|Popui2O0

Used Inpu

in
Keadal-e

Balik
Notathgns Poisson Pattern(M)
(aie)
ofCustomers
Bebeviours
Negmtve
ErExponent
lang(E(ró))inl(M)
General
:(e)

Limited (N) Customer's

Zone
Unlimited (
Renege Configur
ation Queue
Single
Muli

Jockey First n Fist Out (FTF)

Last in First Out
Select in Random(LIFO)
Order (SIRO)
Pre-Emptive
-Non-Pre-Bmpive
System
Queue

Service Server Stationary

of Customer Moving
Customer Stationary
Server Moving

Poisson Pattecn(M)
Exponential(M)
Erlang(Er)
General(G)
Server's
Zone

-Singie (1) V
Multi (C)

Configur
Service Parallel
ation
Series

rre
Output
Unsatisiod
Unstcady
men Service
State Transient
Satisfied Steady

Dissatisfied
DNVTÐ VIVAHLdVHD
Queue Models (Waiting Lines)
423
80Introduction
Oneof the common
system. Rather, there will notsituations
in our daily life
be any person on this occurs with waiting or queue
moubles due to waiting such as at bus stops, ticket earth who did not experience the
bank counters, traffic lights andso On. booths, doctor'sclinic, petrol bunks,
shops where the machines wait for Queues are very common in industries also, in
mterial waiting for operations etc. These
may even cost hugely in some
repair/maintenance,
queues ruin
operators wait at tool cribs,
the time of customers and it
occasions.
costs are the primary concerns of these The minimisation of waiting time and thus
queue models in Operation Research.
81 Terminology
The following terms are commonly used in queue
models.
1. Customers : The persons or objects that require certain service are called
customers.

2. Server : The person or an object or a machine that

service is known as server.
provides certain defined
3. Service : The activity between server and customer is called service that
Consumnes some time.

4. Queue or Waiting Line :A systematicor a disciplined arrangement of a group

of persons or objects that wait for a service is called queue or waiting line.
5. Arrival : The process of customers coming towards service facility or server to
receive certain service is Arrival.

6. Mean Rate of Arrival : The average number of customers arriving per unit
time is called mean rate of arrival and is denoted by (read as LAMBDA).
A= Total no. of customersarriving/total time taken.
7. Mean Inter-arrival Time : It is the average time gap between two consecutive
rate i.e.,
arrivals of customers. It is the inverse of mean arrival
1/2= Total arrival time
Number of arrivals
of customers served per unit time
O. Rate of Service : It is the average number
and is denoted by u (read as MUE).
Total number of customers served
Total service time
customers
Service Time : It is the average time taken by the server to serve a
Mean Total service time
rate i.e., 1/4= Number of customers served
and is equalto inverse of service
 cice
8.2 When Does A Queue Result?
Suppose a queue system whose arrival rate is A
and service rate is u.
Arrival rate Service rate
(a) ()

Arrivals Service Served customers

facility
FIGURE S.: QUEUE DR WAITING LINE
Now, in this system any of the following three relations must
trichotomy). occur (by law of
(i) A> (ü) A= (ii) A<
i) When ,> u, i.e., the rate of arrival is
greater than rate of service, i.e., number
of customers arriving is greater than number
(time kept constant), the system results in pilingofup
customers served per unit time
of arrived customerswaiting
for service and thus a queue results.
(ü) If A=u, ie., number of customners
arrived is equal to customers served, there
will not be any waiting, and this queue system has cent percent eficiency.
(iü) 2<u, i.e., number of customers
If
served per unit time. This means that arriving is less than number of customerS
server will be free, waiting for customers
to arrive. Thus the queue results in server's
point of view.
Thus quee results where there is an inperfect matching
the rate of seruice. between the rate of amival and

8.3 Costs Associated with Waiting Lines

Queuing theory, can be
(Queuing)
applied to variety of operational situations where there
is an imperfect matching between
of arrival and the rate of arrivals and the rate of service. The rate
service
matching usually occurshowever,
due to
depend on the probability studies and the imperfect
service time. These may cost oninability of server to predict accuratey the arrival and
service facility. Thus the levelof ustomer
service
due to waiting or on server to
(either service operate the
facilities) determined by cost anaBysis of the
is rate or the number of service
1. Cost of offeringservice. following conflicting costs.
two
2. Cost incurred due to delay in
waiting. offering service or due to keeping customer
Now, we can analyse these as
1. Cost of Service: If
follows.
of service facilities existing service level in terms of rate of service or
number
(channels) reduced, the cost of operation comes dow.
is
In other words, inrease in level of
facilities. service increases the cost of operating serice
Queue Models (Waiting Lines) 425
2. Waiting Costs :In contrast to the above, the
will hike the customers waiting time or mayreduction
result in
in existing service level
long queues. Thus the
increase in service level will decrease the waiting costs and
level will increase the waiting costs. decrease in service
Thus the optimum level of service can be obtained by
achieving a balance
between the above two costs or at the level where the total cost
This isinterpreted through a graph shown below. shows minimum.
Total cost
Service cost
(to service)

