Operations Management

Lecture 3: Waiting Line

Management

Dr. Gyan Prakash

Capacity Planning
 Capacity
 Max. level of value-added activity over time for a process under normal conditions
 Scale of operations
 Processing capacities of investments
 Capacity implications
 Costs, revenue, working capital, service level
 Objectives of capacity management
 Measurement of aggregate capacity and demand
 Identification of the alternative capacity plans
 Choice of the most appropriate capacity plan
Capacity Planning…

Capacity planning and Operations

control strategy
Demand
The market requires … availability Operations
of products and services management
Design Improvement
Supply
The operation supplies ... capacity to
deliver products and services Planning and
control
Aggregate Capacity
 Service operations: aggregate capacity of a hotel
 Rooms per night

 Manufacturing: aggregate capacity of an aluminium producer

 tonnes per month
 Demand Fluctuations

Present Future

Forecast demand
Aggregated output

Estimate of current capacity

Demand

Time
Demand Fluctuations…
Only 5% chance of demand being higher
than supply

Service level: % of
Distribution of demand time demand will
be fulfilled
DEMAND

DEMAND
Only 5% chance of demand being lower
than supply

TIME TIME
Equipment Effectiveness to Serve Work
Not worked Availability rate = a
Loading time (unplanned) = total operating time/
loading time
Set-up and
changeovers
Availability
Total operating time
losses Breakdown Performance rate = p
failure = net operating time/
total operating time
Speed Equipment
Net operating time ‘idling’
losses

Slow-running
equipment Quality rate = q
Valuable = valuable operating time/
Quality
operating time net operating time
losses
Quality
losses
 The time equipment is available for operation
 The speed, or throughput rate of the equipment Equipment effectiveness= a* p *
q
 The quality of the product or service
Measuring Demand and Capacity
Actual output
Efficiency =
Effective capacity
(168-59-58)/109= 51/109

Planned loss of 59
Design capacity hours
Unavoidable loss
168 hours per
week Avoidable loss
58 hours per week
Effective capacity
109 hours per
week Actual output – 51
hours per week

Actual output
Utilization = (168-59-58)/168= 51/168
Design capacity
Matching Capacity and Demand

Job/
customer
Demand Demand Demand

Capacity Capacity Capacity

Server with
a capacity

.
Demand
Level capacity Chase demand
management

Long-term strategy Medium-term strategy

Elements of Waiting Lines
 “Queue” is another name for a waiting line ( A. K.Erlang, 1913).
 Useful for capacity planning decisions (service operations)
 A waiting line system consists of two components
 The customer population
 (people or objects/tasks which needs to be processed )
 The process or service system
 Waiting lines occur due to …
 Arrival rate greater than servicing rate
 Customers usually arrive at random intervals (randomness)
 Some orders take longer than others (variability)
 Service system- tradeoff between cost and service level.
 Decision problem
 Balance cost of providing good service with cost of customers waiting.
 Queuing Analysis

Total Customer Capacity

cost = waiting cost + cost

Total cost Cost of service

capacity
Cost

Cost of customers
waiting

Service capacity Optimum

 Simple Queuing System
Service operations can not
Distribution of arrival times Distribution of store their services!
Random/appointment processing times Server
1

Rejecting Balking Reneging

Server 2
Source of
customers
Served
customers
Jockeying Server m

servers
Channels: Parallel
Finite versus Infinite populations
Arrival depend/does not depend on how
many already served
Boundary Queue or
of system ‘waiting line’
Queue discipline
Phases: Sequential stages in service
Adapted from Slack, Chambers and
Queue Implications: Performance
Measures
 Service system utilization
 Average number of customers waiting
 Average customer time in system
 Waiting time + processing time
 Average customer waiting time
 Probability of excessive waiting
 Customer waiting costs
 Service costs
 Probability of lost sales
Arrival & Service Patterns
 Arrival rate
 The average number of customers arriving per time period

