Capacity Management

What is capacity management?

Capacity management is the activity of understanding the nature of

product demand or service demand, and effectively planning and
controlling capacity.”

- Maintaining customer satisfaction.

- Need for operations to maintain efficiency by minimizing the costs
of excess capacity.
Capacity management performance objectives
 Costs will be affected by the balance between demand and capacity.
Capacity levels in excess of demand could mean underutilisation of
capacity and therefore high unit-cost.

 Revenues will also be affected by the balance between demand and

capacity, but in the opposite way. Capacity levels equal to or higher
than demand at any point in time will ensure that all demand is
satisfied, and no revenue lost.

 Working capital will be affected if an operation decides to build up

finished product inventory prior to demand. This might allow
demand to be satisfied, but the organisation will have to fund the
inventory until it can be sold.

 Quality of services might be affected by a capacity plan that involves

large fluctuations in capacity levels, by hiring temporary staff for
example. The new staff and the disruption to the routine working of
the operation could increase the probability of errors being made.

 Speed of response to customer demand could be enhanced either by

the deliberate provision of surplus capacity to avoid queuing, or
through the build-up of product inventories.

 Dependability of supply will also be affected by how close demand

levels are to capacity. The closer demand gets to the operation’s
capacity ceiling, the less able it is to cope with any unexpected
disruptions.

2
  Flexibility, especially volume flexibility, will be enhanced by
surplus capacity. If demand and capacity are in balance, the
operation will not be able to respond to any unexpected increase in
demand.

A framework for capacity management

1) The most common first step on the demand side of the ‘equation’ is
to measure (forecast) demand for services and products over
different time periods. This involves selecting from a range of
qualitative (panel, Delphi and scenario planning) and quantitative
(time series and causal models) tools to support more accurate
prediction of demand.

2) The second step is typically on the supply side of the framework and
involves measuring the capacity to deliver services and products.

3) The third step is to consider if and how to manage demand using

demand management and yield management techniques.

4) The fourth step is to manage the supply side by determining the

appropriate level of average capacity and then deciding whether to
either keep this constant (level capacity plan) or to adjust capacity in
line with changing demand patterns (chase capacity plan).

3
Qualitative approaches to forecasting

Panel approach
The panel acts like a focus group allowing everyone to talk openly.
Although there is the great advantage of several brains being better than
one, it can be difficult to reach a consensus, or sometimes the views of the
loudest or highest status may emerge (the bandwagon effect).

Delphi method
The best-known approach to generating forecasts using experts is the
Delphi method. It employs a survey of experts where replies are analysed,
and anonymous summaries are sent back to all experts. The experts are
then asked to re-consider their original response in the light of the replies
and arguments put forward by the other experts. This process is repeated
several times to conclude either with a consensus or at least a narrower
range of decisions.

Scenario planning

4
This is usually applied to long-range forecasting, again using a panel. The
panel members are usually asked to devise a range of future scenarios.
Each scenario can then be discussed, with inherent risks considered.
Unlike the Delphi method, scenario planning is not necessarily concerned
with arriving at a consensus, but looking at a range of options and putting
plans in place to try to avoid the ones that are least desired and taking
action to follow the most desired.

Quantitative approaches to forecasting

Two key approaches are time series analysis and causal modelling
techniques.

Time series examines the pattern of past behaviours to forecast future

behaviour. Causal modelling is an approach that describes and evaluates
the cause–effect relationships between key variables.

Time series analysis

Times series analysis is a method of forecasting that examines the
pattern of time series data and, by removing underlying variations with
assignable causes, extrapolates future behaviour.

 Time series methods of forecasting use past patterns of demand to make predictions.

Simple moving-average forecasting

The simple moving-average is used to estimate demand for a future time
period by averaging the demand for the n most recent time periods. The
value of n can be set at any level but is usually in the range of 3 to 7.