8Waiting cost Minimum total cost

(to customer)
Optimum level
of service

Level of service

FIGURE 8.2: CoST ANALYSIS OF QUEUE SYSTEM

8.4 Queue System : Salient Features

Aqueue system may be studied by diving into two subsystems viz arrival system
and service system. The arrival system is based on customer's behaviours and their
process of arrival while the service system is dependent on the service mechanism and
server's behaviour.
These are diagrammatically showm in the bloch diagram given at the beginning of this
chapler. here below in
The essential features shown in the block diagram are discussed
detail.
set that
input source or population is aThus
1. Input Source or Population : The
customers to come out for service. the
Contains the probable potential
may consist of sub populations. The
nput source need not be homogenous andpatients for the queue system at a
Input source may contain the subset of
doctor'sclinic, a sub set of vehicle owners for queue system of petrol bunk, a
passengers who may enter a waiting line for a bus and so on. Another
as
Subset of
noted here is that all the population need not be customers, butare
point to be defined service
come out of population with an intention to get the
ney
considered as the customers to that service system.
customers are expected to arrive at their own convenience
S ATnyal Pattern:TheHowever. the time gaps between any two consecutive arrivals,
and conditions. pattern. These patterns of arrival times
Tnoticed, will be fit in certain fashion or distribution.
often follow one of the following probability
(i) Poisson distribution (represented by M)
(1) Exponential distribuion (represented by M)
 distribution (represented by E, )
(ii) Erlang
General pattern (represented by G)
(iv) queues) The limit of
(Restricted/unrestricted queueserver,i the
: server's bechaviour. If the limit is imposed by the
of Queue by
Limit restricted
3. times
for a given period. For example, ifa doctor has
arrivals
that will belimited
he would treat 50 out-patients only per day, the queue will be limited g
finite and 51st out-patient will have to be in queue to enter the queue i.., he
conditon,
will be in secondary queue while the first 50 will be in primary queue. Howeve,
if there is no restriction, the queue is said to be unlimited or infinite. The limitet
queue is denoted by Nwhile unlimited queue by oo,
Limited queue may be adopted by space restrictions or other conditions ls
For instance, a service facility such as a cinema hall, has restrictei
accommodation which can not accommodate more than its space. In such caset
the server will not allow further arrivals unless some space is vacated o
accommodate the new entry. In some other cases, the limited queue system ma
be adopted to prevent a formal queue and to avoid nuisances. These queue
are often referred to as 'take-a-number or 'take-appointment' or 'reservaion
policy.However, the queue may be unlimited for geting into this 'reservaion
4. Queue Discipline : The queue discipline is the order in which the customer is
selected from the queue for service. There are numerous ways in whki
customers in queue can be served of which some are listed below.
(a) First In First Out (FIFO) or First Come First Serve (FCFS) : It is the
discipline in which the customers are served in the chronological orde
of their arrivals.
e.g. Tichets at a cinema hall; sales al a
platform etc. grocery shop, trains om a (singleline
(b) Last In First Out (LIFO) or Last
is made in opposite order of Come First Serve (LCES) : If the secomes
arrivalsgoes
last is served first and first obviously of to
customers, ever
lastittiis i.e.,ledwhoLIFO or L
LCES
system.
e.g. Stack of plales; loadirg and fing!
papers in chronological orders;unloading
wearing socksa truck
andorshoes; dressngofice
go-down; ashirtad
coat over u, packtng systems etc.
(c) Service In Random Order
is picked up at (SIRO) : By this rule, the for servt
random, irespective
eg. Lottery system from whch one is pickedof up,
their
the arrivals.
customer
dresses waiting na wardro
from which one is to be
commodues or vegelables etc.chosen, food stuffs in a buffet, sales
(d) Priority Service : Under
his ale, the certain
customer(s) due osoIe server gives to
priornty cost grouip
of the customers.
e.g. A telephone urgent call
mportance or
prestigeous or hign
guen to a customer is charged at
separate counler for cheques at a who buys
at (APSRTC) reservaton counter higher
electricity bill a CATIConcessiniona giv
Annsal
priority
Ticket) cardelc. payment counter;
dels (Waiting Lihes)

highest priority given to

iscustomer
(e) Pre-cmptie Priority Rule : Under this rule,
arrival and costs. The
iscertain to enterirespective
customer(s)
allowed of their
into service iumediately after entering into queue
system, even if another (lower priority) customer is already in service.
1.e, the lower prnority customer who is in service will be interruptea
(pre-empted) to facilitate the special customer.
eg An emergeny casearriving at a doctor's clinic who is attending to a reguar
out-fpahent. (The doctor unll stop his sernce to the regular patient ana
immediately rushes to emergency case,)
Aminister or VIP coming to receive a serice at counter is given highest
priority,
An expediting in production shop.
) Nom Pre-emptive Priority Rule:This is also arule by priority to the special
customer but the priority will not emptivate the current service. The
service to the special customer starts immediately after the completion
of current service.

