Queueing Theory

14.3
Steady State: A
behavio of the system when the
ISis said
said to in Steady State
probabilit that system is t0. be denote the
state there are 'n' indepenn
pendent of time. Let P,()
units in
nts i the system at time t. Then in steady

lim
I ) = 0
14.4. Kendal's Notation for
Generally Queueing Modelrepresenting Queueing models
be completely specified in
following symbol form: (a|b|c): may
(d\ e):
he

where a =

b
probability law for the arrival.
probability law according to which
customers are served.
C
Number of channels (or Service stations).
d =
capacity of the system.
e =
Queue discipline.

14.5. Distribution of Arrivals "The Poisson Process" Arrival

Distribution Theorem. (Pure Birth Process)
If the arrivals are completely random, then the probability
follows
distribution of a number of arrivals in a fixed time interval
a Poisson distribution.
Proof: We make the following assumptions:
and the
units in the system at time t,
1) Assume that there are n
e small time
arrival (birth) will
occur during t
probability that exactly one

interval Ar be given by AA.

of more than one

to be so small that the probability

) Al is considered
arTival in time At is almost zero.
determine the
increments. We wish to
has independent denoted by P,
n e process interval of length
i,
probab of n arrivals in a time
exclusive ways of
two mutually
Case 1: When
en n > 0, there may be

having n units at time+ Ar. at time t and

no arrival
takes place

units in the s
ystem
t n e r e are n

during time interval A.

 14.4
Applied Operations
(b) there are (n 1) unitsi the system at time t,
Researchh forfor Man
Management 14.5
and one
yeueng neory

place during Ar. Nent =2

arrival takes ePj(7) + AP1()]
Adding the 2 probabilities, Now v
have

P,t+Ar) a
=
P,M-AAn) + P,n-1 ) AAr d l PiI
=

of
Case 2: When n = 0 both sides
w.r.t '
ntegrating
Po(+A)= Po) (1 AAr) -
e P ) = at+B
Rewriting (1). (2) and dividing thro' out putting=0 > B =0
by At we
get
(2
PAt+Ar)-P,) . P)= 1
Ar --AP,(G)+
P(t) +AP-
AP,-1 ()
0
and Pot+ Ar)-Po() L) e for n=2eetc.,
Similarly P2) =
2!
as
At0,
Ar
-APot)
Po'(t) =-APo(t) for P() for n m
Ingeneral
=
and
n =
0 =

m!
P) =
-P,) +AP-1 3) hypothesis, the result
(3) can be written ), n>0 Wecan prove this for n = m+ 1.. By induction
as 4) value of n.
is true for general
Po') d
Po) =
Aor [ log P,() = A" which is a Poisson distribution formula.
Pot)] =
a n!
Integrating w.r.t 'r Model
log Po) =
-At
14.6. Modell: (M | M |): (»/ FCFS)- Birth and Death
+ A probability distribution of Queue
The With usual notation, show that
constant 'A' can be .5)
found out ength is given by p" (1- p) where p=<1and n 2 0.
using boundary conditions
P,0)=n=0 = 0 , n>0
IMU. BE. Apr 96]
Substituting n =
0, Po0) =
1, The probability that there
will be n units in the system
at

Using this in (5) A

=0
OLution:
independent cOmpound

logPo) ne
(t+Ar) may be expressed as the sum of four Poisson
=-t (or) fundamental properties
of probability,
Putting n =
l in Po(t) =e PODabilities, by using the
(4) artival and of exponential service times

P) (a + ) At] + Pa-10) AA
=
-AP (0) +APo) e,
P (t + Ar) =
P,(t) [1
-

+ P,()
(AuNA).. (1)
or
P) +P+1)Ar
or
-AP^)++le-t
This is
P) +AP) le-u = For there are following 4 cases Time + A )
No. of units
Service
differential equation of order timet Arrival
This can be one. No. of units
solved by
multiplying thro' out by integrating
I.F = erd = eht aeictor

42
4.6
Applied Operations Re 14.7

From (1) Research forfor Managerment eueing

Theory
2

lim P,(+ Ar)- P,() PoPo*

( Po+
Ar
-

(A +
)
Ar0
P,() + AP,-1
In the steady state P,'() 0 )+ P.
n+1 (t)

P,) P or Po =

n+1
Probability that there will no unit at time (t+ Ar) will .2 in LHS is infinite Geo. Series whose I
following independent probabilities be the and the infinite series
sum
um ofof For
(i) Prob [that there is no unitin the system at timet and and C.
Ratio
erm=
time Ar
in
Pol) (1 -AA) or Po 1- =1-
Question of any service in Ar will not arise as there
at time }}
are na .
units As P -() P
(i) Prob [that there is one unit in the
system at time t,
one unit
unit
serviced in Ar, and no arrival in Ar]
= P ( ) uAr (1 - AAr)
(Proved)
P = p" (1 -p)
adding these probabilities,
Pot+Ar) =
Po) (1 AAr) + P,() uAr (1 AA)
-

Measure of Model the system,L

of units in
-

number
(expected)
Pgt + A)Fol 1. To find the average
of Expected
value
as Ar>00 Ar --JPolt) + uP () for n Solution: By definition
In steady state 0 =
-APo + uP nP