 Modeled using the Poisson distribution

 Arrival rate usually denoted by lambda ()

 Inter-arrival time is denoted by (1/), exponential distribution

 Example: = 50 customers/hour
 Inter-arrival time (1/) = 0.02 hours or 1.2 minutes between customer arrivals

 Poisson Arrival: Orderliness, stationary, independence

e  t  t k
 Probability of k arrivals within time interval t. P(k ) 
k!
 λ = is mean arrival rate per time unit
 t = the length of time interval
 e= 2.71 and K!= k(k-1)(k-2)………3.2.1
Arrival & Service Patterns
 Service rate
 The average number of customers served during the period of time

 Service times are random, usually modeled using the exponential distribution

 Service rate usually denoted by mu (µ)

 Average service time is denoted by (1/µ)

 Example: µ= 70 customers/hour;
 Average service time (1/µ) =0.014 hours per customer or (0.857 minutes per customer).
 Even if the service rate is larger than the arrival rate, waiting lines form!
 Reason is the variation in specific customer arrival and variation service times.
Queue Discipline/Priority Rules
 First come, first served
 Best customers first (reward loyalty)
 Highest profit customers first
 Quickest service requirements first
 Largest service requirements first
 Earliest reservation first
 Emergencies first
 Etc.
Service Configuration
 Single channel, single phase.
 One server, one phase of service.
 Single channel, multi-phase.
 One server, multiple phases in service.
 Multi-channel, single phase.
 Multiple servers, one phase of service.
 Multi-channel, multi-phase.
 Multiple servers, multiple phases of service.
 Single Channel, Single Phase
Service system
Served units
Arrivals Queue

Service facility

Ships at sea Ship unloading system Empty ships

Waiting ship line

Dockyard
 Single Channel, Multi-Phase
Service system
Served units
Arrivals Queue
Service Service
facility facility

Cars in area Fast food drive-through

Cars & food
Waiting cars

Pay Pick-up
 Multi-Channel, Single Phase
Service system
Served units

Service
Queue
Arrivals facility

Service
facility
 Multi-Channel, Multi-Phase

Service system
Served
units
Service Service
Arrivals Queue facility facility

Service Service
facility facility
Assumptions in the Basic Model
 Customer population is homogeneous and infinite.
 Queue capacity is infinite.
 Customers are well behaved (no balking, or reneging).
 Arrivals are served FCFS (FIFO).
 Poisson arrivals.
 The time between arrivals follows a negative exponential distribution
 Exponential service times: Services are described by the negative exponential
distribution.
Steady State Assumptions
 Mean arrival rate , mean service rate and the number of servers are
constant.

 The service rate is greater than the arrival rate.

 These conditions have existed for a long time.

Queuing Model Notation

A/B/C
Number of servers or channels.

Service time distribution.

Arrival time distribution.

 M = Exponential distribution (Poisson arrivals).

 G = General distribution
 D = Deterministic (scheduled)
 E k = Erlang distribution
 C = Number of Servers
Nomenclature
Kendall’s notation

A/B/C/K/N/D
K = Capacity of the system (considered infinite if not specified)

N = Calling population (considered infinite if not specified)

D = Discipline of the queue (considered as FCFS if not specified)

Examples
M/M/1/∞/∞/FCFS (denoted just as: M/M/1)

M/D/1/∞/∞/FCFS (denoted just as: M/D/1)

M/M/s/∞/∞/FCFS (denoted just as: M/M/s)

M/M/s/∞/N/FCFS
Basic Queue Performance
 System utilization
 ρ = λ/μ

 Average number of customers being served

 r = λ/μ

 Average number of customers …

 Waiting in line for service, L = (λ)2/{μ(μ-λ)}
q
 In the system, L = L + r
s q
 Average times …
 Waiting in line, W = L /λ
q q
 In the system, W = W + 1/μ = L /λ
s q s
 The Classic Queue: M/M/1
Server

