You are on page 1of 11

13.

WAITING LINE MODELS


Introduction
 Waiting in queues (waiting lines) is a common, everyday occurrence
o E.g. shopping, movies, the bank
 Not only do people wait in lines, other aspects are also dependent on a
waiting system
o E.g. products at a production plant, machinery that must be serviced,
planes wait to take off
 Reduction of waiting time is very important, because every minute spent waiting
in line means a minute that is not bringing in revenue
 The improvement of service with respect to waiting time has also become
more important in recent years because of the increased emphasis on quality,
especially in service related operations
 Customers do not like to wait for service and often results in the customer
leaving and supporting a rival store/business
 More and more companies focus on reducing waiting time and provide faster
service by increasing their service capacity, which usually means adding more
servers
o E.g. tellers at the bank, mechanics at a car dealership, checkout clerks
at a grocery store
 However, increasing service capacity also increases the cost to the company
 Herein lies the basis of waiting line analysis: the trade-off between the cost of
improved service and the cost of making customers wait

 Like decision analysis, queuing analysis is a probabilistic form of analysis, not a


deterministic technique
 Thus, the results of queuing analysis (called operating characteristics) are
probabilistic
 These operating statistics (e.g. the average time a person must wait in line to
be served) are used by the manager of the operation containing the queue to
make decisions

 A number of different queuing models exist, but we will focus on 2:


 Single-server system
 Multiple-server system

1
13.1. Elements of Waiting Line Analysis
 Waiting lines form because people or things arrive at the servicing function, or
server, faster than they can be served
 Not necessarily because the service operation is understaffed or does not
have the overall capacity to handle the influx of customers - most businesses
and organizations have sufficient serving capacity available to handle their
customers in the long run
 Waiting lines result because customers do not arrive at a constant, evenly paced
rate, nor are they all served in an equal amount of time
 Customers arrive at random times, and the time required to serve them
individually is not the same
  a waiting line is continually increasing and decreasing in length (and is
sometimes empty), and it approaches an average rate of customer arrivals
and an average time to serve the customer in the long run
 Decisions about waiting lines and the management of waiting lines are based on
these averages for customer arrivals and service times

13.2. The Single-Server Waiting Line System


 A single server with a single waiting line is the simplest form of queuing system
 One checkout counter and one employee who operates the cash register at
the checkout counter
 The combination of the cash register and the operator is the server (or service
facility) in this queuing system
 The customers who line up at the counter to pay for their selections form the
waiting line (queue)

 Most important factors to consider in analyzing a single-server queuing system:


 The queue discipline (in what order customers are served)
 The nature of the calling population (where customers come from)
 The arrival rate (how often customers arrive at the queue)
 The service rate (how fast customers are served)

13.2.1. The Queue Discipline


 The queue discipline is the order in which waiting customers are served
 First-come, first-served: the first person in line in served first
 Most common type
 Last-in, first-out: the last part is on top of the stack and will be selected
first
 Commonly used by machine operators
 Random
 E.g. machine operator simply reaches into a box full of parts and
select one at random
 Prearranged schedule: according to a schedule regardless of the order of
arrival
 Such as patients at a doctor's or dentist's office or diners at a restau-
rant where reservations are required
 Alphabetically
 Customers are served according to their last names, such as at school
registration or at job interviews

2
13.2.2. The Calling Population (where customers come from)
 The calling population is the source of the customers
 It may he assumed to be infinite
 There is such a large number of possible customers in the area that the
number of potential customers is assumed to be infinite
 Some queuing systems have finite calling populations
 E.g. the repair garage of a trucking firm that has 20 trucks has a finite calling
population
 (The queue is the number of trucks waiting to be repaired)