So, for example, if n is set at four, the next period’s demand is forecast by
taking the moving average of the previous four periods’ actual demand.
Thus, if the forecast demand for week t is Ft and the actual demand for
week t is At, then:

5
Simple exponential smoothing
The main disadvantage of moving averages is that they do not use data
from beyond n periods in forecasting. The exponential smoothing
approach forecasts demand in the next period by taking into account the
actual demand in the current time period and the forecast that was
previously made. It does so according to the following formula:

Where

Ft = new forecast

At‒1 = previous period’s actual demand

Ft‒1 = previous period’s forecast demand

α = smoothing constant

- The main disadvantage of simple exponential smoothing is that it

assumes a stable underlying average.

Trend-adjusted exponential smoothing

When there is a trend in the average, exponentially smoothed forecasts lag
behind the changes in underlying demand. While higher smoothing
constants (>0.5) help to reduce forecast errors, there may still be a lag if
the average is systematically changing. Therefore, it is possible to include
a trend within exponentially smoothed forecasts to improve accuracy. The
new formula is:

6
where
FITt = forecast including trend
Ft = exponentially smoothed forecast

Tt = exponentially smoothed trend.

For a trend-adjusted forecast, we must smooth both the average (Ft) and
the trend (Tt). The smoothing constant is shown with the α symbol for the
average and the β symbol for the trend. To arrive at the forecast including
(FITt) we must compute the two parts of the equation:

where
Ft = exponentially smoothed forecast for period t

Tt = exponentially smoothed trend for period t

At = actual demand for period t

α = smoothing constant for the average

β = smoothing constant for the trend.

Seasonality in forecasting
Most organisations experience seasonal patterns in their demand.

- Climatic (Holidays)

- Festive (gift purchases)

- Financial (tax processing)

- we typically think of seasonality in annual terms.

7
However, in forecasting the term is used to describe any regularly
repeating changes in demand (quarterly, monthly, weekly, daily or hourly).

!!!!!!) In demand forecasting, ‘seasonality’ refers to any repeating pattern

of demand – annual, quarterly, monthly, weekly, daily, or even hourly.

A popular technique for incorporating seasonality in forecasting is the

multiplicative seasonal model, which involves the following five steps (for
simplicity, here we assume that there is no other trend in the data, apart
from seasonality):

1) Find the average demand for each ‘season’ by summing the demand
for that season and dividing by the number of seasons available. For
example, if in March, we have had sales of 80, 75 and 100 over the
last three years, average March demand equals (80 + 75 + 100) / 3 =
85.

2) Calculate average demand over all ‘seasons’ by dividing total

average demand by the number of seasons. For example, if total
average annual demand is 1320 and there are 12 seasons (months),
average demand equals 1320 / 12 = 110.

3) Compute seasonal index by dividing average season demand (step 1)

over average demand (step 2). For example, March seasonal index
equals 85 / 110 = 0.77.

4) Estimate the next time period’s (in this case, annual) total demand
using one or more of the qualitative or quantitative methods
described in this section.

8
 5) Divide this estimate by the number of seasons (in this case, 12
months) and multiply by the seasonal index to provide a seasonal
forecast.

Causal models
Causal models often employ complex techniques to understand the
strength of relationships between the network of variables and the impact
they have on each other.

‘Simple regression’ models try to determine the ‘best fit’ expression

between two variables.

!!!!!) Causal models make predictions by examining the impact that one or
more variables have on demand.

How is capacity measured?

The capacity of an operation is the maximum level of value-added activity
over a period of time that the process can achieve under normal operating
conditions.

Measuring capacity may sound simple but can in fact be relatively hard to
define unambiguously unless the operation is standardised and repetitive.

The effect of activity mix on capacity measurement

An operation’s ability to supply is partly dependent on what it is being
required to do. For example, a hospital may have a problem in measuring
its capacity because the nature of the service varies significantly. Output
depends on the mix of activities in which the hospital is engaged and,
because most hospitals perform many different types of activities, output is
difficult (though not impossible!) to predict. Some of the problems caused
by variation mix can be partially overcome by using aggregated capacity
measures (remember that ‘aggregated’ means that different products and
services are bundled together in order to get a broad view of demand and
capacity).

