Queueing

Queuing Theory Customers arrive λ = 15 persons/ho ur every 4 min.

Casher processes 3 min./perso n µ = 20 persons/hour
1. Queuing Theory Regular Arrival & Regular Service
2. Poisson Arrival and Exponential Service
3. M/M/1/∞
4. M/M/1/K (Queue with Truncation) No Delay
Random Arrival & Random Service
5. M/M/c (M/M/c/∞)
Assignments

λ < µ &Random

1 2

1. Queuing Theory Notation by Kendall (1953)

…... Service
Demand
Demand Pattern/ Service Number of Servers/ Capacity/ Population size
Queue [Arrival Pattern] Pattern/ [Service Units] (Omit)
Customers: people, telephone call, …
arrive for service, wait for service if it is not immediate, and Demand and Service Patterns
having waited for service, leave the system after being served.
Customers arrive in time and served at a constant rate
Service Providers: of arriving intervals and service duration
Provide service, treating customers in fair.
Service time depends on the skill and capability
Ex.)
Importance of queuing theory in practice
Costs of providing the service Costs with waiting for the service
Fast Expensive Minimum
Slow Less expensive Longer queues
3 4
 Poisson
Queue Discipline 2. M/M/1 (λ∆t ) k e −λ∆t
• FIFO Poisson Arrival and Exponential Service P(k ) =
k!
λ: the mean arrival rate of the demand
µ: the mean service rate of the system
n: the queue length
• LIFO
Probability of the queue length at t+∆t
n ≥1
• SIRO
Pn ( t + ∆ t ) =
• PRI
Pn (t + ∆t ) − Pn (t )
=
∆t
n −1 n n +1 µn : as a system
with m serviceunits
not for each unit
(1 − λn ∆t )(1 − µ n ∆t )= 1 − λn ∆t − µ n ∆t + λn µ n ∆t 2 ≅
5 6

M/M/1 n=0
λ: the mean arrival rate of the demand n − 1 n n +1 P0 (t + ∆t ) = (1 − λ0 ∆t ) P0 (t ) + µ1∆tP1 (t )
µ: the mean service rate of the system
n: the queue length P0 ( t + ∆ t ) − P0 ( t )
=
∆t → 0 ∆t
d
Pn (t ) =
dt 0 1
dPn (t )
Steady State =0
dt Steady State
µ∆t
(1 − λ∆t )

Especially λ0 = λ1 = λ2 = ... = λ µ1 = µ 2 = ... = µ Especially λ0 = λ1 = λ2 = ... = λ µ1 = µ 2 = ... = µ

ρ＝λ／µ: Facility utilization

7 8
 λ0 Especially λ0 = λ1 = λ2 = ... = λ µ1 = µ 2 = ... = µ
P1 = P0
µ1
λ0 ρ=
λ
λ + µ1 λ λ + µ1 λ0 λ P1 = P0
P2 = 1 P1 − 0 P0 = 1 P0 − 0 P0 = µ1 µ
µ2 µ2 µ 2 µ1 µ2
λ1λ0 λ
P2 = P0 = 1 P1
λ + µ2 λ λ + µ 2 λ1λ0 λ λ µ 2 µ1 µ2
P3 = 2 P2 − 1 P1 = 2 P0 − 1 0 P0 =
µ3 µ3 µ3 µ 2 µ1 µ3 µ1
λ2 λ1λ0 λ
n −1 P3 = P0 = 2 P2
∏λ P0 = µ3 µ 2 µ1 µ3
λn −1λn − 2 ......λ1λ0 i
Pn = P0 = i =0
P0 =
µ n µ n −1.......µ 2 µ1 n
λn −1λn − 2 ......λ1λ0 λ
∏µ i Pn = P0 = n −1 Pn −1
i =1
µ n µ n −1.......µ 2 µ1 µn
n −1 ∞
∞ ∞ ∏λ i
∑P n =1
∑ Pn = 1, P0 + P0 ∑ i =0
n
=1 n =0
n =0 n =1
∏µ
i =1
i

9 10

Ex. 3.18 Landing of Airplanes

M/M/1 Treating 2 controllers as one system Expected Queue Length
λ 40
λ0 = λ1 = λ2 = ... = ρ= = = 1.3333 ∞
µ 30 n : number in system L = E (n) = ∑ nPn m
µ1 = µ2 = µn = m : number waiting
n =1
∞
Lq = E (m) = ∑ mPm
P1 = P2 = m =1