13.2.3. The Arrival Rate


 The arrival rate is the rate at which customers arrive at the service facility during
a specified period of time
 This rate can be estimated from empirical data derived from studying the
system or a similar system, or it can be an average of these empirical data
 However, it would be impossible to determine the exact amount for each
hour
 Given these assumptions, it is further assumed that arrivals at a service facility
conform to some probability distribution
 Although arrivals could be described by any distribution, it has been
determined (through years of research and the practical experience of people
in the field of queuing) that the number of arrivals per unit of time at a service
facility can frequently be defined by a Poisson distribution
 Poisson distribution: a probability distribution that describes the
occurrence of a relatively rare event in a fixed period of time; often used to
define arrivals at a service facility in a queuing system

13.2.4. Service rate


 The service rate is the average number of customers who can be served during
a specified period of time
 For the first example (in the next section), 30 customers can be checked out
(served) in 1 hour
 Random variable
 Influences on service rate include:
 Different sizes of customer purchases
 The amount of change the cashier must count out
 Different forms of payment
 It is an average – not constant from one hour to the next
 To analyze a queuing system, both arrivals and service must be in compatible
units of measure, thus service time must be expressed as a service rate to
correspond with an arrival rate

3
13.2.5. The Single-Server Model
 The Fast Shop Market checkout counter is an example of a single-server queuing
system with the following characteristics:
1. An infinite calling population (potential customers)
2. A first-come, first-served queue discipline
3. Poisson arrival rate
4. Exponential service times
 Variables:
 λ – arrival rate (average number of arrivals per time period)
 µ - service rate (average number served per time period)
 λ < µ - customers must be served faster that they arrive

 Formulas for operating characteristics of a single-server model:

( µλ )
 Probability that no customers are in the queuing system 1: P0= 1−
n
λ
 Probability that n customers are in the queuing system: P =( ) . P
n 0
µ
n
λ λ
P =( ) . ( 1− )
n
µ µ
λ
 Average number of customers in the queuing system: L=
µ−λ
λ2
 Average number of customers in the waiting line: q
L =
µ(µ−λ)
1
 Average time a customer spends in the total queuing system: W =
µ− λ
L
W=
λ
 Average time a customer spends waiting in the queue to be served:
λ
W q=
µ( µ− λ)
λ
 Utilization factor2 (customer has to wait): U =
µ
 Probability that the server is idle (customer can be served): I =1−U
λ
I =1−
µ
 I =P0 – when there is no-one waiting in line, then the customer can be
served
1
Waiting in line as well as being served
2
Probability that server is busy

4
Example: Fast Shop Market
24 customers per hour arrive at the checkout counter, while 30 customers per hour
can be checked out. Compute the various operating characteristics.
λ
( )
 P0= 1− =0,2 probability of no customers∈the system
µ
 Pn  not calculated, because the estimated number of customers in the queue is unknown
λ
 L= =4 customers , on average ,∈thequeuing system
µ−λ
λ2
 Lq = =3,2 customers , on average,∈the waiting line
µ(µ−λ)
1
 W= =0,167 hour ( 10 minutes ) averagetime∈the system per customer
µ− λ
λ
 W q= =0,133 hour ( 8 minutes ) average time∈the waiting line per customer
µ( µ− λ)
λ
 U = =0,8 probability that the server will be busy ∧the customer must wait
µ
λ
 I =1− =0,2 probability that the server will be idle∧a customer can be served
µ

Hint: always used given values instead of calculated values (if possible)

Aspects of both the general model and the example


 The operating characteristics are averages, and they are assumed to be steady
state averages
 Steady state averages: constant average level that the system realizes after
a period of time
 For a queuing system, the steady state is represented by the average
operating statistics (determined over time)
 Utilization factor must be less than 1, implying that it is possible to serve all the
customers that arrive at the shop
 The server must be able to serve customers faster than they arrive, else the
line will grow to an infinite state, and the system will never reach a steady
state

13.2.6. The Effect of Operating Characteristics on Management Decisions


 24 customers come into the shop every hour
60
  2,5 customers arrive per minute
24 ( )
 Busy store!
 Waiting 8 minutes in line before being served and then