Use n 1 in
P Po -

(2)
then P2
--9)E-
P
Po for n 20
-(D Using Bionomial

seres

Since
n = 0
 14.8 Applied Operations Research for U E V E N gT h e o r y
14.9
Man. robability [Waiting tiine in the Queue 2
Management = J p(u-A)eu-Awdw

where p = < 1 0 . T r a f f i cI n t e n s i t y = 4

2. To find the
average length of Queue, ample 1In a railway Marshalling yard, goods train arrive at
Example
L
adeof 30 Trains per day. Assuming that inter arrival
a rate. time follows
an expor
nonential distribution and the service time
distribution is also
xnonential, with an average of 36 minutes. Calculate the
following:
1-p t h e mean Queue size (line length)
3.
Expected waiting time in the () the probability that Queue size exceeds 10
system (i) If the input of the Train increases to an
Ws what will be the
average 33 per day,
changes in (i), (ii) ? [MU. BE. 1990]
Little's formula] [MBA. Nov '97 MKU]
-
4. 30
Waiting time in the Solution: 1
Queue, 60x24 48 6 trains per
minute
p = =0.75.
48
= -
5. H(H-A) (i) L = 0.75
-0.75 3 trains
Expected waiting time of a
customer who has to (ii) P(210) = (0.75)10 =0.056
wait
(W|W>0) (n when the input increases to 33 trains per day,
6. Expected length of the
non - empty 33 480 and = 6
we have = 60x24 trains per minute.
Queue, (L|L>0)
H-
Probability of Queue size 2 N Now, where p- :p 0825
is p
Probability [Waiting time in . s 0825 = 5 trains (app)
-0.825
the
= J - 2 ) u AWdw
system2 Also P(2 10). = pl0 = (0.825)

= 0.1460
Ans
 14.10 Applied Operations 14.11
Research for q y e v e i n gh e o r y

arriving customer will

have to
Example 2 Customers arrive at a
one
window dri
Management ii Probabili ility that
an

according to Poisson distribution with mean 10 ban

the i n d i c a t e d s p a c e

time per customer is exponential with mean 5 mi

Der bank = 1-042 0.58
our. Service
he space in in the Queue
front of the window including that for the time of a customer
seru iced Spacein
car can Average
waiting t
accommodate a maximum of 3 cars. Other c a n ed car (ii)
rait outside
space. this
) What is the probability that an
arriving customer can driveco. -

directly to the space in front of the window?

)What is the probability that an = 0.417 hours [Ans
arriving customer will hae
to wait outside the Indicated space ? have arrival rate
market, the
average
In a super The
(ii) How long the arriving customer is expected to wait
Example 3 foilowing
Poisson process.
before 30 minutes
the
customer's

starting service? is 10 in every calculate

[MU. BE. Nov '93] customer
the cashier to list and What is
time taken by
distribution.

exponetial
Solution: We know that p, denotes the average following
probability of n units in the purchases
2.5 minutes,
is
length exceeds6?
system and that the Queue the system?
a customerin
the probability
Nov 95)
time spent by [MU. MBA
is the expected

--
What
Pn
arrival rate
mean
Here the
() The Solution:
probability that an
arriving customet can drive directly to the per minute

space in front of the window

minute
per
=25
PotP1 +P2 and mean
service
rate

Po 1- = 0.8333

>>n)
= p"
Queue
size
(0.8333)° = 0.3348 [Ans
of
PotP1+ P2 =
(1- (i)
) (The
probability

When
n
= 6

Here 10
per hr
- 2499
60 W,A
=

12 per hr . =
10
12
(ii) 0.833
-0.833
3 0 .167
Ans]

PotPP 9119-) 14.96 minutes

= 0.42.
 14.12
Applied Operations 14.13

ample 4 In public Telephone booth Research for Mar

Q e v e n g
i h e o r y

arrivals arertManrononysrethet
a

average 15 per hour. A call on the i the

the Now L'
just one phone, find (i) expected average takes 3
number of callers minutes. IfIfethe
time (i) the
proportion
of the time the booth is in the there b or 2

expected tobooth alat amy

bebe ldlk?
s

Solution: Mean arrival rate 15 per

[MU. BE. MBA or2-4
-
32 0

hour Apr 961 0

Mean service
rate =
or (u-8) (u+4)=
(i)
60 =
20 per hr. 8 patiens/ hr as ' = -4 illogical
Expected length of the or
non-empty
Average time required by each patient
Queue -1
220-15
0 4.
(ii) The service is Ans hrs minutes
busy means =

required by each patient

:
the booth expected to ldle Decrease in the time

for 1- hrs =10-- minutes

Example 5 On =15 minutes
patients per 24 hour dayAns
an The budget required for each patient
theservice of an average 96
10 emergency clinic. Also on
minutes of active require =100 1 0
only one attention. Assume thataverage, a patient requires
per emergency at a time. the facility
patient can handie = Rs.125
and that treated to
each minute obtain anSuppose that it costs the
clinic Rs. 100 patient should be
average servici Todecrease the of the Queue, the budget per
10 per
patient
of
decrease ricing time of 10 minutes size

Clinic to
Clini treated, how much in this averagage time would cost Rs incTeased from Rs. 100 to Rs.I25.
decrease the would have
ave to be
patient? average size of the budgeted byDy the time spent on his job
Queue
ueue from
115 patien Example 6 A T.V repairman finds that the
patients to 30 minutes. If he repairs
Solution: Here IMU. BE. Mech. Apr 95 a n exponential distribution with
in and if the arrival
of sets is
mean