Time in System (ν)= Waiting (ｗ)+ in Service (s)

P3 = P4 =

4
ρ Relatioship between L and Lq
(1 + ) P0 = 1 (1 + 3 )P = 1
1 1 4 0
1− ρ 1− ×
2 2 3
L ≠ Lq + ρ ?
Probability of airplane immediately landing Very useful when difficult check M/M/1/K, M/M/c
11 to estimate L/Lq 12
M/M/1 (M/M/1/∞) • Delay in System
υ = w + s (Time in System = Waiting + Service )
For simplicity
P0 =
1
=
1
= 1− ρ L = E (n) = λE (υ )
λ0 = λ1 = λ2 ...... = λ , ∞
ρ λ
1+ ∑ ρ n
1+ ρ= <1
µ1 = µ 2 .......µ n = µ n =1 1− ρ µ E(υ) =

Pn = ρ n P0 = ρ n (1 − ρ ) (n = 0,1,2....) m=n−1
E(w) =
• Expected Queue Length in the system
directly E(w)=
L=

E(m)
E(m | m > 0) = =
P(m > 0)
E(m) E(m)
Lq = = = 2 =
1− (1− ρ) − (1− ρ)ρ ρ
13 14

Ex. 3.18 Landing of Airplanes

• Distribution of Waiting Time: T M/M/1 Treating 2 controllers as one system
λ 40
M/M/1 n customers in the system λ0 = λ1 = λ2 = ... = λ = 40 /hour ρ= = = 1.3333
µ 30
T = T1 + T2 + ........ + Tn −1 + Tn
f Ti (t) = µe -µt
µ1 = µ = 30 / hour µ2 = 2µ = 60 µn = 2µ = 60 (n ≥ 2)
µ n t n -1e - µt t µ nτ n -1e - µτ λ0 λ λ1 λ ρ ρ2
f T| N (t) = P(T ≤ t | N = n - 1) = FT| N (t) = ∫ dτ P1 = P0 = P0 = ρP0 P2 = P1 = P1 = P1 = P0
(n − 1)! 0 (n − 1)! µ1 µ µ2 2µ 2 2
Steady State P0 = 0.2 P1 = 0.267 P2 = 0.178
∞
FT (t) = P(T ≤ t ) = ∑ p × P(T ≤ t | N = n - 1)
*
n -1 Probability of airplane immediately landing
n =1
∞ t µτ e n n -1 - µτ ∞ t µτ en n -1 - µτ P 0 + P1 = 0.2 + 0.267 = 0.467
= ∑ (1 − ρ ) ρ n −1 ∫ dτ = (1 − ρ )∑ ρ n −1 ∫ dτ
n =1
0 (n − 1)! n =1
0 (n − 1)! Expected Number of Planes in the System
n n -1 - µt
dFT (t) ∞
µ t e ∞
( µρt ) n -1
ρ c +1
f T (t) = = (1 − ρ )∑ ρ n −1 = µ (1 − ρ )e - µt ∑ Lq = P0 =
 ρ
2
dt n =1 ( n − 1)! n =1 ( n − 1)! c!c1 − 
∞
( µρt ) r  c
= µ (1 − ρ )e - µt ∑ =µ (1 − ρ )e - µt e - µρt = µ (1 − ρ )e - µ (1− ρ ) t
r =0 r!

∴ E(T) =
1
= E ( w)
Waiting time is L=
µ (1 − ρ ) exponentially distributed. 15 16
 n P(n) nP(n) (n-c)P(n)
ρn ρ ρ ρ ρ ρ ρ 2−ρ
P0 = ρP0 , ρP0 , P3 = P2 = ( ) ρP0 , P4 = P3 = ( ) ρP0 ,.....
0 0.2 0
P1 = P2 = P1 = 2 3
P0 = 1 0.266667 0.266667
n! 2 2 2 2 2 2 2+ ρ 2
3
0.177778
0.118519
0.355556
0.355556
0
0.118519
∞
ρ ρ ρ
4 0.079012 0.316049 0.158025
ρ c +1
L= ∑ nP = 1 × ρ P0 + 2 × ρ P0 + 3 × ( ) 2 ρ P0 + 4 × ( ) 3 ρ P0 + ...
5 0.052675 0.263374 0.158025
n Lq = P0 = 6 0.035117 0.2107 0.140466
n =1 2 2 2  ρ
2
0.023411 0.163877 0.117055
c!c1 − 
7