eg. A medcal representative will be given appointment immediately afler the

cuTrent service to an out-patient at a doctor's clinic.
Ina production shop, backlog planning, chase planning is in acordance with
non pre-emptive in many occasions.
Dte : In the above disciplines (a) and (b) are called static queue disciplines while the
rest are dynamic queue disciplines.
5. Customer's Behaviour : Acustomer coming out ofpopulation with an intention
to receive certain service may be prepared to wait till he gets service. If he waits
hoping to get the service, he is called 'patient customer' or 'positive customer"
or 'optimisticcustomer'.Where as acustomer may walk out of thequeue without
getting service due to various reasons such as long queue in front of him or not
having patience to wait etc., he is said to be an 'impatient customer' or 'negative
customer' or 'pessimistic customer'. Various types of negative behaviours are
defined here below:

(a) Balking : A customer who gets discouraged by seeing the length of the
queue before him and thinks that he may not get service, may walk out
or may not join the queue. He is said to be 'Balking'.
(b) Reneging :Acustomer who joins the queue and waits for some time but
leaves the queue due to intolerable delay or impatient to wait any longer
is said to be 'Reneging'.
(e) Jockeying : Acustomer who moves from one queue to anoher hoping to
Jockeying'.
receive a more quick service is said to be
Thus, the qustomers who come out of their original queue showing a
negative behaviour that he may not get service. However, a customer who
receives service also may have negative bahaviour as given below :
 Operations Researt tice
customer who is not satisfied by the by service
(d) Unsatisfied Customerr:Aunsatisfied. For exarmple, a customer of an hotel
the quantity is saidto be the quantity of food supplied to him it though
may not be satisfied with
is tasty.
is not satisfiedi by
the quality of
(e) Dissatisfied Customer ::Acustomer who
dissatisfied. For instance, a customer getting tasteless
service is said to be is dissatisfied.
food at an hotel though supplied in huge quantities
certain level of
The customers unsatisfied or dissatisfied usually expect
service and when they do not have the actual, equivalent to their
expectation they will not have satisfaction. However, if a customer has
the match between his expectation and the actual receipt, he will be
satisfied and ifthe actual is more than his expectation, he will be 'delighted'
6. Type of Service : The service system may be in two ways as given below :
(a) Customer stationary and server moving.
e.g. (i) Aranged meal at which seruer brings what customer desires.
(u) A machine waiting for repair.
(b) Server stationary and customer moving.
e.g. (i) Abuffet meal where customer goes to buffet table and gels what he wants.
(ü) An aeroplane waiting for a run way for landing.
7. Service Patterns : Similar to that of inter arrival tines, the
by the server to serve each customers varies. These service time taken
rate of arrival and server's behaviour. However, thesevariations depend on the
of the following probability distribution patterns. times can be fit into one
(a) Poisson distribution (denoted by M)
(b) Exponential distribution (denoted by M)
(c) Erlang distribution (denoted by E,)
(d) General fashion (denotd by G)
8. Number of Servers : We
associated costs depend on have already learnt that a queue system and its
the level of service which
service and number of service facilities. On includes both the rate of
dependent on the service time one hand the rate of service is
the level of service can be distribution patterns while on
are two ways, of assumption manipulated by arranging number ofthe other hand
service facilities. to the queue problem with servers. There
regard to number ot
(a) Single server system
(b) Multiserver represented by 1.
9.
system represented by S.
Service Configurations : The
increased by arranging the service efficiency and capacity of aqueue system can be
canmulti-service
be facilities are available.systems in an
Various effective way, particularly when
classified into following three ways. arrangements of
service facilities
 Queue Models (Waiting Lines) 429

(a) Series Configuration :

Arrival
Service facilities Departure

Fuel Cash
Customers
½ service ½ service i Served
customners
FIGURE 8.3: SERIES CONFIGURATION
Here service is divided into certain parts
(6) Parallel Configuration : served sequentially
Service facilities
Arrival
Cash Departure
Fuel
Fuel and Cash
Full service
Customers Cash Served
customers
Fuel
Fuel and Cash
Full service
FIGURE 8.4: PARALLEL CONFIGURATION
Here arrivals are divided into parts and separated in to different queues.
(c) Combined Configuration : This configuration is a combination of the
above two configuration.
Series S,1

Input Output

Series S.
Parallel (S, vS,)
HGURE S.5: cOMBINATION CONFIGURATION
10. State of Service : We have learnt that the service systemis influenced by service
time distribution and on server's behaviour. These are inter dependent on
arrival patterns andcustormer behaviour. On close observation of the changes
that occur in service svstem, we can find three sates of service viz.,

(a) Unsteady state. (b) Transient state. (c) Steady state.

Unstesdy Transient Steady
state state
state

RGURE 8.6: UEUE

STATES
430 Operations Research: Theory and Practica
When the service system is just started, the server will not be able estimate how
much time he has to take per customer. Also, he may be in a bit of confusion and
non-unifomity due to initial conditions and hence takes more time for service. This
state in whikh the server does non-uniform service is unsteay stale.
From this state he slowty moves towards attaining uniformity. This consumes
Some time and the server during this time is said to be in transient state.
After getting enough hands on experience. the server picks up and acquires
complete knowBedge on the service system and takes almost same times at uniform
rate to serve cach customer. This state is called steady statle. Further, this steady state
can exist when the ratio of arrival rate to service rate is less than unity.
Steady state condition :
lI. Departure and Output : Aserved customer goes out of the service zone and
emerges as output. This activity is often known as departure in queuing
terminology. Acustomer departed trom the service zone may have been
satisied or unsatisfied or disatisied. This has already been explained in fifth
point of this section. In the above eleven features, first five are inchuded in
customer zone while the rest in server zone.