= 96 Es in the order, in which they came

= 4 patientshr n with an average rate of 10 per 8 hour day, what Is his
set
jobs are ahead of the average
pected Idle time day ? How many
Science Apr 931
Average number of 10 60 6 patients/ =
just brought in ? IMU. BSc App.
patients in the Queue hr MBA. Nov.96,
MKU. BE. Apr 971
JMU.
Solution:
This number is to be per minute
Mean service rate
reduced from 1
to
to
x60 =2 sets per hour
 14.14 Applied Operations 14.15
Research forfor Management
Man QUeUeng
Theory

Mean arrival rate

10 time in the Queue
8 per hr waiting
Average
(i) = 0.5
10 =0.05 hr
where H =
2 per
hr.
3 minutes.
= per hr.
Average
Queue length,
(i)
Theutilisation factor is
5 -

= 0.5 car
For 8 hr day, Repairman's busy time 8 5 L 1 -p
=
x =
5 hrs 1
-2
. Idle time of repairman 8-5 hrs =
3 hrs
=

units in the system P, P, (1 p)

-

The number of jobs ahead = of n

No. of units in the
system [Ans (iv) Probability

8 When n =2, P2 : [Ans

-

in Post Office arrivals

Public Telephone booth
a
= 2 app, TV sets Example 8 At a
inter-arrival time of 12
Ans are be Poisson with a n average
considered to
assumed to be
Example 7 Cars arrive at a petrol pump,
having one petrol unit, minutes. The length of the phone
call may be
the
4 minutes. Calculate
in Poisson fashion with an an average of
average of 10 cars per hour. The service distributed exponentially with
time is distributed
exponentially with a mean of 3 minutes. Find following: will not have to
() average number of cars in the
system (ü) average waiting time in that a fresh arrival
the Queue (üi) average (a) What is the probability
Queue length (iv) the probability that the wait for the phone ?
number of cars in the system is 2. arrival will have
to wait more

that an
(6) What is the probability is free?
Solution: before the phone
than 10 minutes formed from
time to

Mean arrival rate, A (C)What is the average length

of Queues
=
10 per hour
time ?
Mean service rate, u =
x 60= 20 per hr
Solution:
A
Mean arrival rate,
=

P
(i) Mean service rate.
- 0.33
Average no of cars in the system,
to wait
not have
arrival will
a fresh
that
Prabability
i - p (a) = 1 - = l-0.33
0.67
2
= = I car

1-2
 14.16 Applied Operations Research for Man- 14.17
(b) Probability that an arrival will have to wait for
atle
tleast Managemert
10 min (ueveing
1heory

inutes
10
-3)eu-¥ dt Solution:

a =O
10
0.10 = =
0.33 person per minute
la)P ( W > 0 ) = 1 - P o

(0.33) (0.25 -0.083) e0.167 dt

10
-1-(1-
05311 0167/ 10
= 0.05511|o.167J
A 0.100.3
0.0621 justified if the arrival
(b) The Installation of the second booth will be
(c) The average length of time. Then the length of the Queue will be
Queues from time to time rate is greater than the waiting
L>0) Increasing

0.25 Now W (-2)*°0.33(0.33 -)

025-0.085 Where W = 3 and a = (say) for second booth.
= 1.5

Example 9 Arrivals at
Ans Onsimplification ' =0.16.
Poisson with
a
Telephone booth are considered to be the
an
average time of 10 minutes Arrival rate should become 0.16 person per minute to justify
next. The between One arrival and
duration of the second booth.
distributed with mean 3 phone call is assumed to be exponentauy arrivals per
minutes. Increase in the arrival rate is 0.16 -0.10 0.06
(a) What is the
have to wait probability that a person minute.
BRU. BE. Mech. Novarriving
at the will
? boou 9| (c) Average number of units in the system is given t
()The telephone '96, MKU. BE. Apr
Department
convinced that an arrival will Install a second hen
would
boothtleast 3 Ls 1-p
minutes for phone. By how much expect waiting i as
inerease in order to should the flow of ar 0.43 = 0.43customers.
(c) Find the justify the second booth ?
(d) Estimate average number of units in the that the phone will be busy = traffic Intensity
the fraction of system a) The fraction of a day
the day that the
the phone
phone will
w be i
uSe.

(e) What is 0 p 0.3

the IBRU. BE. Mech. Nov
altogether probability that it will take more
to wait 1ore than 10
ninutes

P(w2 10) = (H-A) eu-A d)

for
phone and mplete the tn (e)
completehe call ?
ca
B.Sc App.Maths 1980
1980 10
IMKU. = 0.10. [Ans]
14.18
Applied Operations 14.19

Example 10 People arrive at

Research for Ma Q u e u e i n gh e o r y

boo Managemen
a
Theatre ticket
arrival hour. Service booth in Poisson
distributed rate of 25 customers arrive according too
per barber shop,
2 minutes. Calculate tim
time is consta 1.
At a - man
one

mean arrival rate of 5 per hour and the

distribution with
a
(a) The mean number in
the
stant a n distributed with an average hair
time was exponentially
(b) The mean waiting time. waiting line. ir cutting
hair It is assumed that because of
excellent
minutes.
10 minu
(c) The utilisation factor. cut
taking
were always willing to wait. Calculate the
putation, customers
custon
reputation,
Solution: = 25 per hr,
following
number of customers in
the shop and the average
x 60 30 per hr. (a) Average
for a hair cut.
number of customers waiting
arrival walk in right without
: p 30 =0.833
-
(b) The
having to wait.
percent of time an can

