8 Queue Models

Chapter
(Waiting Lines)
CHAPTER OUTLINE

8.0 Introduction
8.1 Terminology
8.2 When Does A queue Result?
8.3 Costs Associated with Waiting Lines (Queuing)
8.4 Queue System: Salient Features
8.5 Queuing Models: Kendall - Lee Notations
8.6 Parameters of Queue System
8.7 Axioms of Poisson Process
8.8 OperatingCharacteristics of Queue System
8.9 Model - 1: (M/M/1):(o/FIFO)
Practice Problems
8.10 Model - 2:(M/M/1): (oo/SIRO)
8.11 MODEL - 3: (M/M/1): (N/FCFS)
Practice Problems
8.12 Model -4: (M/M/S): (o/FIFO)
Practice Problems
Review Questions
Objective Type Questions
Fillin the Blanks
Answers
 Queue System
(SIRO)
Customer's Zone Server's Zone
(FlFO)Order
(LIRO)
Poisson
Psten
M) Sttioneyy
Customer Poisson
Psten(M)
Exponentinl(M) Out OutRadon
Bnpive Statiorary
Moving
Customer
Server Erponential)
(N) Unlimited
(O
General(G) Limited
Erang(Er) FintFng Pre-Emptire
Moving
in Eriang(Er)
Genera(G) Singie
(1) -Mutti
Singie in Select Non-Pro
Fista Server
(C)
Paralel
Unsteacdy
Traosiect
Muli Last Series Steady

Input Queue
Service
Population
Configur Configur Service
ation State
ation
Onenuaygoqeja
Buuogncggu Output
Balk Ronege Jockey
6oniFIME SatisfiedDissatisfed
sLSC SLSC ObeL Unsaiafcd

Ueodia Keodal Loo Notatlons (a b) : (die)

8
ServedCestomners
 423
80 Introduction
One of the
common
Rather, there will notsituations
1be any in our daily life ocaurs with waiting or
system. R
troubles due to waiting such as at bus person on this earth who did not queue
bank counters, stops, ticket booths, doctor's clinic,experience the
traffic lights and so on. Queues are petrol bunks,
shops where the machines wait very common in industries also, in
for
nay even cost hugely in some
repair/maintenance,
material waiting for operations etc. These queues ruin theoperators wait at tool cribs,
time of
occasions. The minimisation of waitingcustormers
time
and it
and thus
cOsts are the primary concerns of these queue
models in Operation Research.
8.1 Terminology aninellsdi3o
The following terms are commonly used in queue models.
1, Customers : The persons or objects that require certain service are
called
TCustomers.
2. Server The person or an object or a machine that provides certain defined
n Service 1s known as server.notsto o oduuT
3. Service :. The activity between server and customer is caled service that
Consumes some time.sDOes to 9dmssi >A31 ( )

4. Queue or Waiting Iine: Asystematic or adisciplined arrangement of agroup

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.
per unit
6. Mean Rate of Arrival : The average number of customers arriving
as LAMBDA).
timeis called mean rate of arrival and is denoted by '2 (read
eogts = Total no. of ustomersarriving/total time taken.ns
7. Mean Inter-arival Time : It is the average
time gap between two consecutive
of mean arrival rate i.e.,
Po o arrivals of customers. It is the inverse
Total arrival time
o atotto Number of arrivals ivlss J2009sonb
number of customers served per unit
time
Rate of Service : It is the average
icandis denoted by u ( read as MUE). n ub bexont o0 S
served autisw
Total number of customers
Total service time
the average time taken by the server to serve a customers
1Mean Service Time: It is Total service time
served
and is equal to inverse of servicerate i.e, 1/=Number of customers
 8.2 When Does A Queue Result?
service tate is u.
sp 0 pitle BupOse a queue system wlose artival rate is 4 and
udloilyea ArrivalArival tate
rate Service r a t e o o l
(0)
las
Served customers
oService
facility
iodeatlsbot
FIGIRE .Ls GrUE GR WAITIHGLME
three relations nust occur (by law f
Now, in thís system arny of the following
tríchotomy).
than rate of service, ie., nurber
the greater
rate of arrivalísnurnber served peT unit tine
per
0) When A> 4, i.c, than i n n of T v
ustoners e d
las s of custorners arrivingísgreater
systern reulre
results in piling up of arrivedaustómerswaiting
keptconstant), thesystern
the
(timene results.
for service and thus a queue equal to customers served, there
arrived is
of custorneTs system has cent percent efficiency.
(i) 1fA , i.c.,numberwaitúng, and thís queue
will not be any of customers
of custorners arriving ís less thannumber
(iii) If <u, ie., umber server willbe free, waiting for customers
Thís neans that
served per unít túme.queue
the results inserver's point of view.
o to arrive. Thus between the rale of amival and
so an imperfect matching
Ths qune Tesults where there is
the rale ofservice.

with Waiting Lines (Queuing)

