Goa Institute of Management

PGDM-FT 2024-2025

Quantitative Analysis for

Management
Dr. Deepti Mohan
 • Design of Service Systems
• Queuing Models

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 2

Design of service systems
• Design of operations systems
• Making choices wrt location, layout, capacity and technology
• Manufacturing system vs Service system
• Extent of customer participation in the conversion process
• Key design choices in a service system:
1. Degree of customer contact
2. Degree of complexity
3. Degree of divergence

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 3

Degree of Customer contact
• Customer is an integral element in the design process of service systems

• Degree of Customer contact

• Low: quasi-manufacturing
• Medium: mixed service
• High: pure service

• Higher the customer contact,

✓it gets more difficult to pursue efficiency goals. Effectiveness will have to be
prioritized over efficiency.
✓higher investments of capacity is required as capacity exceeding average
requirements have to be built.
✓locating the facilities in a convenient place is crucial.
✓Training and skill development of service personnel is critical

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 4

Degree of Complexity and Divergence
• Degree of complexity: The number and intricacies of the steps and sequences in the process
• Degree of divergence: The number of alternate options available at each step of the service
delivery process
Low complexity/Divergence Process Description High Complexity/Divergence
No reservation Table Reservation Specific table selection options offered
Recite Menu, Describe in detail all starters &
Self seating, Menu on the board Seating Guests, Offering Menus
other special items in offer for the day
Serve Water & Starters at the beginning of the Assortment of salads, chips and fruit juices
Eliminate
service offered
Order takers interact with the customers at the
Customer calls out his requirements Order taking process
table in constructing the menu for the guests
Individually prepared and served at the table as
Pre-prepared: Fixed set of offering Salads, Papads
per request
Just 2 or 3 choices offered Starters Customers can choose from 20 alternatives
South Indian, Jain, Tandoori, Chinese,
Only South Indian (Vegetarian) Main Menu Continental, Brazilian (Both vegetarian & non-
vegetarian)
Payment at the counter while leaving (Cash Multiple choices of payments (Cash, Card,
Cash Payment
only) Coupons etc.)
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 5
 Service Positioning
Service positioning is the strategic choice a firm makes on the three
parameters (customer contact, complexity and divergence)

04/10/2024 6
Service Positioning
Low Degree of Interaction High
Low
No frill Fast-food
Airlines Joints
Restaurants
Online Airlines
Degree of Labour Intensity

Retailers
Budget
Internet & Phone Hotels Ethnic & High End
Banking
Eating Places

For-profit Five Star

Hospitals Hotels
Traditional
Retailing Traditional
Banking
Boutique
Personal Hospitals
Banking

High
Capacity Planning in Services using Queuing
Analysis
• In service systems, waiting time is an important operational measure
that determines the service quality
• In waiting-line models, analysing the adequacy of the capacity needs
assessment of performance measures such as
1. Server utilization
2. Average waiting time of a customer in queue
3. Average time a customer spends in the system
4. Average number of customers in queue
5. Average number of customers in the system

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 8

Queuing system – basic structure

Calling
Population
Arrivals
Waiting Line Server Served
customers
• Calling population: Potential customers that can request service
• Customer: An arriving unit that needs some service
• Queue / Waiting Line: Point where customers accumulate waiting for the service.
• Doesn’t include the customers who have been served or going through serving process.
• Server / Service Channel: Process or system (could be human or machine) that
performs services for the customers.
• Queue Length: Number of customers present in the queue.
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 9
Single-server (channel)

Single-server, single-stage

Waiting line Server

Single-server, multiple stages

Waiting line Servers

Multi-server (channel)
Multiple-servers, single stage

Servers

Multiple-servers, multiple-stages

Waiting line
Servers
Performance metrics
Key performance measures to analyse the adequacy of the capacity in
waiting-line models:
1. Server utilization (𝜌) (The percentage of time the servers are busy)
2. Average waiting time of a customer in queue (𝑊𝑞 )
3. Average time a customer spends in the system (𝑊𝑠 )
4. Average number of customers in queue (𝐿𝑞 )
5. Average number of customers in the system (𝐿𝑠 )

