Int. J.

Production Economics 85 (2003) 331–346

Strategy not tactics drives aggregate planning

Geoff Buxey*
Faculty of Business and Law, Deakin University, Geelong, Vic. 3217, Australia

Abstract

The aggregate planning model supposedly shows how manufacturers cope with seasonally biased sales.
Unfortunately, industry has failed to embrace any of the sophisticated algorithms that were developed to solve the
corresponding resources allocation problem. This paper reveals why such methods have proven so unattractive.
Aggregate planning is a chimera. In practice, planners construct the master production schedule directly, in line with a
preferred production strategy. A ‘‘chase’’ plan is the most popular choice. When this option proves infeasible,
management plumps for another predetermined strategy. The resultant stockpiling brings certain financial risks.
However, companies take various measures to reduce their exposure.
r 2003 Elsevier Science B.V. All rights reserved.

Keywords: Aggregate planning; Seasonal sales; Strategy; Case studies

1. Aggregate planning theory expected monthly sales, and computes the

corre- sponding direct labour requirements.
The demand for many products varies in a Plant and equipment is a given entity. Therefore,
predictable manner over a 12 months cycle, due the relevant options to raise or lower production
to the weather, Xmas shopping, the sporting capacity are overtime or undertime, hiring or
calen- dar, etc. Therefore, manufacturers firing to alter manning levels or the number of
experience difficulties in scheduling production shifts that operate, and subcontracting work
to satisfy anticipated sales, whilst ensuring that out. The best plan incurs the minimum total
sufficient resources will always be on hand to marginal cost over a calendar year. It achieves
execute the plan. Holt et al. (1955) defined this end by balancing stockholding charges
the aggregate planning model, using a paint against additional labour expenses and
factory as an example. The basic concept is premiums. All sales forecasts, inventory
that the most economical solution comprises a levels, labour inputs and production rates are
mix of two pure production strategies. A level expressed in a suitable aggregate format. The
plan maintains a steady production rate over purpose is to simplify the myriad of calcula-
the entire year, using finished goods tions.
(smoothing/anticipation) stocks to absorb This model has been accepted in the literature
ongoing differences between output and sales. with few reservations. Indeed, researchers
The chase alternative simply tracks the continue to use it as a platform to demonstrate
the efficacy of various analytic methods.
*Tel.: +61-3-5227-1421; fax: +61-3-5227-2151. Examples include (the original) linear decision
E-mail address: [email protected] (G. Buxey). rules derived from differential calculus (Holt
et al., 1955), and
0925-5273/03/$ - see front matter r 2003 Elsevier Science B.V. All rights
reserved. doi:10.1016/S0925-5273(03)00120-8
33 G. Buxey / Int. J. Production Economics 85 (2003) 331–346 33
2 2
simulation (Jones, 1967; Taubert, 1968), produc- ished goods inventory levels. To yield a tangible
tion switching rules (Elmaleh and Eilon, 1974; schedule, the aggregate production totals should
Mellichamp and Love, 1978), and several kinds of be decomposed back into the original
mathematical programming (Bowman, 1956; constituent products. At the same time, the
Akinc and Roodman, 1986). An apparent lack of master production schedule (MPS) must
any real applications has been reported (Nam and conform to the previously configured
Logendran, 1992), and this has provided grounds (aggregate) marginal cost structure. For
for minor modifications. For instance, goal complex assemblies, materials requirements
programming (Deckro and Hebert, 1984; planning (MRP) software ‘‘explodes’’ the MPS,
Gilgeous, 1989) permits ‘‘soft’’ policy objectives generating detailed purchasing and production
to be included in the model. Two postal surveys schedules for all dependent parts and raw materi-
were undertaken to discover why the als. Finally, just-in-time (JIT) production is lauded
practitioners have consistently ignored the as the ideal system for repetitive manufacture. The
purported benefits of the different mathematical implementation of a JIT system drastically reduces
techniques. The find- ings led Gilgeous (1987) to raw materials and intermediate stocks. It is
propose extending the original model to noteworthy that the underlying operations man-
accommodate the interests of other functional agement philosophy confounds the long estab-
areas like human resources, marketing, and lished, cost balancing, economic order quantity
finance. On the other hand, DuBois and Oliff (EOQ) model.
(1991) concluded that potential industrial users The modern books previously cited
did not possess reliable enough sales forecasts recommend that firms influence the market in
and cost information to make rigorous a way that ameliorates their seasonal sales
algorithms worthwhile. Also, managers lack the patterns. The relevant steps include slashing
necessary mathematical expertise. prices during quiet periods, spending more on
Three later papers assume that the operational advertising, or, devel- oping a counter-seasonal
bounds of the model are set by fixed product range. A good example of the latter is
management policies. Hwang and Cha (1995) greetings cards, designed for different occasions
modified the switching rule to stay within spread throughout the year (Slack et al., 2001).
desirable inventory levels, and Silva et al. (2000) The texts explain conventional aggregate planning
adapted the linear decision rule to find the logic, and go on to describe several well-known
optimum number of permanent employees. A analytical techniques. Never- theless, the majority
flexible labour contract underpins another stable recommends a trial and error approach for
workforce model (Hung, practitioners. There is a distinct absence of
1999). The method provides a pure ‘‘chase’’ guidelines though for extracting the MPS from
solution. the aggregate plan, and any worked examples
are trivial. Typically, the MPS is conjured
up from nowhere in particular, and used as a
2. The steps of production vehicle to demonstrate the logic of MRP. No
planning links are established between aggregate planning
and JIT production. Nor are they even suggested.
When standard texts (Schmenner, 1993; Slack Hierarchical production planning (HPP) has
et al., 2001; Heizer and Render, 1999; Gaither and succeeded in integrating the development of the
Frazier, 1999; Schroeder, 2000; Dilworth, MPS with the aggregate plan (Bitran et al., 1982;
2000) discuss production planning they commence Hennet, 1999). The methods concentrate on
with a long-range business plan. This matches capacity planning and are normally restricted to
capital expenditures against projected broad one or two-stage manufacturing processes. A 0–1
market requirements. The exercise determines a integer-programming model by Chung and
plant’s maximum capacity limits and expected Krajewski (1987) expresses the aggregate plan in
average utilisation, and confirms its economic terms of product families. Families are groups of
viability. Next, the medium term aggregate plan products that take advantage of common machine
sets the overall monthly production rates, and
the corre- sponding monthly workforce
numbers and fin-
setups, and the model incorporates the corre- rithm is an extension to the production-switching
sponding costs. A second LP model breaks the rule, which is not recommended for pronounced
initial monthly schedule into a weekly MPS. seasonal sales swings. Management nominated
Ozdamar et al. (1996) merely state the aggregate several feasible output/capacity configurations,
plan. Then their algorithm focuses on minimising and set an appropriate inventory range. Occasion-
family batch changeover costs. After enforcing a ally, the impending month’s sales forecast triggers
level workforce constraint, Venkataraman and the allocation of more (or less) labour to opera-
Smith (1996) found an aggregate LP solution for tions. A second algorithm compiles a daily produc-
another paint factory problem. An integer goal- tion schedule. It assigns woven fibreglass
programming model applied minimum batch size product families to a number of semi-dedicated
rules, and translated it into a production schedule looms, and finds combinations for efficient
comprising 17 different product families. simultaneous batch production. The mat
The rubric of the American Production and fibreglass product groups are directed to
Inventory Control Society (APICS) is somewhat at appropriate production lines, whence HPP finds
odds with the aforementioned planning the best batch sequences. A giant apparel
methodol- ogy (Everdell, 1984). There is no manufacturer used to carry excessive finished
aggregate plan. Instead, the weekly MPS is the goods inventories (Edwards et al., 1985).