=
8 0.015607 0.124859 0.093644
 c 9 0.010405 0.093644 0.072834
10 0.006937 0.069366 0.055493

2−ρ ρ2
11 0.004624 0.050868 0.04162

ρ Lq + ρ = ρ +ρ =
12 0.003083 0.036995 0.030829

L= 2 + ρ 0  ρ 2
13 0.002055 0.026719 0.022608

41 − 
14 0.00137 0.019183 0.016442
2
 2
15 0.000913 0.013702 0.011875
ρ 16 0.000609 0.009744 0.008526
(1 − )L = 17 0.000406 0.006902 0.00609
2 18
19
0.000271
0.00018
0.004872
0.003428
0.00433
0.003067
1 0.00012 0.002406 0.002165
L = ρ P0
20
1 =
L = ρ P0 ρ
0.00008 0.00168 0.00152
=
21

ρ (1 −
0.00005 0.00118 0.00107
)2
22

(1 − )2
23 0.00004 0.00082 0.00075
2 24 0.00002 0.00057 0.00052
2 25 0.00002 0.00040 0.00036
26 0.00001 0.00027 0.00025
ρ c +1 ρ2 27 0.00001 0.00019 0.00018
L= Lq = P0 = ρP0 28 0.00000 0.00013 0.00012
 ρ  ρ
2 2

c!c1 −  41 − 
29 0.00000 0.00009 0.00008
0.00000 0.00006 0.00006
 c  2
30
17 1.000 2.400 1.067 18

∞)
3. M/M/c (M/M/c/∞ ∞

1≤ n < c ∑ Pn = 1
λn = λ n=0

nµ (1 ≤ n < c) µ n = nµ c
ρ n ∞
ρ n
ρ λ
µn =  λn = λ ∑ n!
P0 + ∑ c n−c
c!
P0 = 1 ( =
c cµ
< 1)
 cµ (n ≥ c) n=0 n = c +1

M/M/1 with λn and µn n≥c

∞
ρ n
1  ρ c +1 ρ c+2 ρn 
µ n = cµ ∑ c n − c c!
=  + 2
+ ...... + n−c
+ .... 
Pn = n = c +1 c!  c c c 
λn = λ ρ
ρc ρ ρ 2
ρ n−c
 ρc c
P0 = =  + 2 + ...... + n − c + ....  =
µ: the mean service rate of one server c !  c c c  c ! 1−
ρ
ρ =λ/µ ρ c
λ / cµ = < 1
1≤ n ≤ c c ρ
  ∴ P0 =
λ λ ......λ1λ0 λλ ......λλ  c ρn ρc 
Pn = n −1 n − 2 P0 = P0 = ∑ + c  P0 = 1
µ n µ n −1.......µ 2 µ1 (nµ )(n − 1) µ .....(2 µ ) µ  n=0 n ! c ! ρ 
 1− 
n≥c  c 
λn −1λn − 2 ......λ1λ0 λλ ......λλ
Pn = P = P =
µ n µ n −1.......µ 2 µ1 0 (cµ )(cµ )..(cµ )(c − 1) µ.......(2 µ ) µ 0 19 20
1≤ n ≤ c Ex. 3.18 Landing of Airplanes λ 40
ρ n
ρ×ρ n −1
ρ ρ= = = 1.3333
Pn = P0 = P0 = Pn −1 M/M/2 λ=40 µ=30 µ 30
n! n × (n − 1)! n 1 ≤ n ≤ c Pn = ρ Pn −1
n
n≥c
P3 = P4 = P5 =

