Chapter 4: Waiting Line Management

4.1 Introduction

Understanding waiting lines or queues and learning how to manage them is one of the most
important areas in operations management. In our service economy we wait in line every day,
from driving to work to checking out at the supermarket. We also encounter waiting lines at
factories-jobs wait in lines to be worked on at different machines, and machines themselves wait
their turn to be overhauled. In short, waiting lines are pervasive.

• Queue: a line of waiting customers, jobs, equipment or the like, who require service from
one or more service providers.

• Queuing system: waiting room + customers + service provider

Figure 4.1 queue system

Arrival rate: the expected number of customers that arrive each period.

Service rate: the capacity of a server measured in number of units that can be processed over a
given time period.

1
Waiting lines are most likely to occur when arrivals are bunched or when service times are
particularly lengthy, and they are also very likely to occur when both factors are present.

4.2 Economics of Waiting Line Problem

The central problem in virtually every waiting line situation is a trade-off decision. The manager
must weigh the added cost of providing more rapid service against the inherent cost of waiting.
For example, if we find that the total time our employees spend in the line waiting to use a
copying machine would otherwise be spent in productive activities; we could compare the cost of
installing one additional machine to the value of employee time saved. The decision could then
be reduced to dollar terms and the choice easily made.

Cost-Effectiveness Balance

Exhibit TN 4.1 shows the essential trade-off relationship under typical (steady-state) customer
traffic conditions. Initially, with minimal service capacity, the waiting line cost is at a maximum.
As service capacity is increased, there is a reduction in the number of customers in the line and
in their waiting times, which decreases waiting line cost. The variation in this function is often
represented by the negative exponential curve. The cost of installing service capacity is shown
simplistically as a linear rather than step function.

The aggregate or total cost is shown as a U-shaped curve, a common approximation in such
equilibrium problems. The idealized optimal cost is found at the crossover point between the
service capacity and waiting line curves.

Exhibit TN 4.1

2
4.3 Suggestions for Managing Queues (Qualitative)

The following are some useful suggestions for managing queues that go beyond the quantitative
waiting line models.

1. Determine an acceptable waiting time for your customers. How long do your customers expect
to wait? Set operational objectives based on what is acceptable.

2. Try to divert your customer’s attention when waiting. Providing music, a video, or some other
form of entertainment may help to distract the customers from the fact they are waiting.

3. Inform your customers of what to expect. This is especially important when the waiting time
will be longer than normal. Tell them why the waiting time is longer than normal and what you
are doing to alleviate the queue.

4. Keep employees not serving the customers out of sight. Nothing is more frustrating to
someone waiting in line than to see employees, who potentially could be serving those in line,
working on other activities.

5. Segment customers. If a group of customers needs some-thing that can be done very quickly,
give them a special line so they do not have to wait for the slower customers.

6. Train your servers to be friendly. Greeting the customer by name, or providing some other
special attention, can go a long way toward overcoming the negative feeling of a long wait.
[Hint: Rather than servers being told to just “be friendly,” psychologists suggest they be told
when to invoke specific friendly actions such as smiling-when greeting customers, when taking
orders, and when giving change (in a convenience store). Tests using such specific behavioral
actions have shown significant increases in perceived friendliness of the servers in the eyes of
the customer.]

7. Encourage customers to come during the slack periods. Inform customers of times when they
usually would not have to wait; also tell them when the peak periods are-this may help to smooth
the load.

8 Take a long-term perspective toward getting rid of the queues. Develop plans for alternative
ways to serve your customers. Where appropriate, develop plans for automating or speeding up

3
the process in some manner. This is not to say you want to eliminate personal attention; some
customers expect this.

4.4 The Queuing System

The queuing system consists essentially of three major components: (1) the source population
and the way customers arrive at the system, (2) the servicing system, and (3) the condition of the
customers exiting the system.

4.4.1 Customer Arrivals

Arrivals at a service system may be drawn from a finite or an infinite population. The distinction
is important because the analyses are based on different premises and require different equations
for their solution.

Finite Population: A finite population refers to the limited-size customer, machines etc. pool
that will use the service. As an example, consider a group of six machines maintained by one
repairperson. Breakdown rate may represent arrival rate.