9
The effect of time frame on capacity measurement
The level of activity and output that may be achievable over short periods
of time is not the same as the capacity that is sustainable on a regular
basis.

Design capacity is the theoretical capacity of an operation that its

technical designers had in mind when they commissioned it.

Effective capacity is the capacity of an operation after planned losses

are accounted for.

Actual output is the capacity of an operation after both planned and

unplanned losses are accounted for. (Quality problems, machine
breakdowns, absenteeism)

10
Capacity ‘leakage’
This reduction in capacity, caused by both predictable and unpredictable
losses, is sometimes called ‘capacity leakage’ and one popular method of
assessing this leakage is the overall equipment effectiveness (OEE)
measure that is calculated as follows:

OEE = a x p x q

where a is the availability of a process, p is the performance or speed of a

process and q is the quality of product or services that the process creates.

OEE works on the assumption that some capacity leakage occurs, causing
reduced availability.

For processes to operate effectively, they need to achieve high levels of

performance against all three dimensions – availability, performance
(speed) and quality.

Viewed in isolation, these individual metrics are important indicators of

performance, but they do not give a complete picture of the process’s
overall effectiveness.

11
Understanding changes in capacity
Capacity management decisions should reflect both predictable and
unpredictable variations in capacity and demand.

How is the demand side managed?

Demand patterns clearly have a big influence on the way operations
function and therefore many organisations will seek to influence them in
some way. Referred to as demand management, this involves changing the
pattern of demand to bring it closer to available capacity.

This achieved – either by stimulating off-peak demand or by constraining

peak demand. There are a number of methods used to manage demand:

12
  Price differentials – adjusting price to reflect demand. (Surge
pricing = dynamic pricing method where prices are temporarily
increased as a reaction to increased demand and a limited supply)

 Scheduling promotion – varying the degree of market stimulation

through promotion and advertising in order to encourage demand
during normally low periods.

 Constraining customer access – customers may only be allowed

access to the operation’s products or services at particular times.

 Service differentials – allowing service levels to reflect demand

(implicitly or explicitly) by letting service deteriorate in periods of
high demand and increase in periods of low demand.

 Creating alternative products or services – developing services or

products aimed at filling capacity in quiet periods.

Yield management
In operations that have relatively inflexible capacities, such as airlines and
hotels, it is important to use the capacity of the operation for generating
revenue to its full potential. One approach used by such operations is
called yield management. This is really a collection of methods, some of
which we have already discussed, which can be used to ensure that an
operation maximises its potential to generate profit.

Yield management is especially useful where capacity is relatively fixed;

the market can be fairly clearly segmented; the service cannot be stored in

13
any way; the service is sold in advance; and the marginal cost of making a
sale is relatively low.

How is the supply side managed?

Here, decisions include setting the base capacity level, and then using two
key methods of managing supply;

1) level capacity plans , where nominal capacity is kept constant.

2) chase capacity plans, where capacity is adjusted to ‘chase’

fluctuations in demand over time.

Setting base capacity

The most common starting point in managing the supply side is to decide
the ‘base level’ of capacity and then adjust it periodically up or down to
reflect base level:
fluctuations in demand. Three factors are important to consider in setting
this base level:

1) The operation’s performance objectives.

2) Perishability of the operation’s outputs.

3) Variability in demand or supply.

The effect of performance objectives on the base level

Base levels of capacity should be set primarily to reflect an operation’s
performance objectives.

Setting the base level of capacity high above average demand will result in
relatively low levels of utilisation of capacity When an operation’s fixed
costs are high, underutilisation has significant detrimental effects.

14
Conversely, high base levels of capacity result in a capacity ‘cushion’ for
much of the time, so the ability to flex output to give responsive customer
service will be enhanced.

When the output from the operation is capable of being stored, there may
also be a trade-off between fixed capital and working capital in where base
capacity level is set.