n≥c
ρn ρ × ρ n −1 ρ
Pn = n −c
P0 = n −1− c
P0 = Pn −1
c c! c×c c! c

Probability of airplane immediately landing

21 22

Ex. 3.20 Waste-treatment process Expected Queue Length: Lq

λ = 10 veh/hr µ = 5 veh/hr n ≥ c
M/M/m ∞ ∞
ρn ∞
ρ r+c
 nµ = 5n n≤m λ λ 10
Lq = ∑ (n − c) P n = ∑ (n − c)
c n − c c!
P0 = ∑r c r c!
P0
µn =  ρ =λ/µ =2 < 1 ∴m > = =2 n = c +1 n = c +1 r =1
mµ µ 5
mµ = 5m n≥m
1≤ n ≤ m ρ2 ρ3 ρ4 ρ5
P0 + ρP0 + P0 + P+ P0 + P0 + .... = 1
2 6 18 54
ρ
n≥m Lq =
c
ρ
ρ ρc ρ ρ2 ρ3 ρc c
(1 − ) L q = P0 ( + + + ...... + ...) = P0
m=3 c c! c c2 c3 c! 1 − ρ
c
ρ
ρc c
L q = P0 =
ρ
2
c! 
1 − 
 c 

23 24
Expected Number in the System: L Ex. 3.20 Waste-treatment process
∞ c
ρ n ∞
ρ n
M/M/m λ = 10 veh/hr µ = 5 veh/hr
L= ∑ nP n = ∑n P0 + ∑ n n−c
P0
 nµ = 5n n≤m λ λ 10
n =1 n =1 n! n = c +1 c c! µn =  ρ =λ/µ =2 < 1 ∴m > = =2
mµ µ 5
c
ρn ∞
ρ r+c mµ = 5m n≥m
= ∑
n =1
n
n!
P0 + ∑ ( r + c )
r =1 c r c!
P0 1≤ n ≤ m P0 =
1
22 1
= 0.111
ρ [1 + 2 + ]
Pn = Pn −1 2 1− 2
c
ρ n
ρ c ∞
ρ r n
= Lq + ∑ n P0 + c ∑ P0 n≥m
2
3
 2 
n! c! cr ρ P1 = P0 = 0.222  P4 = P3 = 0.099 
n =1 r =1
ρ Pn = Pn −1 1

3

c
ρn ρ c P0 m 2  2 
= Lq + ∑ n P0 + c c L = Lq + ρ ? P2 = P1 = 0.222  P5 = P4 = 0.066
n! c! ρ m=3
ρ ρ2
2 
3 
n =1
(1 − ) P1 = ρP0 P2 = P1 = P0 2  2 
P6 = P5 = 0.044 
c 2 2 P3 = P2 = 0.148
ρ ρ 3 3  3 
υ = w+s (Time in System = Waiting + Service ) P3 = P2 = P0
3 6
ρ ρ4
E (υ ) = L / λ P4 = P3 =
E(w) = Lq / λ 1 P0
E (υ ) = E(w) + ? 3
ρ
185
ρ
µ P5 = P4 = P0
3 54
25 26

4.M/M/1/K (Queue with Truncation)

K K
∞ n −1 n −1 n n +1
Lq = ∑ (n − R ) P = 0 . 888 L= ∑ nP = 2 . 888 = L q + ρ • n<K: same as M/M/1/∞
∏λ
n n
n = R +1 n =1
λn −1λn − 2 ......λ1λ0 i
Pn = P0 = i =0
P0
µ n µ n −1.......µ 2 µ1 n
m P(0) ｍ χm Lq 80m 100Lq T
i =1
∏ µi
1 • n=K
2 K −1 K
3 0.111 15 0.6667 0.8880 240 88.800 328.8 PK (t + ∆t ) = λK −1∆tPK −1 (t ) + (1 − µ K ∆t ) PK (t )
PK (t + ∆t ) − PK (t )
4 0.130 20 0.5000 0.1733 320 17.333 337.3
= λK −1 PK −1 (t ) − µ K PK (t )
5 0.134 25 0.4000 0.0397 400 3.970 404.0
∆t λK −1∆t
d (1 − µ K ∆t )
PK (t ) =
dt
For steady state

