UNIT 5 PRODUCTION PLANNING AND

CONTROL
Forecasting – Qualitative and Quantitative forecasting techniques – Types of production
– Process planning – Economic Batch Quantity– Loading – Scheduling and control of
production – Dispatching–Progress control.
5.1 INTRODUCTION
Forecasting is the first major activity in the planning. It involves careful study of past data and present
scenario. The main purpose of forecasting is to estimate the occurrence, timing, or magnitude of future
events. For example, the trend of past ten years in the demand of cars and corresponding purchasing
power of the consumers may form a basis of forecasting the demand of cars during next year. Once, the
reliable forecast for the demand is available, a good planning of activities is needed to meet the future
demand. Forecasting, thus, provides the input to the planning and scheduling process.
Precise forecasting of economic activities, such as product demand, is almost impossible because of
many interactive factors, which are difficult to model. Despite the fact that highly reliable forecast is
unrealistic, the approximate estimate forms the basis of planning process.
5.2 BENEFITS OF FORECASTING
Good forecast of material, labour and other resources for operation are essentially needed by the
managers.
If good projection of future demand is available, the management may take suitable action regarding
inventory. Similarly, if production activities are accurately forecasted, then balanced work-load may be
planned. Good labour relations may be maintained as there would be lesser hiring and firing activities
by the management with better manpower planning. Therefore, forecasting is useful due to following
benefits:
1. Effective handling of uncertainty
2. Better labour relations
3. Balanced work-load
4. Minimisation in the fluctuations of production
5. Better use of production facilities
6. Better material management
7. Better customer service
8. Better utilisation of capital and resources
9. Better design of facilities and production system.
Efforts in forecasting activity involve two types of costs. While more effort in forecasting causes
increased cost due to data collection and analysis, lesser forecasting activity involves lost revenue,
which may be due to unplanned labour, unplanned material or unplanned capital cost (Figure 5.1).
Therefore, each firm should maintain a balance in its forecasting effort and stick to a zone near to
accuracy cost trade off (Figure 5.2).
Difference between Forecasting and Prediction
Forecasting and prediction are two different things (Table 5.1). Forecasting is objective, scientific,
reproduce, free from individual bias and error analysis is possible in it. Prediction, on the other hand, is
subjective, mostly intuitive, non-reproducible and contains individual bias. Only limited error analysis
is possible in prediction.
Table 5.1 Difference between Forecasting and Prediction