5
taking another 2 minutes to be served is too slow
 This problem can be solved by:
 Adding an employee
 Adding a new checkout counter

Original single-server
model

Alternative 1: Adding an Employee


 Cost: R1500 per week
 According to research for the shop, the store avoids
losing R750 per week in sales for every minute that
waiting time is reduced
 The store loses money if customers have to wait,
because they leave without buying anything, or if they
do buy, they do not return
 If a new employee is hired, customers can be served in
less time
 Service rate (the number of customers served
per time period) will increase
 µ = 30 becomes µ = 40 1 new employee
added
 If the arrival rate remains 24, then:
Operating characteristic µ = 30 µ = 40
Probability that no customers are in the queuing
0,2 0,4
system ( P0)
Average number of customers in the queuing
4 1,5
system (L)
Average number of customers in the waiting line ( Lq) 3,2 0,9
Average time a customer spends in the total queuing 0,167 hr 0,063 hr
system (W ) (10 min) (3,75 min)
Average time a customer spends waiting in the queue 0,133 hr 0,038 hr
to be served (W q ) (8 min) (2,25 min)
Utilization factor (U ) 0,8 0,6
Probability that the server is idle (I) 0,2 0,4

 The average waiting time per customer has been reduced from 8 minutes to 2,25
minute – a significant amount
 The savings (i.e. decrease in lost sales) is computed as:
 8 min−2,25 min ¿ 5,75 min
 5,75 min × R 750/min=R 4312,50
 The extra employee costs R1500 per week, so the total savings will be:

6
 R 4312,50−R 1500=R 2812,50 per week

 Note: increasing the service rate might, in the long run, increase future arrivals

Alternative 2: Adding a New Checkout Counter


 The total cost of the project would be R60 000, plus an
extra R2 000 per week for an additional cashier
 The checkout counter would be opposite the present
counter (so that the servers would have their backs to
each other in an enclosed counter area)
 There would be several display cases and racks
between the two lines so that customers waiting in line
would not move back and forth between the lines (called
jockeying) 1 new checkout counter
 Jockeying would invalidate the queuing formulas added
 It is assumed that the customers would divide themselves equally between the two
lines, so the arrival rate for each line would be half of the prior arrival rate for a single
checkout counter (λ = 12)
 If the service rate (µ) remains 30, then:
Operating characteristic λ = 24 λ = 12
Probability that no customers are in the queuing
0,2 0,6
system ( P0)
Average number of customers in the queuing
4 0,67
system (L)
Average number of customers in the waiting line ( Lq) 3,2 0,27
Average time a customer spends in the total queuing 0,167 hr 0,055 hr
system (W ) (10 min) (3,33 min)
Average time a customer spends waiting in the queue 0,133 hr 0,022 hr
to be served (W q ) (8 min) (1,33 min)
Utilization factor (U ) 0,8 0,4
Probability that the server is idle (I) 0,2 0,6

 If the same sales savings of R750 per week for each minute’s reduction in waiting
time is assumed, then we would save:
 8 min−1,33 min ¿ 6,67 min
 6,67 min × R 750/min ¿ R 5 002,50 per week
 Cost of a new cashier is R2000 per week, so the amount saved per week is:
 R 5 002,50−R 2000=R 3 002,50 per week
 The capital outlay of this project is R60 000, so the amount of weeks it would take
to recoup the initial cost is:

7
R 60 000
 =20 weeks
R3 000
 Once the cost has been recovered, the weekly saving will be:
 R 3 002,50−R 2 812,50=R 190 per week
 It seems much more advantages to add a new checkout counter than hiring a
new employee, but it must be considered that during the 20 week recovery
period, there will not be a “R190 saving”
 When comparing the alternative improvements, other aspects should also be
considered:
 Employee idle time is 0,4 for alternative 1 and 0,6 for alternative 2
 Loss of space resulting from a new checkout counter

 Note: the results of queuing analysis only provide information for decision
making, not an actual recommended decision (like an optimization model)