27 28
 λ0 Especially λ0 = λ1 = λ2 = ... = λ µ1 = µ 2 = ... = µ
P1 = P0
µ1
n −1

λn −1λn − 2 ......λ1λ0 ∏λ i
λn −1
Pn = P0 = i =0
P0 = Pn −1 (1 < n < K ) 1 1
µ n µ n −1.......µ 2 µ1 n
µn P0 = =
∏ µii =1 K ∏λ
n −1

i
1+ D
1+ ∑ i =0

λK −1 λ λ ......λ1λ0
n

PK = PK −1 = K −1 K −2 P0 K
n =1
∏µ i

µK µK µK −1.......µ2 µ1 ∑
n=0
Pn = 1 i =1

P0 = D = ρ + ρ 2 + ρ 3 + ........ + ρ K
n −1
P0 + ρ P0 + ρ 2 P0 ...... + ρ K P0 = 1
K K ∏λ i
P0 (1 + ρ + ρ 2 + ...... + ρ K ) = 1 1
∑ Pn = 1, P0 + P0 ∑ i =0
n
=1 P0 =
1 + ρ + ρ 2 + ρ 3 + ........ + ρ K
n =0 n =1
∏µi =1
i

λ0 λ1λ0 λ2 λ1λ0 λ λ .....λ2 λ1λ0

D= + + + ........ + K −1 K − 2
µ1 µ 2 µ1 µ3 µ 2 µ1 µ K µ K −1....µ3 µ 2 µ1
29 30

Ex. 3.19 Earth Excavation Operation

Shovel
M/M/1/5 with λn and µ P0 =
Total Number of Trucks 5
λn = (5 − n)λ n = 0, 1, ..., 5 λ0
Fill Area
P1 = P0 =
λ = 7.5trucks/hr µ1
For a certain truck, it comes 7.5
µ = 30 trucks/hr λ1
times/hour to the shovel P2 = P1 =
µ2
Pn = ρ = λ / µ = 7.5 / 30 = 1 / 4 λ2
P3 = P2 =
P1 = P2 = P3 = µ3
λ3
P4 = P5 = P4 = P3 =
P6 = P7 = ... = 0 µ4

5λ 5 × 4λ2 5 × 4 × 3λ3 5 × 4 × 3 × 2λ4 5 × 4 × 3 × 2 × 1λ5 λ4

(1 + + + + + ) P0 = 1 P5 = P4 =
µ µ 2
µ 3
µ 4
µ 5 µ5

31 32
M/M/1/K (Queue with Truncation) • Average Number in System
P1 = ρP0 in case of
K 1− ρK −(1− ρ)KρK
(1− ρ)E(n) = P0ρ 1−(1+ K)ρK + KρK+1
λ0 = λ1 = λ2 ...... = λ , 1− ρ = P ρ
Pn = ρ n P0 (n < K ) 1− ρ
0

µ1 = µ 2 .......µ n = µ 1− (1+ K)ρ + Kρ

K K +1
1− (1+ K)ρ + Kρ K+1
K

PK = ρ PK −1 = ρ K P0 L = E(n) = P0ρ = ρ
(1− ρ)2 (1− ρ)(1− ρ K+1)
1 − ρ K −1 K
P0 =
1 − ρ K +1
ρ＝λ／µ Alternatively
K K K K
λ∆t d n d K
• Average Number in System L = E(n) = ∑nPn = ∑nρ n P0 = P0 ρ∑nρ n−1 = P0 ρ∑ (ρ ) = P0 ρ ∑ρ n
K K n=0 n=0 n=0 n=0 dρ dρ n=0
L = E(n) = ∑nPn = ∑nρnP0 = P0 (ρ + 2ρ2 +3ρ3 + 4ρ4 +......+ KρK )
n=1 n=1 = P0 ρ
d  1− ρ

K +1
 ( ) (
 = P0 ρ − (1+ K)ρ (1− ρ) − 1− ρ (−1)
K K +1
)
dρ  1− ρ  (1− ρ) 2
ρE(n) =
1− (1+ K)ρ K + Kρ K +1 1− (1+ K)ρ K + Kρ K +1
(1−ρ)E(n) = P0( ρ+ρ +ρ +ρ +.......+ρ −Kρ ) =
2 3 4 K K+1
= P0 ρ = ρ
(1− ρ)2 (1− ρ)(1− ρ K +1)
1−(1+ K)ρK + KρK+1
L = E(n) = P0ρ =
(1− ρ)2
34