vehicle to set stocks and manning levels, reconcile Single product groups like jeans break down
workloads with capacities, meet productivity into lines (e.g. men’s corduroy), styles (e.g.
targets, and, satisfy market demands. Real straight leg), lots (e.g. blue), and sizes. There are
products should be speci- fied right away, up to 75 lots per line, leading to more than
unless there are more than one hundred. If so, 10,000 stock keeping units (SKUs). Nowadays
it is permissible to use a more manageable seven distinct (management science) models are
number of broad types at first. Every applied. However, the company insisted on a
4 weeks, at least, the rolling MPS is regenerated, in level workforce strategy from the outset.
order to react to fresh sales and production data. Annual monthly sales forecasts are prepared for
The right degree of flexibility should always be each lot and corresponding safety stock figures
maintained. Accordingly, only the ‘‘live’’ portion calculated. Using aggregate data, a simulation
is (semi) frozen. As the furthest sections gradually model pro- duces a level production plan for
move towards the front end they become more each complete product line. The next model
product specific and the MPS becomes less translates the target monthly line inventories
flexible. Even then minor product details like (smoothing stock plus safety stock) back into lot
colour and packaging are ignored, until week 1 is format. Staple lots are allocated proportionally
partitioned into a daily production schedule. A more smoothing stocks than their less
set of rigid, fully integrated, optimum plans is predictable, fashion counterparts. A further step
clearly not what is wanted. Furthermore, product divides the lots into the constituent sizes, before
aggregation and disaggregation never occurs. a weekly production schedule is generated that
Two recent articles (Enns, 2001; Shtub, 2001) incorporates economic batch quan- tities. The
describe how to design an integrated, basic MRP final (LP) model selects a cost-effective set of
I system, and a more elaborate enterprise weekly markers for cutting the required
resource planning (ERP) system, respectively. fabrics. In this instance, a somewhat limited degree
Both schemes dispense with aggregate planning. of model aggregation and disaggregation does take
place. However, the main aim of the exercise is to
level production. After that the idea is to
3. Relevant applications minimise the risks associated with accumulating
smoothing stocks. At no stage are any labour–
A lone HPP implementation (Burch et al., 1987) inventory type cost balancing calculations
demonstrates the importance of batch changeovers involved.
in process industries. The aggregate planning algo- The second edition of Slack et al. (1998)
showcases a highly automated canned soft drinks
plant. Production has to cope with both seasonal
sales swings and short-term, weather-related highly constrained process of product disaggrega-
peaks. The warehouse holds a mere 5-day’s output tion a realistic proposition? How does the
and the beverages have a limited shelf life. aggre- gate planning model fit into the overall
Management adopts a pure chase policy. How- scheme of things? Can we reconcile its medium
ever, accepting some idle time during winter allows term rigidities with the need for short-term
the operation to maintain a constant workforce. scheduling flexibility? Finally, why is there no
Another authentic situation is described in a later nexus to JIT production? The JIT credo treats
edition of the book (Slack et al., 2001). This inventories as ‘‘waste’’. So why give planners
factory produces cheese. The output is compelled carte blanche regarding smooth- ing stock
to ‘‘chase’’ the seasonal variations that occur in deployment? After all, finished goods are the most
the milk supply. Management negotiated an expensive form of inventory.
annual- ised working hours arrangement, so as to To provide some answers, a study was made of
stabilise the employment level. 42 manufacturers that experience seasonally
Several articles (Silva et al., 2000; Vergin, 1966; biased demand. The aim was to cover a wide
Lee and Khumawala, 1974) offer proof that range of industries and organisations. Taking a
mathematical techniques will outperform current case based approach means that the sample is
industrial practices. The former are fed a test much richer in information content than both of
company’s actual aggregate forecast or historical the surveys previously cited. Nevertheless, it is of
demand figures. For cost comparison purposes, comparable size. An interview with a senior
the researchers transform the relevant real life manager(s) was followed up by a plant tour in
production schedule into an assumed aggregate order to make observations and prompt further
plan. Thus, they avoid having to generate their questions. No substantive body of evidence exists
own MPS. On top of this, the methodology to indicate that the aggregate planning model
overlooks the fact that it is much easier to represents normal industrial practice. Therefore,
predict aggregate demand than actual (model) the discussions avoided preconceptions. The topics
sales. The typical MPS is also subject to further encompassed the company’s business environ-
batch size and sequence constraints. These tend to ment, strategy for dealing with seasonal
inflate the levels of finished goods stocks. Paint variations, the objectives and broad format of
manufacture schedules are prone to this production planning, and resources scheduling.
debilitating effect (Hill et al., 2000). Also included were the development of the
MPS, shop floor constraints, management issues,
and the nature of the volume and mix flexibility.
4. Research objectives, methodology and Plans can have a strategic, tactical or opera-
summary findings tional dimension. Orthodox aggregate planning
focuses exclusively on the compromise ‘‘mixed’’
A worrying gap obviously exists between strategy option. As a consequence it is only
aggregate planning theory and industrial practice. concerned with tactics. A number of possible
For the literature the main issue is whether or not strategies can be inferred from the literature and
the current mathematical methods are robust the study (level; chase; mixed; demand manage-
enough. In other words, can they accommodate ment; modified chase; modified level; level work-
all the factors that managers routinely take into force; level core workforce). The ‘‘modified’’
account? A broader consideration of production strategies stem from case-specific operational and
planning suggests a more fundamental question. business constraints. These factors may eliminate
Perhaps the original aggregate planning model is any chance of matching outputs to peak monthly
at fault, rather than a host of derived algorithms. demands, or prevent a factory from completely
This would explain the dearth of published smoothing out its workload.
applications, covering a span of almost 50 years. Faced with a seasonal production problem,
There are some secondary issues as well. Is a each firm in the sample abides by a
predeter- mined strategy. Remarkably, about
three quarters
Table 1 weather is notoriously fickle, and if sales fall well
Strategies selected by sa mple firms below peak forecasts the enterprise could be stuck
Strategy Number of % Cumulative % with surplus goods for the best part of a year.