(i) Length of the Queue of customers who have to wait prior to getting
(c) The percentage [MU. BE. '91]
into the barber's chair.
Lg 1-p 16.7 %]
[Ans: (a) 4.8, 4 (app) (b) 83.3%, (c)
0.833)2 centre is 3 per hour. The
2. The mean arrival rate to a service
1-0.833 mean service time is found to be
10 minutes per service. Assuming
0.693889 Poisson arrival and exponential service time,
find
0.167
(i) Utilisation factor for this service facility
= 4 (app)
(ii) Probability of 2 units in the system
(ii) Mean waiting time = (i) Expected number of units in the Queue
customer has to spend
in the
(E) Expected time in minutes that a
9.6 minutes
(iüi) system. Ans: (i). (ii) z. (i) 1. (iv) l
Utilisation factor p
= 0.833. [BRU. BE. '85]
according to Poisson
Cars arrive at a toll gate on a frequency
3,
EXERCISE time for passing thro' the
aistribution with m e a n 90/hr. Average time.
1. 1. is 38 seconds. Drivers complain of long waiting
Explain the terms ates decrease the passing
time through the gate
willing to
queue discipline
pline and system Authorities are
automatic devices.
These can be
syste: capacity.
2.
Describe the capacuy490) to 30 seconds
Justified only if under the old system,
by introducing
the number of
new
waiting car
queueing models M/M/1 and(M.U., M.B.A., Apr exceeds 5. In addition, the
percentage of the gate's
idle time under the

3.
M/M.1/lc. new system should not
exceed 10%. Can new
device be justified.

Briefly discuss the IM.U., M.B.A., Nov 90

[Ans New device justified
0.25 >0.10]
not as Po =

management. application queueing theory

theory
of queuein
in industr to repair
machines which
4. What is the M.U., inr
M.B.A. 981
Apr '987 4. A repairman is to be hired breakdown follows

queueing theory? breakdown at an average

rate of 3/hr.
The
 14.20 Applied Operations Research for eveingTheory 14.21
poisson distribution. Non-productive time of a machin
is considered
Management long plane would be in the stack and in the process of
a

to cost Rs. 10 per

hour. Two repairmen have
been intervic (b) Ho
landing in good and bad weather ?
is slow but cheap, while the other is fast but expenewed
Sve. The
wed- one lawmuch stack and landing time to allow so that priority to
repairman charges Rs. 5 per hour and he servico The slow land out of order would have to be requested only once in
machines at the rate of 4 per hour. The fast Repairman d. break
twenty?
per hour and he services at an
average rate of 6 per hon nds Rs.7
repairman should be hired ? Which [Ans () Lg = 6 ingood weather = in bad weather
[Ans: Fast repairman since it results in savings
Rs. 144 per day.]
of Rs. (280 (ii) W = 40 hr's in good weather
136) =
5. A branch of a National bank has
only one typist. Since
typing work varies in
length, the typing rate is ice the s
hr's
0 bad weather
aDproximating Poisson distribution with mean randomly
rate of 8 distributed (il) Waiting time w 4.5 minutes in good weather
hour. The letter arrive at a rate of 5
per hour during the
letters per
hour work day. If the entire 8- 18 minutes in bad weather.
typewrite valued at Rs. 1.50 per hour.
is 8. inbank there is only one window. A solitary empioyee
a
Determine performs all the service required and the window continuously
1. equipment utilisation open from 7.00 am to 1.00 pm. It has been discovered that
remains
2. The percent time
arriving letter has to
an
wait
theaverage number of clients is 54 during the day and that the
average service time is of five minutes per person. Calculate
3. Average system time the
4.
average number of clients in the system (including one

Average idle time cost of the typewriter being served)

per day. () the average number of clients in the waiting line (excuding
[Ans: (i) 0.625, (iü) 62.5%, (iü) 20
6. At
minutes, (iv) Rs. 4.50) the one being served) and
what
average rate must a clerk at a supermarket work in (ti) the
average.waiting time.
order to ensure a
to wait
probability of 0.90 that the customer will
not ha [Ans: (i) 3. (ii) 2.25, (ii) 20 minutes. per client
longer
than 12 minutes ? It is
assurmed that there is only one
counter, to which customer arrive in 14.7. Model I: { Multi Service Modely
Poisson fashion at an averag
-

a
rate of 15 per hour.
The length of service by the clerk has au (M M S: (co/FCFS)
exponential distribution. Dtain the systenm of steady state equations and hence tind the

u e of P in usual notations where

Ans: P(212) = 0.10 ;2.48 minutes per servi
7. The mean rate
of arrival of
(i)n <s ii) n 2s
peak hour period is 20 planes at an airport duringt Olution: We obtain overall
service rate, when
there are n units n the

per hour, but the number of arrivals

any hour follows a Poisson distribution.actual
The airport ca
nd 60 system as follows
may be served simultaneously.
There
hour Ttnss; all the customers
lanes per hour on an average in good weather and 30 per io (
be Queue, (s - ) number ot
serverS may reman dle and then

number landed in any hour follows

no
a weather, but the actual follo
, 0, , ,
Poisson distribution with
thereb
there b n =

the serversare busy, maxmum nuuder ot

customers

respective averages. When

the i n the stac (u) I f n S , all
ongestion, the planes are forced to fly over the field *the su lling in the Queue
will be n , then u,,=su.

waiting the landing of other planes that arrival also A

earier i n t h es t a c

(a)How many
planes would be flying over the 1eld ,
on an 43
average, in good weather and in bad weae
 14.22 Applied Operations
The steady state difference
Research for Manano QyeveingTheeory
14.23
equations are S+2
Pa(t +A) Pof |1 = -

AI+ P,() HAr for n =0;

Po
P( +An)= P, (1 (% + nu)) Ar)
P,2sP
+Pr-1(t) AA
-

+Pp+1 ) (n+ ) pAr for n Again, in general

and P, (t+ Ar) P,/(l -(+sp)) Ar) + P,-jt) AI 1,2,3,...

