Solutions for Student exercises:reliability theory

[Xiuya Li]

Hi All,

 

Please comment on my solutions for the last homework -- Student exercises: reliability theory.

 

Good luck for us all in the final!
Ted Li

[alex smirnov]

Hi Ted,

I got all the same answers as you, except #5.

I think the answers should simply be  17.78% and 10% respectively

your're calculating lamda, which is the failure rate.  I believe the question is asking for probability of an engine failure

so if you have .001 chance of loosing all 4 engines, the probability of an engine failure that would produce that, is just a 4th root of that

am I wrong?

Alex

[Nicholas Woo]

I agree with Alex.

[Xiuya Li]

I have no enough information to judge which one is correct. This came
back to the clarification of the problems. Can some one send email to
professor and ask (professor cc'ed here).
Thanks!
Ted

[Savitha Gandikota]

I got the same answers as Alex. Below is how I reasoned it.

Prob of losing all 4 engines is 1 in 1000 0.0010 Given
Prob of losing all 4 engines = (Prob of losing an engine)^4
Prob of losing single engine = (Prob of losing all engines) ^¼
Prob of losing single engine = (1/1000)^(1/4) 0.17783

Prob of losing all 4 engines is 1 in 10000 0.0001 Given
Prob of losing single engine = (Prob of losing all engines) ^¼
Prob of losing single engine = (1/10000)^(1/4) 0.10000

Savitha

On Mon, May 30, 2011 at 10:30 AM, alex smirnov <alex.s...@gmail.com> wrote:

[Ajay Goyal]

Hi All,

I think the answer to #3 should be 0.99696089

 

As we calculated the reliability of a single harddisk is ~95%

So when we have two disks in parallel the reliability of the system should go up and not come down (to 89%) as agreed by all till now.

 

We need to use the permutation/combination formula for K out of N failures.

 

thanks

ajay

[alex smirnov]

Hi Ajay,

I think the question is asking for serial reliability, not parallel

I read "one or both" as "if one fails => fail" & "if two fail => fail"

Alex

[Xiuya Li]

Talked with professor, #5 should be just F(t), Alex was right on that.

 

Other problems should be fine.

 

Thanks!

Ted

[alex smirnov]

it's good to be right  =)

[Jonathan Carandang]

Good one, Alex

Jonathan