cases In addition, many consumer products run the risk
Chase 19 45.2 45.2 of obsolescence. A chase plan implies minimum
Modified chase 12 28.6 73.8 financial exposure. Also, it is important to be able
Demand management 4 9.5 83.3 to react quickly, in the event that demand has been
Level 4 9.5 92.8 underestimated. Retailers will not accept back-
Other 3 7.1 99.9
Mixed 0 0.0 99.9 orders, once their own critical sales period has
Total 42 100 100 passed. Certain companies make a high variety of
similar products, or, their designs embody an
element of fashion. These factors compromise
selected a chase or modified chase option, and more accurate sales forecasting and amplify the risks
than 80% have adopted a policy that tries to of stockpiling. A ‘‘make to order’’ arrangement
keep minimum quantities of stocks (Table 1). negates this particular issue, but completely pre-
Not a single enterprise has implemented a mixed cludes building up anticipation stocks. Some
strategy, or even attempted to balance marginal wares are bulky or valuable, perishable, or
labour costs against imputed charges for hard to handle. These characteristics make it
smoothing inventories. Appendix A presents a list impossible or undesirable to store them for long
of all 42 manufacturers, with a brief outline of the or in mean- ingful quantities. Contemporary
most pertinent factors as far as responding to business is well aware of the benefits of the JIT
seasonal sales is concerned. The cases are approach. In the present climate, substantial
classified and sorted according to their chosen quantities of finished goods stocks are an
strategy, which makes it easier to discuss the anathema.
pertinent underlying reasons. Some detailed A manufacturer of jewellery offers an extremely
descriptions are included as well. These serve wide selection to chain stores and buyer
to illustrate the overall methodology, the groups that represent the small retailers. Standard
idiosyncratic nature of the investigations, and the pieces include earrings, rings, lockets, brooches,
complexity of the results. The ways that bracelets, etc., made of 9, 14 and 18-carat gold,
different aspects of specific cases interact add silver, or plated metal and set with various
some sub- stance to the general arguments precious stones. Cash flow is a problem, due to
presented. the high value of the raw materials. Customers are
It would be most convenient if a certain often late settling accounts, but some suppliers
product or production category could always be want their money immediately. A debt-collecting
associated with a specific seasonal strategy. agency ensures that 75% of clients pay within
Unfortunately, this is not so. All plants possess 60 days. Diamonds and other raw
limited volume flexibility, due in the main to high materials are purchased at a few days notice.
average capacity utilisation or the extensive Gold can be obtained the next day. However,
training that new labour requires. The first certain precious stones must be sourced
constraint depends on the fixed cost of the overseas with a 4–6 weeks lead time. The
production equipment. Nevertheless, the two invoice situation means that gold is sometimes
factors are not related. The magnitude of the bought in less than ideal quantities for
seasonal sales swing has to be considered too. melting. Manufacturing is a batch process
This effect is definitely indepen- dent of any involving several departments. Hinges, brackets,
normal industrial classification. pins and similar components are pro- duced in
economic lot sizes and kept in stock, since casting
and press setups are time consuming. Some semi-
5. The chase strategy finished pieces are held in stores too, minus their
expensive stones. This cuts the time to fill
A strong seasonal demand pattern magnifies the customers’ orders.
consequences of over stocking. For example, the
 Demand peaks when shops stock up for Xmas Table 2
and December’s sales are about eight times the Reasons for selecting modified chase
January figure. There is relatively little work from Constraint Number of cases %
May through August, despite buyers being offered
Equipment 6 50.0
deals to level the profile and plenty of long-range Training time 4 33.3
orders being received. The majority of the items Training capacity 1 8.3
are made to order, although stock replenishment Seasonal raw materials 1 8.3
applies to the higher-volume, lower-variety, cheap- Total 12 100

er lines. However, no seasonal smoothing stocks

are maintained. The planning system promises to
dispatch a regular order within 14 days of clock. The situation reflects a pre-existing financial
receipt and also accepts rush jobs. Typically, trade off between capital investment, the efficient
an order comprises a mix of different items, and utilisation of capacity, and money tied up in
customers often stagger their delivery dates. For warehouse stock. Almost as many firms find it
budgetary purposes, a rolling 12 months forecast difficult to ramp up their labour input for short
is compiled by product group. It is quite reliable, periods. Principally, this is because skilled
apart from new products, but due to the need to workers are unwilling to accept limited seasonal
turn orders around quickly the actual production employ- ment and inexperienced recruits need
plan only covers the first 20 days. Every extensive training. In addition, the training
morning (consoli- dated) orders are released to department may not be able to cope with the
the shop floor. They will have arrived in the required intake rate. At one particular factory
previous day’s mail. production must respond to the seasonal
The factory employs 34 people on a single shift, availability of perishable crops. Therefore, the
including 8 jewelers and one apprentice. A 1-week enterprise cannot avoid keeping large
training period suffices for the packers and many inventories of the corresponding finished
other operations can be learnt in a day. Approxi- goods.
mately half of the workforce is employed on a Planners acting under modified chase con-
casual basis. These employees can be called in for straints commonly engage in forms of stockpiling
just a morning’s work although their hourly wage that carry negligible financial risk. For example,
is higher. At weekends, bench assembly tasks can customers lodge export orders, indent orders and
be performed at home, paid at the normal rate. very big orders well in advance of requested
Nevertheless, home-based local subcontractors delivery dates. Such jobs can be slotted into the
provide the greatest volume flexibility. This mode MPS earlier than necessary. This adjustment
can be cheaper too because no overheads are transfers work away from overloaded periods
incurred. Overtime is never planned but it is and into the slack ones. The resultant stockpiles
available in emergency situations. A 3 weeks still tie up cash temporarily, but there is no
vacation shutdown occurs in the January trough, danger of producing unwanted or obsolete
and a further week of holidays is scheduled for products.
whatever is expected to be the next quietest period. A supplier of pharmaceutical cre" mes,
liquids, sterile products, tablets and capsules
sells brand name products from stock to medical
6. The modified chase strategy wholesalers. In addition, 35–40% of the
business is contract manufacture using the
Frequently, the chase option is abandoned in clients’ own recipes and labels. Aggregate sales
favour of a modified chase strategy. The reasons peak during November and December because the
for this are summarised in Table 2. In 50% of public buys pharmaceutical
these operations there is insufficient equipment to items to take on holiday. Also, pensioners use up
accommodate maximum monthly demand, given their free scripts at the end of the year. Sales in
that not all firms are able to run right round the January are correspondingly flat, compounded by
the reluctance of vacationers to consult a different
doctor. About 60 products (10% of total) suffer
more pronounced (2–3-fold) monthly swings. For production schedule, specified in weeks. The
instance, strong spring and autumn winds lift the minimum manufacturing lead time is 1 week for
sales of eye drops and ointments, while more rush orders. A degree of flexibility is gained by
antibiotics and cough mixtures are bought in maintaining a 2–3 week safety stock buffer of
winter. The actual sales volumes depend greatly proprietary products. Unplanned overtime can
on the weather, and the random spread of germs address forecast errors, if it proves necessary.
and viruses. For branded products, Most tasks entail machine minding, materials
unpredictabil- ity is the biggest planning handling or packing. These activities necessitate
problem. On the other hand, the established just a few hours of training. New operatives are
contract customers generally give the factory 3 assigned to the easy positions. Later, they are
months to supply firm orders, and provide a rotated to build up their experience. This permits
requirements forecast for a further flexibility in planning. However, nobody works in
9 months. Occasionally, the business obtains an the sterile products area until they have been
unexpected, labour-intensive contract. with the firm for a considerable length of time.