8.3 CostsAssociated
of operational situations where there
18M2iesOueuíng theory, can be applied to variety of service. The rate
rate of arrivals and the rate
is an imperfect matching
between the studies and the imperfect
depend on the probability
arival and service however,
ofofmatchíng arr0val and
of server to predict accurately the operate
uRially occurssdue inability
to
customer due on server to the
These may cost on to waiting or
5.1 978evice tíme.
Thusthe level of service (eíther service rate or the number of service
acility,
service
facilities) ís determínedIby cost analysis of the
following two conflicting costs.
ia 1. Cost of ofleríng servíce.
eni siny keeping customer
2, Cost incurred due to delay in offering service or due to
waitíng.
Now, we can analyse these as follows.
s 1 . Cost of Service : If existing service level in terms of rate of service or number
fservice facilities (channels) is reduced, the cost of operation comes down.
In other words, increase in level ofservice increases the cost ofoperating servce
facilitíes.
 2.
Waiting Costs : In contrast to the
will hike the
customers above, the reduction in existing service level
increase waiting
in service level will time or may result in long
agop o Ievel willincrease the queues. inThus the
decrease the waiting costs and decrease
waiting costs. service
Thus the optimumn level of service can be
obtained
between the above two costs or at the level where by achieving a balance
This is interpreted through a the total cost shows
minimum.
graph shown below.
ail.psng
ofintni obatinu Total cost
Service cost
(to service)

891 88rsi Sn9Waiting cost

Minimum total cost
(to customer)
Optimum level
of service
SSI6boce008
ousup lsnot s3svs1cbs sd
Seist osi gtu o.ssinil Level of serviceoilog
FIGURE 8.2: CoST ANALYSIS OF OUEUE
SYSTEMatiapeib SU9UO 9T: ssilkioai s9s

8.4 Queue System : Salient Features malu

loigolo Aqueue system may be studied bydiving into two subsystems viz arrival system
and service system. The arr0val system is based on customer's behaviours and their
sgri) process of arrival while the service system is dependent on the service mechanis1m and
server's behaviour.
These are diagrammaticaly shown in the block diagram given at the beginning of this
chapier.
o0Iba
detail.
The essentialfeatures shown in the block diagran are discussed here below in
o1. Input Source or Population : The input source or population is a set that
contains the probable potential customers to come out for service. Thus the
input source need not be homogenous and may consist of sub populations. The
Tao input source may contain the subset of patients for the queue system at a
doctor's clinic, a sub set of vehicleowners for queue system of petrol bunk, a
subset of passengers who may enter awaiting line for a bus and so on. Another
point to be noted here is that all the population need not be customers, but as
they come out of population with an intention to get the defined service are
considered as the customers to that service system.
2. Arrival Pattern:The customersare expected to arrive at their own convenience
and conditions. However, the time gaps between any two c¡nsecutive arrivals,
ifnoticed, will be fit in certain fashion or pattern. These patterns ofarrival time
often follow one of the following probability distribution.
() Poisson distribution (represented by M)
() Exponential distribution (represented by M)
 427
(ii) Erlang distribution (represented by E, )o) gaiiewS (e)
(v) General pattern (represented by G) sn r
50
od c qg
Pre-emptive Priority
certain customer(s) Rule : Under this rule,
is allowed to iTespective highest priority is given to
3. Limit of Queue: (Restricted/unrestricted queues) The limit of enter into service of their arrival and costs. The
times restricted by server's behaviour. Ifthe limit isimposed queue is some
by the system,
i.e., the even
if another (lower iuumediately after entering intocustomer
priority) queue
arrivals will be limitedfor a given period. For exxample, ifa doctor has server,
the
(pre-empted)
lower priority customer who is customer is already in service.
to facilitate the special in service will be interrupted
that he would treat 50 out-patients only per queue
will be in secondary queue whilethe
to be in
finite and 51st out-patient will have first
day, the queue will be
to enter the
50 will be in primary queue.
licondi
mitedtion
queue i.e., heor e-g. An emergency case customer.
arriving at a doctor's cinic who is attending to a
ut -patient. (The doctor
will stop regular
eis said to be unlimited or infinite. However Solsvipa sss immediately rushes to his service to the regular patient and
if thereis no restriction, the queue i The limited emergency case,)
queue is denoted by N while unlimited queue by oo, Aminister or VIlP
coming to receive a service at counter is given
Limited queue may be adopted by space restrictions or other conditions also,
cinema hall. has
priority,r
An expediting in production shop.
highest
For. instance, a service facility such as a such restricted () Non Pre-emptive Priority Rule: This is also arule by
accommodation which can not accommodate more than its space. Invacated customer but the priority will not emptivate the priority to the special
unless Some space is
the server will not allow further arrivals
accommodate the new entry. In somne other cases, the limitedi queue system may service to the special customer starts immediately current service. The
after the completion
be adopted to prevent a formal queue and to avoid nuisances. These queues of current service.
reservabion
are often referred to as 'take-a-number' or 'take-appointment' orreservation' e.g. A medical reresentative will be given appointment immedialeky afer the
policy. However, the queue may be unlimited for getting into this uTrent servnce to an out patient at a doctor's clinic.
is the.order in which the customer ie
4. Queue Discipline : The queue disciplineThere
s tndis Inaproduction shop, backlog planing, chase planning is in accordance with
selected from the queue for service. are numerous ways in which non pre-emptive in many occasions.
listed below.
customers in queue can be served of which some are et sonit Note : In theabove disciplines (a) and (b) are called static queue disciplines while the
(a) First In First Out (FIFO) or First Come First Serve (FCES) : It is the ODnqpD z rest are dynamic queue disciplines.o o E s
Ssra9 discipline in which the customers are served in the chronological order DLUol e ns leveo sisT
isgs s
of their arrivals.e atere s3vebs 5. Customer's Behaviour:Acustomer coming out ofpopulation with an intenion
If he waits
e.g. Tickets at a cinema hall; sales at a grocery sho, trains on a (single line) to receive certain service may be prepared tO wait till he gets service.
customer'
platorm etc. hoping to get the service, he is called'patient customer' or positive queue without
customer. Where as a customer may walk out of the
(b) Last In First Out (LIFO) or Last Come First Serve (LCFS) : If the service or 'optimistic
reasons such as long queue in front of him or not
is made in opposite order of arrivals of customers, i.e., who ever comes getting service due to various
austomer' or 'negative
e last is served first and first obviously goes to last/t is aled LIFO or LCFS having patience to wait etc., he is said to be an impatient
types of negative behaviours are
system. customer' or pessimistic customer'. Various
seg. Stack of plates; loading and unloading a truck or go-down; office filing bos ye Sdefined here below:
a shirt and
papers in chronological orders; wearing socks and shoes; dressing of the
: A customer who getsthatdiscouraged by seeing the length
coat over it, packingsskemas elc. bos 5ooe 3bgnt o sisy Si (a) Balking him and thinks he may not get service, may walk out
(c) Service In Random Order (SIRO): By this rule, the customer for service io sdt oo slfw gueue before to be Balking'.
is picked up at random, irrespective of their arrivals. o9duor may not join the queue. He is said
queue and waits for some time but
e.g Lottery systemfrom which one is picked up, the dresses waiting in a wara Tou Reneging: A customer who joins the or impatient to wait any longer
from which one is to be chosen, food stufs in a bufjet, sales counter (b) intolerable delay
commodities or vegetables et. leaves the queue due to
S a t is said to be 'Reneging'. hoping to
(a) Priority Service one queue toanother
customer(s) due :to Under this rule, theor server
some importance gives priority
prestigeous to certait
or high cost grouP Jockeying: A customerwho moves frombe Jockeying'.
of the customers. (c) service is said to showing a
receive a more quick come out of their
original queue
customer who
e-g. Atelephone urgent call given to a customer is charged at higher price; a customers who However, a
separate counlerreservation
for chequescounter ggiven Thus, the notgetservice. as given below :
thathemaynegative
at (APSRTC) at aelectricity bill payment
who buys counter; priority
a CAT( Concessinional. Annual negativebehaviour have
may
bahaviour
Ticket) card etc. receives service also
 dOnS Que