=

+P+1)uAr for n =
s, s +1, s +L,..
2,..
P P s +(n-s) P Po forn 2s

Now dividing the equation by At and

difference cquations are, taking limit as Ar »0. the
the
Pg) =-APg() + uP, (1) forn
Now, find P using the fact
Pn=0
=0
P,0 =-(à+ nu)) P,)+2 .(1)
P,-1) +(n +1)uPn+1 () .(2)
s-1

for n =
1, 2, 3, s
P+ P, = I
P,) =
-(à+
sp)) P,)+A P,-)+ su Pa+i () -1 n=0 n=s

.(3)
for n =
s, s +1, s+2,. or
Considering the case of steady state (i.e.) when t n=0
(independent of 1) and hence P, , P,) P
0
()>0 for all n, above
equations become
-Po+ uP for n =0 or 1

-( +
np)) P, +\ Pn-1 + (n +1)
n=s

p Pn+l
Or
0 for n= 1 <n Ss-1
= -

(à+ su)) P + à
Here P-1 + SH Pn+1 for n 2 s n s

Po=Po (initially) where p S

P Po
or Po (
P P 22 Po Ln=0
n!

P 3uP2 =3 123
Po
Thus steady state distribution of arrivals (n) is

In general
s-1
P,
P Po ifn 0,1.
= . . .

Po for P

P P = . o if n = s,s +1.

P
Ps P +1

Po
 14.24 Applied Operations
Research for 14.25
Measures of Model II
Management
veueing Theory

1. Length of the Queue, ample11

Example A Telephone exchange has two long distance
ors. The telephone company finds that during the peak load,
rators.

operance calls arrive in a Poisson fashion at an average rate of 15

P
g =

Pd-where P, = hour. The length of service on these calls is

approximately
s
onentiallydistributed with mean length minutes.
p e r

2. Length of the system (L)

ial What is the probability that a subscriber will have to wait for
L his long distance call during the peak hours of the day?
3. Waiting time in the Queue IMU. BE (Mech) Oct 98
(b) If the subscribers will wait and are serviced in turn, what is
Wa theexpected waiting time? [MU. BE. Civil 1991]
4. Waiting time in the system
W
Solution: s =2, = =

5. The mean
number of ps
given by (L|L>0) waiting individuals, who
actually wait is
sp_ 7-
Po +
1-p Ln=0 n s!(1-P
6. The mean
waiting ime in the Queue for
is given by (wjw>0) those who actually wait
21(1518)
su-
7.
Prob (w>0) 13
8.
=1-p
Probability that there will be
some one
waiting (a) Prob (w > 0) s!(-p o
9.
1-p
Average number of Idle
servers
S-(average number of customers served)
10. Efticiency of
M/ M/S model - 0.48 A
Average number of
Total customers
number of served
customers served
14.26
Applie 14.27
Operations Research for Q e v e i n gT h e o r y

-I
(b) Managetet 4
9

= P(sp)2
as!(1 p)2 Po

ThusProb (W>0) 2!(1 - 'l/2)

[Ans)
4

1 (b) The
fraction of the time,
the service is busy

3.2 minutes.
Example12 A supermarket has two girls SH
counters. If the service time for ringing up sales at tho time service remains
mean 4 each customer is . The fraction of the
minutes, and if the people arrive exponential with
rate of 10 per hour. in a Poisson 67% (nearly) [Ans
=I-
=

fashion at the idle =

(a) What is the

probability of having to wait for service?
(b) What is the IMU. BE. Mech. Apr 97| (c) (W/W>0) su-A
expected percentage of idle time for
each
girl? [Ans]
(c) If a
customer has to
IMU. BE. Mech. Apr 971 3 minutes
wait, what is the
waiting time? expected length of his
A petrol station has
two pumps. The
service time
[MU. B.Sc. Maths 90 Example 13 4 minutes and
cars
[MU. MBA April 98] distribution with m e a n
Solution: (a) Tollows the exponential
rate of ten cars per hour.
Probability of having to wait for service arrive for service in a
Poisson process at the
has to wait for
service. What
customer
Find the probability that a Nov 95,
idle ? IMU. MCA.
proportion time the pump remains of
is MU. BE. Mech. Nov 97)
P(w>0) =

s! (1 p)
s! (1 --

Solution: Here s = 2, = 10 per hr

Here ==4.=2, p per minute
=
4 15 per hour
== H

Po -
sp) sp)2 7-
n! the pumps remain busy
Ln=0 s!(1 -p) (ii) The proportion of time,

(2./3)" n +
(2./)2 7-
1
2!(1 / 3 ) The proportion of time, the pump
remains idle =
-
% of idle period = 67% (app)
14.28 Applied Operations 14.29
Research
arch forfor Management Q v e u e i n gT h e o r y