Arrivals
Waiting Line Server
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 12
Implications of operating characteristics
• Short lines could mean either good customer service or too much capacity.
• Long lines could indicate either low server efficiency or the need to increase capacity.
• Long lines do not always mean long waiting times.
• If the service rate is fast, a long line can be served efficiently. However, when waiting time seems
long, customers perceive the quality of service to be poor.
• Managers may try to change the arrival rate of customers or design the system to make long wait
times seem shorter than they really are. Walt Disney World, customers in line for an attraction are
entertained by videos and also are informed about expected waiting times, which seems to help
them endure the wait.
• A large number of customers in the system causes congestion and may result in
customer dissatisfaction, unless more capacity is added.
• Total time in system (the total elapsed time from entry into the system until exit from the
system) may indicate problems with customers, server efficiency, or capacity.
• In general, very high levels of utilization (over 85%) is undesirable in service systems.
Why?

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 13

Server utilization

Server utilisation
• In the case of single server: 𝜌 = 𝜆/𝜇
• In the case of multiple, say 𝑠 servers: 𝜌 = 𝜆/𝑠𝜇

• When 𝜌<1, the system can handle the customers well, and waiting times will be shorter.
• When 𝜌→1, the system becomes congested, and waiting times increase dramatically.
• If 𝜌 ≥ 1, the system is overloaded, and waiting times would approach infinity

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 14

Capacity Design: Flexibility/Utilization Trade-
off
High utilization
Low cost of operation
Poor service
Operational Performance
Measures

Low utilization
High cost of operation
Good service

0 Utilization 100%
Capacity Design: Cost Relationship
Expected costs

Total cost

Service
cost

Waiting Costs

Level of service
Queue behaviors
• Jockeying
• Customers switching between queues
• Balking
• Customers refusing to join the queue because of its length
• Reneging
• Customer in the queue who leave in anticipation of longer waiting time

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 17

Kendall’s Notation

• System of notation defining the characteristics of a queueing model

• A General Queueing Model: (a / b /c ) : ( d / e / f )
• a: Type of distribution for Inter-arrival Time
• b: Type of distribution for Service Time
• c: Number of servers (service channels) in the system
• d: Maximum number of customers allowed in the system (System Capacity)
• e: Size of the calling population (Source) of customers
• f: Queue Discipline

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 18

 Poisson Distribution Exponential Distribution
• A discrete probability distribution • A continuous probability
distribution
• Models the number of events
occurring in a given time period • models the time between
events.
• 𝑚 be the average number of
events in a unit time • 𝑚 be the average number of
events in a unit time
𝑒 −𝑚 𝑚𝑥
• 𝑃 𝑋=𝑥 = ; • 𝑓 𝑥 = 𝑚𝑒 −𝑚𝑥
𝑥!
(𝑥 = 0, 1, 2, …) • Mean = 1/𝑚 (average time
• Mean = 𝑚 between events)
• variance = 𝑚 • Variance = 1/𝑚2

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 19

Kendall’s Notation - description
A General Queueing Model: (a / b /c ) : ( d / e / f )
• a: Type of distribution for Inter-arrival Time
• b: Type of distribution for Service Time
• c: Number of servers (service channels) in the system
• d: Maximum number of customers allowed in the system (System Capacity)
• e: Size of the calling population (Source) of customers
• f: Queue Discipline

• Type of distribution for the inter-arrival time

• M: Markov (poisson distributed arrival rate ֞ exponentially distributed inter-arrival times)
• D: Deterministic inter-arrival times
• G: General distribution of inter-arrival times
• Type of distribution for the service time
• M: Markov (poisson distributed service rate ֞ exponentially distributed service times)
• D: Deterministic service times
• G: General distribution of service times
• Number of servers: s
• System capacity: infinite or finite
• Calling population size: infinite or finite
• Queue discipline (FCFS/FIFO/LIFO/SIRO/Priority)
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 20
Little’s Law
• A fundamental law that relates the number of customers in a queue system to
the arrival rate and waiting time of customers.