Production capacity is spread over two The seasonal staff are picked from a pool of
separate sites that employ around 200 people. former employees who proved satisfactory. They
One of the plants is split between can be called in at a day’s notice, although not all
cre" mes/liquids and sterile products. A single at once. These people know that their tenure
shift operates for both labour- and capital- will last between 1 and 3 months. When it
intensive processes. There is a mixing room with expires the worker can be put off for a day and
various vats. The resultant cre" mes and then rehired. There is an 18 days shutdown over
liquids are pumped to 9 semi-automatic filling and the Xmas and January holiday period.
packing lines. Sterile products are made much
the same way, except that the 5 mixing/filling
lines are located in separate rooms to avoid 7. The demand management strategy
cross- contamination. Only the critical lines are
kept fully occupied. The situation could be On the surface it appears that demand
classed as medium-volume batch production. manage- ment is the ideal way to tackle the
Setups take from 2 hours to 2 days and the seasonal problem. This strategy affords the
longest production run lasts 4 days (the prospect of combining zero smoothing stocks
majority consumes 1 day). Consequently, the with the luxury of level or stable production.
jobs are sequenced in Despite the obvious attractions there is a
‘‘families’’, and all changeovers occur at the paucity of applications. Clearly, it is a hard
termination of a shift. strategy to implement. The apparel, footwear
Each year an MPS is constructed in monthly and greetings card cases contain an element of this
buckets for the 600 end products. It stretches approach. However, there is still enough residual
over a 2 years horizon, but gradually shrinks to a seasonal bias to call for a different strategy. On
1-year plan before the next complete regeneration. the other hand, the barbecues business
A very developed, from scratch, a complemen- tary
lengthy plan is necessary, despite substantial product range (heaters). In addition, two
forecast errors, because many of the 3000–4000 manufacturers simply purchased another company
ingredients are imported on very long lead times with a suitable line of products and consolidated
(manufacturing offsets range from 6 weeks to 9 operations at a joint plant. The lawn bowls
months). The upshot is that the initial 3 months producer supplies several big export markets
are updated in detail, whereas the next 6 located in different parts of the world. A fairly
buckets are just subject to general revisions. A even work schedule results.
modified chase strategy applies. The contract At the air conditioners/gas heaters factory
clients’ orders are shunted within their lead there is sufficient process and labour flexibility to
times until the planned output is leveled over adjust the product mix on a daily basis. This
each individual allows the
3 months block. MRP utilises monthly time
buckets and fortnightly renewals, but each plant
manually prepares its own detailed, short-term
monthly production schedule to chase two com- and modest levels of smoothing stocks suffice. The
pletely different sales profiles all year round. Such cricket ball manufacturer is an exception because it
a pure chase response is impossible to suffers a big seasonal sales swing. In response,
implement for the fans/heaters and management designed an attractive invoicing
barbecues/heaters plants. So, their strategies are policy, which actively encourages customers to
based on separate, semi- independent, mini place indent orders. Their objective was to put a
schedules for each product range. The limit on the financial effects of stockpiling.
production facilities are switched over twice per The radiators factory breaks the low variety
year to the other product type and the mould. A fairly steady rate of production for the
corresponding level resources plans cover 6
different parts of the year. There has to be a OE models is designed to fill the order schedules of
small amount of stockpiling to cover the sales two big motor vehicle builders. This side repre-
troughs. Little risk is attached. Fans and heaters’ sents only 30% of total production. The rest
sales contain a high proportion of indent orders. consists of radiator assemblies and cores for the
The other organisa- tion includes an extensive general aftermarket. Approximately 1200 catalo-
retail arm. If it wants to move dormant stocks, gue items are made for stock. Older cars require
the stores discount their barbecues or heaters another 300–400 cores. These are made to order
in order to stimulate the market. on a 24 hours service basis. Spares sales exhibit a
Unfortunately, the principal export summer peak (monthly variation is roughly 50%)
destinations (New Zealand, the Pacific Islands, and a demand forecasting exercise is carried out
and South Africa) reinforce Australia’s for all the stocked items. Predictions are quite
domestic seasonal bias. Foreign multinationals accurate for the top sellers. On the other hand, the
prevent their local divisions from targeting the low-volume sales are erratic, leading to
northern hemisphere. Anyway, the substantial sales forecast errors.
counterbalancing effects of overseas sales are There are dedicated production lines for the
somewhat distorted by long shipping times. As high-volume OE models. The rest of the items
a consequence, exports are no panacea, although are made in a general workshop area. Spares
6 out of the 26 exporters do dampen their are normally produced in monthly lot sizes.
seasonal sales variations to a certain extent. A However, to economise on set ups, some low-
manufacturer made several failed attempts to demand items are made less frequently, with
export electric wall heaters. Eventually, the firm minimum batch quantities of 4 cores and 20–25
turned to subcontract work for clients in the radiators. Each day the plant produces 50–60
automotive industry. This keeps the skilled work- different cores and 12 different radiators. A
ers in key sections of the plant busy throughout significant part of the labour force is engaged in
summer. hand soldering. These workers take 3 months to
reach regulation speed. A full day shift runs, plus
a small afternoon shift. The latter group is
8. The level strategy assigned to the current bottleneck equipment.
Along with the modular nature of radiator
A mere four manufacturers adhere to the design, this short-term action provides sufficient
level strategy. The main driver is the long time it model mix flexibility.
takes new operators to become proficient at A 6 months rolling production plan is generated
critical production tasks. This means that there for the aftermarket in the middle of each month.
is little room to manoeuvre as far as short-term The first month is ‘‘frozen’’ and the next 5
capacity is concerned. Besides, 24 hours per day periods incorporate purchases of raw materials
operations is the norm for the capital-intensive from overseas. Fixed manufacturing lead time
battery plant. The ‘‘level’’ group generally makes a offsets equal 6 weeks. The resultant workloads
low variety of stable products, while the seasonal are fitted around the weekly or daily shipment
factor is fairly benign. Thus, sales forecasts are requirements of the repetitive low-volume OE
quite reliable, there is little chance of creating orders. Pareto’s law applies, and 9 popular
obsolete products, models account for
60–70% of the aftermarket sales. The policy is that piling is irrelevant. Trainee machinists take from
the smoothing inventories should comprise just 3 to 6 months to become fully productive and
these few best selling catalogue items. This allows weavers spend 2 years learning their jobs.
the shop to level production, without taking any Not surprisingly, the workforce is permanent.
undue risks concerning unsatisfactory stock turn- A modified level strategy is pursued. The school
over. Overtime is considered to be expensive. uniforms side of the business is highly seasonal
Therefore, it always appears as an unplanned and represents 60% of the sales volume. It has a
reaction to rush orders or machine breakdowns. fixed customer base, which always places indent
orders. These jobs are spread out to try and
level the workload as far as possible. In the overall
9. The other strategies peak season the daily output rate is lifted
slightly. A minor alteration of the duty roster
The three remaining cases are dominated by allows the weavers to run bottleneck equipment
labour considerations. Pyrotechnics workers need for longer hours each day. Certain operatives
6 months to master certain tasks and 2 years to undertake ancillary duties during the low
achieve complete job rotation. This firm’s former season, plus extra training to improve the
chase policy depended on ‘‘hiring and firing’’, and scheduling flexibility. Besides this some
it was very wasteful. There are legal limits on voluntary labour turnover permits a modest
storage capacity and the magazine holds just 4– downward adjustment to the workforce.
