Waiting Lines and Queuing Models-

Numerical

Sachin Modgil IMI-K

 Introduction
A popular system classifies parallel-server queuing
models using the following notation in which three
features are identified: A / B / C.
• A represents the distribution of time between arrivals
• B the distribution of service times, and
• C the number of parallel servers (e.g., cashiers at a
supermarket).
A/B/C : Arrival distribution/Service time
distribution/Number of service channels open
The basic three-symbol is called as Kendall notation (D.G
Kendall developed this notation)
 Queuing Models
• M = exponential interarrival or service time distribution
(or the equivalent Poisson distribution of arrival or
service rate)
• D = deterministic or constant interarrival or service
time
• E k = Erlang distribution with shape parameter k (if k =
1, then Erlang is equivalent to exponential; if k = ∞
then Erlang is equivalent to deterministic)
• G = general distribution with mean and variance (e.g.,
normal, uniform, or any empirical distribution)
Queuing Models
 Single-Server Waiting Line (M/M/1)
• Assumptions
– Customers are patient (no balking, reneging, or
jockeying)
– Arrivals follow a Poisson distribution with a mean
arrival rate of . This means that the time between
successive customer arrivals follows an exponential
distribution with an average of 1/ 
– The service rate is described by a Poisson distribution
with a mean service rate of µ. This means that the
service time for one customer follows an exponential
distribution with an average of 1/µ
– The waiting line priority rule is first-come, first-served
– Infinite population
Formulas: Single-Server Case (M/M/1)

  lambda  mean arrival rate

  mu  mean service rate

p   average system utilizatio n

Note :    for system stability. If this is not the case,
an infinitly long line will eventually form.
 Formulas: Single-Server (M/M/1)


Ls   average number of customers in system
 
Lq  pLs  average number of customers in line
1
Ws   average time in system  including service
 
Wq  pWs  average time spent waiting
Pn  1  p  p n  probability of n customers in the system
at a given point in time
 Formulas: Single-Server M/M/1

The probability that the number of customers

in the system is greater than k
k 1
Pn  k  (λ /  )
 Example 1 M/M/1

• A help desk in the computer lab serves students on

a first-come, first served basis. On average, 15
students need help every hour. The help desk can
serve an average of 20 students per hour.
• Based on this description, we know:
– µ = 20 students/hour (average service time is 3
minutes)
–  = 15 students/hour (average time between
student arrivals is 4 minutes)
 Example 1 M/M/1

Calculate
• (a) The average utilization of the help desk server
• (b) The average number of students in the system
• (c) The average number of students waiting in line
• (d) The average time a student spends in the system
• (e) The average time a student spends waiting in line
• (f) The probability of having more than 4 students in
the system
 a) Average Utilization

 15
p   0.75 or 75%
 20
(b) and (c)Average Number of Students
in the System, and in Line

 15
Ls    3 students
   20  15

Lq  pLs  0.75 3  2.25 students

(d) and (e)Average Time in the System & in Line

1 1
Ws    0.2 hours
   20  15
or 12 minutes

Wq  pWs  0.75 0.2   0.15 hours

or 9 minutes
 (f) Probability of having more than 4
Students in the system
4
P 1  Pn
0
4
P 1  0
(1  p ) P n

 1  0.25(1  0.75  0.752  0.753  0.754 )

 1  0.7626
 0.2374
or ( / u ) k 1
(0.75)5 23.74 percent) chance of having more than four
students in the system
 Example 2
• The SRC is concerned with its waiting line system.
Currently, the SRC uses a single-server, single-line,
single-phase system when processing license renewals.
Based on historical evidence, the average number of
customers arriving per hour is 9 and is described by a
Poisson distribution. The service rate is 12 customers
per hour with the service times following an
exponential distribution. The customers are patient
and come from an infinite population. The manager of
the SRC would like you to calculate the operational
characteristics of the waiting line system.
 Example 2
• (a) What is the average system utilization?
• (b) What is the average number of customers in the
system?
• (c) What is the average number of customers waiting
in line?
• (d) What is the average time a customer spends in the
system?
• (e) What is the average time a customer spends
waiting in line?
 Example 2
(a) Average utilization is 0.75, or 75 percent.
p = λ/μ
= 9/12 = 0.75
(b) Average number of customers in the system is 3.