(d) Unsatisfied Customer : Acustomer who is not satisfied by the

Pracibyce Series Configaration : 429

the quantity is said to be unsatisfied. For example, a customer service

ofan ho o oi Arrival
may not be satisfied with the quantity of food supplied to him though i Service facilities
is tasty. Departure
(e) Dissatisfied Customer : A ustomer who is not satisfied by the quality of
service is said to be dissatisfied. For instance, acustormer getting tasteless Fuel Cash
dissatisied.
food at an hotel though supplied in huge quantities is Customners 4service service Served
The customers unsatisfied or dissatisfied usualy expect certain level r
actual, equivalent to the: CONFIGURATON
customers
service and when they do not have the However, if a customer ha FIGURE 8.3: SERIES
expectation they will not have satisfaction. Here service is divided into certain parts served sequenialy
match between his expectation and the actual receipt, he will
the expectation, he will be 'deliphl (6) Parallel Configuration :
satisfied and ifthe actual is more than his 3an Service faciities htT
may be in two ways as given below :
6. Type of Service : The service system esi ain Departure
(a) Customer stationary and server moving.
brings what customer desires. Fuel
eg. i) Arranged meal at which server Fuel and Cash
(ü) A machie waitingfor repair. Full service
reservaton
(b) Server stationary and customer
moving. Customers to Served
want.ne
goes to bufet table and gets what he customers
eg. () A buffet meal where customer landing.
which Fuel
(ü) An aeroplane waiting for aTun way for Fuel and Cash
Full service
time taken
that of inter arTival times, the service
7. Service Patterns : Similar to These variations depend on the
customers varies.
by the server to serve eachbehaviour. can be fit into one FIGURE &.4: PARALLA CONFIGURATION
However, these times different queues.
rate of arrival and server's Here arrivals are divided into parts and separated in to combination of the
of the following probability distribution patterns.
M)5voTo (c) Combined Configuration : This
configuration is a
(a) Poisson distribution (denoted by igod
above two configuration.
(b) Exponential distribution (denoted by M) SeriesS.
() Erlang distribution (denoted by E, )
(d) General fashion (denoted by G) sbsq ga Output
) laitog Input
that a queue system aad is
8. Number of Servers : We have already learntwhich incudes both the rate of
associated costs depend on the level of service
hand the rate of service is Series S , !
service and number of service facilities. On one
dependent on the service time distribution patterns while on the other nand Parallel (S, v S,)
the leveB of servicecan be manipulated by arranging number of servers.
to number of
are two ways, of assumption to the queue problem with regard HGURE 8.5: COMBNATION CONFIGURATION systemisinfluenced by service
service facilities.
have learnt that theserviceThese are inter dependent on
(a) Single server system represented by 1. 40 State of Service: We
and on server's behaviour. observationofthechanges
ne distribution behaviour. Onclose service viz.,
(b) Multiserver system represented by S. patterns and customer can ind three states of (c)Steadystate.
aTival system, we
that Occur in service (b)Transientstate.
9. Service Configurations : The efhciency and capacity of a queue system can (a) Unsteady state.
inreased by arrangingtheservicesystemsin an effective way. particularly
service acilities Transicnt
Steady
state
multi-service facilities are available. Various arrangements of Unstcady state

