8/16/2019 Modul S20 Teknik Antrian
1/42
Waiting Line Models forService Improvement
Model Antrian untuk
Perbaikan Pelayanan
8/16/2019 Modul S20 Teknik Antrian
2/42
Elements Of
Waiting Line Analysis Queue
a single waiting line
Waiting line system
arrivals
servers
waiting line structures
8/16/2019 Modul S20 Teknik Antrian
3/42
Calling population – source of customers
– an infinite population assumes suc a largenum!er of customers tat it is always possi!lefor one more customer to arrive to !e served
–
a finite population consists of a counta!lenum!er of potential customers
Arrival rate" λ
– fre#uency of customer arrivals to system – typically follows $oisson distri!ution
8/16/2019 Modul S20 Teknik Antrian
4/42
Service time – often follows negative e%ponential distri!ution
– average service rate & µ
Arrival rate must !e less tan service rate
or system never clears out
'λ < µ)
8/16/2019 Modul S20 Teknik Antrian
5/42
Components Of A Queuing
System
Source of
customers Arrivals Server Served
customersWaiting Line
or #ueue
8/16/2019 Modul S20 Teknik Antrian
6/42
Queue (iscipline And Lengt
Queue discipline – order in wic customers are served
– first come" first served is most common
Lengt can !e infinite or finite
– infinite is most common – finite is limited !y some pysical
structure
8/16/2019 Modul S20 Teknik Antrian
7/42
)asic Waiting Line Structures
Channels are te num!er of parallel servers
Phases denote num!er of se#uentialservers te customer must go troug
8/16/2019 Modul S20 Teknik Antrian
8/42
Single*Cannel Structures
ServersWaiting line
Single*cannel" multiple pases
Waiting line Server
Single*cannel" single*pase
8/16/2019 Modul S20 Teknik Antrian
9/42
Multi*Cannel Structures
Servers
Servers
Waiting line
Multiple*cannel" single pase
Multiple*cannel" multiple*pase
Waiting line
8/16/2019 Modul S20 Teknik Antrian
10/42
Operating Caracteristics
+e matematics of #ueuing teory does notprovide optimal or !est solutions
Instead" operating caracteristics arecomputed tat descri!e system performance
Steady state provides te average value of
performance caracteristics tat te systemwill reac after a long time
8/16/2019 Modul S20 Teknik Antrian
11/42
Operating Caracteristics
Notation Description
L Average num!er of customers in te system
'waiting and !eing served,L# Average num!er of customers in te waiting line
W Average time a customer spends in te system'waiting and !eing served,
W# Average time a customer spends waiting in line
8/16/2019 Modul S20 Teknik Antrian
12/42
$- $ro!a!ility of no '.ero, customers in te system
$n $ro!a!ility of n customers in te system
ρ /tili.ation rate0 te proportion of time tesystem is in use
8/16/2019 Modul S20 Teknik Antrian
13/42
Cost 1elationsip In Waiting Line
Analysis
E % p e c t e d
c o s t s
Level of service
+otal cost
Servicecost
Waiting Costs
8/16/2019 Modul S20 Teknik Antrian
14/42
Waiting Line Analysis
And Quality +raditional view * te level of service
sould coincide wit te minimum point on
te total cost curve
+QM view * a!solute #uality service is temost cost*effective in te long run
8/16/2019 Modul S20 Teknik Antrian
15/42
Single*Cannel" Single*$ase
Models All assume $oisson arrival rate
2ariations
– e%ponential service times – general 'or un3nown, distri!ution of service times
– constant service times
– e%ponential service times wit finite #ueue lengt
– e%ponential service times wit finite callingpopulation
8/16/2019 Modul S20 Teknik Antrian
16/42
)asic Single*Server Model
Assumptions4 – $oisson arrival rate
– e%ponential service times – first*come" first*served #ueue discipline