6 weeks’ production. Since this makes the
corre-
sponding level output infeasible, management 10. Labour issues
switched to a stable (core) workforce strategy. In
spring and summer temporary recruits fill the First and foremost, ‘‘hiring and firing’’ is a
unskilled positions. A pure chase policy can also management issue. Mathematical cost balancing
be discounted under this labour regime, because exercises never influence such crucial decisions.
hazardous operations are confined to daylight Although many Australian companies adjust their
hours. The smoothing stocks are all fireworks, due workforce to match seasonal sales they do not
to the fact that flares have a limited storage life incur a double cost penalty. Instead, various
(and the date of manufacture is printed on the standing arrangements are invoked. By and large
packaging). these depend on the individual circumstances.
Most work at the wood heater plant entails Factors like the magnitude of the required
semi-skilled welding. Training lasts 4 weeks. The volume flexibility and the length of the job
old chase system was based on ‘‘hiring and firing’’ training are especially germane. Short-term
and spawned unwarranted quality and productiv- contracts are rife, for both full and part time
ity losses. A switch to a stable workforce employees. A typical seasonal contract lasts for
strategy cut the average labour content per unit three months. Beyond this period trade union
from 8.5 to agreements dictate that casual employees must
3 hours. The warehouse is not big enough to be converted to permanent staff.
facilitate a pure level alternative for such bulky Seasonal employment appears to be a viable
products, and by the start of the high season it is proposition, provided a maximum of 3 weeks
already full. Extensive overtime is scheduled training applies. Many companies maintain a
throughout the rest of the peak period. During register, and ask experienced staff (housewives,
the off-season the employees make jigs and retirees, students, etc.) to return at the same time
fixtures for the new models, but some idle time each year. Some foods are perishable, like meat
is unavoidable. However, this is an expensive pies and crumpets. Their sales also exhibit
brand with high profit margins. significant daily fluctuations, which repeat over
The club/corporate wear and school uniforms a 1-week cycle. The corresponding production
are custom designed. Production consists
entirely of firm orders. So, any recourse to
regular stock-
 schedules are based on delivering customers’ peak overloads. At the same time, a high utilisa-
orders the very next day. To achieve such a high tion factor is guaranteed for the existing equip-
degree of flexibility, part of the casual workforce ment. A manufacturer of bulky pine garden
may be called in at extremely short notice for just products licenses several inter-state subcontrac-
a few hours of continuous work. The nail plate tors, including sheltered workshops. It even
firm circumvents its training barrier by using a supplies these organisations with cheaper, but
specia- list employment contractor. For a small less efficient, capital equipment. The
premium, it can obtain the required number of subcontractors are strategically located and
appropriately skilled workers on a shift-by-shift service the more distant customers. This saves a
basis. A single day of advance notice must be considerable sum on order delivery costs. If
given to the agency. Firms must attempt to they are a practical proposition, outworkers or
retain their skilled workers. Nevertheless, sheltered workshops may actually represent
during sales troughs they may not replace any the cheapest form of labour. A multinational
personnel that depart of their own volition. Slack toy corporation minimises full product costs by
periods can be utilised for further training, outsourcing the entire production function.
retooling for new models, main- tenance and Another justification is that each year this
other indirect operations. A popular ploy is to enterprise designs and markets a number of
selectively close the plant down for holidays vastly different product ranges. Such uninhibited
(Australia’s winter weather is quite clement). product variety calls for extreme process
Several operations share a site with a much flexibility.
bigger (non-seasonal) department. Recently hired Firms that are experts in casting, stamping,
workers, who would otherwise be surplus to enameling, and moulding seek to become subcon-
requirements, are simply transferred at the end of tractors themselves. The main objectives for
the peak seasonal period. Thus, due to combined soliciting such work are to maximise revenue and
labour turnover, there is no need to hire staff on keep skilled operatives fully occupied
short-term contracts. When all else fails, a throughout the year. On the other hand, lowering
moderate amount of idle time may be acceptable the marginal costs of seasonal production is not
for skilled workers. Especially when production is really a relevant issue.
capital-intensive.
Overtime is never deployed in a cost-
balancing role. Rather, it helps production to
chase predicted demand in a flexible manner. 11. Developing the master production schedule
Furthermore, sales forecasts are far from
perfect, and unplanned overtime facilitates a The companies in the sample tailor the MPS
quick response to unforeseen events. This means format to suit their own purposes. None
setting a reasonable limit to planned overtime. construct an ‘‘optimum’’ aggregate plan first.
There comes a point where it is better to Many of the cases must cope with extremely
augment capacity by hiring additional workers. high product variety and a fluctuating sales
The need to leave room for flexible late mix. Unit labour and materials costs can vary
adjustments to capacity is particularly apposite for significantly across the different items as well.
products with volatile, weather-dependent sales. Thus, it would be virtually impossible to
Prime examples are pharmaceuticals, garden convert an aggregate plan into a feasible model-
hoses, lawn mowers, fertilisers, wetsuits, ice cream, based schedule, while pre- serving a specific
meat pies and beer. balance of monthly labour and inventory costs.
Another legitimate tactic that has been mis- Further methodological discre- pancies are
construed is subcontracting. In a ‘‘make to order’’ apparent when discrete summer and winter
environment it provides enough flexibility to product ranges must be scheduled. The critical
absorb sudden demand spikes and rush jobs. decisions are not so much about the size of the
Some firms avoid purchasing expensive duplicate workforce as the exact timing of the product
machinery by allowing subcontractors to deal changeovers.
with
 When a full 12 months planning horizon is used line setups. Accordingly, the algorithm schedules
a major portion of the MPS is speculative. The ‘‘clusters’’ of compatible products. Often the
more distant months may be couched in terms of a production of some inexpensive items is pulled
few product groups. As each time bucket forward, in order to avoid a premature change of
advances towards the current date further containers.
decisions are taken. Ideally, specific product Quarterly workforce planning relies on a trial
identities do not evolve until firm orders or and error approach. Labour allocations are
accurate (model-based) sales forecasts become modified over the furthest 9 months of the
available. The MPS triggers the timely acquisition planning horizon until the MPS algorithm yields
of labour and raw materials. However, for certain a satisfactory result. The output is augmented in
derived activities on long lead times (e.g. overseas discreet weekly steps as sales start to rise, initially
purchase orders) the plan may be forced to rely via overtime and then by recruiting extra day and
on aggregate forecasts. Modular design helps evening shift crews. These casual workers are
here. Later on these pre-existing contracted for a stipulated minimum period, since
commitments may constrain the composition of most jobs merely entail surveillance, removing
the plan’s front end. A pure chase strategy, along misaligned containers, and packing. Usually,
with preferred ways to implement it, there is excess plant capacity, and the new crews
streamlines the entire planning process. In the can be assigned to any line that is still free
absence of stockpiling, the MPS needs to on their particular shift. The planners convert
stretch just far enough to incorporate the the MPS’s front 2 weeks into a daily production
longest cumulative (resources) lead time. The plan. This procedure entails manipulating batch
most detailed part of the plan is even shorter. sequences until all of the ‘‘upstairs’’ constraints
Relatively few calculations and decisions are are satisfied.
involved, although the schedule is frequently
regenerated.