• The average number of items in a system is equal to the product of the average
arrival rate of items and the average time an item spends in the system
𝑳𝒔 = 𝝀 𝑾𝒔
• The average number of items in a queue is equal to the product of the average
arrival rate of items and the average time an item spends in the queue
𝑳𝒒 = 𝝀 𝑾𝒒
• Little's Law applies to any stable system, regardless of the distribution of arrival
times or service times.

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 21

(M/M/1): (ꝏ/ ꝏ /FIFO) Queuing system
A General Queueing Model: (a / b /c ) : ( d / e / f )
A system where •
•
a: Type of distribution for Inter-arrival Time
b: Type of distribution for Service Time

✓both inter-arrival times and service times are •

•
c: Number of servers (service channels) in the system
d: Maximum number of customers allowed in the system (System Capacity)

exponentially distributed (Markovian), •

•
e: Size of the calling population (Source) of customers
f: Queue Discipline

✓There is a single-server,
✓there is no limit to the number of customers
allowed in the system (infinite system
capacity),
✓the calling population is infinite, and
✓the queue discipline is First-In-First-Out.
Additionally,
✓the system operates in a steady state (𝜆<𝜇).
✓no Balking or Reneging.
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 22
(M/M/1): (ꝏ/ ꝏ /FIFO) Queuing system
❑λ : Mean arrival rate
❑μ: Mean service rate
• 𝜌 = 𝜆/𝜇 ❑ 𝜌: Server utilization
1 ❑𝐿𝑠 : Average number of customers in the system (waiting and being served)
• 𝑊𝑠 = ❑𝐿𝑞 : Average number of customers in the waiting line (waiting to be served)
𝜇−𝜆
𝟏 ❑𝑊𝑠 : Average time a customer spends in the system (waiting and being served)
• 𝑾𝒔 = 𝑾𝒒 + ❑𝑊𝑞 : Average time a customer spends waiting in line (waiting to be served)
𝝁
❑𝑃𝑛 : The probability that there are n customers in the system
(Avg waiting time = Avg waiting time in queue + Avg service time)

• 𝑳 𝒔 = 𝝀 𝑾𝒔
Little’s Law
• 𝑳𝒒 = 𝝀 𝑾𝒒
𝜆
• 𝐿𝑠 = 𝐿𝑞 +
𝜇
(Avg no. of customers in the system = Avg queue length + Avg no. of customers being served)
• 𝑃𝑛 = 1 − 𝜌 𝜌𝑛
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 25
(M/M/1): (ꝏ/ ꝏ /FIFO) Queuing system
• 𝜌 = 𝜆/𝜇 ❑λ : Mean arrival rate
❑μ: Mean service rate
• 𝑃𝑛 = 1 − 𝜌 𝜌𝑛 ❑ 𝜌: Server utilization
❑𝐿𝑠 : Average number of customers in the system (waiting and being served)
1 ❑𝐿𝑞 : Average number of customers in the waiting line (waiting to be served)
• 𝑊𝑠 = ❑𝑊𝑠 : Average time a customer spends in the system (waiting and being served)
𝜇−𝜆 ❑𝑊𝑞 : Average time a customer spends waiting in line (waiting to be served)
𝜆 𝜌 ❑𝑃𝑛 : The probability that there are n customers in the system
• 𝐿𝑠 = 𝜆𝑊𝑠 = =
𝜇−𝜆1−𝜌
𝜆 𝜌
• 𝑊𝑞 = 𝑊𝑠 − 1/𝜇 = =
𝜇(𝜇−𝜆) 𝜇−𝜆
𝜆𝜌 𝜌2
• 𝐿𝑞 = 𝜆 𝑊𝑞 = =
𝜇−𝜆 1−𝜌