Infinite Population: An infinite population is large enough in relation to the service system so
that the population size caused by subtractions or additions to the population (a customer needing
service or a serviced customer returning to the population) does not significantly affect the
system probabilities. e.g. a department store with 10,000 customers.

Distribution of Arrivals

When describing a waiting system, we need to define the manner in which customers or the
waiting units are arrived for service.

Arrival rate: the expected number of units that arrive per period (2 customers per hour). A
constant arrival distribution is periodic, with exactly the same time between successive arrivals.
In productive systems, the only arrivals that truly approach a constant interval period are those
subject to machine control. Much more common are variable (random) arrival distributions.

4
In observing arrivals at a service facility, we can look at them from two viewpoints:

First, we can analyze the time between successive arrivals to see if the times follow some
statistical distribution. Usually we assume that the time between arrivals is exponentially
distributed. Second, we can set some time length (T) and try to determine how many arrivals
might enter the system within T. We typically assume that the number of arrivals per time unit is
Poisson distributed.

Poisson Distribution: In the second case, where one is interested in the number of arrivals
during some time period T, the distribution appears as in Exhibit TN4.3 and is obtained by
finding the probability of exactly n arrivals during T. If the arrival process is random, the
distribution is the Poisson, and the formula is:

The formula shows the probability of exactly n arrivals in time T. For example, if the mean
arrival rate of units into a system is three per minute (λ=3) and we want to find the probability
that exactly five units will arrive within a one-minute period (n=5, T=1), we have

That is, there is a 10.1 percent chance that there will be five arrivals in any one-minute interval.
The Poisson distribution is discrete because n refers, in our example, to the number of arrivals in
a system, and this must be an integer. (For example, there cannot be 1.5 arrivals.)

5
Exhibit TN 4.3: Poisson distribution

Exponential Distribution: In the first case, when arrivals at a service facility occur in a purely
random fashion, a plot of the inter-arrival times yields an exponential distribution such as that
shown in Exhibit TN4.2. Exhibit TN4.2: Exponential Distribution

-λt
The probability density function is f(t)=λe , where λ is the mean number of arrivals per time

period.

The formula to determine the probability that the inter-arrival time is greater than t is the cumulative
-λt
density function F(T > t) = e .

The formula to determine the probability that the inter-arrival time is less than equal to t is the cumulative
-λt
density function F(T < = t)= 1- e .

6
Example: inter-arrival time with λ=1

Exercise: find the probability that the time between arrivals is between 1 and 4 minutes.

Hint: P(1 <=t <= 4) = 1-[(p>=4) +p(t<=1)] or the integral of the function between 1 and 4.

Therefore, the arrival rate, inter-arrival time and service time are not deterministic rather they are
stochastic.

Remember that the time between arrivals is exponentially distributed and the number of arrivals
per unit of time is Poisson distributed.

Other arrival characteristics include arrival patterns, size of arrival units, and degree of patience.
(See Exhibit TN4.4.)

Exhibit TN4.4: arrival characteristics

7
Balking: arrive, survey both the service facility and the length of the line, and then decide to
leave.

Reneging: view the situation, join the waiting line expecting that the line will move fast, and
then, after some period of time, if the line moves very slowly and then decide to leave.

Jockeying: shifts from one line to the other, depending on speed of service of the other. It
happens in multiple server queue system with dedicated queue lines.

The Queuing system: Factors

The queuing system consists primarily of the waiting line(s) and the available number of servers.
Factors to consider with waiting lines include the line length, number of lines, and queue
discipline.

Queue Length: the number of units in a line

Finite queue length: In this case, there is a restriction on the queue length. If queue length
exceeds some threshold limit, then the new arrival beyond the limit will not join the queue and
return back without getting service. Example, consider a maximum service capacity of garage
with single server is six machines per day. If a 7th machine arrives, then it will not join the queue.

Infinite queue length: In this case, there is a restriction on the queue length. Example, queue in
bank

Number of lines: single or multiple lines

Queue discipline: A queue discipline is a priority rule or set of rules for determining the order of
service to customers in a waiting line. The rules selected can have a dramatic effect on the
system’s overall performance. The number of customers in line, the average waiting time, the
range of variability in waiting time, and the efficiency of the service facility are just a few of the
factors affected by the choice of priority rules.