can be classified into following three ways. state

HGURE 8.6: QUEUE STATES

 Queue
130 Whentheservicesystemisjuststarted, the server will not be able estimate hD i blend of the
Aqueue modellis
much time he has to take per customer. Also, he may be in a bit of confusion and above five points, which issdiagramaticallyshown
below.
hencetakes: more time for
conditions andservice is unsteady state.
non-uniformitydue to initial non-uniform
the server does moves towards attaining uniformity.
service. This
stateinwhich
From this state
he slowly be
time is said to in transient state.
This consumes
and the server during this
Some time experience, the server picks up and
hands on
complete
After getting enough
knowledge onthe servicesysterm and takes almost same times at unifor
steadystate.
called
customer. This stateisservice rate is less Further, this acqires
rateto serve each
the ratio of arrival rate to than unity. steady statr n 231 General (G)
can exist when
:
Steady statecondition
goes out of the
Departureasand
11. emerges
Output
output. This
served customer
: Aactivity serviceinzone
is often known as departure and
queing
from the service zone may have ailidedcSa
departed
terminology. Acustomerdissatisfied. This has already been explained inbeen fth
satisfied or unsatisfiedInorthe above eleven features, irst five are inchuded in
point of this section. SERL
OF
in server zone. iidedo
customer zone while the rest
Kendall Lce *Single()
NotationS
8.5 Queuing Models : Kendall - Lee
Notations "Multi (S)
(ablc): (de))9
Wio Jtobalhni
five most significant features of the abowe
Queuing models can be defined by1953 has noiced three of them in the forn
explained eleven points. D.G. Kendall in more in the form (de) to describe a gueue
(abc) and later A.M Lee in 1966 added two standard format as (a/bc): (dA
model. These are known as Kendall-Lee notations in the
where
basase oU bab st Poisson(M) itaso88
oISCIPLINE
QUEUE
Exponential (M)
a= Arrival distribution * FIFO
Erlang (Er) oosit
*LIFO
General (G)
-Poisson (M) * SIRO *Limited (N)
Exponential (M) (*Pre-cmptive *Unimited (æ)
b= Service time distribution oth theraeo rd baro
" Non-prc
Erlang (Er) emptive
IMITOF OUEU
Kendall -Lee General (G)thet ba
notations for Single (1)srversThet
qucue model c = No. of service channels (servers) ber
Multi (S 2,3,...) FIGURE 8.7 : aUEUE MODEL: KENDALL -LEE NOTATIONS distribution,exponential
composed of Poisson arrival hence
Limited (N) RUON! vd39 Thus if queue system is of these follow Markovian' property and
d= Limit of customers
dService time
distribution (both
unlimited number of customers in queue can enter
Unlimited (--) Tepresented byM), single channel, Serve, the queue is
described as
isqe and follow a discipline, First Come First
bas suup ntotail g(M/M/1):(/FIFO)b0stqse0 service time
9ethsoi=Queue discipline -LIFO
(N/SIRO) means Poisson arrivals, Erlangian
allowed, selected
SIRO Omlarly (M/E/S): limited number
ofcustonmersare
Pre-emptive number of channels,
distribution,:s
in random order.
Non pre-cmptive
 Quot
ractlca It is
server
utilisation
5. Thethe proportion of factor or
the fraction of busy
In other words it is the time that aserver period
433

8.6 Parameters of Queue System probable period of actually

or traffic
sspends with the intensity :
Every queue system has two parameters upon which entire queue systen
with a greek letter p(rho). server being busy. It is customers.
designated
depends. These are
Average service completion time (1/)
A=mean arrival rate (or l/h = mean inter arrival time) General
Averageinter arrival time (1/a)
Relationship Between
u= mean service rate (or l/y=
mean service tùme) Operating Characteristics :
1. Between L, and L,
L. = No. of
8.7 Axioms ofPoisson Process = No.-of
customers in the system
The arrivals in a Poisson's process are characterised by the following three customers in queue + number of customers in service.
L, =L,t
axioms.
Ariom l: In anon-overlapping interval, the number of arrivals are statistical 2. Between W,and W,
increments.
independent. This means Poisson has independent between the time W. = Waiting time in system.
one arrival
Axiom 2: The probability of occuring more thanprobability
the of occuringtwo or =Waiting time in queue + waiting time in service.
interval,. tand (t + A) is 0(A) This means thatnegligible. Hence we have 1
arTivals during the sall time interval At is
more W, = Wt
Po (A) +PË (AT) +0 (A) =1
between time, t, and time 3. Between L, and W,
Axiom : The probability that an arrival occurs
(t+ A), is equal toÀ (A) + A). No. of customers in queue
Hence, we have, P(A) = à(49) +0 (A) Wa Waiting time in queue
where À =a constantand independent of N() No. of arrivals into queue
N) = otal number of arrivals upto time t Total time of arrivals into queue
=0
N(O)
|, = W
above
4. Between L, and W, :Similar to the
L, = W,
8.8 Operating Characteristics of Queue System 5. Expected number of customers served
per busy period.
interest to estimate
Sorme of the operational characteristics, which are of general
queue system
L
the performance of the existing queue system orand to design a new L; = P(n 2s)
are listed below:
P.P(n2s)
(or length) of queue ) :
1. Expected number of customers in queue or size service L, = Ly.P(n2s) = -