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 26

Multi-Server Queues: Approximation based
on data
• This formula can be used to compute the queue length if the mean and standard
deviation (sd) of Inter-arrival time (IAT) and service time (ST) are computed from
observed data for a queuing system with 𝑠 servers.
• Let 𝑋_𝑎 and 𝑆𝑎 be the mean and the sd of IAT, respectively and 𝑋_𝑠 and 𝑆𝑠 be the mean
and sd of ST computed from observed data for a queuing system with 𝑠 servers, then
𝜌 2 𝑠+1 𝐶𝑎2 +𝐶𝑠2
• 𝐿𝑞 = ( )
1−𝜌 2
• Where,
• Mean arrival rate, 𝜆 = 1/𝑋_𝑎
• Mean service rate, 𝜇 = 1/𝑋_𝑠
• Server utilization, 𝜌 = 𝜆/(𝑠𝜇)
• Coefficient of variation of IAT, 𝐶𝑎 = 𝑆𝑎 / 𝑋_𝑎
• Coefficient of variation of ST, 𝐶𝑠 = 𝑆𝑠 / 𝑋_𝑠

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 28

Problem-1
The arrival rate of customers at a bank with single cash counter follows
Poisson distribution with mean 45 customers/hour. The service rate of the
counter clerk also follows Poisson distribution with mean 60 customers/hour.
i) What is percentage of the time the counter clerk is busy?
ii) What is the probability of having 0 customers in the system?
iii) What is the probability of having 5 customers in the system?
iv) Average number of customers in the system
v) Average queue length
vi) Average time spent in the system
vii) Average time spent in the queue

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 29

Problem-1 This system can be modelled as an (M/M/1): (ꝏ/ ꝏ /FIFO)
The arrival rate of customers at a bank with queuing system where
single cash counter follows Poisson 𝜆 = 45 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑝𝑒𝑟 ℎ𝑜𝑢𝑟
distribution with mean 45 customers/hour.
The service rate of the counter clerk also 𝜇 = 60 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠 𝑝𝑒𝑟 ℎ𝑜𝑢𝑟
follows Poisson distribution with mean 60
customers/hour.
Percentage of the time the counter clerk is busy is same as the
i) What is percentage of the time the server utilization
counter clerk is busy?
𝜆 45
ii) What is the probability of having 0 Sever utilization, 𝜌 = 𝜇 = 60 = 0.75 𝑜𝑟 75%
customers in the system?
iii) What is the probability of having 5 𝑃0 = 1 − 𝜌 = 25%
customers in the system?
𝑃5 = 1 − 𝜌 𝜌5 = 5.93%
iv) Average number of customers in the 1 1 1
system 𝑊𝑠 = = = ℎ𝑜𝑢𝑟𝑠 (4 𝑚𝑖𝑛𝑢𝑡𝑒𝑠)
v) Average queue length 𝜇 − 𝜆 60 − 45 15
1 1 1 3 1
vi) Average time spent in the system 𝑊𝑞 = 𝑊𝑠 − = − = = ℎ𝑜𝑢𝑟𝑠 3 𝑚𝑖𝑛𝑢𝑡𝑒𝑠
𝜇 15 60 60 20
vii) Average time spent in the queue 1
𝐿𝑠 = 𝜆𝑊𝑠 = 45 ∗ = 3 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠
15
1
𝐿𝑞 = 𝜆𝑊𝑞 = 45 ∗ = 2.25 𝑐𝑢𝑠𝑡𝑜𝑚𝑒𝑟𝑠
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 20 30
Problem-1
The arrival rate of customers at a bank with single cash counter follows
Poisson distribution with mean 45 customers/hour. The service rate of the
counter clerk also follows Poisson distribution with mean 60 customers/hour.
i) What is percentage of the time the counter clerk is busy? 75%
ii) What is the probability of having 0 customers in the system? 25%
iii) What is the probability of having 5 customers in the system? 5.93%
iv) Average number of customers in the system: 3 customers
v) Average queue length: 2.25 customers
vi) Average time spent in the system: 4 minutes
vii) Average time spent in the queue: 3 minutes

