ISyE 3232C

Z. She, YL. Chang

Stochastic Manufacturing and Service Systems

Fall 2015

Homework 4 - Solutions
September 20, 2015
1. Consider a single server queue. Initially, there is no customer in the system. Suppose that the interarrival time of the first 15 customers are: 2; 5; 2; 4; 5; 3; 9; 3; 10; 8; 3; 2; 16; 1; 8
In other words, the first customer will arrive in 2 minutes, and the second will arrive in 2+5 minutes,
and so on. Also, the service time of the first 15 customers are 3; 1; 1; 4; 3; 7; 5; 2; 6; 11; 9; 2; 1; 7; 6
(a) Compute the average waiting time (the time customer spend in buffer) of the first 10 departed
customers.
One can calculate the waiting time for each specific customer from the following table.
ui
vi
wi
zi

2
3
0
3

5
1
0
1

2
1
0
1

4
4
0
4

5
3
0
3

3
7
0
7

9
5
0
5

3
2
2
4

10
6
0
6

8
11
0
11

For example, consider the 1st customer. He arrives at time u1 = 2, and is served in v1 = 3 minutes.
Thus, he departs at time u1 + v1 = 5. New customer, the second customer in the queue, arrives
at time u1 + u2 = 7, so his waiting time is 0, because u1 + v1 = 5 < 7 = u1 + u2 . For the third
customer in the queue, he arrives at time u1 +u2 +u3 = 9 and get served immediately, because the
second customer has already leaved in time u1 +u2 +v2 = 8 and u1 +u2 +v2 = 8 < 9 = u1 +u2 +u3 .
The
arrives at time
P7 only customer who needs to wait is the 8th customer. The 7th customer P
7
u
=
30,
and
is
served
for
v
=
5
minutes.
Thus,
he
departs
at
time
i
7
i=0
i=0 ui + v7 =
P8
35. However, the 8th customer arrives at time
u
=
33.
Therefore,
he
has
to wait for
i=0 i
P7
P8
( i=0 ui + v7 ) i=0 ui = 35 33 = 2 minutes.
One can also refer to the figure in Problem (c) for a pictorial understanding of this process.
The average system time for of the first 10 departed customers is then
E[W ] = Total waiting time / number of customers =

2
= 0.2
10

.
(b) Compute the average system time (waiting time plus service time) of the first 10 departed customers.
The system time zi is given in the table above.
Thus, the average system time:
E[W ] =

3 + 1 + 1 + 4 + 3 + 7 + 5 + 4 + 6 + 11
= 4.5
10

(c) Draw the number of customers in the system (including in service) with respect to time t during
the first 20 minutes.
number of customers

3
2
1

9 10 11 12 13 14 15 16 17 18 19 20

time, t

Figure 1: Number of customers in the first 20 minutes

(d) Compute the average queue size during the first 20 minutes.
During the first 20 minutes, none of the customers are on the queue. Thus, the average queue
size is 0.
(e) Identify the time intervals when the server is busy and Compute the average server utilization
during the first 20 minutes (The fraction of time that the server is busy).
(See Figure 1)
The system is busy during the intervals [2,5), [7,8), [9,10), [13,17), and [18,20].
So the average utilization in the first 20 minutes is
11
3+1+1+4+2
=
20
20
(f) Does the Littles law hold for the average queue size in the first 19 minutes?
No. Littles law characterize quantities in the long-run average, in this problem, the given time is
too short.
2. We want to decide whether to employ a human operator or buy a machine to paint steel beams with a
rust inhibitor. Steel beams are produced at a constant rate of one every 15 minutes. A skilled human
operator takes an average time of 800 seconds to paint a steel beam, with a standard deviation of 300
seconds. An automatic painter takes on average 40 seconds more than the human painter to paint a
beam, but with a standard deviation of only 100 seconds. Estimate the expected waiting time in queue
of a steel beam for each of the operators. Comment on the effect of variability in service time.
By the Kingmans formula, for a G/G/1 queue, the expected waiting time in the queue is equal to
E[W ] =

1

c2a + c2s

1
2

where c2a and c2s are the squared coefficients of variations for the inter-arrival times and the service
times, respectively, is the arrival rate, is the service rate and is the utilization of the server. Since
each inter-arrival time is constant c2a = 0 and = 1/E[u1 ] = 60/15 = 4 beams/hour. Let us consider
the two possible options: human operator and automatic painter.
2

 Human operator:
= 1/E[v1 ] = 3600/800 = 4.5 beams/hour,
c2s =
=

V ar(V1 )
E[v1 ]2

= 90000/(800)2 0.14

= 4/4.5 = 0.88

Automatic painter:
= 1/E[v1 ] = 3600/840 4.286 beams/hour,
c2s =
=

V ar(V1 )
E[v1 ]2

= 10000/(840)2 0.014

= 4/4.286 = 0.933

Therefore, the average waiting time in the queue for the human operator is
E[W ] =

1
0.88
0.14

= 0.114
4.5 1 0.88
2

The average waiting time in the queue for the automatic painter is
E[W ] =

0.933
0.014

1
= 0.023
4.286 1 0.933
2

Even though the service rate of the human operator is greater than the service rate of the automatic
painter, because the service times of the automatic painter are less variable, the waiting time in the
queue of the automatic painter is a lot smaller.