5.3 TYPES OF FORECASTING

The forecasting may be classified on the basis of time span or range of forecast. Three categories may
be identified as follows:
5.3.1 Long-Range Forecasting
Long-range forecasting consists of time period of more than 5 years. Characteristics: Normally, it is
difficult to model and foresee events for more than five years. It is mainly due to economic uncertainty
and variation in the behaviour of the interrelated processes. For example, state of economy and
technology may completely change in next five years and, therefore, the trend of data during past few
years may not be sufficient.
Applications: Long-range forecasting is useful in the following areas:
• Capital planning
• Plant location
• Plant layout or expansion
• New product planning
• Research and development planning
• Technology management, etc.
Method of Forecasting: Long-range planning is a difficult task. Generally, these forecasts are broad in
nature and general type in characteristic. Mostly, qualitative techniques are used. Studies, related to
technological break-through, economic studies, marketing survey, demographic projections, etc., are
used to make judgemental estimate of the future event.
5.3.2 Medium-Range Forecasting
The range for medium-range forecasting is generally 1 to 5 years.
Characteristics: As the range of forecast shortens from .5 to 1 year, the accuracy of forecast increases.
This is due to better understanding of future and relatively lesser uncertainty. For these forecasts, more
numerical estimates are needed. Estimate of reliability of forecast may be useful in medium-range
forecast.
Applications: Medium range forecast is vary useful in following areas:
• Sales planning and sales force decisions
• Productions planning
• Capital and cash planning
• Inventory planning
• Enrollment of students in a college, etc.
Method of frecasting: Medium-range forecasting needs judgement as well as time series analysis.
Combination of collective opinion, regression analysis, correlation of different index and inflation, etc.,
may be useful in forecasting.
5.3.3 Short-Range Forecasting
The ange for short-range forecasting is typically less—from one hour to one year. In most cases, it is
for one season, a few months or a few weeks.
Characteristics: The short-range forecasting is needed at detailed level, such as demand of specific
items. This forecast may affect the purchasing activity. Specific value of forecast is needed. There is
very less sLope of judgement in short-range forecast and, therefore, past data are mainly projected into
future.
Applications: Short-range forecasting is commonly used in immediate control of activities. Some
related applications are:
• Purchasing
• Overtime decisions
• Scheduling of job
• Machine maintenance, etc.
Methods of Forecasting: Short-range forecasting is based on past data. The trend of data is projected
or extrapolated into future. For this exponential smoothing, graphical projections, part explosion into
product family, etc., are used. For example, monthly forecast of sales may form the basis of production
planning activities.
5.4 COMMONLY OBSERVED DEMAND PATTERN
Forecasting demand of a product is generally associated with the pattern of its past demand. Different
products have different demands. With growing population, the demand for housing is likely to
increase.
But the demand for some products, for examples raincoat or umbrellas also depends upon season. In
rainy season, their demand is quite high as compared to winter season. This is a seasonal demand
pattern.
Various demand patterns are shown in Figure 5.3.
Some demand patterns are abnormal for which forecasting is difficult. For example, transient impulse,
sudden rise or sudden fall in the pattern. These may occur due to some unforeseen reasons like war or
natural calamities.
5.5 QUALITATIVE METHODS OF FORECASTING
Qualitative methods are needed in forecasting when data, necessary to use time series or causal model,
are not available. For example, when a new product is to be launched in the market, its past demand
data are not available. Therefore, time series trend analysis is impractical. Qualitative techniques, which
incorporate human judgement, expert opinion, management intuition, market research, historical
analogy or grass root forecasting are useful in such cases. Some approaches are as follows:
5.5.1 Delphi Method
In this method, a panel of outside experts is identified. They are given a series of structured
questionnaires.
The answers of each questionnaire are used as input for the design of the next questionnaire. The identity
of experts is not disclosed. This is for the purpose that nobody should influence the opinion of others.
The coordinator of the project prepares the statistical summary of responses. This, along with the
support for the responses, is provided to the experts in the next round. The participants are asked if they
want to modify their previous response. In this way, after few rounds of questionnaires, the final forecast
is derived.
Delphi is used for long-range forecast. It is generally used for new product demand, technological
forecast for new technology, effect of scientific advances, changes in society, changes in competitive
environment etc. For example, the effect of intemet/intranet or information-highway in the educational
system of India in next 25 years may be forecasted through this approach.
Advantages of Delphi Method: The advantages of Delphi methd are:
• It is effective, when past data is absent.
• It does not require experts to meet in person.
• It is extremely useful for the forecast of new technology or new product.
Limitations of Delphi Method: The limitations of Delphi method are as follows:
• It is a time consuming process, which may be around one year. During this period, the experts may
change their perception. Sometimes, the very need of forecasting loses its significance due to the delay.
• As experts are not accountable, their response may be less meaningful.
• If the questionnaires are poorly designed, Delphi would be ineffective.
• Accuracy or reliability of forecast is relatively poor in Delphi method. Therefore, it should only be
used when past trend is absent and quantitative models are difficult to use.
5.5.2 Market Research
It is used to determine consumer liking in a product or service. A set of hypothesis is tested through
the data, which is generated in the survey.
5.5.3 Salesforce Forecast
Salesforce is a team that is closest to the customer. Their estimates are compiled to assess the future
demand. 1, pharmaceutical market, the estimates given by medical representatives of all territories are
oftenly us to determine the sales forecast of a particular medicine.
5.5.4 Historical Analogy
It is used when the new product or new technology is strongly similar to an established product whose
demand dat-1 is known. This. approach is effective for medium to long-range forecast and it is quite
cost effective.

