OPERATIONS MANAGEMENT

UNIT 3. LONG TERM CAPACITY PLANNING

What is Capacity ? The máximum rate of output of a process or a system in normal conditions over a
period of time.

The measurement of capacity presents particular difficulties in service companies and industrial
companies that produce a wide range of heterogeneous products. In these cases, capacity if defined as
the amount of resources available over a period of time.

Examples:

- Airline: seats per month

- Hospital: beds per month
- Hotel: rooms per month
High investment associated with it.

Close relationship with the organization’s long-term strategy.

Long-term capacity plannings’ importance
Impact on operational efficiency.

Restrictions it imposes on other decisions, once it is determined.

LACK of capacity can EXCESS of capacity can

 Decrease in service  Decrease prices

 Customer and market losses  Unused labor and machinery
 Decrease in quality level  Excess inventory

Frequent mistakes in the definition of capacity:

Measures that do not take into account the time factor

- Capacity of an oil refinery: 2000 barrels… per day? Per month?

Confusing capacity with volume

- Production volume: quantity actually produced

- Production capacity: the maximum that can be produced under normal conditions
- Solution: define degree of utilization

Production Volume
Utilization =
Production Capacity
 Confusing peak capacity and sustainable capacity

- Peak capacity: maximum production valume that can be obtained over a period of time under
ideal conditions, using overtime and even using the machinery above its recommended
capacity
- Sustainable capacity: maximum production colume that can be sustained in normal operating
conditions.

We will refer to sustainable capacity but we will use the denomination of nominal, theoretical or normal
( not related to environmental sustainability ).

Manufacturing Capacity Vs Service Capacity

Manufacturing capacity Service capacity

Goods can be stored for later use. Capacity must be available when service is
needed – cannot be stored.

Goods can be shipped to other locations. Service must be available at customer demand
point.

Volatility of demand is relatively low. Much higher volatility is typical.

Capacity management

CAPACITY PLANNING (LONG- CONSTRAINT MANAGEMENT

TERM) (SHORT-TERM)

- Economies and diseconomies of scale - Theory of constraints

- Capacity timing and sizing strategies - Identification and management of
- Systematic approach to capacity decisions bottlenecks
- Product mix decisions using bottlenecks

DETERMINATIONS OF THE PRODUCTIVE CAPACITY

CAPACITY PLANNING AND CONTROL

It translates production plans into capacity needs, estimate the available capacity, its future projection
and try to adapt both.

- Contraction  inadvisable option, only as a last reort (implies loss of accumulated knowledge)
 Alternative: use existing capacity to meet new demands -> take advantage of economies of
scope ( when it is cheaper to manufacture two products in the same factory than to do it in
separate factories, by sharing some assets in the ,a ufacturing process). Search new
geaographic, functional markets…

- Expansion  when capacity is added. Related to the previous question: Are we using ALL the
current available capacity well?

The concept ofeconomies of scale refer to the reduction of the avcerge cost per unit as the volume of
production increases. How does this happen?

- As production volume increases, fixed costs are distributes over a larger number of units
- Possibility of obtaining discounts for volume of purchases

However, increasing the capacity of the plant can also generate additional costs, leading to the
appearance of diseconomies of scale ( the average unit cost rises with increase in production)

- They are problems due to communication, coordination, complexity, oversight and loss of
flexibility, loss of focys and inefficiencies which may have it unfeasible to increase capacity.

Economies and Diseconomies of scale

Internal: exclusive to a specific company

Extermal: all industry / market

*look at the slide’s example*

FACTORS DETERMINING LONG TERM CAPACITY

Not always the best option is to invest in new plants. Available options:

- Possible existence of bottlenecks (improve them)

- Possibility to redesign the product (timing of operations and preparation)
- Improve the reliability of components (defective and stopped)
- Possible changes in product mix (standardization)
- Possibility of changing technology (automation)
- Changes in programming techniques (improving execution, waiting times…)çfinally, it will be
necessary to take into account the reatcion of competitors

STAGES/PHASES OF LONG TERM CAPACITY PLANNING

1 ESTIMATE DEMAND FOR PRODUCTS

- Estimate the total demand for the product in the sector

- Estimate the market share for our company (% of total demand that we expect to achieve)

- The reliability of the estimates decreases tas the planning horizon expands
- Responsibility: Marketing Research Department

2 CALCULATE CAPACITY REQUIREMENTS

Forecast current capacity

- Future capacity can be reduced due to the aging of facilities and equipment (breakdowns,
increase of defective firms, lower production rate…)
- The future capacity can be increased due to the learning effect: reduction of the time a process
takes as the experience is the accomplishment of the different activities of the productive
process increases

Estimate gaps comparing forecasted capacity with the expected demand

- Take into account installed capacity of the competition: reaction potential

- Capacity may not match demand for several reasons, such as
Not having enough financial resources, which forces to give up part of that demand.
Given the lack of accuracy of demnd forecasts, more/less capacity is installed (capacity cushion)

*look at the slides’s example*

 3 ESTABLISH ALTERNATIVES

Alternatives when forecasted capacity /=/ Expected demand

EXPANSION CONTRACTION

- Give another use to the facilities or put

- Obtain or acquire new facilities
them in reserve
- Expand, modify and update existing
- Sell facilities and inventories and dismiss
facilities or/and their way of use
or teansfer labor
- Establish subcontracting networks
- Develop new products or servuces that
- Reopen inactive facilities
absorb excess capacity
- Rent excess capacity

4 EVALUATION AND SELECTION

o Break-Even Point Graphics

o Net Present Value
o Internal Rate of Return

o Decision trees
o Queuing Theory
o Simulation
 o Degree of compatibility with existing staff
o Possible reactions of the public opinion
o Degree of reaction competition
o Obsolescence risk
o Multi-criteria/attribute techniques

(PART II). QUEUING (WAITING LINE) THEORY

QUEUING THEORY

The study and analysis of the phenomena of waiting lines is done through the Queuing Theory. The
formation of waiting lines occurs whenever the current demand for a service exceeds the current
capacity to provide it.

