Suppose we have a system like the one in the picture. Each component works and fails independently, and each component's duration is given by a random variable with exponential distribution with $\lambda = 0.002$.

What's the probability that the system will last more than $500$ hours?

enter image description here

I tried the following, but I'm not sure if its correct

Let $T_i$ be the duration of each component. We know that the system fails if component $1$ fails, or if both, components $2$ and $3$ fail (the two in parallel).

Therefore $P(T<t)= P(T_1<t) + P(T_2,T_3<t)- P(T_1,T_2,T_3<t)$, this yields (through independence), $P(T<t)=(1-\exp(-\lambda t))+\left(1-\exp(-\lambda t)\right)^2-(1-\exp(-\lambda t))^3$.

And now, I just do $P(T>500)=1-P(T<500)\approx 0.22$.

Is this correct?


Let $T_i$ be the lifetime of component $i$. Then the lifetime of the system is $$T = T_1\wedge(T_2\vee T_3)$$ The $T_i$ are independent with distribution function $F(t)=1-e^{-\lambda t}$, so $T_2\vee T_3$ has distribution function $F^2(t)=\left(1-e^{-\lambda t}\right)^2$. It follows then that $T$ has survivor function \begin{align} \overline G(t) &=(1-F(t))(1-F^2(t))\\ &= e^{-\lambda t}\left(1-\left(1-e^{-\lambda t}\right)\right)\\ &= 2e^{-2\lambda t} - e^{-3\lambda t} \end{align} A straightforward computation yields $$\mathbb P(T>500) = \overline G(500) = 2e^{-2}-e^{-3}\approx 0.220883.$$


Comment: Verification, by simulation (2 or3-place accuracy).

m = 10^6
a = rexp(m, .002)
b = rexp(m, .002)
c = rexp(m, .002)
mean((a > 500) & (b > 500 | c > 500))     ## '&' means 'and';  '|' means 'or'
## 0.221315

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.