Queue
Identical
Customers

 M/M/1 queueing model

 Assumptions
 Population of customers is infinite or very large
 Arrival rate Poisson
 Single waiting line with infinite length, but no balking or reneging
 Rule: First-Come, First Served
 Poisson service times
M/M/1 Queuing Formulas
 Average number of customers waiting in line
 Lq = λ2/μ(μ - λ)
 Probability of zero units in the system
 P0 = 1 – (λ/μ)
 Probability of n units in the system
 Pn = P0(λ/μ)n
 Probability of less than n units in the system
 P<n = 1 - (λ/μ)n
 M/D/1 Queue
Server

Queue
Identical
Customers

 M/D/1 queueing model

 Assumptions

 Constant Service Time

 Average number of customers waiting in the line
 L = λ2/2μ(μ - λ)
q
 Priority Queues
Pool of Servers

Fatal Accident

Priority Queue

Broken Arm

Non-Priority Queue
Finite-source Queuing Model
 Appropriate for situations when there is a calling population that is finite
(countable and small)

 Arrival rate of customers in a finite situation is affected by the length of the

waiting line
 The more customers are already waiting, the fewer customer can arrive

 If everyone is already in waiting line, then no additional customers can

arrive … everyone is already there.

Finite-source Queuing Model Groups

Not waiting or Being

Waiting
being served served

J L H

U W T

Efficiency factor to J H
F 
measure the effect of wait J  LH
For a service factor Interpolated from Finite queuing tables
Finite-Source Queueing Formulas

T
Service factor X 
T U
Average number waiting L  N (1  F )
Average waiting time L( T  U ) T (1  F )
W 
N L XF
Average number running J  NF (1  X )
Average number being served H  FNX
Number in population N J  L  H
Little’s Law
The long-term average number of customers in a stable system Ls
long-term average arrival rate, λ,
long-term average time a customer spends in the system, Ws

Ls = λ Ws
Holds good for a wide variety of systems
Does not require knowledge about distribution of arrivals and service times

Little’s Law cannot be applied to systems with ‘finite’ calling population

Remember:  & 

  = Mean number of arrivals per time period. If average service time is 15

minutes, then μ is 4
 Example: 3 units/hour.
customers/hour

  = Mean number of arrivals served per time period.

 Example: 4 units/hour.
 1/ = 15 minutes/unit.
Managerial Implications
 Waiting line concepts for capacity planning
 Managing demand with supply
 Demand is arrival rate and supply is service rate
 In service system context desired variables are
 High level of quality of service ( QoS)
 Ability to manage volatility in demand
 Ability to provide reasonable level of responsiveness
 Service delivery systems
 Supporting facility, facilitating goods, information, explicit benefits, implicit benefits
Flexibility/Utilization Trade-off
Trade off benefits of high utilization levels with benefits of flexibility and service

High utilization
Low cost of operation
Low flexibility
Poor service

Low utilization
High cost of operation
High flexibility
Good quality of service
Basis of differentiation

= 0.0 = 1.0 or 100%

Utilization 
Implications…

Issue for consideration Peak Hour Non-Peak Hour

Operations strategy Standard offering Customized order

Service portfolio Narrow offering Wide-range of offering
Demand management Reservations Special tariffs
Resource management Multi-skilled labour, Adding Dedicated tasks, specialized
temporary work force training and development
Other Considerations
 Variance in arrivals and service rates causes long waits
 Wait time & queue length increase rapidly for  >0.7
 Queue is small until system is about 70% busy; then queue grows very quickly.
 Cost of waiting is non-linear.
 Customer ‘s unsatisfaction may grow exponentially
 Word-of-mouth
 Pooling servers is usually advantageous.
 Airport check-in
More Considerations
 Reduce effect of waiting.
 Distract customers with something to do, look at or listen to.

 Music, art -paintings, sculptures, mirrors, etc.

 Provide feedback on expected length of wait in queue.

 “Your call will be answered in 6 minutes…”

 Use self-service servers.

 Reduce system load
Thank you