L
 
9
L
12  9
= 9/3 = 3
 Example 2
(c) Average number of customers waiting in line is 2.25.
Lq = p * L
= 0.75 * 3
= 2.25
(d) Average time a customer spends in the system is 0.33
hours, or 20 minutes.
1
W 
 
1
W   0.33hours
12  9
 Example 2
(e) Average time a customer spends waiting in line is
0.25 hours, or 15 minutes.
Wq= p *W
= 0.75 *0.33= 0.25 hours
 Example 2a
λ= 24 customers per hour arrive at checkout counter
μ= 30 customers per hour can be checked out
(a) What is the average system utilization?
(b) What is the average number of customers in the
system?
(c) What is the average number of customers waiting in
line?
(d) What is the average time a customer spends in the
system?
(e) What is the average time a customer spends waiting in
line?
Example 2a
Example 2a
Example 2a
 Example 3
Naved is running a service center in New Town. His
mechanic, Romita , is able to install new mufflers at an
average rate of 3 per hour, or about 1 every 20 minutes.
Customers needing this service arrive at the service
center on the average of 2 per hour. Naved, the shop
owner, studied queuing models in an MBA program and
want to calculate the numerical values of the preceding
operating characteristics.
 Example 3
Given
λ = 2 cars arriving per hour
μ=3 cars serviced per hour
 Example 3
Given
λ = 2cars arriving per hour
μ=3 cars serviced per hour
Example 3
 Example 3- Cost of Waiting
Now that the characteristics of the queuing system have
been computed, Naved decides to do an economic
analysis of their impact. The waiting line model was
valuable in predicting potential waiting times, queue
lengths, idle times, and so on. But it did not identify
optimal decisions or consider cost factors. As stated
earlier, the solution to a queuing problem may require
management to make a trade-off between the increased
cost of providing better service and the decreased
waiting costs derived from providing that service. These
two costs are called the waiting cost and the service cost.
Cost of Waiting
 Example 3- Cost of Waiting
Total service cost = (Number of channels)(Cost per
channel)
Total service cost= m * Cs

Where
m= number of channels
Cs=service cost (labor cost) of each channel
 Example 3- Cost of Waiting
The waiting cost when the waiting time cost is based on
time in the system is
Total waiting cost = (Total time spent waiting by all
arrivals)(Cost of waiting)
= (Number of arrivals)(Average wait per arrival) *Cw
Total waiting cost = ( λ*W)*Cw
If the waiting time cost is based on time in the queue, this
becomes
Total waiting cost = ( λ*Wq)*Cw
 Example 3- Cost of Waiting
Total cost = Total service cost + Total waiting cost

= m * Cs + ( λ*W)*Cw
When the waiting cost is based on time in the queue, the
total cost is
= m * Cs + ( λ*Wq)*Cw
 Example 3- Cost of Waiting
Naved estimates that the cost of customer waiting time, in
terms of customer dissatisfaction and lost goodwill, is Rs.
50 per hour of time spent waiting in line. (After customers’
cars are actually being serviced on the rack, customers
don’t seem to mind waiting). Because on the average a car
has a 2/3 hour wait and there are approximately 16 cars
serviced per day (2 per hour times 8 working hours per
day), the total number of hours that customers spend
waiting for mufflers to be installed each day is 2/3 *16 or
32/3 or 10 2/3 hours or 10 hour 40 minutes.
 Example 3- Cost of Waiting
Total daily waiting cost = (8 hours per day)λ* Wq* Cw
= 8* (2) *(2/3)*(50)
= Rs. 533
The only other cost that Naved can identify in this queuing
situation is the pay rate of Romita Das, the mechanic. Das is
paid Rs.15 per hour

Total daily service cost = (8 hours per day)*m *Cs = 8* (1)

*(15) = Rs. 120
Total daily cost of the queuing system = 533.33 + 120 =
653.33
 Example 3- Cost of Waiting
Now comes a decision. Naved finds out through the muffler
business grapevine that the Rusty Muffler, a cross-town
competitor, employs a mechanic named Rajesh Khanna
who can efficiently install new mufflers at the rate of 4 per
hour. Naved contacts Rajesh and inquires as to his interest
in switching employers. Rajesh says that he would consider
leaving the Rusty Muffler but only if he were paid a Rs. 20
per hour salary. Naved, being a crafty businessman, decides
that it may be worthwhile to fire Romita and replace him
with the speedier but more expensive Rajesh.
 Example 4
Given
λ = 2 cars arriving per hour
μ= 4 cars serviced per hour
 Example 4
Given
λ = 2 cars arriving per hour
μ= 4 cars serviced per hour
Example 4
 Example 4