8.5 Queuing Models : Kendall -Lee Notations

Queuing models can be defined by five most significant features of the above
explained deven points D.G. Kendall in 1953 has noiced three of them in the form
aic) and later AM Lee in 1966 added two more in the form (de) to describe a queue
model. These are known as Kendall-Lee notations in the standard format as (abc): (de)
where
Poisson (M) V

G = Arrival distnbution
Exponential (M)
Erlang (Er)
General (G)
-Poisson (M)
b= Service time distribution Exponential (M)
Erlang (Er)
Kendall - Lee General (G)
noations for Single (1)
queue model c = No.of service channels (servers).
Multi (S 2, S, ...
Limited (N)
d = Limit of customers
Unlimited (~)
FIFO
LIFO
e= Queue discipl1ne SIRO

Pre-emptive
Non pre-emptive
QueueModels

Aqueue modelis blend of the above five points, which is diagramatically snow
below.

SERVICE)

PAT ERNG
ARRIVAL,

TTERNS * Poisson (M)

*Exponnetial (M)
* Erlang (E)
* General (G)

ER
OFSER
VERS

Kendall Lee
Notations * Single ()
(ablc) :(de) * Multi (S)

WArTINGY
KINE
MOD

QUEUEpISCIPLINE

* FIFO
* LIFO

*SIRO
*Pre-emptive
* Non-pre
*Limited(N)
* Unlimited (¢)
emptive
LIMIT OF QUEUE

FIGURE &.7: QUEUE MODEL:KENDALL -LEE NOTATIONS

Ihusif queue system iscomposed of Poisson arrival distribution, exponential
service time distribution (both of these follow 'Markovian' property and hence
CDresented by M), single channel, unlimited number of customers in queue can enter
cad ToloW a discipline, First Come First Serve, the queue is described as
(M/M/l) : (o/FIFO)
service time
omlarly (M/E,/s) : (N/SIRO)means Poisson arrivals, Erlangian allowed, selected
distribution,
s number of channels. limited number of customers are:
in random order.
 8.6Parameters of Queue System
systenm has two parameters upon which entire queue system
Every queue
depends. These are mean inter arrival tirne)
= mean arrival rate (or 1/a =
mean service rate (or l/å= mean service time)
=

8.7 Arioms of PoissOn Process

three
The arrivals in a Poisson's process are characterised by the following
axioms.
Axiom 1: In a non-overlapping interval, the number of arrivals are statistically
independent. This means Poisson has independent increments.
Axiom 2: The probability of occuring more than one arrival between the time
interval, t and (t +A) is 0(4) This means that the probability of occuring two or
more arrivals during thesmall 4ime interval At is negligible. Hence we have
Po(A) + P, (aT) + 0 (A4) = 1
Axiom 3: The probability that an arrival occurs between time, 4, and time
(t+ A), is equal to A(A) + 0A1).
Hence, we have, P(A) = (A) + 0 (A)
where 2 =aconstant and independent of N()
N(C) = total number of arrivals upto time t
N(0) = 0

0(A) 0(A) =0
At’0

8.8 Operating Characteristics of Queue System

Some of the operational characteristics, which are of general interest to estimate
the performance of the existing queue system or/and to design a new queue system
are listed below:
1. Expected number of customers in queue or size (or length) of queue (L,):
This is the average number of ustomers waiting for service and is denoted by L
2. Expected number of customers in system or size (or length) of the system
(L) :It isthe average number of customers expected to be waiting in queue and
being in service. It is denoted by L,
3. Expected waiting time in queue (W): This is average time spent by a customer
inqueue before the commencement of the service. This is denoted by Wa
4. Expected Waiting time in the System (W) : It is the total time spent by a
customer in the system, whcih includes the waiting timne in queue and the service
time. It is represented with W,.
W, =Wa t mean service tÉme =W +
 433
5. The server utilisation
Itis the proportion of
factor or fraction of busy period or trafic intensity :
the
In otherwords it is he time that a server actually spends with the customers.
probable period of server being busy. It is
with a greek letter p (rho). designated
Average service completion time (1/u)
p= 2
Averageinter arrival time (1/)
General Relationship Between Operating Characteristics :
1. Between L,and L,
L, = No. of
customers in the
= No.of customers in
system
queue + number of customers in
+
serVIce.
2. Between W,and W,
W, = Waiting time in system.
= Waiting time in queue + waiting time in
1 service.
W, = W, +
3. Between L, and Wa
La No. of customers in queue
Wa Waiting time in queue
No. of arrivals into queue
= i.e.,
Total time of arrivals into queue
|, =2 W,
4. Between L, and W, :Similar to the above
L, = W,
5. Expected number of customers served per busy period.
L P (n 2s)
L, = L;.P(n2s)= - . P (n 2s)

for (M/M/i):(»/FIFO)
6. Expected waiting time during busy period,
W, =
Pb
Wa = W,.P,= P,