• Waiting Number in System Ex. 3.19 Earth Excavation Operation

Shovel
K K K K
M/M/1/5 with λn and µ
Lq = ∑ ( n − 1) P
n =1
n = ∑ nP − ∑ P
n =1
n
n =1
n = ∑ nP
n =1
n − (1 − P0 )
Total Number of Trucks 5
λn = (5 − n)λ n = 0, 1, ..., 5 Fill Area
1− ρ ρ − ρ K +1 1− ρ K
= L − (1 − ) = L − = L − ρ λ = 7.5trucks/hr
1 − ρ K +1 1 − ρ K +1 1 − ρ K +1 For a certain truck, it comes 7.5
µ = 30 trucks/hr times/hour to the shovel
L = Lq + ρ ? λn −1
Pn = Pn −1 ρ = λ / µ = 7.5 / 30 = 1 / 4
• Expected Queue Length µn
L = E ( n ) = λ E (υ ) P0 = 1 /(1 + 4.0234) = 0.19907
∴ E (υ ) = E ( n ) / λ
1
Lq = E(m) = λE(w) E(w) = E (υ ) − E ( s ) = E (υ ) − ?? Efficiency of shovel
µ
Lq
∴ E(w) = Expected production per hour
λ

35 36
 Ex. 3.21 Storm
k: number of trucks is optimum or not ? M/M/1/1 λ=0.1/day µ=1/day ρ=0.1
Total cost per hour G( k )
E[TC( k )] = =
Expected production per hour Y(k)
G( k) = cost of truck + cost of shovel etc. = 14.8k + 35.6
Y(k) = (1 - P0 ( k))µTC

k P0(k) Y(k) E[TC(k)]

1 0.8 48 1.05
2 0.615 92.5 0.705
3 0.451 132 0.599
4 0.311 165 0.568
5 0.199 192 0.567
6 0.117 211.5 0.581
7 0.063 224.5 0.618

37 38

5. M/M/c/K Expected Queue Length: Lq

λ (1≤ n < K) nµ (1≤ n < c) n ≥ c
ρn ρ r+c
λn =  µn = 
K K K

0 (n ≥ K) cµ (c ≤ n ≤ K)
ρ＝λ／µ Lq = ∑ (n − c) P
n = c +1
n = ∑
n = c +1
(n − c)
c n − c c!
P0 = ∑r
r =1 c r c!
P0

1≤ n ≤ c ρc ρ ρ2 ρ3 ρK
= P0 ( +2 +3 + ...... + K )
λn −1λn − 2 ......λ1λ0 λλ ......λλ c! c c2 c3 cK
Pn = P0 = P0 =
µ n µ n −1.......µ 2 µ1 (nµ )(n − 1) µ .......( 2µ ) µ ρ ρc ρ2 ρ3 ρ4 ρ K +1
L q = P0 ( +2 +3 + ....... + K )
c≤n≤K c c! c2 c3 c4 c K +1
λ λ ......λ1λ0 λλ ......λλ ρ ρc ρ ρ2 ρ3 ρ K −c ρ K − c +1
Pn = n −1 n − 2 P0 = P0 = (1 − ) L q = P0 ( + + + ...... + − (K − c) )
µ n µ n −1.......µ 2 µ1 (cµ )..(cµ )(c − 1) µ .......(2 µ ) µ c c! c c2 c3 c K −c c K − c +1
ρ K −c
M/M/c/c ρ ρ c ρ 1− ( c ) ρ
P0 =
1 (1 − ) L q = P0 [ − ( K − c )( ) K − c ]
1≤ n ≤ c c c! c ρ c
c
ρ n K
ρn 1−
1+ ∑ + ∑c n−c ρ ρn c
n =1 n! n = c +1 c! Pn = Pn −1 = P0 ρ
1− ( ) K −c
n n!
ρc ρ c ρ
1 L q = P0 [ − ( K − c )( ) K −c ]
P0 = ρ ρ
c
ρn c! (1 − ) c 1− c
1+ ∑ c c
n =1 n! 40
Expected Number in the System: L M/M/c/c
K c
ρ n K
ρ n
1≤ n ≤ c
L= ∑ nP
n =1
n = ∑n
n =1 n!
P0 + ∑
n = c +1
n
c n−c
c!
P0
ρn ρρ n −1 ρ
Pn = P0 Pn = P0 = Pn −1
c
ρ n K −c
ρ r+c
r = n−c n! n(n − 1)! n
= ∑n n!
P0 + ∑ (r + c) c r c!
P0
P0 =
1
n =1 r =1 c
ρn
1+ ∑
c
ρn ρ c P0 K −c
ρr n =1 n!
= Lq + ∑ n P0 + c ∑ c c
ρn c
ρ n−1
n =1 n! c! r =1 cr K −c
L = E(n) = ∑ nPn = ∑ n P0 = P0 ρ ∑
ρ n=0 n=1 n! n=1 (n −1)!
1−  
c
ρ n
ρ P0
c +1
c  L = Lq + ρ ?
c
= Lq + ∑ n P0 + = ρ ∑ Pn−1 =
n ! c ! ρ
n =1
(1 − ) n=1
c
ρ c+1P0 ρ c+1P0
υ = w + s (Time in System = Waiting + Service ) Lq = 0 L=ρ− = Lq + ρ −
c! c!
E(w) = Lq / λ E (υ ) = L / λ E (υ ) = E(w) +
1
?
µ E(w) = 0 E (υ ) = L / λ
41 42