The decision of what amount of service to provice is usually complicated:

- Excessive services would increase system costs

- Lacking sufficient capacity would generate excessively long queues that increase costs, and may
cause customers to leave with the consequent loss of sales

The ultimate goal  achieve economic balance between the cost of providing the service and the cost
associated with waiting for customers for that service.

Problems :

o ANALYTICS: how long does a customer wait in the queue? How many customers are waiting in
the line? How many clients are in the line?
o OF DESIGN: number of people who must provide the service, if there must be one or more
waiting lines, the space that must exist for customers to wait until their return, etc.
PARTS OF A WAITING LINE

- Customers come in
- Customers are served
- Customers leave

CHARACTERISTICS OF A WAITING LINE SYSTEM

Population size  is the potential population that may need the service. The sixe of this input source
will be the total number of leads. The size can be:

- Finite population: limited-size customer pool that will use the service and, at times, form a line.
- Infinite population: unlimited, population large enough so that the population size caused by
substractions or additions to the population does not significantly affect the system
probabilities.

Regarding the way in which customer are generated over time, it can be analyzed according to two
equivalent approaches:

- By the number of clients arriving per unit of time (arrival rate) is represented with the Greek
letter alpha
- For the time between the arrival of two consecutive customers, 1 entre alpha

The queue  is characterized by the average number an the maximum of clients that can admit.

Queue Discipline: refers to the order in which clients are selected to receive the service: it can be
randomly, FIFO, by means of a priority procedure, etc.

Service Mechanism: there may be one or more service stations with one or more servers.

The time between the beginning of the service to a customer and its termination is called ‘service time’.
The number of clients served per unit of time is represented by the letter u.

Service time distribution:

- Constants: service provided by automation

- Variable: service provided by humans. Described using negative exponential distribution

POISSON DISTRIBUTION

Where one is interested in the number of arrivals during some time period. The probability function is:
NEGATIVE EXPONENTIAL DISTRIBUTION

When service time at a service facility pccurs in a purely random fashion. The probability function is:

BASIC STRUCTURE FOR THE MODELS

There are several types of models depending on the layout of the servers:

A. SINGLE-SERVER, SINGLE-PHASE SYSTEM

It is the most elementary. EXAMPLE a bank with a single window to serve clients, a
barber shop with one single employee providing haircuts.
 B. SINGLE-SERVER, MULTIPLE-PHASE SYSTEM
EXAMPLE an assembly line, a car wash center where several servuces are performed
C. MULTIPLE-SERVER, SINGLE-PHASE SYSTEM
The number of service stations increases, but there is only one waiting line. EXAMPLE a
bank with multiple open window that depend on a single waiting line.
D. MULTIPLE-SERVER, MULTIPLE-PHASE SYSTEM
EXAMPLE two or more assembly lines in parallel.

DISTRIBUTIONS

Distribution of the time between two consecutive arrivals

- Deterministic: when it is known with certainty the time that elapses between the consecutive
arrival of two clients
- Exponential: when said time is considered a random variable with Poisson distribution
- Generic: when this time between arrivals is represented through the realization of a random
variable with probability distribution other than exponential.

Distribution of the time that elapses in the provision of the service (time of service)

- Deterministic: when it is known with certainty the time that elapses between the service of two
consecutive clients.
- Exponential: when said time is considered a random variable with negative exponential
distribution.
- Generic: when said time between arrivals is represented by a random variable other than
exponential

Number of servers: indicates the number of individuals or servers that serve the clients of a facility.

The queueing systems are represented by the so called kendall notation

Thus and M/M/3 model indicates a queuing system in which the time between two consecutive arrivals,
although unknown, can be simulated through a random variable with Posison distribution, as well as the
service time, and in the service facilities of the system there will be three servers that will attend to the
customers that arrive. In this topic we will focus on the M/M/c type models.

The four queuing models that follow all assume:

1. Poisson distribution arrivals

2. FIFO discipline
3. A single-service phase
 Arrivals are served on a FIFO basis, every arrival waits to be served regardless of the length of
the queue
 Arrivals are independent of preceding arrivals, the averaged number of arrivals does not
change over time
 Arrivals are described by a Poisson probability distribution and come from an infinite
population
 Service times vary from one customer to the next and are independent of one another, but
their average rate is unknown
 Service times occur according to the negative exponential distribution
 The service rate is faster than the arrival rate ( u > alpha )  STABLE PHASE
Examples on slides 26-27

Examples on slides 30-34

LITTLE’S LAW

“ The average number L of customers in a system in steady state (inventory) is equal to the average
effective arrival rate landa multiplied by the average time W that a customer spends in the system.”

- Once two of the parameters are known, the other can be easily founds.
- It makes no assumptions about the probability distribution of arrival and services times.
- Applies to all queuing models except the finite population model.

Example 3 on slide 36 and example 4 slides 37-40

QUEUING COSTS

The incorporation of the costs in the queuing models allows to analyse the calculation of the
OPTIMAL SERVICE LEVEL.

Objective  find balance between the cost of providing a good service, which is usually related to
the cost of waiting for customers, and the cost derived from the number of servers. One way to
evaluate a queuing system is through its total cost: the sum of service costs plus the costs of
waiting.

The queuing model with lower total costs is preferred.

Rate of service utilization and service quality are directly linked.

ECONOMIC APPLICATIONS ( ALL EXAMPLES LOOK AT SLIDES