The processed food factory produces 300 12. Conclusions
different items (Buxey, 1988). A complex, heuristic
computer algorithm compiles the MPS. Four An up to date article on aggregate planning
customers take 85% of sales and the planning (Piper and Vachon, 2001) acknowledges the fact
system is ‘‘make for stock’’. The ingredients are that the overwhelming trend throughout industry
prepared, blended and cooked in bulk on the is to follow the chase strategy. This phenomenon
first floor of the plant. After that the required has been fuelled by the spread of JIT
mixture is piped downstairs to 6 high-speed, production. The authors go on to warn that the
automated filling/packaging lines, each dedicated regulation aggregate planning model is
to cans, jars, or herb/spice containers. The inadequate, and its outputs seriously flawed.
designated batch quantities equate to a whole Any chase plan has unstable labour inputs,
number of shifts. So a change of food contents which they believe are associated with
does not halt production. Notwithstanding, it significant productivity and capa- city losses.
can take up to 2 shifts to adjust the equipment Moreover, these effects are overlooked in
to accommodate a different sized vessel. This regulation aggregate planning. A new LP model
setup process also causes considerable wastage rectifies the deficiency and promulgates a
of raw materials. better (mixed strategy) solution for a
Mostly, a chase (sales) philosophy prevails. This benchmark pro- blem. Naturally enough, it leans
outlook is modified in order to run certain more towards a pure level plan. This paper
products whenever seasonal crops are picked. serves a perfect illustration of the mindset that
The 52-weeks MPS is revised at 4-weekly intervals. continues to dog the aggregate planning
It obeys a set of constraints, which include literature.
maintaining designated safety stock levels and A chase plan is clearly the best policy. It impacts
not making products during their particular sales positively on a wide range of costs and
troughs. The objective is to minimise the improves a company’s overall cash flow
combined costs of finished goods inventories plus situation. Most importantly, it minimises the
packaging firm’s financial
exposure. Also, there is less reliance on distant constraints, penalties, etc. will not put matters to
sales forecasts, and management is forced to right, although some mathematical models are
explore flexible avenues to adjust factory capacity. germane for lower level batching problems in
In the field, a chase strategy is a realistic option, process type industries. For a manufacturer,
provided that the production tasks are fairly the correct response to a seasonal sales profile
easy to master. Alternatively, an enterprise may is to implement the most appropriate production
be able to attract, on a temporary basis, various strategy.
categories of suitably experienced people. Very few actual real life examples that are
Despite all this there are plenty of cases that relevant to seasonal production seem to appear in
call for extensive training of any new the literature, although a pair of ‘‘aggregate
employees. Then the real remedy is to planning’’ vignettes is featured in one textbook.
implement a different, yet predeter- mined, non- In addition, a journal article describes how an
chase strategy. American company uses a suite of sophisticated
No obvious guidelines exist for selecting another computer models to plan the production of
strategy. The choice depends on a business’s garments. Happily, those practical details that
unique amalgam of fixed costs, training times are given tend to coincide with many of the
and degree of seasonal bias. Management plumps findings of this case-based research, and therefore
for the alternative that it expects will be most lend their support to its radical proposition.
effective in their particular environment. The
‘‘modified chase’’ strategy is the most appealing
fall back option. This substantiates the view that a Appendix A
pure ‘‘chase’’ policy is the best. A modified
chase regime bows to certain labour or The case studies: Unless stated the plant runs on
equipment constraints. Thereafter, it attempts 5 days per week and each shift lasts around
to chase the demand profile to the fullest 8 hours. Summary information for each operation
extent that is still possible. All non-chase plans covers product portfolio; product variety;
necessitate a certain amount of stockpiling, and in produc- tion mode; minimum labour input; main
general the planning horizons are longer than means of achieving volume flexibility (where
they would be for chase. appropriate); and germane comments regarding
This does not mean that manufacturers simply the feasibility of the chase option.
tolerate the negative consequences of carrying
smoothing stocks. Rather, only specific types of A.1. Chase strategy
item are stored. The particular categories are
designed to eliminate or contain most of the Bulk/bagged cement; none; for stock, but big
attendant financial risks. contracts to order; continuous production; spare
In this kind of planning there are three prime capacity at final milling stage; level production of
considerations. Productivity and costs are taken clinker intermediate ensures maximum utilisation
care of by selecting the right overarching of the most capital intensive stage (kilns).
medium- Petrol and oils; 10 products from common
term strategy. At the same time, suitable tactics crude feedstock; for stock; continuous
(for adjusting labour inputs and accumulating production; none; export sales (15–20%) maintain
smoothing stocks) are evaluated and laid down in critical production units at full utilisation.
advance. The enterprise satisfies the demands of Product mix is adjusted to match individual
the customers via the composition of the MPS. seasonal demand patterns.
Future sales are uncertain, so schedules are Industrial paints and varnishes; approximately
designed to be reliable, robust and flexible. All of 400 types; to unique order; 2 shifts; overtime;
this leads to the stark realisation that the theory production of some orders pulled
of aggregate planning is fundamentally wrong. forwards.
It makes unwarranted assumptions about Domestic paints and finishes; 417 SKUs from
strategy and focuses on the tactics instead. 28 types, colour range, and 5 can sizes; for
Tinkering around the edges of this model stock;
with extra
1 shift; increased manning levels, overtime, then Nail plate products for the building industry;
second shift; JIT production and purchasing. 800 items in 60 family groups; for stock; specials
Beer; 2 types; for stock; 2 shifts; normal (and exports); to order; 1 shift; increased
capacity sufficient for predictable summer manning levels; core workforce supplemented
weather and by skilled packers from labour hire company.
overtime available during a heat wave; some idle Jewellery; 1200 standard items approx. plus
time accepted as labour costs are insignificant. occasional custom orders; majority of items to
Ice cream; 102 SKUs including flavours and order, higher-volume, cheaper lines for stock; 1
packaging; for stock; 1 shift but not all lines shift; increased manning levels and (home based)
running; increased manning levels, overtime, subcontractors.