This isthe zverage umber of austomers waiting for and is denoted by Lã

L-Pfor(M/M/1):(oo/FIFO)
2. Expected number of customers in system or size (or length) of the system
L) :it is he average number of customers expected to be waiting in queue ana busy period,
being in service. It is denoted by L, 6. Expected waiting time during
3. Expected waiting time in queue (W,J: This is average time spent by a custonie Wa
in queue before the commencement of the service. This is denoted by Wa P -
4. Expected Waiting ime in the System (W :It is the total time spent by W, =W,.Ps
Customer in the system, whcih inciudes the waiting tine in queue and the serv
time. It is represented with W, for(M/M/1):(»/FIFO0).
W, = W,+ mean service time = W,+.
 Queu
:(M/M/1):(o/FIFO) Calculate the %oftimethe?Telleris busy and
8.9 Model-1 expressions are used characterise this model. () Alsocalculatethe average time acustomer is expected to wait.
The following JNTU (Mech.) 98/S|
Function P, Solution :
1. Probability Density Tn the given problem, arTival process is in
=p(1 -p) where p<l;n=0, 1, 9 Poisson
distribution is in exponential process (M), No. of servers is 1,manner (M), service time
No. of custoners allowed
2. Probability of system being empty (.e, no cçstomer waiting) in queue processiso, and the service discipline is FCFS.
Po=l-=|-0 Hence the queue model is (MM/1) : (-/ FCFS) (according to Kendall - Lee
notation).
l-P =/u.
8. Probability of serverbeing busy (Py) = Arrival rate (2)= 30 customershr.
4. Expected size of system oG 3
Serivce time
-min min

=*B brs, =hours.

40
5. Expected length of queue Service rate () 40 customers/hr.
4-2) 1-p <, hence the model-I implies at steady state
6. Expected waiting time in queue
()Traffic intensity (p) =
W, (u-2) 30 3
7. Expected waiting time in system (or) factor of busy period = 400.75
The percentage of the time the Teller is busy is
0.75 x 100 = 759%
8. Expected waiting time in the queue for busy system 30
1 =3
-=W,
(ü) Average length of queue ( L) = l-p 40- 30
expected to wait
9. Expected number of customers served per busy period. Average waiting time that a customer is
3 hours = 6 min.
Py l-P -À1-p 30 10
10. Probability that k or more custormers waiting in the system.
k+1 ILLUSTRATION 2 mean service time is
P(azk)=( and P(n> i)=|| a service centre is 3 per hour. Theservicing time,
1he mean arrival rate to
Le.,
rateandexponential
11. Probability of waiting mínutes or more in quèue 10minutes.AssumingPoisson arrival
(1) Utilisationfuctor. system.
P(W2) (i) Probability oftwo units in the
in the system.
(ii) Expected number ofunits in the queue.
) Expectednumber ofunits system.
LUSTRATION 1 customerhas to waitin the
( time in minutes,
Expected
In a bank, cheques are cashed ata atthe
single 'Teller' counter. Customers arrive, The
counter in a Poisson manner at an
Teller takes, on an average a minuteaverage
has been shown to be
and a half
rateto of
cash
30 cheque. perhour
customersTheservicetime Solution : above problem let us first identify the queue
system.
In the distribution i.e., M
exponentially distributed. Arrival pattern is Poisson
exponentially distributed i.e., M
Service pattern is centres is 1.
Number of service
 Also given that,
limit is imposed) = o0,
Limit of customers arrival (No Mean arrival rate =,=30 per hour
Queue discipline =FIFO(When
there is no mention of disciple, itmay
to be FIFO)
Thus the above system can be described as (M/M/1) : (o/FIFOS) according o
assumed 30
60x 60 PeT Second
120 Per seCond
Kendall-Lee. identified.
system are to be
Next, parameters of the and service time = 90 seconds
hour.
Mean rate of arrival = = 3per 10
Mean service time = 10Min = 60 hrs.
6 . mean service rate (P) = 90 per second
H=6 per hour. -|&|-|8 90
3 1 Fraction of Busy period or Traffic intensity (p) = -= 0.75
Traffic intensity (p) ==~, 1 120 4

exist.
As A/u <l, steady state can characteristics that govern the queue svste Average numbe of customersin the systern (L.)
Now, we can find all the operating 0.75 0.75
2 3 1-p 1-0.75 0.25 =3 customers
-
or