ACCURACY OF FORECAST.
It is almost impossible to obtain an exactly right forecast every time. This is due to many factors, which
affect the trend in data. It is difficult to capture the exact interrelation of these influencing factors.
Therefore, some error in forecasted value and actual value is quite common.
Sometimes, it is important to know if the forecaster (a forecasting technique) is unbiased or not. An
unbiased model should overestimate or underestimate the forecast in almost equal ratio.
5.6.1 Measures of Forecasting Error
1. Mean Absolute Deviation (MAD):' This is calculated as the average of absolute value of difference
between actual and forecasted value. The negative sign in this difference is ignored as overestimate as
well as underestimate are both off-target and thus undesirable.
2. Mean Sum of Square Error (MSE): The average of square of all errors in the forecast is termed as
MSE. Its interpretation is same as MAD.

3. Bias: Bias is measure of overestimation or underestimation. A positive bias indicates under-

estimation while a negative bias indicates overestimation.

4. Tracking Signal (TS): It is used to identify those items, which do not keep pace with either positive
or negative bias or trend.

5.7 QUANTITATIVE METHODS OF FORECASTING

5.7.1 Extrapolation
Extrapolation is one of the easiest ways to forecast. For example, based on the past few values of a
production capacity, next value may be extrapolated on a graph paper. This may be done by extending
the curve (or line) joining the already known values. For example, if the production capacity of a firm
has been 445, 545 and 645, then in the next year one may expect a production capacity requirement of
745 units (Figure 5.4).
The limitation with the extrapolation method is its inability to deal with non-linear trend and swing
in the pattern of past data.
Time Series Analysis: There are some models in forecasting which involve analysis of past data
or happenings. These models are as follows.
1. Simple moving average
2. Weighted moving average
3. Exponential smoothing
4. Double exponential smoothing.
5.7.2 Simple Moving Average (SMA)
Following approach is followed in simple moving average method:
Compute the mean of only a specified number of consecutive data which are most recent values in
series. Call this Ft. This F would be the forecast for next period.
For example, a 5-month forecast of moving average should account for the last five values of demand
in the demand forecasting. In general, the forecast at the end of t periods, the n period, simple moving
average forecast for (t + 1) th period is given by

where, Di is the actual demand for the i th period n is the number of periods included in each average.
Example 5.1 Forecast for 3-Month Simple Moving Average.
It may be observed in SMA approach:
1. The longer the moving-average period, the greater the random elements smoothening.
2. In case of trend (increasing/decreasing), the SMA has adverse trend. This is due to lagging trend.
3. The longer is the time span, the smoother is the forecast but with lagging trend.
4. Simple moving average method involves quite large data handling as we go fol. large period average
(Figure 5.5).
5.7.3 Weighted Moving Average
This approach is based on the principle that more weightage should be given to relatively newer data.
The forecast is the weighted • average of data. Thus:
It may be noted that when more weight is given to the recent values, the forecast is nearer to the likely
trend. Weighted moving average is advantageous as compared to simple moving average as it is able to
give more importance to recent data.
5.7.4 Exponential Smoothing
In exponential smoothing method, the weightage of the data diminishes exponentially as the data
become older. In simple moving average, the only few past data are accounted. However, in exponential
smoothing, all past data are accounted. The weightage of every previous data decreases by (1 - a) times,
where a is called as exponential smoothing constant. For example, when a is equal to 0.3, the weightage
of last period data is 0.3 and weightage of last to last period data is 0.3 (1 - 0.3) or 0.21.
Example 5.4 Comparison of Exponential Smoothing Forecast for Different Values of Smoothing
Constant. Let us consider demand of an item for ten weeks, given in the following table. The forecast
for different value of a is also shown.

Following observations pertain to the exponential smoothing constant (a):

• Smaller is the value of a, more is the smoothing effect in forecast.
• Higher value of a gives more robust forecast and response more quickly to changes.
• Higher value of a gives more weightage to past data as compared to smaller value
of a.