second shift, then weekend overtime; some idle Greetings cards and related products: everyday
time accepted for small core workforce. cards; 2500 designs; for stock; seasonal cards; 1000
35 mm photographic films; 21 families and 150 designs for Xmas alone; for stock, but 95% pre-
packages; to intra-company order (unique packa- sold; specials; to order; 2 shifts; overtime;
ging); 3 shifts; weekend overtime; customers will everyday card sales 2–2.5 times greater than
accept orders a month early-domestic and various seasonal cards. Specials are 30% of total
export sales patterns tend to cancel out; JIT workload.
production. Crumpets and scones; 1 type each, some custo-
Ladies’ wear; 500 styles compounded by mised wrappers; to order; 1 shift 2 days, not all
fabrics and sizes; to unique order; 1 shift but not lines running; increased manning levels, overtime,
all lines manned; parallel 6 hours shift with up to 3 shifts 6 days; shelf life 5 days, sales
workforce increased in steps; counter seasonal depend on day of the
styles and modifications—order book priority week.
juggling—some idle time accepted for regular Meat pies, pasties and sausage rolls; 30 items; to
workforce. order; 2 shifts 6.5 hours 6 days, staggered for
Ladies’ footwear; winter range 287 styles, different processes; increased manning levels and
summer range 100 styles, in 60 colours, 8 flexible shift lengths; short shelf life, sales
materials and different sizes; to order; 1 shift; depend on day of the week.
orders pulled Toys; typically, 25–30 items made locally but
forward, overtime, up to 4 subcontractors for most are imported; for stock; local production
stitching; no room to expand capacity—natural subcontracted out; contract arrangements provide
attrition reduces workforce slightly in troughs and process and volume flexibility; main problem is to
some idle time accepted. get sales forecasts right.
Tumble dryers; 10 domestic models; for stock; 10
(low volume) export ‘specials’; to order; 1 shift; A.2. Modified chase strategy
increased manning levels; excess workers trans-
ferred to larger, (non-seasonal) washing machine Reinforced garden and industrial hoses; 50 types;
facility—JIT purchasing and production. small orders supplied ex stock and major custo-
Lawnmowers, edgers, outdoor vacuum cleaners; 8 mers supplied to order; 1 shift, not all lines
models; for stock; 1 shift plus a small, selective running; increased manning levels, overtime, up
second shift; increased manning levels, overtime, to 3 shifts 7 days; limited stockpiling in own
second shift; JIT purchasing and production. warehouse preferred to Sunday working, renting
Refrigerators; many models, based on 15 outside storage space is a last resort.
cabinet sizes, 3 brands, and 3 colours; for stock; Ice cream; 240 SKUs including flavours and
1 shift; increased manning levels; JIT packs, 30–50 families; for stock; 2 shifts but not
purchasing and all lines running; increased manning levels,
production. overtime; limited stockpiling necessary due
Electric wall heaters; 30 models; for stock; 1 to training capacity restrictions.
shift; increased manning levels, overtime, Processed foods; 300 items, including sizes; for
second shift in assembly; subcontracting stock; 1 shift, not all lines running; increased
maintains level workload in engineering
workshops.
manning levels, overtime, second shift; stockpiling Bicycle safety helmets; 10 models, 5 colours, 3
necessary for products derived from seasonal sizes; for stock; 1 shift; increased manning levels;
crops. stockpiling avoids second shift—2 models depend
Chocolate Easter eggs, Xmas novelties, etc.; 300 on mould shop for a component but jobs easily
items, 75% are customer specific (usually packa- accommodated (this facility runs continuously,
ging); mix of make for stock (70–80% of producing non-seasonal, contract orders for
forecast annual sales) in an initial run and make parts).
to order (the balance) in a second batch; 2 Garden tools, axes and block splitters, and
shifts, not all lines running; increased manning non- seasonal hardware; 2500 items; for stock; 2
levels, 3 shifts in critical areas; stockpiling shifts; overtime, increased manning levels in
necessary due to equip- ment constraints, but all assembly; chase MPS but stockpiling of forged
production scheduled as late as possible. components due to skilled labour constraint.
Pine landscaping products (borders, pavers); Ammunition; 260 items, belonging to a smaller
4 types, 9 products; for stock; 1 shift; overtime, number of families; 2 or 3 shifts depending on
second shift 6 hours, more overtime, process; mainly for stock, but approximately 40%
subcontrac- tors top up supplies to interstate to order (exports and low volume industrial
distribution centres; subcontracting saves on range); overtime; capital-intensive facilities kept
capital equipment and transport costs; fully loaded by pulling long lead time orders
stockpiling required to cope with sales spike at forward.
beginning of peak season.
Surfer’s wetsuits; 50 models in 12 sizes; for A.3. Level production strategy
stock, but 45–70% pre-sold (indents); 1 shift;
labour turnover trims workforce in troughs, Motor vehicle batteries; 4 sizes; for stock;
(a little) overtime; counter-seasonal product mod- 2 12 hours shifts on 6 days; marketing
ifications—indent orders pulled forward to main- discounts prices to achieve sales targets; lengthy
tain labour productivity. training and high plant utilisation.
Boardshorts; 5 types, 40 fabrics and 20 sizes; Colour television sets; 12 models; for stock; 1
to order; 1 shift; labour turnover trims workforce shift; overtime during pre-Xmas peak to
in counter
troughs, overtime; JIT production—indent orders forecasting errors; lengthy training.
pulled forward to maintain labour productivity— Cricket balls; 4 models; for stock; 1 shift;
counter-seasonal styles and products, exports to customer credit extended for early orders; lengthy
northern hemisphere. training for stitching operation.
Pharmaceuticals (cremes, liquids and sterile Motor car radiators; 6 OE models; to order;
products); 600 items including packaging, 400 1200 catalogue items for general aftermarket
product types; own products for stock, contract (mainly cores); for stock; 300–400 older model
manufacture (35–40%) to order; 1 shift, not all cores; to order; 1 full shift plus 1 very small shift;
lines running; increased manning levels, flexible second shift relieves bottlenecks; signifi-
overtime; cant training for welding operations.
60 seasonal items do not peak in same months—
level production over initial 3 months after A.4. Demand management strategy
juggling contract order priorities; forecast errors
a bigger problem than seasonal bias. Cooling fans and electric strip heaters; 6
Bulk/bagged fertilisers; blends 3000 products models each; for stock but >60% pre-sold to
from 1 manufactured and 1 imported type; to major customers; 1 shift; whenever regular
order; 2 shifts 12 hours 3 days in production employees switch to heaters line a temporary shift
and 1 8 hours shift in mix/dispatch; up to 6 mans fans line; JIT purchasing and production.
days Gas barbecues; 70 types expands to 200 with
in production via overtime, and up to 3 shifts in finishes and options; solid fuel heaters; 10 models;
mix/dispatch by increasing workforce; stockpiling
of manufactured fertilizer type due to capacity
constraints.
for stock; 1 shift plus constant overtime; more References
overtime, tooling gradually switched to other
product range; retail chain stimulates sales via Akinc, U., Roodman, G.M., 1986. A new approach to
discounts and promotions. aggregate production planning. IIE Transactions 18 (1),
84–94.