A general surance company has three claim

ple 14
(i) Prob (w >0) =

1-p where Ps= s! Po in t branch

its
adjusters in i office. People with claims against the company
to
arrive in Poisson fashion at an average rate of 20 per 8
found adjuster spends with a
amount of time that an
andPo=2 sp_7-1 are The
Ln=0 n s!(1 -p)| hourntis found to have negative exponential distribution with mean
time 40 minutes. Claimants are processed in the order of their
service
psn! (ps)2 1-1 a p p e a r a n c e .

week
Ln=0 2!(1 p) How many hours a can an adjuster expect to spend
(a)
with claimants ?
s (ps)2- b) How
much time, on the average, does claimant spend in the
n!
branch office?
arrivals per hour
Solution: Here = =

- 1 + 2 + 21(1-7 service per minute = per hour

S 3

1+3 Theservice is busy is given byp

=

'For 5 working days (5 x 8 =40 hrs)

with claimants
Prob (W>0)= Expected weekly time an adjuster spends
where P
1-P = x 40

=22.2 hours. [Ans

P 2! in the system is W
2! time an adjsster spends
(ii) The average

Prob where L, = L,
(W> 0) = 9
P P where P su
and Lg (1-p
= 0.167 (app)
[Ans
where Ps Po
14.30 Applied Operations
Research for
Managerment
Q v e v e r n gT h e o r y

and Po (Sp)" (sp) 1

n=0
n! s' (l -p) (0.0908) 9 0.0908x0.555
2

G-7 (
n=0
n
81x 0.0908 x 0.555
16
4.0801914
16

= 0.255 + 1.666 = 1.921

1921x 2 0.7684 » 46 minutes [Ans

W
=
6x4 5

Example 15 A bank has two tellers working on savings accounts

withdrawals only. The second teller handles
The first teller handles distribution for
found that the service time
deposits only. It has been with mean service
withdrawals both are exponential
the deposits and arrive in a
are found to
time 3 minutes per customer. Depositors
mean arrival rate
16 per
with
Poisson fashion throughout the day arrival
Poisson fashion with mean
hour. Withdrawers also arrive in
a
the average waiting
11.666+ 1.388 + 1.041 rate 14 per hour. (i) What would
be the effect on
teller could handle
both
withdrawers if each
ume for depositors and if this could
the effect
What would be
5.095= 0.1962 withdrawals and deposits? (iü) minutes
service time to 3.5
De accomplished by increasing (BRU. ME. 1981)

Ps Po s!
Solution: In the first case

s=3, Po=0.1962 (a) W (A)

5
x0.1962
P 0.1962
6
5 15
0.545 = 0.0908
12minutes.

Ls PP
(1 -p2
14.32

(b) In the ADCApplied Operations Re

second case
2 = 14 Research for Manaqe 14.33
60 3030 = =
yeing Theory

Ifu
=

(d)
i)
W2
H(-) 30 then P 8

7 minutes.
(c) Now Po n!
combining the two cases
Since this =
1
n=0
2:1-
becomes the case of but
4+ 167
p =
two
service stations, 60 4
s=2
SH
_p(sp)
Wa s!(1 -p) Po
. Po se
n!+ (sp)s ]-1
Ln=0
s!(1 -P)]
=2 15

=
21-
n=0 n! 30 = 11.4 minutes.

1- The effect is that average waiting time is increased.

being run on the frontier of a

Example 16| Four counters are
of the tourists.
country to check the passports and necessary papers
counter at random. If the
arrival at the frontier
he tourist chooses a is exponential with
Wa psp-p)2 o s
Poisson at the rate of a and the service time
s!(1 Po parameter, what is the steady state average queue at each

counter 2

2 Solution: Here s
= 4,

The effect is
that Po
average waiting time is
n=0

= =3.86 minutes. reducedd.

 14.34
Applied Operations
Research for Q u e u e n g1 n e o r y

14.35
Management
S
(1 p)2 Po
n
n=0
21-
= 0.047

L= Lt

Example 17 TAns]
Given an
average arrival rate of 20 PsP and P, =Po
better for a
customer to get service at
per hour, is it
where
Lg(1 -p)
service rate of 22 customer or
single
channel with mean
at one of the 1.478
with mean service channels in parallel, two
rate of 11
Assume that both customer for each
Queues are M /M/S types. of the two channels W
Solution: For a single channel = 0.43 hours
On comparison we see that it is better for a customer to wait for 2
=
20 arrivals
per hour channels since W, = 0.43 hours as against 0.45 hours in case of a single
and = 22 customers
channel. Ans
Po 1-p
.
Probability of no customer is 0.09.
=

1- =1 -22 0.09 EXERCISSE

arrival has a mean

A two waiting line with Poisson
channel
L
-

rate of 50 per hour and exponential service

with mean rate of 75 per
H- ()
the probability of an empty system
20 nOur for each channel. Find (i) will have to wait.
22 2 0 = 10 arrival in the system
ne
probability that an
[MU. BE. 1979)
Average waiting time of a [Ans: (i) 0.5. (iü) 0.111)
customer in the
Queue= 2. Find the probability
that there are no
customers in the system
is 3, mean arrival rate =

channels in parallel
Bven that number of l0/hour. Compare
H(-2.) service rate in each channel =
4/hour and mean for the following 2
is in the system
a customer
h e average time that