() Uülisation factor =p =7=

-Average number of customers in the system ( L,)=3ustomers
system P (n =2) =p²=
(iü) Probability oftwo units in the Average queue length (L, J= p(-) 1-p-=0.75 x 3 = 2.25
Average waiting time of acustomer in queue ( W,)
=

the systerm L, = - 4 6-8

(ü) Expected number of unitsin 9 9 2.25
=4.5 minutes or 270 seconds.
queue L, = (u-2) 6(6-3) 6x32 V120
(iv) Expected number of units in the system
minutes customer has to wait in the system W, or
(v) Expected time in 60 60 Average waiting time of acustomer in the seconds
1 x 60 = = 20minutes. time =6minutes or 360
W,= 6-3 Wa + service
ILLUSTRATION 4 with an
USTRATIONS 3
managed by a single individual, customers arrive according to a poisson process
Customers arrive at a box office window, being 30 per hour. The time Atacertain petrolpump, arrivals. The service time is exponentially
input process with mean rate of 5 minutes between the average queuetimelength()the
according toa Poisson has an exponential distribution
90
with a mean of ayerage time ofnean time of2 minutes. Find (i) spentbya
averagecustomer
required to serve a customer determine the average distributed with queueing system (iüi) before
seconds. Find the average waiting time ofa customer. Also ofcustomersin the
average numberpetrol (iv) average waiting time of a pump will
length.
number of customers in the system and average queue |JNTU pumnp and queue is 4 min, a second
(Mech., ECE) 1999/CCC)) CUStomer in the ifhe waiting timein
receiving the service. (b)be the arrival rate foropening second pump. BRAOU (MBA) 97]|
be opened. Whatshould Dr.
olution :
Arrival prOcess is Poisson i.e., Markovian represented by 'M.
Service distribution is Exponential which can be represetnedas 'M (Markovian)
Solution : (oo/FCFS) queue model.
This is an (M/M/1):
Number of servers =1 = 5 min=
12
inter arrival time=
No. of people (customers) allowed to enter queue = oandIqueue
desciplineis Average
understood as FCFS (First Come First Serve). or = 12/hr 1
Hence, the given queue model is represented as =-=2 min =30 hr
Average service time
(MIMI1): (oofFCFS) 30/hr
or ! =
 drive directly to Solution :
(i) What is probability that acustomers can window. =20;
have to wait outside =60 in good weather and 30 in
{ü) What is probability arriving ustomer indicated space, bad weather.
trucks per day. (6) L,-in good weather
(üi) Expected waiting time of customers
4
in bad weather
40 hour in good weather
olution : 1
(ü) hour in bad weather
hour.
5 12 /
A= 10/kr; u== B Find the (i) average queue length, (i) average time
therewould be two customers in the queue in a salesspent, (üi) probability that
counter nanned by one
sales person where the service time is about 6 minute and arrival rate is one
i) Po+ Pi +Pz person every 10minutes.
Solution:
60 60
= 042 10 =6/ hour, =10 /hour
(1) Lo= 0.9 customers
(ü) P(r23) =(/u=(10/12) =0.48 SV2enreserd

(i) W=5/2 hour or 25 minutes. (iü) W,=hour

4
or 15 minutes
3. A scooter medhanic has an average of four customers an hour. The average (iii) P = (/)* (1- M)=0.144 or 1.44%
service time is six minutes. Arrivals are at Poisson fashion and service is done
arrival rates, negative exponential
at exponential pattern. Find the 6. A maintenance service facility has Poisson
a first come frst serve queue discipline Break
(i) Proportion of time during which shop is empty. service times and operates on
sotaio2 day with a range of zero to eight. The
downs occur on an average of three per
(ü) Probability of finding at least l customer. maintenance crew can service on an average siX machines per day with a range
the
(iüi) Average time spent (service time inchuding) from zero to seven. Find
(iv) Average number of customers in system. service facility.
(i) Uilisation factor of the
Solution: (ii) Mean time in the system
60 the system in break down.
A=4/ hour =10
6 /hour; p==0.4 (ü) Mean number in
system.
(iv) Mean waiting time in the machines in the system.
() P,=l-p= 0.6; (ü) P(n2 l) =0.4; finding two
(v) Probability of
the queue.
(vi) Expected number in
2
(ii) W,= hour or 10 minutes (iv) W,=
Solution:
4. The average rate of arrival of trains at arailway station during the busyperiod l=6/day
is 20 per hour and actual nunber of arrivals in any hour follows a Poisson A= 3/day;
odistribution. The maximum capacity of the railway stationtis sperhour p 3or50%
on an average in good weather and 30 trains i 60trains butactual ()
bad weatherrwith these
number of arrivals in any hour follows aper hour indistribution
averages. When there is congestion the trainsPoisson
are made to wait for clearance
outside thestation. (i) W,= l nachine (vi) L= machine
i) How many trains would be waiting outside on an average in good weather ( / ( 1 - )=0.125
and bad weather. (v) Pg=
(ii) How long would atrain be waiting outside before getting into stationin -p= 50%%.
good and bad weather. (vii) Idle time = l
8.10 Model -2: (M/M/): (oo/SIRO) LLUSTRATIOM 8
pr Raju's outpatientt clinic can
Single Channel- Unrestricted Queue: This model is identical with model -1 accommodate
The patients whoarrive when hall is full, balk sixKpeople only inthe
all respects except for the queue discipline. Since the operating waiting hall.
queue model are independent of queue discipline(as the derivation for charactpreroibabisticlsity
independent of queue discipline), all the results remain unchanged. Therefore is
of fashion at an average rate of 3per
Doctor's chamber which is e
Theprobabilityt
hour and away.
thataeponentially
The
spend an patients
patient can getdistributed.
Findaverage arrive in Poisson
of 15 minutes in
problems under this model may be solved using same formulae as those hisarrival directly into the doctor's chamber
model -1. (ii) Expectednumberrofpatients waitingfor
upon
8.11 Model -3: (M/M/):(N/FCFS) ti) Effective arrival rate. treatment
) The time apattentcan epectto spend in the clinic
Single Channel - Restricted Queue Model : This model differs with model
respect of limit of queue. Here the capacity of the system is limited to N..The relevant Solation:
formuBae are given below : From the above problem,
Arrival : Poisson (M); service : Exponential (M;
1. P, for 0 S n sN, p l ie, A*us Number of servers :One; (1) discipline : PCFS;
af But the queue is limmited
for p=l or à= The model is (M/M/1): (N/FCFS)
N+1 Now capacity of the system
and P = 1-p plandp< 1 N = hall capacity + doctor's chamber capacity
nie tool cra =6+l=7
Where
Aeaetobe ses Arrival rate (^) = 3 per hour.
t crataina s
N - limit of queue system. Service rate (2) = per min = 4 per hours.
n =number of customers arrived.
2. Expeted size of system or expected number of customers in the system p=,=0.75 i.e., < 1 (stead state exists)
arrival.
(N+ 1) pN+1-for p*1 (i) Patient will directly enter doctor's chamber if the hall is empty on his
l-pN+1 The probability at this situation is P
for p=1 1-p 0.2778
Po
3. Expected length (size) of queue or expected number of customers waiting in
queue.
loadiog tise
waiting for reatment is (
(u) Expected number of patients 8
4. Expected waiting time of custorner in the 3
system 8
4 =2.1
8
W = L,