Especially  ρ 
M/M/1 M/M/c  c ρn ρc 
λ 0 = λ 1 = ... = λ
n =1 λn = λ P0 = 1 /  ∑ + c 
µ 1 = µ 2 = ... = µ  n = 0 n! c! 1 − ρ 
λ nµ (1 ≤ n < c)  
P1 = 0 P0 n =1 µn =   c 
µ1
λ  cµ (n ≥ c) ρn ρ
n >1 P1 = ρP0 where ρ = 1 ≤ n ≤ c Pn = P0 = Pn−1
µ n =1 n! n
(λ n + µ n ) λ
Pn +1 = Pn − n −1 Pn −1 n >1 λ
µ n +1 µ n +1 P1 = ρn ρ
Pn +1 = (1 + ρ ) Pn − ρPn −1 µ
P0 n≥c Pn = P0 = Pn−1
λ n −c
c c! c
Pn = n −1 Pn −1 (n = 1,2,....) 1< n < c
µn P0 = 1 − ρ ρ c +1
n −1
(λ + nµ ) λ Lq = P0
λn −1λn − 2 ......λ1λ0 ∏λ i Pn = ρPn −1 = ρ n P0 Pn +1 = Pn − Pn −1  ρ
2

Pn = P0 = i =0 nµ ( n + 1) µ c!c1 − 
µ n µ n −1.......µ 2 µ1 n
P0 ρ λ  c ρ
∏µ L= = n≥c
i
1− ρ µ − λ c
ρn ρ c P0
1
i =1

1ρ2 Pn +1 =
(λ + cµ )
Pn −
λ
Pn −1
L = Lq + ∑n n!
P0 + c
c!
c
ρ
P0 = = L = cµ cµ n =1
(1 − )
n −1
λ λ λ λ λλ q
1− ρ c
∞ ∏ λi 1 + 0 + 0 1 + 0 1 2 + ....
µ1 µ1µ 2 µ1µ 2 µ3 1 E(w)= Lq / λ
1+ ∑ i =0 E(υ ) = L / λ =
n µ (1 − ρ )
n =1
∏µ i ρ E(υ) = L/ λ
i =1 E(w)= Lq / λ =
µ(1− ρ ) 43 44
 Especially
M/M/1/K M/M/c/K M/M/c/c
λ 0 = λ 1 = ... = λ 1
n =1 P0 =
µ 1 = µ 2 = ... = µ c
ρn K
ρn 1
λ 1+ ∑ + ∑c P0 =
P1 = 0 P0 n =1 P1 = ρ P0 n! n−c
c! c
ρn
µ1 n =1 n = c +1
1+ ∑
1< n < K 1< n < K 1≤ n ≤ c n =1 n!
(λn + µ n ) λ n −1 Pn +1 = (1 + ρ ) Pn − ρ Pn −1 ρn ρ 1≤ n ≤ c
Pn +1 = Pn − P Pn = P0 = Pn −1
µ n +1 µ n + 1 n −1 n=K n! n
PK = ρ PK −1 c≤n≤K ρn ρ
n=K Pn = P0 = Pn −1
λ 1 − ρ ρn ρ n! n
PK = K −1 PK −1 P0 = Pn = P0 = Pn −1
µK 1 − ρ K +1
c n −c c! c Lq = 0
λ λ ......λ1λ0 Pn = ρPn −1 = ρ P0 n ρ K −c ρc
Pn = n −1 n − 2 1− ( ) L = ρ (1−
P0 ρ c
ρ ρ P0 )
µ n µ n −1.......µ 2 µ1 1−(1+K)ρK +KρK+1 L q = P0 [ c − ( K − c )( ) K −c ] c!
L= ρ ρ ρ
λ (1−ρ)(1−ρK+1) c! (1 − ) c 1− c
E(w)= 0
Pn = n −1 Pn −1 (n = 1,2,...K ) c
 ρ 
c K −c
µn ρ −Kρ +(K−1)ρ
K K+1 1−  
1 Lq = ρ
c
ρn ρ c + 1 P0  c 
P0 =
λ0 λ0 λ1 λ λ ...λ (1−ρ)(1−ρK+1)
L = Lq + ∑ n
n!
P0 +
c! ρ E(υ) = L/ λ
1+ + + .. + 0 1 K −1
n =1
(1 − )
µ1 µ1µ 2 µ1µ 2 ..µ K E(w)= Lq / λ
c
E(υ) = L/ λ E(w)= Lq / λ
45 E(υ) = L/ λ 46