– infinite #ueue lengt
– infinite calling population
λ & mean arrival rate
µ & mean service rate
8/16/2019 Modul S20 Teknik Antrian
17/42
5ormulas for Single*Server Model
$- &λ
µ'6 * ,
$n &λ
µ
n$-
&λ
µ' ,λ
µ'6 * ,
L# &
L &λ
µ − λ
λ2
µ(µ − λ)
$ro!a!ility tat no customersare in system
$ro!a!ility of e%actly n
customers in system
Average num!er ofcustomers in system
Average num!er ofcustomers in #ueue
n,'
8/16/2019 Modul S20 Teknik Antrian
18/42
1
(µ − λ)
W & &L
λ
W# &λ
µ(µ − λ)
λ µ
ρ &
1 − ρΙ & & & $- λ
µ'6 * ,
Average time customer
spends in system
Average time customerspends in #ueue
$ro!a!ility tat server is !usy" utili.ation factor
$ro!a!ility tat server isidle 7 customer can !e served
8/16/2019 Modul S20 Teknik Antrian
19/42
Single*Server E%ample
8iven4 λ & 9: per our" µ & ;- customers per our
$- &λ
µ
'6 * ,
L# &
L &λ
µ − λ
λ2
µ(µ − λ)
$ro!a!ility tat no customers
are in system
Average num!er ofcustomers in system
Average num!er ofcustomers in #ueue
& 6 * '9:
8/16/2019 Modul S20 Teknik Antrian
20/42
1
(µ − λ)W &
W# &λ
µ(µ − λ)
λ µ
ρ &
1 − ρI &
Average time customer spends in system
Average time customerspends in #ueue
$ro!a!ility tat server is !usy" utili.ation factor
$ro!a!ility tat server isidle 7 customer can !e served
& 6';-*9:, & -=6>? r & 6- min
& 6 * -=@- & -=9-
& 9:
8/16/2019 Modul S20 Teknik Antrian
21/42
Waiting Line Cost Analysis
Management wants to test two alternatives toreduce customer waiting time4
6= ire anoter employee to pac3 uppurcases
9= Open anoter cec3out counter
8/16/2019 Modul S20 Teknik Antrian
22/42
Alternative 6
E%tra employee costs B6- < wee3
Eac one*minute reduction in customer waiting time avoidsB? in lost sales
E%tra employee will increase service rate to :- customers perour
1ecompute operating caracteristics
W# & -=-;@ ours & 9=9 minutes" originally was @ minutes
@=-- * 9=9 & =? minutes =? % B?
8/16/2019 Modul S20 Teknik Antrian
23/42
Alternative II
Dew counter costs B>--- plus B9-- per wee3 for cec3er
Customers divide temselves !etween two cec3out lines
Arrival rate is reduced from λ & 9: to λ & 69
Service rate for eac cec3er is µ & ;- 1ecompute operating caracteristics
W# & -=-99 ours & 6=;; minutes" originally was @ minutes
@=-- * 6=;; & >=>? minutes
>=>? % B?
8/16/2019 Modul S20 Teknik Antrian
24/42
Constant Service +imes
Constant service times occur witmacinery and automated e#uipment
Constant service times are a specialcase of te single*server model wit
general or undefined service times
8/16/2019 Modul S20 Teknik Antrian
25/42
Operating Caracteristics 5or
Constant Service +imes
W# &L#
λ
λ
µ'6 * ,$- &
L# &λ2 σ2 + (λ / µ) 2
2 ( 1 − λ / µ )
L & L# λ
µ
$ro!a!ility tat no customersare in system
Average num!er ofcustomers in system
Average num!er ofcustomers in #ueue
Average time customerspends in #ueue
8/16/2019 Modul S20 Teknik Antrian
26/42
$ro!a!ility tat server is !usy" utili.ation factor
1
µW & W#
Average time customer spends in system
λ
µρ &
Wen service time is
constant and σ & -"formula can !e simplified
L# &λ2 σ2 + (λ / µ) 2
2 ( 1 − λ / µ )
λ2 0 + (λ / µ) 2
2 ( 1 − λ / µ )
&
& (λ / µ) 2
2 ( 1 − λ / µ )
λ 2
2 µ(µ − λ)&
8/16/2019 Modul S20 Teknik Antrian
27/42
Constant Service +ime E%ample
Automated car was wit service time & := min
Cars arrive at rate λ & 6-
8/16/2019 Modul S20 Teknik Antrian
28/42
5inite Queue Lengt
A pysical limit e%ists on lengt of #ueue
M & ma%imum num!er in #ueue
Service rate does not ave to e%ceed arrival rate to
o!tain steady*state conditions 'µ > λ,
$- &
L & λ / µ
1 − λ / µ
1 − λ / µ
1 − (λ / µ)M+1
