter 3, problem 3-41 (page no.134)

Found 2 solutions by Theo, ikleyn:
Answer by Theo(13342) About Me  (Show Source):
You can put this solution on YOUR website!
it looks like 89% reliability is the magic number.

that will give you an overall probability of 2 or more engines working at the same time of .995115.

anything less than that will result in an overall probability of less than .995.

i used the binomial theorem and kept dropping the individual reliability percent down by 1 until the overall probability became less than 99.5%.

in other words, i started at p = .92, then .91, then .90, then .89, then .88.

.89 was above 99.5% overall.
.88 was below 99.5% overall.

the calculations are shown in the following spreadsheet printouts.

$$$

$$$

$$$

i'm not exactly sure if this is the right formula to use, but if it is, there's a high probability that the answer will be correct.

the binomial formula tells you the probability of x occurrences out of n, given the probability that the event will occur or not occur.

the formula is p(x) = p^x * q^(n-x) * c(n,x).

n is 4 in this problem.
x ranges from 0 to 4.
p is .92 or .91 or .90 or .89 or .88
q is 1 - p
c(n,x) is the numbe of ways you can get sets of x elements out of a set of n elements when order is not important.
that formula is c(n,x) = n! / (x! * (n-x)!)











Answer by ikleyn(52794) About Me  (Show Source):
You can put this solution on YOUR website!
.
The probability that all 4 engines are in order and work properly is p%5E4.


The probability that 3 engines of 4 are in order and work properly is C%5B4%5D%5E3%2Ap%5E3%2A%281-p%29 = 4%2Ap%5E3%281-p%29.


The probability that 2 engines of 4 are in order and work properly is C%5B4%5D%5E2%2Ap%5E2%2A%281-p%29%5E2 = 6%2Ap%5E2%2A%281-p%29%5E2


The probability to have one of these cases is the sum


P(p) = p%5E4 + 4%2Ap%5E3%2A%281-p%29 + 6%2Ap%5E2%2A%281-p%29%5E2.


They ask you to find minimal "p" such that  P(p) is still greater than  0.995.


It is easy calculation which can be done with MS Excel, for example.


Below is the table which I got in this way:


      Table


 p       P(p)
---------------------
0.99	1.0000
0.98	1.0000
0.97	0.9999
0.96	0.9998
0.95	0.9995
0.94	0.9992
0.93	0.9987
0.92	0.9981
0.91	0.9973
0.90    0.9963
0.89	0.9951    <<<---===
0.88	0.9937
0.87	0.9921
0.86	0.9902
0.85	0.9880
0.84	0.9856
0.83	0.9829


It shows that the minimal "p" under the question is about p = 0.89.

I hope that this solution and this explanation is more clear than that of the other tutor.


--------------------
Comment from student :  Hello Ikleyn,  Thanks for your generous help,  really appreciate it.
I don't want to push it hard to greater lengths,  but could you please help me figure out a way or a method
without actually using the brute force technique.  That is to say,  finding P(p) without trying every single  p - from 0.99
down to 0.88  Anyways,  thanks for you help.  You have a good time ahead.
--------------------


My response :   Yes,  such a method does exist.

It is Calculus.  You need to take the derivative  P'(p) and equate it to zero.

The derivative will be a polynomial of the degree 3 without the constant term.

After equating to zero you can cancel  "p"  in degree one - then you will get a quadratic equation.

Then you can solve it using the quadratic formula.

In this way you will get the required "minimal"  p  "analytically".