Assignments Assignments
3.13 A parking space is provided for customers of a local post office. Suppose customers arrive according to a Poisson process
with a mean rate of 10 per hour; the time taken by each customer in the post office is exponentially distributed with a mean of 3.16 Figure 3.16 represents a multi-server operation where R parallel shovels are involved in a queuing system of hauling trucks.
five minutes. Assume that a customer will wait for a parking space if necessary.
(a) What percentage of the time would parking spaces be available to incoming customers?
(b) How would the answer to part (a) change if an additional parking is provided ? Consider two cases separately:
(i) The post office can only handle one customer at a time.
(ii) The post office can serve more than one customer simultaneously.
(c) Repeat part (a) if the service time is not exponentially distributed but has coefficient of variation of 60% and
a mean of five minutes.

Suppose that the arrival of a particular hauling truck (not in the queue) follows a Poisson process with mean rate λ, and the
service time of each shovel follows an exponential distribution with a mean shoveling time of 1/μ. Trucks are served on a first-
come-first-serve basis.
3.14 Consider a garage with capacity for four trucks. Suppose trucks arrive at garage according to a Poisson process with a (a) For a fleet of k trucks, derive the expressions for the probabilities of the queue length at the steady-state condition.
mean rate of one per day. The amount of time truck spends in the garage is exponentially distributed with a mean of two day. (b) Determine the production per hour if four shovels are serving seven trucks. The average cycle time of each shovel is 1.5
If an arriving truck finds the garage already fully occupied, it will go elsewhere. minutes and the average traveling time for each truck is four minutes. The capacity of each truck is 16 cu yd.
(c) Determine the optimum number of trucks required to work with three shovels, such that the cost per cu yd of earth moved it minimized.
(a) What fraction of the time is the garage fully occupied?
The average traveling time for each truck is minutes and the average loading time is three minutes. A shovel costs $100 per hour including
(b) Suppose each truck weight 50tons. What is the average load applied to the garage’s foundation over a long time period? operator and oiler, whereas a truck costs $40 per hour including a driver. The capacity of each truck is 16 cu yd.
(c) If the weight of each truck is normally distribute N(50,10) tons, and the weights between trucks are statistically
independent, what is the probability the total load from the trucks in the garage exceeds 200tons at a given time?

3.15 A left-turn bay is provided at a street intersection. The time until a gap long enough for a safe left turn follows an exponential 3.18 A solid waste treatment plant processes wastes at an average rate of 10 minutes per truck load with a c.o.v of 40 %.
distribution with a mean of 20 sec. Suppose the left –turn traffic arrives according to a Poisson process at a mean rate of two per minute. Suppose the truckloads arrive according to a Poisson process with a mean arrival rate of four per hour. Determine
(a) What should the size of the left-turn bay(in terms of the number of car spaces)be such that it will be adequate 96% of the time? (a) The proportion of time in which the treatment plant is idle.
(b) what is the expected waiting time before one can make a left turn at this intersection? (b) The mean queue length at the steady-state condition
(c) What is the probability that the waiting time will exceed on minute? (c) The steady-state probabilities for queue lengths of 1, 2, and 3 truck loads if the service follows a gamma distribution

47 48