8
13.6. The Multiple-Server Waiting Line
 E.g. airline ticket and check-in counters where passengers line up in a single line,
post office line
 Characteristic of the system:
 First-come, first-served queue discipline
 Poisson arrivals
 Exponential service times
 Infinite calling population
 The parameters of the multiple-server model are as follows:
 λ - arrival rate (average number of arrivals per time period)
 µ - service rate (average number served per time period) per server (channel)
 c - number of servers
 cµ - mean effective service rate for the system (must exceed the arrival rate)

 Formulas for the operating characteristics of a multiple-server model:


 Probability that there are no customers in the system:

1
P 0= n=c−1 n c
1 λ 1 λ

[ ∑
n=0
( ) ] ( ) ( cµ−λ
n! µ
+
c! µ)
1 λ n
 Probability that n customers are in the queuing system: Pn= ()
c ! c n−c µ
. P0 [for n>c]

1 λ n
P n=
n µ( ) . P0 for [nc]

λ c
λµ( )
 Average number of customers in the queuing system: µ λ
L= P+
2 0
( c−1 ) !(c µ−λ) µ
λ
 Average number of customers in the waiting line: Lq=L−
µ
L
 Average time a customer spends in the total queuing system: W =
λ
1
 Average time a customer spends waiting in the queue to be served: W q =W −
µ
Lq
W q=
λ
 Probability that a customer arriving in the system must wait for service (all the
1 λ c cµ
servers are busy): wP =
c ! µ c µ−λ
P0()
Note: if c = 1, then all the formulas become the same as those for the single-server model

9
Example: Biggs Department Store (Customer Service)
 The customer service department has a waiting room in which chairs are placed
along the wall, in effect forming a single waiting line
 Customers come to this area with questions or complaints or to clarify matters
regarding credit card bills
 The customers are served by three store representatives, each located in a
partitioned stall
 Customers are served on a first-come, first-served basis
 This queuing system has to be re-evaluated, because excessive waiting time can
make already agitated customers angry enough to shop at other stores

 If a 12 month period is considered, an average of 10 customers per hour arrive at


the customer service department, and an average of 4 customers per hour can
be served by each store representative
 Also, this is a three-server queuing system
  c = 3 store representatives
 To compute the various operating characteristics:
1
P0= n=c−1
n
 1 λ 1 λ c cµ
[ ( )] ( )(
∑ n ! µ c ! µ cµ−λ
n=0
+ )
¿ 0,045 probability that no customers are∈the service department
λ c
λµ( )
 µ λ
L= P + =6 customers , on average ,∈the service department
2 0
( c−1 ) !( c µ−λ) µ
λ
 Lq=L− =3,5 customers , on average , waiting ¿ be served
µ
L
 W = =0,60 hours ( 36 minutes ) average time∈the service department
λ
1
 W q =W − =0,35 hours ( 21 minutes ) averagetime waiting∈line per customer
µ

1 λ c cµ
Pw = ()
c ! µ c µ−λ 0
P =0,703 probability that a customer must wait for service(i . e . there are 3∨more cust

 The department store's management has observed that customers are frustrated
by the relatively long waiting time of 21 minutes and the 0,703 probability of
waiting
 To try to improve matters, management has decided to consider the addition of an
extra service representative
 New operating characteristics for this system must be recomputed with c=4:
Operating characteristic c=3 c=4
Probability that no customers are in the queuing 0,045 0,073

10
system ( P0)
Average number of customers in the queuing
6 3
system (L)
Average number of customers in the waiting line ( Lq) 3,5 0,5
Average time a customer spends in the total queuing 0,6 hr 0,3 hr
system (W ) (36 min) (18 min)
Average time a customer spends waiting in the queue 0,35 hr 0,05 hr
to be served (W q ) (21 min) (3 min)
Probability that a customer must wait for service ( Pw ) 0,703 0,31
.

11

You might also like