A high level of base capacity can require considerable investment, while a

lower base level would reduce the need for capital investment but may
require inventory to be built up to satisfy future demand, thus increasing
working capital.

The effect of perishability on the base level

When either supply or demand is perishable, base capacity will need to be
set at a relatively high level because inputs to the operation or outputs
from the operation cannot be stored for long periods.

- A hotel cannot store its accommodation services. If an individual

hotel room remains unoccupied, the ability to sell for that night has
‘perished’. In fact, unless a hotel is fully occupied every single night,
its capacity is always going to be higher than the average demand for
its services.

15
The effect of demand or supply variability on the base
level
Variability, either in demand or capacity will reduce the ability of an
operation to process its inputs.

As a reminder, the greater the variability in arrival time (demand) or

activity time (supply) at a process, the more the process will suffer both
high throughput times and reduced utilisation.

Because long throughput times mean that queues will build up in the
operation, high variability also affects inventory levels.

The implication of this is that the greater the variability, the more extra
capacity will need to be provided to compensate for the reduced utilisation
of available capacity.

Level capacity plan

Once base capacity is set, the first alternative supply-side approach is a
‘level capacity plan’, where capacity is fixed throughout the planning
period regardless of the fluctuations in forecast demand.

Level capacity plans offer stable employment patterns, high process

utilisation and often high productivity with low unit costs.

Unfortunately, they can also create considerable inventories of materials,

customers or information.

!!!!) Level capacity plans are not well suited to ‘perishable’ products, such
as foods and some pharmaceuticals, for products where fashion changes
rapidly and unpredictably (for example, fashion garments) or for
customised products.

!!!!!!) The higher the base level of capacity, the less capacity fluctuation is
needed to satisfy demand.

 Level Capacity Plan:

• keep output stable

• in times of low demand, store unsold output

• in times of high demand, sell from inventory

16
 • staff schedules fixed

• materials inflow stable

• higher assets utilization

Chase (demand) capacity plan

In contrast to level capacity plans, chase capacity plans attempt to match
demand patterns closely by varying levels of capacity.

Chase capacity strategies are much more challenging than level capacity
plans, as different numbers of staff, different working hours and even
different amounts of equipment may be necessary in each period.

A pure chase plan is more usually adopted by operations that are not able
to store their output, such as some customer-processing operations or
manufacturers of perishable products. It avoids the wasteful provision of
excess staff that occurs with a level capacity plan, and yet should satisfy
customer demand throughout the planned period.

Where output can be stored, the chase demand policy might be adopted
in order to minimise or eliminate finished goods inventory, especially if
the nature of future demand (in terms of volume or mix) is relatively
unpredictable.

17
!!!!!!!) The
‘chase’ (demand) approach is most useful when output cannot
be stored or when demand is both volatile and unpredictable.

 Chase Capacity Plan:

• match capacity to forecast demand
• does not work if storing the product or
service is hard or impossible
• possibly lower asset utilization
• requires staff flexibility
requires active management approach
to manage adjustments

How can operations understand the

consequences of their capacity management
decisions?
When making capacity management decisions, managers are attempting to
balance the need to provide a responsive and customer-oriented service
with the need to minimise costs.

For example, an accounting firm may seek to bring forward some of its
peak demand by offering discounts to selected clients (demand
management plan).

Capacity may also be increased through the use of outsourced suppliers

during the busiest months of the year (chase capacity plan).

18
However, some capacity may be constrained (for example, specialist
advisory services offered by the firm) and therefore clients may still
experience delays during high demand periods (level capacity plan).

!!!!!!) Most organisations mix demand-side (demand management and

yield management) and supply-side (level and chase plans) capacity
management strategies to maximise performance.

Four methods to examine consequences of capacity management

decisions:

 factoring in predictable versus unpredictable demand variation;

 using cumulative representations of demand and capacity;

 using queuing principles to make capacity management decisions;

outlooks.

 taking a longitudinal perspective that considers short- and long-term

outlooks.

19
20
21