$n & '$- , λµ' ,n for n ≤ M
'M 6)'λ / µ, M 6
6 * 'λ / µ ,M6
$ro!a!ility tat nocustomers are in system
$ro!a!ility of e%actly ncustomers in system
Average num!er ofcustomers in system
8/16/2019 Modul S20 Teknik Antrian
29/42
Average num!er of
customers in #ueue
Average time customer spends in system
Average time customerspends in #ueue
L
W & λ '6 * $M,
L# & λ '6* $M,
µ
L
1
µWW# &
Let $M & pro!a!ility a customer will not Foin te system
8/16/2019 Modul S20 Teknik Antrian
30/42
5inite Queue E%ampleQuic3 Lu!e as waiting space for only ; carsλ & 9-" µ & ;-" M & : cars '6 in service ; waiting,
$ro!a!ility tat no
cars are in system
$ro!a!ility of e%actlyn cars in system
Average num!er ofcars in system
$- &1 − λ / µ
1 − (λ / µ)M+1
6 * 9-
L & λ / µ
1 − λ / µ
'M 6)'λ / µ, M 6
6 * 'λ / µ ,M6
& 9-
8/16/2019 Modul S20 Teknik Antrian
31/42
Average num!er ofcars in #ueue
Average time carspends in system
Average time car
spends in #ueue
L
W & λ'6 * $M,
L# & λ '6* $M,
µ
1
µWW# &
L * 9-'6*-=-?>,
;-& 6=9: * & -=>9
6=9:
9- '6*-=-?>,& & -=>? ours
& :=-; min
1
;--=->? *& & -=-;; ours
& 9=-; min
8/16/2019 Modul S20 Teknik Antrian
32/42
5inite Calling $opulation
Arrivals originate from a finite 'counta!le, population
D & population si.e
L# & λ µ
λ D *
$n & $-λ
µ' ,
nDG
'D * n,G
were n & 6" 9" ===" D
'6* $-,
$ro!a!ility tat nocustomers are in system
$ro!a!ility of e%actly ncustomers in system
Average num!er ofcustomers in #ueue
$- &
1
(λ / µ)nDG
'D * n,GΣD
n & -
8/16/2019 Modul S20 Teknik Antrian
33/42
W & W# 1
µ
L# L & '6* $-,
W# & 'D * L, λ
L#
Average time customer spends in system
Average time customerspends in #ueue
Average num!er ofcustomers in system
8/16/2019 Modul S20 Teknik Antrian
34/42
5inite Calling $opHn E%ample
9- macines wic operate an average of 9-- rs!efore !rea3ing down λ & 6
8/16/2019 Modul S20 Teknik Antrian
35/42
L# &
-=-- -=9??@
-=--& 9- '6* -=>9,
Average num!er ofmacines in #ueue
W & W# 6
µ
L & L# '6*$-, & -=6> '6*-=>9, & -=9-
W# & 'D * L, λ
L#
Average time macinespends in system
Average time macinespends in #ueue
Average num!er of
macines in system
& -=6>
λ µ
λD '6* $-,
'9- * -=9-, -=--
-=6>& & 6=?:
6=?: 6
-=9?@& & =;; rs
8/16/2019 Modul S20 Teknik Antrian
36/42
Multiple*Cannel" Single*$ase Models
+wo or more independent servers serve a singlewaiting line
$oisson arrivals" e%ponential service" infinite callingpopulation
sµ J λ
$- &6
[ Σn & s * 6
n & -K
λ
µ' ,n6
nG
6
sG
λ
µ' ,s sµ
sµ * λ ' ,
8/16/2019 Modul S20 Teknik Antrian
37/42
L &
$n & $-"λ
µ' ,n6
sG sn*sfor n J s
$n & $-"λ
µ' ,n6
nG
$w & $-λ
µ' ,s6
sG
sµ
sµ * λ ' ,
λ
µ' ,λµλ µ' ,s
's * 6) G 'sµ * λ)2
$-
$ro!a!ility of e%actly ncustomers in system
Average num!er ofcustomers in system
$ro!a!ility an arrivingcustomer must wait
for n & s
8/16/2019 Modul S20 Teknik Antrian
38/42
λ
8/16/2019 Modul S20 Teknik Antrian
39/42
Multiple*Server E%ample
Customer service area
λ & 6- customers
8/16/2019 Modul S20 Teknik Antrian
40/42
L &λ
µ' ,λµλ µ' ,
s
's * 6) G 'sµ * λ)2 $-
Average num!er ofcustomers in system
'6-,':, '6-- r & ;> min
8/16/2019 Modul S20 Teknik Antrian
41/42
Average num!er ofcustomers in #ueue
Average time customerspends in #ueue
$w & $-λ
µ' ,s6 sµ
sµ * λ ' ,$ro!a!ility an arriving
customer must wait
W# & µ & ;= * 6-
8/16/2019 Modul S20 Teknik Antrian
42/42
Improving Service
Add a :t server to improve service
1ecompute operating caracteristics
$o & -=-?; pro! of no customersL & ;=- customers
W & -=;- our" 6@ min in service
L# & -= customers waiting
W# & -=- ours" ; min waiting" versus 96 earlier
$w & -=;6 pro! tat customer must wait