=
-for (M/M/1): («/EIFO).
 (M]: (oc|FFo)
dArVd

=((- e)".
Probabiah
L tled Ruly
454 QueueMo

8.9 Model -1 :(M/M/1) : (oo/FIFO)

ractice () Calculateethe %oftimethe e 435
Teller is busy
The following expressions are used characterise
this model. () Also caleulate the
average time a andiscustomer expectedto wait.
1. Probability Density Function P, = Solution : JNTU (Mech.) 98/S)
Tn the given
:p -p) where p <l;n=0, 1,2.. problem, arrival process is in Poisson
distribution is in exponential
2. Probability of system being empty (ie, no customer waiting) in queue process is oo, and the process (M), No. of servers is 1,manner
No. of (M), servige time
: Po=l-=l-p Hence the queue modelservice
is
discipline is FCFS. customers allawed
3. Probablity of server being busy (P) = 1- P = u.
notation). (M/M/1) :(n | FCPS) (according to Kendall- Lee
4. Expected size of system
Arrival rate (2)= 30 customershr.
Serivce time
,Iin =min
5. Expected length of queue 60 hrs, =hours.
40
Service rate (u) =40 customers/hr.
6. Expected waiting time in queue A<, hence the model-I implies at steady state
i) Trafic mtensity (p) =
7. Expected waiting time in system
30
(or) factor of busy period = 407=0.75
W= W,--=
8. Expected waiting time in the queue for busy system The percentage of the time the Teller is buISy is
W, 0.75 x 100 = 75%
1
W;P -= W, 30
9. Expected number of (ü) Average length ofqueue ( L)=or I-p 40-30 3
customers served per busy period. -

1
L*P,
10. Probability
1- P.
that kor
l-p
Average waiting time that acustomer is expected to wait
more customers waiting in the ==0.1hours = 6min.
system. 30 10
ie., P (n2 k)= and P (n>) =
11. Probability of
ILLUSTRATION 2
waiting e mínutes or more in queue The mean service time is
1hemean arrival rate to aservice centre is 3 per hour.
P(W2) 10 minutes. Assuming Poisson arrival rate and exponential servicing time,

ILLUSTRATION 1
=u-)-) dt
t
() Utilisation factor.
() Probability oftwo units in the system.
In a bank, cheques are (i) Expected number ofunits in the system.
cashed ata single "Teller' counter.
cOunter in a Poisson manner (IV) Expected number ofunits in the queue.
has to wait in thesystem.
at an average rate of Customers arrive, attne Expectedtime in minutes,customer
Teller takes, on an average a 30 customers per
has beenn
shown to be minute anda half to cash chegue. The hour. 1e
exponentially distributed. service u
Solution us first identify the queue
In the above problem let disuribution i.e., M
Poisson
system.

ATIval pattern is exponenially dËstributed i.e., M

Service pattern is
Number of service centres is 1.
436 QueueModels

Limit of customers arrival (No limit is imposed)

Practice 437
Also given that,
Queue discipline = FIFO(When there is no mentionnof disciple, it may Mean arrival rate ==30 per
to be FIFO) assumed hour
30
Thus the above svstem can be described as (M/M/1) : (oo/FIFO)
Kendall-Lee. according to 60x 60 Per second

Next, parameters of the system are to be identified. 120 Per second

Mean rate of arrival = À= 3 per hour. and service time = 90 seconds
10 1
Mean service time = 10 Min = hrs.
60 6 . mean service rate (p) =per second
H=6per hour. 90
A_3 1
Iraffic intensity(o)= u 6 9 -18|-18
20
Fraction of Busy period or Traffic intensity ( p) == 90
As /u< 1, steady state can exist.
Now, we can find all the operating characteristics that govern the queue system.
u 1
90
120 -2=0.75
Average numbe of customers in the systern (L. )
) Uilisation factor =p==2= r0.75 0.75
-
1-p 1-0.75 0.25 =3customers
ü) Probability of two units in the system P(n=2) =p² =|-6=025 Average number of customers in the system (L, ) = 3austomers
(im) Expected number of units in the sy stem L, = Average queue length (L, )= u(u-) P=075 x3 =2.25
1-p
9 Average waiting time of acustomer in queue ( W,)
(iv) Expected number of units in the queue L, = b 225
(4-4) 6(6 - 3) 6x32 == = 4.5 minutes or 270 seconds.
(v) Expected time in minutes custoner has to wait in the system V120
1 60 60 systern W, =or
W,=x 60 =6-3 = 20 minutes.
Average waiting time of acustomer in the seconds
=6minutes or 360
ILLUSTBATIONS 3 W, + service time
Customers arrive at a box office window, being managed by a single individual, ILLUSTRATION 4 with an
according to a Poisson input process with mean rate of 30 hour. The time customersarrive according to a poisson process
required to serve a customer has an exponential distributioDper,
withh amean of 90 Ata certain petrolpump, arrivals., The service
minutes between Find() the average
ime is eponentially
length (ii)
seconds. Find the average waiting time ofa Customer. Also determine the average erage time of5 ti spent bya
number of customers in the system and average queue length. Qstributed with meantimeof2inminutes.queueing system (ii)average customer before
1JNTU(Mech., ECE) 1999/CCC)) humberofcustomers the average waiting time ofasecondpumpwill
average petrol pump and (ivj queueis 4min,apump.
customer in The waitingtimeinopening
Solution: service.() ifhearrival
eceivingtheWhatshoula bethe rate for second
[Dr. BRAOU (MBA) 97
Arrival process is Poisson i.e.. Markovian represented by 'M. be opened.
Service distribution is Exponential which can be represetned as 'M (Markovia
Nurnber of servers = 1 Solution This is an
(M/M/1):(/FCFS) queue
model.
5min =,,hr
No. of peopie (custouers) allowed to enter queue = « and
understood as FCFS (Fust Come First Serve). queue desciple Average inter arrival
time =
or A= 12/hr
Hence, the given queue model is represented as 1
-2 min=hr
30
(M/MI): (aFCES) Average service time
hr
4 = 30/
or
 Operations Research : Theory and Models (Waiting Lines)
Queue
438