When
II 0.45
there are 2 parallel channels, systems: service rate of 10/hour.
we have . =
20, u in parallel with mean
=
11,s= 2.
(i) 5 channels
service rate of 30/hour.
Po sp) sp)- (ii) I channel
with mean

BTech. 1983
Ln=0 n! minutes (i) l0 minutes) (AU.
s!(1 p) = 0.019, ( ) 0
S LAus: P,
 1436 Applied Operations
3 Arailway goods trafi+C section for Research Ms.
14.37
Customers with claims against the has 4
Railways are oheclaims Managemet
ssistants
peveing Theory

The orobability that a customer has to wait for service

a Poisson fashion at an
average rate of 24
The amount of time the claims
assistant
per 8 observed
hour to arrive (p)
The expected number of idle tax advisers at any specified
found to have an exponential ay for 6
spends with he in
distribution with a th
he cla daysis
claimant time.
40 minutes. Service is
given in the order of themean service
appearance 6.61.(ii) 3.28, (ii) 0.66 hours, 0.33 hours (iv) 33.3 hours ()
customerss time. Ans: (i)
ance of the 065(vi)0.666]
)How many hours/week can a claims
the claimant? Assistant expect toos
How much time does 4.8. Model l: (M | M| ): (N/ FCFS)
goods traffic office ?
a
claimant n
on the
average spends in tthe Here the capacity of the system is limited, say N. Infact arrivals
[Ans: 72 hours (i) 47.2
(i) ill not exceed N in any case. The various measures of this Model are
4. For a
Queueing minutes] [MU. M.Sc.
xponential service time system with k service
distribution stations,
19751 1. Po NT where p l is allowed
Queue with built up arrival with mean
each havin
service rate p feed
ustomers in the rate 2, find
system (ii) average (i) b
ystem if k =2, p =5 per average
waiting time of a customer ot
number 2. Pn N+T P for n =
0. 1.2.. N
minute,À =8 per minute. in the N
3.
5. A
telephone company is [Ans: (i) 4.44, (ii)
0.56 mintes) L, =
Po2n p
new
Airport. It has planning to install telephone booths n=0
ve to
wait more than established the policy that a in
4.
10 person should not
mand for use is estimated percent of the time to use
r
hour. The to be
Poisson with an phone. The
an time of 5
average phone call has an average rate of 30 5.
Ws
minutes. How exponential distribution
many phone looths should be with
6. A tax instale 6.
W=
consulting firm
eive people who have has four service
Ans: Six phones
counters
counters in the office to
oltiee
ome, wealth and sales problems and complaints
complaints apou
about their
r
service day. taxes.
Each Arrivals average 80
ENample 18 If for period of 2 hours in a day (8-10 AMD trains

in tax adviser
servicing the arrivals persons
spends an irregular Ve at the
yard every 20 minutes but the service time continues to

onential distribution. which have

have
amou have
emain 36 minutes, then caleulate for this period

culate The been

been found to (a) the probability that the yard is empty
averagee service
service time is
tum minutes
20 ninutes.
) the average (b) average Queue length, assuming that capacity of the yard is 4
) number of trains only.
average customers in the syste
i) average number of customers waiting
ting for service
servee Solution: Here p= 20
30
8 , N=4
queue waiting time for the
customner
Istomer in the
the syste
)number of hours each (a)
customer week
with
a
consultan
tax consultant appe

44
 (b) lied
average Queue
size
4
Operations Research
for
=

Po 2n p MananBm veueingT h e o r y
14.39
n 0 barbers s e r v i c e time is
negative exponential with an
0.04 (p he average of

2.9 +2p2 3p3+ l.5minutes per customer. Find Po PP..

3 trains. 4p ) Solution: Here N = I0, A = =
Example 19 In a
at a
rate of 30 railway
day. marshalling
follows trains per Aling yard, P
Assume thatyarà, Ans
an
is also exponential distribution
exponential with an average ofand
goods trains
traing
1-1-16
nd the the inter a
arrive -p
(a) the 36 service timearrival Po
(b) probability that the yard is minutes, calculatedieta
distrit ie
0.1667
average Queue empty =0.1926 [Ans
trains. length assuming that the line
0.8655
yard is 9

Solution: Here capacity t the P p"

=p p= 0.75
(a) The
probability that the Queue size is =
(0.1926) x n=0, 1,2, 10 [Ans
zero is
given by
Po Example 21| A car park contains 5 cars. The arrival of cars is
pN+11 where N =9 Poisson at a mean rate of 10 per hour. The length of time each car
spends in the car park is negative exponential distribution with mean
Po = 1-0.75 0.25 of 2 hours. How many cars are in the car park on average ?

(b) I-(0.75)10 0 90 0.2649 [Ans

Average Queue length is Solution: N=5,
10
p-20
given by the formula, A60 2 x60

aNT n=0
np Po p N T
=20
Ls 1-0.75 9
1 n
(0.75)10 n=0 (0.75)" 19 2.962 x 10-7
=
-6399
=
0.28
Example 20 A
x
9.58 3 trains. [Ans N

customers.
comes to
He can barbershop has Space
service space only to accommodate
10
= Po
n=0
p"
his
shop and only
mer

one person at a time. If a cuso

Customers
randomly, arrive at it full,person
finds x 10-y
a
shop. xn(2.9692
he next goes to the (2.9692x 103)
goes an hours
and n=0
average rate
rate A = 10 per
 14.40
Applied Operations
=
(2.9692x 10S) x [0 Research for Managey p e e n g1 n e o r y
14.41