• It is quite evident that Rajesh’s speed will

result in considerably shorter queues and
waiting times. For example, a customer would
now spend an average of 1/2 hour in the
system and 1/4 hour waiting in the queue, as
opposed to 1 hour in the system and 2/3 hour
in the queue with Romita as mechanic. The
total daily waiting time cost with Rajesh as the
mechanic will be
 Example 4

Total daily cost = 8 hours per day * λ*Wq*Cw

= 8 *2*(1/4)*50
= Rs. 200 per day
Notice that the total time spent waiting for the 16 customers
per day is now

16 cars per day x (1/4hour per car) = 4 hours

instead of 10.67 hours with Romita. Thus, the waiting is much

less than half of what it was, even though the service rate
only changed from 3 per hour to 4 per hour.
 Example 4

The service cost will go up due to the higher

salary, but the overall cost will decrease, as we
see here:

Service cost of Rajesh = 8 hours/day x Rs. 20/hr

= Rs. 160
Total expected cost = Waiting cost + Service cost
= Rs. 200 + Rs. 160
= Rs. 360 per day
 Example 4

Because the total daily expected cost with

Romita as mechanic was Rs. 653.33, Naved may
very well decide to hire Rajesh and reduce costs
by Rs. 653.33 – Rs. 360 = Rs. 293.33 per day.
 Example 5
Lake Travis has one launching ramp near the dam
for people who trailer their small boats to the
recreational site. A study of cars arriving with
boats in tow indicates a Poisson distribution with a
mean rate of 6 boats per hour during the morning
launch. A test of the data collected on launch
times suggests that an exponential distribution
with a mean of 6 minutes per boat (equivalent
service rate μ= 10 boats launched per hour) is a
good fit. Calculate the all system characteristics.
Example 5
Example 5
 Cost of Waiting
Admin of Lake estimates that the cost of customer
waiting time, in terms of customer dissatisfaction
and lost goodwill, is Rs. 20 per hour of time spent
waiting in line. (After customers’ boats are being
prepared for launch, customers don’t seem to
mind waiting). Because on the average a customer
has a 0.15 hour wait and there are approximately
12 boats launched per day (1.5 per hour times 8
working hours per day), the total number of hours
that customers spend waiting for boats to be
launched each day is 1/4 *12 or 12/4 or 3 hours.
 Cost of Waiting
Total daily waiting cost = (8 hours per day)λ* Wq* Cw
=

Admin can identify in this queuing situation is the pay

rate of boat mechanic. He is paid Rs. 20 per hour

Total daily service cost = (8 hours per day)*m *Cs =

Total daily cost of the queuing system

Multichannel Queuing Model
Multichannel Queuing Model
Multichannel Queuing Model
 Multichannel Queuing Model
A third option is now explored. Naved finds that at
minimal after-tax cost he can open a second garage bay
in which mufflers can be installed. Instead of firing his
first mechanic, Romita, he would hire a second worker.
The new mechanic would be expected to install
mufflers at the same rate as Romita—about μ =3 per
hour. Customers, who would still arrive at the rate of λ
= 2 per hour, would wait in a single line until one of the
two mechanics is free. To find out how this option
compares with the old single-channel waiting line
system, Naved computes several operating
characteristics for the m=2 channel system:
Multichannel Queuing Model
Multichannel Queuing Model
Multichannel Queuing Model
 Multichannel Queuing Model
To complete his economic analysis, Naved assumes that
the second mechanic would be paid
the same as the current one, Romita, namely, Rs.15 per
hour. The daily waiting cost now will be

Total daily waiting cost = (8 hours per day) λ*WqCw =

(8)(2)(0.0415)(Rs. 50) = Rs. 33.20

Total daily service cost = (8 hours per day)*m*Cs =

(8)2(Rs.15) = Rs. 240

Total daily cost of the queuing system = Rs. 33.20 + Rs. 240 =
Rs. 273.20
 Multichannel Queuing Model
As you recall, total cost with just Romita as mechanic was
found to be Rs. 653 per day. Cost with just Rajesh was just
Rs. 360. Opening a second service bay will save about Rs.
380 per day compared to the current system, and it will save
about Rs. 87 per day compared to the system with the faster
mechanic. Thus, because the after-tax cost of a second bay
is very low, Naved decision is to open a second service bay
and hire a second worker who is paid the same as Romita.
This may have additional benefits because word may spread
about the very short waits at Naved’s Muffler Shop, and this
may increase the number of customers who choose to use
Naved’s.