5.
2(1-PN
Expected waiting time in queue
A=2.1--1.36
Wa(1-PN) = (-Po =2.89 =3 per hour.
(ui) Effective arrival rate (ef =4(1-0.2778)
 Queue
expect to spend in the system 451
(iv) The time the patient can 1l - 0.2778 2. Expected length of queue
L, +(l-P) 1.36+2.89 =0.72 hours
hetf =43.2 minutes.

9 Expected number of customers in the systern

Practice Problems
with Poisson input,
Consider asingle server queuingsystem exponentiheial Serick 4. Expected waiting tine of the custorner in queue
1. rate 3calling units per hour,
is
the mean arrival
times. Suppose hour and the maximum permissible number
calliexpet
ngofuit,etthe
S

in
service time is is0.25
the system two. Derive the steady-state probabiulity distribuiÍn 1
system and then calcualte the expected mumber
number of calling unitsin the 5. Expected waiting time in the system
in the system.
of calling units in the system = 0.81.
Answer : The expected nurmber
2. Ifeveryperiod of 2 hours in the day (8
to 10 a.m.) trains arrive at the
for a20 minutes but the service time continues to remain 36 minutes,yardthena W,=W,* +

6. The probability that a customer has to wait (busy period)

calculate for this period. empty, and
is
(a) The probability that the yard P(n2s) = Po
trains in the system, on the assumption that the
(b) The average number of four trains only.
line capcaity of the yard is limited to a ILLUSTRATION 9
A super market has two girls ringing up sales at the counters. If the service time
Answer:
2= l/20; =l/36: p= 36/20= 1.8 (> l) and n=4 for each customer is exponential with mean 4 minutes, and if people arrive in a
Poisson fashion at the rate of10 hour.
p-l = 0.04 (a) What is theprobability ofhaving to waitfor the service.
(a) P, =
p+] -1
4
(b) Whatis the expected percentageofidle time for each girl?
numberofunits in thesystem ?
(c) Findthe average length and theaverage
(b) L,=) nP, =2.9 =3. Mech/ProdiChem/Mechatronics 2001/S]
JNTUCSE 97/S JNTU-

Solution :
.12 Model -4(M/M/S) : (/FIFO) (2) =|and service rate () =per minute]
No. of servers (s) = 2, Arival rate
Multi Channel - Unrestricted Queue Model: This model can be supposed as ult
server)
(multiplefollowing 1
extension to model-1. Here instead of single channel, multi channel 1
in parallel say'' lines are assumed. With other notations as usual we usethe
formulae for this nodel.
1. for 1sn<s Therefore,

p" 2-1
2
2.4 1

s. Po for n2s
1
Po=
Where n=0
2 1 1
s-l
P, =S or
1(2)sP_ P=Po-3
n=0
 Quel
Now CConsider Combined System:
(a P (n > )= P (W> 0) Arrivals : Poisson (M)
Service : Exponential (M)
n=2
n=2 No. of servers:two (2)
=or) 0.167 Limit of queue : Unlimited (-)
Discipline :FCFS
is l- ie, I-(or)(0
(b) The expectedidle time for eachgirl ofidle time for each girl is Model :(M/M/2) : (oo/ FCFS)
Hence, the expected
percentage 67%.
Mean arrival rate : A= 14+ 16 =30hours
customers waiting time =
(c) Expected length of 1 Mean service rate : =20/hours
0.500.17 3minures
Number of servers (s) = 2 and o = 30 3
sp 2x 20 4
USTRATION 10 accounts. The first teller handles (.. p<l)
working on savingsdeposits
Abank has two tellers
Thesecond teller handles only. Ithas been found that 30
withdrawals onlydistribution and withdrawals both are enoneat
time
the serviceservice for deposits Depositors are fond to arriue
Also 20 9
customer.
time 3 minutes per day with mean arrival rate 16 per hour,
with mean through out the
a Poisson fashionarrive mean arrival rate 14 per bo
in a Poisson fashion with time
S