Probably the most common priority rule is first come, first served (FCFS). This rule states that
customers in line are served on the basis of their chronological arrival. This is popularly accepted

8
as the fairest rule, although in practice it discriminates against the arrival requiring a short
service time.

4.4.2 Service Time Distribution

Another important feature of the waiting structure is the time the customer or unit spends with
the server once the service has started. Waiting line formulas generally specify service rate as the
capacity of the server in number of units per time period (such as 12 completions per hour) and
not as service time, which might average five minutes each. A constant service time rule states
that each service takes exactly the same time. As in constant arrivals, this characteristic is
generally limited to machine-controlled operations.

When service times are random, they can be approximated by the exponential distribution. When
using the exponential distribution as an approximation of the service times, we will refer to µ as
the average number of units or customers that can be served per time period.

Line Structures as Exhibit TN 4.5 shows, the flow of items to be serviced may go through a
single line, multiple lines, or some mixtures of the two.

9
Exhibit TN 4.5: Line Structures

10
4.4.3 Exit

Once a customer is served, two exit fates are possible: (1) The customer may return to the source
population and immediately become a competing candidate for service again or (2) there may be
a low probability of re service. The first case can be illustrated by a machine that has been
routinely repaired and returned to duty but may break down again; the second can be illustrated
by a machine that has been overhauled or modified and has a low probability of re-service over
the near future. In a lighter vein, we might refer to the first as the “recurring-common-cold case”
and to the second as the “appendectomy-only-once case.” It should be apparent that when the
population source is finite, any change in the service performed on customers who return to the
population modifies the arrival rate at the service facility. This, of course, alters the
characteristics of the waiting line under study and necessitates reanalysis of the problem.

11
 4.5 Waiting/queue line models (Quantitative)
Waiting line model is a mathematical model for service design.

 Widely used to estimate desired performance measures of the system

 Typical measures
 Server utilization
 Length of waiting lines
 Delays of customers
 Applications
 Determine the minimum number of servers needed at a service center
 Detection of performance bottleneck or congestion
 Evaluate alternative system designs

The conventional notation for the characteristics of queuing situation can be given in the
following format; it is also called Lee Kendall’s Notation:

Lee Kendall’s Notation→ (a/b/c) :(d/e/f)

Where,

a = Description of the arrivals distribution, Poisson Distribution (M)

b = Description of the departures (service time) distribution, constant (D), Exponential (M)

c = Number of parallel servers; (C)=1, 2,3,4,…

d = Queue discipline; FIFO, LIFO, SIRO, GD (any type)

e = Maximum number allowed in the system (in queue plus in service); finite

(Q numbers), infinite (∞)

f = Size of calling source; finite (N) or infinite (∞)

For example, the model (M/D/2) : ( FIFO/N/∞) uses :

 Poisson arrival time,

 Constant service time or deterministic service time
 2 parallel servers or channels,
 The queue discipline is First come first served,
 There is a limit of N customers in the entire system, and
 The size of the source from which customers arrive is infinite.

There are so many queue models but in this section we will discuss only Model 1.

12
4.5.1 Model 1 (M/M/1) : ( FIFO/∞/∞)
 Arrivals are described by a Poisson probability distribution and come from an infinite
population
 arrivals are served on a FIFO basis and every arrival waits to be served regardless of the
length of the queue
 Arrivals are independent of preceding arrivals but the average number of arrivals does not
change over time
 Service times vary from one customer to the next and are independent of one another, but
their average rate is known
 Service times occur according to the negative exponential distribution

 The service rate is faster than the arrival rate- < 1.

The performance measures are given by these simple formulas:

Ls = Average # of customers in the system = /( - )

LQ = Average # of customers in the queue = Ls - /

Ws = Average customer time in the system = Ls/

WQ = Average customer time in the queue = LQ/

p0 = Probability 0 customers in the system = 1- /

n
pn = Probability n customers in the system = ( / ) p 0

ρ = Average number of busy servers (utilization rate) or Average number customers being served
at any time = /

Example 1: mary’s shoe customer arrival rate and service rate is as follow and calculate all
performance measures.

• Customers arrive according to a Poisson Process about once every 12 minutes

• Service times are exponential and average 8 min.

13
 Solution: Model 1-One server

Given:

– = (60min./hr)/(12 min./customer) = 60/12 = 5/hr.