< u, the steady state solution

exists,
Practlca (ii) Expected number of TV sets inthe
439

As
144 4
system
(1) Average length of queue (L, ) (p-4 ) 30 x 18 15 P 8
12 2 G2 sets approx..
(i) Average length of systein(y-À 18 ILLUSTRATION 6
(ii) Average time spent in the petrol punip (system) In a railway marshallingg
1 yard, goods trains arrive at a
Assumingthatt the inter-arrival
W,= hr = 3.33 min time rate trains per day.
18 service time (the time taken to dump follows an
atrain)dponential
an average of 36 minutes, calculate (i)
df30
istribution and the
is also exponential with
(iv) Average waiting time of customer in queue i.e., before receiving the series (i) probability that the queue size exceeds 10.expected queue size (line length ),
12 1 yerageof 33per day, what will be the change in1fthe input
(i) aná (ii) of trains increase an
to
(WG)= (p-2) 18 45 hr = l.33 min
80
JNTU (Mech.) 97, OU 83
(or W, =W,-=3.33 - 2= 1.33 min) Solaion :
Erom the data, the given queue system has
(v) To open second pump W, '= 4 min service pattern (M,) single track (service facility),exponential arrival (M),
unlimited queue sizeexponential
and FCES
discipline.
W= 4(u-')-= 4 min = hrs. . According to Kendall-Lee, the system is (M/M/1): (oo/FCFS)
We have the parameters as
30(30 à)
2= 20/hr and inter arrival time=3 inin. 30 1
minute
60x 24 48 "ains per
ILLUSTRATION 5
Atelevision repairman finds that the time spent on his jobs h¡s q exponential and =36 trains per minute.
distribution with mean 30 minutes. If he repairssets in ihe order in which they 236 =0.75 as p<l,steady state can exist.
came in, and ifthe arrival of sets followaPoisson distribution approximatelywith Then traffic intensity (p) = 48
an ayerage ráteof 10 per 8 hour day, what is the repairman's expected idle time
each day? How manyjobsare aheadofthe average setjust brought in? We can proceed with the problem as follows :
JNTU(Mech.) 94/P)
(i) Expected queue size (line length)
Solution: 0.75
=3 trains.
Arrival fashion is Poisson distribution (M); service pattern is exponential 1-p 1-0.75
distribution (4), number of servers is 1, no limit is imposed, repair discipline is FIFO.
Therefore, according to Kendall-Lee, the queue system (M/M/1) : (o/FIFO) (ii) Probability that the queue size exceeds 10,
=0.056
and mean arrival rate (^) =
10 5
sets per hour. P (n2 10) =pl= (0.75)0
8 4 per day, then we have
Now, if input increase to 33 trains
and mean service rate (u) = x 60= 2 sets per hour 33 11 munuteand u= 36 rains /min.
30 24480"alnsper
60 x
5
Trafficintensity p =4 480 36-0.83
<l, hence steady state can exist. as p < l, steady state can exist.
(ü), we have
the values of (i) and
(i) Expected idle tine of repairman each day Hence recalculating
number of hours for which the repairer is busy in 8 hour day is 0.83 5 trains (approx.)
(i) L,= 1-p 1- 0.83
8x=8x =5 hours
0.155 (approx.).
Idle time for repairman in an 8 hour day =8-5= 3 hours =p' =(0.83)"=
(u) P (x 10)
 Queue
449
Operations Research : Theory and Practiea
448 LLUSTRATION 8
Dr. Raju's out-patient clinic can accommodate
8.10 Model -2: (M/M/1) :(oo/SIRO)
This model is identical with
model. 1: The patients who arrive when hallis full, six people oxly in
away. The patients the waiting hal.
Single Channel - Unrestricted Oueue
:
operating characteristics of fashion at an average rate of 3 per hourbalk
and arrive in Poisson
discipline. Since the Doctor's chamber which is spend an average of 15 minutes in
all respects except for the queue queue discipline (as the derivation for probabilir i exponentially
queue model are independent of
all the results remain unchanged. Therefore the Theprobability that a patient can getdistributed.
his arrival
Find
directly intothe doctor's chamber upon
independent of queue discipline),
may be solved using sane formulae as
those of model
problems under this model G) Expected number of patients waiting for treatment.
(ii) Efective arrival rate.
8.11 Model - 3: (M/M/1) : (N/FCFS) (i) The time apatient can expect tospend in the clinie
Single Channel - Restricted Oueue Model : This model differs with model-l in
respect of limit of queue. Here the capacity of the system is limited to N. The relevant Solation : From the above problem,
fornulae are given below : Arrival :Poisson (M); service : Exponential (M);
for 0 S n sN, p l ie, À#4 Number of servers:One; (1) discipline : FCFS;
1.
But the queue is limited
-for p = l or , =u The model is (M/M1): (NFCFS)
Now capacity of the system
l-p
and P, = p landp<1 N = hall capacity + doctor's chamber capacity
-6+l=7
Where Arrival rate (^) =3 per hour.
N = limit of queue system.
n = number of austorners zrrived.
Service rate (4) =,E per min =4 per hours.