Withdrawals also waiting and with

for deposttorsWhatwol
What would be the effect on the averagewithdrawals and deposits? Now
drawers, ifeach teller could handle both
accomplished byincreasingthe service timeto
be the effect of this could only be
3.5 minutes? 2
40
lution : systems as follows. +
Initially, we can assume two independent queueing DepositorsSystem 2 40- 30
Withdrawers System Poisson (M)
Arrivals: Poisson (M)
Service : Eponential (M) Exponential (M)
NumberT of serverS : One (1)
Limit of queue :Unimited (o)
Discipline: FCES
One (1)
Unlimited (oo)
FCFS
notulol + 2x1"4* 10

Model : (M/MJ1): (oo/FCFS ) (M/M/1): (oo/ FCFS)

Mean arrival rate : (^) = 14/hours Ag = 16/hours
1
Mean service rate : L1=, per min 20/hours Hg =per min 20/hour Average waiting time of arrivals in the queue
1 --3min) .Po
(P1= =0.7< 1,: service time =3 min) (Mean service time W,
Ag 16 =0.8< 1 1
20
20 (40-30)
Average waiting time (2-1)!
Average waiting time hours. 530
14 16 9 20 1
20(20-16)
HË (y-) 20 (20- 14)
7 W,* He (Hg-g) 4 100 '7 140

60 =7 min hrs = 12 min = 3.86 min

 Que

increasedservice time 2. Two repairmen are attending

Combined waiting time with whens}s breaks down according
five
to a poisson machines ina workshop. Each machine
3.5 mineo2iiEA repair timne per distribution with amean 3
machine is exponential with mean 15 min. per hour. The
A'= 30 hourand
) Find the
60 120perhour repairmanprobability
is idle. that the two repairmen are idle and that one
=.5 7
ie,<l and (i) What is the expected number of
30
su'2(120/7) 8 90/7 g Given an average arrival rate of 20 per
idle machines not being served?
on service, at a single channel with meanhour; is it beter for a customer to get
2x (120/7) of the two channels in paralle with mean.service
service
that of 22 customers or at one
rate of 11
of the two channels? Assume that both the queues are of customers each
(2 (120/7) - 30} for
coLrueib M/M/Stype.
49 2(120/7)
2x1 16X (30/7) Answer : Better to get service at two channels.i*o
B. A telephone exchange has two long distance operators. The telephone
-(15f company fnds that during the peak load, long distance calls arrive in aPoisson
fashion at an average rate of 15 per hour. The length of service on these calls
Average waiting time of arrivals in the queue ol od 9q is approximately exponentially distributed with mean length 5min.
(a) What is the probability that a subscriber will have to wait for his long
distance call during the peak hours of the day?
120/7
Oect (b) If subscribers will wait and are served in turn, what is the expectedwaiting
2-1)! [2 (120/7) -30 15 time?
2
49 120/7 1 49 1207 min.lg ao
oq s won Answer: (a) 0.48 (b) 32
Syete
o boisnileo . A company currently One has two tool cribs, each having a single clerk, in its
343
hours manufacturing area. tool crib handles only the tools for the heavy
for
30 x 60
one handles all other tools. It is observed that
bood ogoda v machinery. While the second witha mean of 20 per
each tool crib the arrivals follow a Poisson distribution
343
min 11433 min. exponential with a mean of
30
hour and the service time distribution is negative
that, if tool cribs are combined in such away
2 minutes. The tool manager feels demand arises, would be more
Practice Problems tool as
os that either clerk can handle any kind of reduced to some extent. Itis believed
I. A post ofice has 3 windows providing the same services. It recieves on an efficient and the waiting problem could be will be 40 per hour, while the servive
tool cribs
average 30 customershours. Arrivals are Poisson distributed and service time that the mean arrival rate at two
exponentially. The post office serves on average 12 customers/hours. time will remain unchanged.
0) What is the probabiity that a customers will be served expected
proposal with respect to the totalnchudng
(ü) What is the probability that a customer will
immediatelyr ig Gompare in status ofqueue and the
tool crib(s), the expectedwaiting time
have to wait? R- A number of machines at the wait for more than
(iü) What is the average no. of and probability that he bas to
customers in the system. service time for each machine
(IW) What is the average total time that five minutes.
office? customer must spend n ue r arriyals, 6 min.oge
s
Answer : () 0.29 Kes-onl Answer: For Separate tool cribs:2
(iü) 0.70 ra .E
(iü) 6customers (approx) (iv)12min. cribs : 2.4 arrivals, 3.6 min.
aeer For combined tool