– (service rate) = (60min/hr)/(8min./customer) = 7.5/hr.

– Formula for average in expo and poison distribution

Mary’s shoes performance measures:

• Avg # of busy servers (utilization rate) or Avg # customers being served

=  = / =(5/7.5) = 2/3 =0.667

• Average # cus.in the system – Ls = /( - ) = 5/(7.5-5) = 2

• Average # cus.in the queue -Lq = Ls - / = 2 - (2/3) = 4/3 = 1.333

• Avg. customer time in the system -Ws = Ls/ = 2/5 hrs = 0.4hr =24 minutes.

• Or , Avg. customer time in the system -Ws = 1/( - )=1/2.5=0.4 hr =24 minutes

• Avg cust.time in the queue - Wq = Lq/ = (4/3)/5 = 4/15 hrs=0.2667hr= 16minutes.

• Prob. 0 customers in the system - p0 = 1- / = 1-(2/3) = 1/3 = 0.3333

• Prob. 3 customers in the system - pn=( / )3 p0 =(2/3)3(1/3) =8/81=0.0988=9.88% of

the time there will be 3 customers in the system (i.e .1 being served, 2 in queue).

Discussion: 1. Prob. (No Queue) =?

2. Probability of at least 2 customers in the system (n>=2) =?

Example 2 In a machine shop, it is observed that the failure of machine tools and equipment

follow Poisson distribution which has an average rate of 6 failures of machines per hour. Two
mechanics are available for taking up repair work. Mechanic A charges 20 Br per hour and his
service rate is 8 per hour. Mechanics B charges 40 Br per hour and he services at the rate of 12
per hour, on an average. Suppose that the down time of a machine tool costs 200 Br per hr and
assume one shift per day. In addition, let it fulfills the model one’s assumptions. Which
mechanic should be hired?

14
Solution: Given

= 6/hr, 1 =8/hr for Mechanic A and 2 =12/hr for Mechanic B

Case ( i) for Mechanic A

Ls = /( 1 - ) WQ= LQ/

=6/(8-6) = 2.25/6

=3 machines =0.375hr *8hr

Ws= Ls/ =3/6=1/2hr =3 hr per day

=1/2*8=4hr/per day

LQ= Ls- / 1 Total cost=hire charges +down-time cost

=3-(6/8) =20*8 + 4*200

= 2.25 machines =960 Br

Case (ii) for Mechanic B

Ls = /( 2 - ) WQ= LQ/

Ls =6/(12-6) = (1/2)/6

=1machines =(1/12)hr *8hr

Ws=Ls/ =1/6hr =2/3 hr per day

1/6*8=8/6hr per day

LQ= Ls- / 2 Total cost=hire charges +down-time cost

=1-(6/12) =40*8 + (8/6)*200

= 1/2 machines =586.7 Br

15
Conclusion: Mechanic B should be hired because 586.7 Br <960 Br.

4.6 MODEL 2 -(M/M/1) : ( FIFO/N/∞)

Where, N is maximum allowed entities in the system. All other notations are the same with
model 1.

P0 =

16
Example

17
NOTE: in model 2,

i. those who join the system are happy because waiting time in the system (Ws), waiting time in
the queue (Wq), expected number in the system (Ls) … decrease extremely. So, from customer
point of view this model is better than model 1.

ii. Server utilization is decreased and those who do not join the system is increased (in the above
example 4.71% of the time, customers do not join the system). So, from company point of view
this model is worse than model 1.

18
4.7multiple Server Model - (M/M/C): (FIFO/∞/∞)

Here, the number of server C >=2.

Exercise: using the following formula, measure the system performance and check (with the
solution given next to the formula) based on the given parameters for a two server model system.

19
Example: two server model

Let C=2 computers, =10cus/hr, =6cus/hr

Then, determine the following:

i) The probability that both the computers are free when a customer arrives.

ans P0 =0.0909

ii) prob.(no queue)= P0 +P1 + P2

ii) prob.(a customer gets a computer)= P 0 +P1

iii) Lq= 3.7878 customers

iv) Wq = = 0.37878hrs

v ) Ls = Lq +ρ =3.7878 + = 3.7878 + = 5.4545customers

vi) System utilization= = 83.33%

20
 The end!

Prepared by Fantahun G.&Dobosha A.2011E.C

21