Evaporative air conditioners and radiant gas Bitran, G.R., Haas, E.A., Hax, A.C., 1982. Hierarchical
heaters; 5 ranges each, 165 model variations; for production planning; a two-stage system. Operations
stock; 2/3 shifts in components production and Research 30 (2), 232–251.
1 full plus 1 part time assembly shift; increased Bowman, E.H., 1956. Production scheduling by the trans-
portation method of LP. Operations Research 4 (1),
manning levels occupy up to 6 assembly lines, 100–103.
2 lines switch between air conditioners/heaters to Burch, E.E., Oliff, M.D., Sumichrast, R.T., 1987. Linking level
chase demand; JIT purchasing and production. requirements in production planning and scheduling.
Lawn bowls; 650 models with 2000 decorative Production and Inventory Management 28 (2), 123–131.
patterns; domestic models for stock, exports and Buxey, G.M., 1988. Production planning under seasonal
demand: A case study perspective. OMEGA International
Australian ‘specials’ to order; 1 shift; level
Journal of Management Science 16 (5), 447–455.
production; 85% share of world market and Chung, C.H., Krajewski, L.J., 1987. Interfacing aggregate plans
exports smooth demand (plus very modest and master production schedules via a rolling horizon
antici- pation stocks for Australia and UK); feedback procedure. OMEGA International Journal of
significant training and difficulties recruiting Management Science 15 (5), 401–409.
Deckro, R.F., Hebert, J.E., 1984. Goal programming ap-
staff. proaches to solving linear decision rule based aggregate
production planning models. IIE Transactions 16 (4),
A.5. Other strategies 308–315.
Dilworth, J.B., 2000. Operations Management, 3rd Edition.
The Dryden Press, Orlando, FL.
Marine distress flares and fireworks; 2 flares
Dubois, F.L., Oliff, M.D., 1991. Aggregate production plan-
types; for stock; 1 shift; increased manning levels; ning in practice. Production and Inventory Management
stable (core) workforce policy implemented by 32 (3), 26–30.
scheduling fireworks (20% of total production) in Edwards, J.R., Wagner, H.M., Wood, W.P., 1985. Blue bell
the slack period and making maximum use of trims its inventory. Interfaces 15 (1), 34–52.
Elmaleh, J., Eilon, S., 1974. A new approach to production
limited storage capacity; lengthy training for smoothing. International Journal of Production Research
certain tasks. 12 (6), 673–681.
Wood heaters; 12 models in 2 colours; for stock; Enns, S.T., 2001. MRP performance effects due to lot size and
1 shift; overtime and enhanced performance rate, planned lead time settings. International Journal of
Production Research 39 (3), 461–480.
some press work subcontracted out; stable
Everdell, R., 1984. Master Production Scheduling: APICS
work- force policy with insufficient warehouse training aid. American Production and Inventory Control
capacity to level production; some idle time Society, Washington, DC.
accepted- welded cabinets stockpiled with final Gaither, N., Frazier, G., 1999. Production and Operations
assembly scheduled during peak season Management, 8th Edition. South-Western College Publish-
ing, Cincinnati, OH.
(components like the electric fan are expensive); Gilgeous, V., 1987. Aggregate planning in UK manufacturing
significant training for welding. companies. International Journal of Operations and Pro-
School uniforms and club/corporate wear (ties duction Management 7 (1), 50–61.
and tie fabrics, scarves, pullovers, tunics, blazers); Gilgeous, V., 1989. Modelling realism in aggregate planning: A
goal-search approach. International Journal of Production
custom designed; to order; 1 shift in labour-
Research 27 (7), 1179–1194.
intensive areas (cutting, making, finishing) and 2 Heizer, J., Render, B., 1999. Principles of Operations Manage-
overlapping shifts in capital-intensive areas ment, 3rd Edition. Prentice-Hall, Upper Saddle River, NJ.
(weav- ing, knitting); varying the degree of shift Hennet, J.-C., 1999. From the aggregate plan to lot-sizing in
overlap so equipment runs longer; encourage multi-level production planning. In: Brandimarte, P., Villa,
A. (Eds.), Modeling Manufacturing Systems: From Aggre-
(school) indent orders and level production as gate Planning to Real-time Control. Springer, Berlin,
far as possible—reduce workforce via labour pp. 5–23.
turnover, operatives assigned to training and
non-produc- tion tasks; very long training.
Hill, J.A., Berry, W.L., Leong, G.K., Schilling, D.A., 2000. ning systems. International Journal of Production Research
Master production scheduling in capacitated sequence- 34 (9), 2613–2628.
dependent process industries. International Journal of Piper, C.J., Vachon, S., 2001. Accounting for productivity
Production Research 38 (18), 4743–4761. losses in aggregate planning. International Journal of
Holt, C.C., Modigliani, F., Simon, H.A., 1955. A linear Production Research 39 (17), 4001–4012.
decision rule for production and employment scheduling. Schmenner, R.W., 1993. Production/Operations Management,
Management Science 2 (1), 1–30. 5th Edition. Macmillan, New York.
Hung, R., 1999. Scheduling a workforce under annualized Schroeder, R.G., 2000. Operations Management. McGraw-Hill
hours. International Journal of Production Research International Edition, New York.
37 (11), 2419–2427. Shtub, A., 2001. A framework for teaching and training in the
Hwang, H., Cha, C.N., 1995. An improved version of the enterprise resource planning (ERP) era. International
production switching heuristic for the aggregate Journal of Production Research 39 (3), 567–576.
production planning problem. International Journal of Silva, J.P., Lisboa, J., Huang, P., 2000. A labour-constrained
Production Research 33 (9), 2567–2577. model for aggregate production planning. International
Jones, C.H., 1967. Parametric production planning. Manage- Journal of Production Research 38 (9), 2143–2152.
ment Science 13 (11), 843–866. Slack, N., Chambers, S., Harland, C., Harrison, A., Johnston,
Lee, W.B., Khumawala, B.M., 1974. Simulation testing R., 1998. Operations Management, 2nd Edition. Pitman,
of aggregate production planning models in an imple- London.
mentation methodology. Management Science 20 (6), Slack, N., Chambers, S., Johnston, R., 2001. Operations
903–911. Management, 3rd Edition. Prentice-Hall, Harlow.
Mellichamp, J.M., Love, R.M., 1978. Production switching Taubert, W.H., 1968. A search decision rule for the aggregate
heuristics for the aggregate planning problem. scheduling problem. Management Science 14 (6), 343–359.
Management Science 24 (12), 1242–1251. Venkataraman, R., Smith, S.B., 1996. Disaggregation to a
Nam, S., Logendran, R., 1992. Aggregate production plan- rolling horizon master production schedule with minimum
ning—a survey of models and methodologies. European batch-size production restrictions. International Journal of
Journal of Operational Research 61, 255–272. Production Research 34 (6), 1517–1537.
Ozdamar, L., Atli, A.O., Bozyel, M.A., 1996. Heuristic family Vergin, R.C., 1966. Production scheduling under seasonal
disaggregation techniques for hierarchical production plan- demand. Industrial Engineering 17 (5), 260–266.