2. Expeed size of systern or expected number of custorners in the system ==0.75 i.e., < 1(stead state exists)
u 4
doctor's chamber if the hall is empty on his arrival.
-for p 1 (1) Patient willdirectly enter
L,1-p for
1-p is P
The probability at this situation
p=l o740.2778
l-p
3. Expected length (size) of quee or expected number of customers waiting in Po
- is (L
waiting for treatment
(u) Expected number ofpatients
4. Expeed wasting tine of austoTIer in the systen
8 -2.1
L
L,=
5. Expeced waiting tisne in
quee
L 4-4--21-3-1.36
W,0-os0 (ef
=u(1-P)
3 per hour.
(iüi) Efective. arrivalIrate =4(| -0.2778)= 2.89 =
 2. Expected length of queue
expectto spend in the systenn
(tv) 1he time the patientcan 1.36+|0.2778 0.72 hours
I, t(l )
9 89 P
Att
-43.2 minutes.
(
4 Expected number of custoners in the svlem

Practice Problems
serviee 4. Expeeted waiting tune of the customer in queue
1. Considera single server queuing system with Poisson input, exponenialexpeCted
arrival rate is calling units per hour, the
times Suppose the mean maximum pernmissible number calline
service timeis 0.25 hourand thesteady-state probabulity distribution of tbe
in the system is wo,. Derive the munber
number of calling unitsin the system and hen calualte the expected 5. Expected waiting tine in the system
in the system.
0,81.
Answer : The expeted number ofcallng units in the system
9. tf for a period of 2 hours in the day (8 to 10 a.m.) rains arrive at the yard at W, - W,
every 20 minutes but the service time continues to rcmain 36 minutes, then 6. The probability that a customer has to wait (busy period)
calculate for this period.
(a) The probability that the yard is empty, and
P(n2s) .Po
(b) The average number of trains in the system, on the assumption that the
line capcaity of the yard is limited to a four trains only. ILLUSTRATION 9
counters. Ifthe service timne
Asuper market has two girls ringing up sales at the
Answer :
if people arrive ina
for each customer is exponential with mean 4 minutes, and
À= /20; =/36: p= 36/20 =1.8 (> 1) and n=4
Poissonfashion at the rate of 10hour.
(a) P 0.04
(a) What is theprobability ofhaving to waitfor the service.
(b) Whatis the expectedpercentage ofidle time foreach ginl? the system ?
4

(b) L, =) nP, 2.9 =3.

() Findthe average length and theaverage numberofunits in Mech/Prod/Chem/Mechatronics 2001/Ss)
[JNTUCSE 97/S JNTU-

8.12 Model -4 (M/M/S):(oo/FIFO) Solution minute],

=;and servicerate (4)=per
4
Multi Channel - Unrestricted Queue Model: This model can be supposed as the Arival rate ()
No. of servers(s)= 2, 6
extension to model- 1. Here instead of singlechannel, multi channel (multiple servers)
in parallel say 's' lines are assumed. With other notations asusual we use the followg 1
formulae for this model. 6 1
1.
2 4
for 1sn<S
Therefore, T!
Po for
s. 2-1 4 27
Where p=
Po=
n=0 2-46
Po = (sp 1 sp S
2 1 1
n-0
Or
P, =Po-3
 Operations Research : Theory and Practie Combined System :
S2 Now Consider
Arrivals : Poisson (M)
(a) P(n2 2)=P(W> 0
Service : Exponential (M)
-
n=2 No. of servers : two (2)
Limit of queue : Unlimited ()
= (or) 0.1l67 2 Discipline :FCFS
idle time for each girl is 1-ie., 1- (or) (= 0.67).
(b) The expected Model : M/M/2) : (o/ FCES)
Hence, the expected percentage of idle time for each girl is 67%.
1 Mean arrival rate :=14+ 16= 30hours
time =
(c) Expected length of customers waiting Mean service rate : =20hours
1
or minutes. 30
0.50-0.17
Number of servers (5) = 2 andp =su 2x 20 4
LLUSTRATION 10
accounts. The first teller handles (: p<l)
Abank has teo tellers working on savings deposits only. lthas been found that
withdrawals oniy. The second teller handles 30 3
the service time distribution for deposits and withdrawals both are erponential Also
with mean service time 3 minutes per customer. Depositors are found to arivein 20
a Poisson fashion through out the day with mean arrival rate 16 per hour.
Withdrawals also arrive in a Poisson fashion with mean arrival rate 14 perhour.
What would be the effect on the average waiting time for depositors and with
drawers, ifeach teller couldhandle both withdrawals and deposits? What would
bethe efject of this could onty be accomplished by increasing the service time to Now Po =

