SOLUTIONS MANUAL for Stochastic Modeling: Analysis and ...
SOLUTIONS MANUAL for Stochastic Modeling: Analysis and ...
SOLUTIONS MANUAL for Stochastic Modeling: Analysis and ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
It appears that there will be very little queueing even with the minimum of 2 <strong>for</strong>klifts.<br />
31. We first approximate the drive-up window as an M/M/s/4 queue with λ 1 = 1/2<br />
customer/minute <strong>and</strong> µ 1 = 1/1.8 customer/minute. (This approximation is rough<br />
because the true service-time distribution is more nearly normal, <strong>and</strong> ε s < 1.)<br />
117<br />
s π 4<br />
1 0.16<br />
2 0.03<br />
π 4 is the probability that an arriving car will find the queue full <strong>and</strong> thus have to park.<br />
Adding a window reduces this dramatically.<br />
The rate at which customers are turned away from the drive-up window is λ 1 π 4 =<br />
(1/2)π 4 . There<strong>for</strong>e, the overall arrival rate into the bank is λ 2 =1+(1/2) π 4 customers<br />
per minute. This process will not be a Poisson process, but we approximate it as one.<br />
As a first cut we model the tellers inside the bank as an M/M/s 2 queue with µ 2 =1/1.4<br />
customer/minute.<br />
(s 1 ,s 2 ) λ 2 w q (minutes)<br />
(1, 2) 1.08 1.9<br />
(2, 2) 1.02 1.5<br />
(1, 3) 1.08 0.2<br />
The bank can now decide which improvement in per<strong>for</strong>mance is more valuable.<br />
Comment: The approximation <strong>for</strong> the inside tellers can be improved by using the<br />
GI/G/s adjustment. Clearly, ε s =1.0/(1.4) 2 ≈ 0.5.Theexactvalueofε a can also be<br />
computed, but requires tools not used in this book, so set ε a = 1 <strong>for</strong> a Poisson process.<br />
32. No answer provided.<br />
33. No answer provided.<br />
34. No answer provided.<br />
35. No answer provided.