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 31

Problem-2
• A harbour has single dock to unload the containers from the incoming
ships. The arrival rate of ships at the harbour follows Poisson distribution
and the unloading time for the ships follows exponential distribution and
hence the service rate also follows Poisson distribution. The arrival and
service rate are 8 ships/week and 14 ships/week respectively. Find
utilization factor, Ls, Lq, Ws and Wq.
• 𝜌 = 57.14%
• 𝑊𝑠 = 1.17 𝑑𝑎𝑦𝑠
• 𝑊𝑞 = 0.67 𝑑𝑎𝑦𝑠 = 16 ℎ𝑜𝑢𝑟𝑠
• 𝐿𝑠 = 1.33 𝑠ℎ𝑖𝑝𝑠
• 𝐿𝑞 = 0.76 𝑠ℎ𝑖𝑝𝑠

04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 32

Problem-3
A Maruti Suzuki Service Centre has a central store, where mechanics arrive to take spare parts for the jobs that
they are working upon. The mechanics wait in queue, if the storekeeper is busy serving other mechanics,
whenever required. The service is on FCFS basis.
Presently, the store is manned by one storekeeper and he can attend on average 10 mechanics per hour. The
arrival rate of mechanics is 8 mechanics per hour. Assuming that the arrival pattern of mechanics is Poisson
distributed and servicing time is exponentially distributed, determine the time spent by the mechanics in the
system, time spent by the mechanics in waiting and the average queue length.
Suppose the service mechanics are paid an hourly wages of Rs. 500 and the store keeper is paid an hourly
wage of Rs. 300, would you support the decision to recruit one additional store keeper to help, if the service
rate would improve to 14 mechanics/hour with the additional storekeeper? Assume the number of mechanics
to be sufficiently large and duration of the shift is 8 hours.
• Serving cost = Number of store keepers * 300* 8
• Waiting cost = 𝑊𝑞 ∗ 𝐿𝑞 ∗ 500 ∗ 8
• Base case cost: Rs. 2400 + Rs. 5120 = Rs. 7520
• With additional store keeper: Rs. 4800 + Rs. 290 = Rs. 5090
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 33
Problem-4
Prolonged observation of a multi-server facility in 𝑋_𝑎 = 0.75 𝑚𝑖𝑛 ; 𝑆𝑎 = 0.25 𝑚𝑖𝑛;
a super market resulted in the collection of some
data with respect to customer arrivals at the 𝑋_𝑠 = 3.25 𝑚𝑖𝑛 ; 𝑆𝑠 = 2.25 𝑚𝑖𝑛;
supermarket and the service time at the checkout 2( S +1)
counter.   C a2 + C s2 
Lq = * 
Based on 100 observations, the mean interarrival (1 −  )  2 
time was 0.75 minutes with a standard deviation 
of 0.25 minutes and the mean service time was Reduced
3.25 minutes with a standard deviation of 2.25
minutes. The data is representative of the Variability Added
Original (5
situation prevailing in the supermarket. Currently, (25%) Counter (6
there are five checkout counters in the Measure Counters)
𝑆𝑠 Counters)
supermarket. S=5
= 0.75 S=6
The supermarket wants to know if reducing the ∗ 2.25 𝑚𝑖𝑛
variability of service process by 25 percent is
better than adding one more counter at the shop. Utilization
Compute the relevant operational performance 86.67% 86.67% 72.22%
(ρ)
measures for the supermarket and advise.
(Lq​) 1.348 0.870 0.314
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 34
04/10/2024 Dr. Deepti Mohan | Goa Institute of Management 35