3.5 minutes?
2
Solution : 1(3) 1(3) 40
Initially, we can assume two independent queueing systems as folows.
Withdrawers System Depositors System
Arrivals : Poisson (M) Poisson (M)
Service : Eponential (M) Exponential (M) 9 40;
Number of servers :One (1) One (1)
Limit of queue :Unlimited (o) Unlimited (oo)
+ 10
Discipline:FCFS FCFS
Model : (MM1): ( «| FCFS ) (M/M/1):(o/ FCFS)
Mean arrTival rate : () = 14hours hg =16hours
queue
Mean service rate : =per min 20/hours Hyper min 20/hour arrivalsinthe
Average waiting time of
min)
(P1==0.7< 1, service time =3 min ) (Mean service time

P2
hg 16 =0.8 <1
20
-
(s-1)!

3
20
(40-30)2'7
Average waiting time Average waiting time
14 16 (2- 1)! 9 hours.
Hj 4j -.) 20 (20 14)
7
W, Ha ug- hg) 20 (20-
16) 9
4100
20 I
7
==7min - hrs =12 min =3.86 min
 Operations Research : Theory and Practice
2. Two reparmen are 455

Combined waiting time with increased service time when

breaks down attendin
to a five
according g machines in a workshop. Each
repair ime per machine is poisson
X=30/hour and =3.5 min
() Find the probability exponentialstrwithibutimean
di on with a mean 3per bour.machiThene
15 min.
repairman is idle. that the two
60
3.5
120 per hour repairmen are idle and that one
(iü) What is the expected i
30 7,
F. 30 number of idle machines not being
S 2(120/7) 8 ie,<l and 120/7 4 Given an average arrival rate of 20 per served?
on service, at asingle channel with
hour; is it better for a
mean service that of 22 customer to get
2 x (120/7) of the two channels in paralle with mean customers or at one
2(120/7)-30 ofthe two channels? Assume that both theservice rate of ll customers for each
queues are of M/MIS vDe.
(JNTU (Mech.) 2001]
+
49 2(120/7) Answer : Better to get service at two channels.
2x1 16 (30/7)
-1 5. A telephone exchange has two long distance operators. The telephone
49
=[15! =
company finds that during the peak load, long distance call arrive in aPoisson
4 4 15
Average waiting time of arrivals in the queue
fashion at an average rate of 15 per hour. The length of service on these calls
1
is approximately exponentially distributed with mean length 5min.
(6- 1)! su'-)
Po (a) What is the probability that a subscriber will have to wait for his long
2 distance call during the peak hours of the day?
1 120/7 1
what is the expected waiting
(2-1)! [2 (120/7) -3015 (b) If subscribers will wait and are served in turn,
2 time?
49 120/7 1 49 120
16 (30/72*15 16 7 30X Answer: (a) 0.48 (b) 32 min.
having a single clerk, in is
343 currently has two tool cribs, eachonly the tools for the heavy
hours O. A company crib handles for
30 x 60
llanutacturing area. One tool handles all other tools. Itis observedthat
343 the second one with a mean of 20 per
min = l1.433 min. machinery. While follow a Poisson
distribution
with a mean of
each tool crib the arrivals distributionis negativeexponential
30
combinedinsuchaway
hour and the service time feelsthat,iftoolcribsare arises, wouldbe more
Practice Problems 2 minutes. The tool manager tool as demand Itis believed
handleany kindof someextent.
1. A post office has 3 windows providing the same services. It recieves on a that either clerk can gproblem couldbereducedtoper hour, whiletheservive
be40
average 30 customers/hours. Arrivals are Poisson distributed and service tme efficient and the waiting i toolcribs will
exponentially. The post office serves on average 12 customers/hours. arrival rate attwo
that the mean totalexpected
() What is the probability that acustomers will be served immediatelyr time will remain unchanged. respecttothe including
proposal with waitingtime than
(ü) What is the probability that a customer will have to wait? queue andthe the
expected for more
Compare in status of atthetoolcrib(s), hastowait
(ü) What is the average no. of customers in the probabilitythathe
system. number of machines machineandd
(Iv) What is the average total tine that customer must spend in
the Po service time foreach
office?
five minutes. arrivals, 6 min.
Answer : () 0.29 (ü)0.70 (iv) 12 min. cribs: 2
(ii)6 customers (approx)
Answer: For separate
tool arrivals,3.6 min.
cribs:2.4
tool
For combined