+(2.9692 x
10-3 ement 0.1666 0.1666
+2 x
(2.9692 10-3)2x
1-0.0001 1
+3 x
+4 x (2.9692 10-3)3x =0.1666
(2.9692 10-3y4x
+5
=
(2.9384x 10-3)
x
(2.9692 10-3)Sx =Ws
[O+(2.9692 10-3
x
x

+2 5

+3x
(2.9692 x 10-3)
x

where Ls =0.166 x npn

+4
(2.9692 x 10-3)2 n=0
x
(2.9692 10-3 x

+5
=
5
(app). [Ans]
x
(2.9692 10-34 x
=0.166 x
n=0

Example 22 At a
one-man barber
= 0.166 [p +2p2+3p +4pt +5p1
following Poisson process at an shop, the customers arrive
are served average rate of 5
service rate according
of 10 exponential to
distribution
per hour and
they
for minutes. Assuming that with an
average =0.166 [0.833+ (2x 0.694)
waiting customers, find only 5 seats are
average time a customer average time the available
a
customers, +(3 x 0.5782) +(4 x 0.4816) + 5(0.4012)
spends in the system. find the
O.166 [0.8331.388 1.7346 +1.9264 +2.006]
[MU. MBA Apr 95 April 97) W
Solution: W
np n=0
0.166x7.88
5
Here 5 per hr
= 1.3094 5

10 60 =0.26 hrs [Ans)

16 minutes.
per hr and N = 5 EXERCISE

w h o she performs
1. A stenographer has 5 persons for service time is
A stenos Arrival rate is
Poissou and
graphic work. 4 hour with an average
P 1- ponential. Average
arrival rate
is per

I-p6 Find
1-i rvice tinme of 10 minutes.

( the average
waiting time
of au
of waiting
line
arrival

length
(1) the a v e r a g e spends in the
system

1-1 6 arrival
time au
average one (i)
22.4 minutes)
T- (üi) the (ü) O.79
1.07 x 104 Ans: (i)
12.4 minutes,
 14.42

2.
Applied Operations Research for 14.43
Consider a
single
server queu
N e U e n gn e o r y

exponential service times. ueueing system with Manaae sp P,[1-pN-s+l -(1 -p)(N-s+ Dp-]x Po
units per hour, the
Suppose the mean arrival Poisson input La s!(1 -p)
maximum permissible expected service time is 0.25 rate is 3
calling (s-n)(sp)
expected number in the system.units in the system is two.honrS d n
the L=L + S-Po n=0 n!
3. At Calculate the
Calcul:
a
railway station, only Ans: 081)
railway yard sufficient only for train is handled at a
is
one
where =a (1-PE)
given signal to two trains to
wait
time Th
leave the station.
average rate of 6 per hour and Trains arrive at while the other i
the
average 12 per hour. railway station can handlestation at an W W-
them on an
service distribution, findAssuming Poisson arrivals and
the steady state exponential for
number of trains in the two barbers and three chairs
-

Also find the probabilities of various shop has

A barber
system. xample23
a new
train coming into the average waiting time of arrive in Poisson fashion at
a
yard. customers. Assume that the customers
according
Ans: P 0.27, Po =
0.53, P2 Hint: N=3 rate of 5 per hour and that
each barber services
customers
if a
=
0.13, Pz =
0.07, and w, = distribution with mean of
15 minutes. Further
minutes] 3.8 to an exponential
chairs in the shop, he will
Customer arrives and there a r e no empty
14.9. Model IV What is the
(M leave. What is the probability that
the shop is empty ?
M/S): FCFSIN) expected number of customers in the shop?
This model is
essentially the
maximum number of customers in thesame as mode II except that the
=012
customer/minute

SIS =
no.
of channels). Therefore system limited to N, where N >
is oLution: Here S 2 , N=3,

5 perminute
n :0 nzN
0snSN
-

and
Pn
= nuOsnss
cus Sn SN Po
n= 0

Po 0Sn ss
PR
Po s S n<N

Po [Ans]
= 0.28
n=c 901
 14.44 Applied Operations Research for
Managem
x0.28, 0sn <2
P
026 2n-2 2!
x
0.28, 2 n
s3
(1.25" x 0.28, 0sn< 2
n

2n-2 21 (1.25" x 0.28, 2 n3

c-n)
Ls = La+s-Po2 n!
n=0

=
(n-2) Pn + 2- Po 2-1-25
2 n
n=2 n=0

= P3 +2-3.2P0

C 2.2 (1.25)x 0.28+ 2-3.2 x 0.28

(1.25) x 0.28 2-3.2 x 0.28

4

= 1.226 customers [Ans]

EXERCISE
situation with three
Let there be an inspection
automobile
when a
that wait in such a way that
inspection stalls. Assume
c a r s
to it. 1ne
the head of the line pulls up
stall becomes vacant, the c a r at statio) a
accommodate waiting (seven in
at most 4 cars
station can car every
Poisson with a mean of one
one time. The arrival pattern is with mea"
The service time is exponential
hours.
minute during peak tem
the syste
the average number of customers in
6 minutes. Find
the average waiting time and the average numo
during peak hours,
enter h e station because of full capacity.
ner
that cannot
hour

Ans: L 3 . 0 9cars, L=6.06 cars, W, =

12.3 minutes
EXDected number of cars per hour that cannot enter the
30.3 cars per hour.)
station is
=