Sales and operations planning optimisation.

Contribution
and limits of linear programming
André Thomas, Patrick Genin, Samir Lamouri

To cite this version:

André Thomas, Patrick Genin, Samir Lamouri. Sales and operations planning optimisation. Contribution and
limits of linear programming. Alexandre Dolgui, Jerzy Soldek and Oleg Zaikin. Supply Chain Optimisation,
Springer, pp.191-204, 2005, ⟨10.1007/0-387-23581-7_14⟩. ⟨hal-00121159⟩

HAL Id: hal-00121159

[Link]
Submitted on 23 Dec 2006

HAL is a multi-disciplinary open access archive L’archive ouverte pluridisciplinaire HAL, est des-
for the deposit and dissemination of scientific re- tinée au dépôt et à la diffusion de documents scien-
search documents, whether they are published or not. tifiques de niveau recherche, publiés ou non, émanant
The documents may come from teaching and research des établissements d’enseignement et de recherche
institutions in France or abroad, or from public or pri- français ou étrangers, des laboratoires publics ou
vate research centers. privés.

HAL Authorization
Chapter 14
SALES AND OPERATIONS PLANNING
OPTIMISATION
Contribution and limits of linear programming

Patrick Genin, Samir Lamouri and André Thomas

Ecole des Mines de Paris, Ecole des Mines de Paris, CRAN Faculté des Sciences Vandoeuvre-
les-Nancy

Abstract: Operations’ planning requires making strategic decisions on inventories levels,

on demands and operations constraints. The importance of these decisions
leads to elaborate and optimise Sales and Operations Plans on a planning time
fence at least as long as the budget. Models using linear programming give the
“optimal” strategy but it does not resist frequent changes in parameters. Other
mathematical tools as well as Taguchi methods are interesting in realising a
simple but robust compromise.

Key words: Sales and Operations Planning, Optimisation, Linear Programming,

Robustness, Planning

1. INTRODUCTION

Today, Supply chain Management becomes the function that chooses the
global level of production and the performance of the other activities in
order to satisfy the actual sales forecasts. Planning production allows to
make arrangements on time to satisfy sales with needed quantities and
promised delays at the smallest cost. These three objectives cannot be
simultaneously achieved. The planning decisions are always the result of a
balance between on-time deliveries, risks on inventories and operations
costs.
The Sales and Operations Plan process (S&OP) builds the sales and
operations strategy that realises the best balance, on a time fence at least as
long as the budget, for product groups [13]. The expected performance for
other activities is deduced on a mid/long term.
2 Patrick Genin, Samir Lamouri and André Thomas

The traditional S&OP calculation is based on graphical techniques or on

Linear Programming models [3]. This chapter sets out how an approach of
S&OP by linear programming, can balance inventories, on-time deliveries
and operations costs but also points out the limits in robustness of the
strategy.

2. THE S&OP FUNCTION

S&OP puts into practice strategic objectives established by management

when dealing with the strategic plan. It is the link between the sales planning
and operations. S&OP is entirely integrated in information and demand
management systems. It drives the execution of the different Master
Planning Schedules (MPS). S&OP is a useful tool for prospective analysis
over the medium to long term.
As the operations system is not flexible enough to follow sales changes
day by day, adjustments are needed at that planning level. Sales are
uncertain data with quick and unpredictable variations. If the demand could
be exactly forecasted, the workload on resources should react the same way.
However this is not always possible. The number of machines is fixed,
training new staff takes time and the negotiations with suppliers have an
impact on lead-time and quantities produced. The firm has to answer the
following question: How can the production system capacity keep up with
fluctuations in the sales volumes? It is the key role of S&OP to answer that
question [8].
The S&OP anticipates the evolution in products families’ sales in order
to adapt the operations and supply chain system to its market. At that level,
budgetary capacities are going to be taken into account. The S&OP will
check cash, inventories, workforces, rough-cut capacities availability to turn
the sales and strategic objectives (market shares…) into activities to
complete on the mid term. The different sub-system are linked together
(Figure 14-1).
SALES AND OPERATIONS PLANNING OPTIMISATION 3

OBJECTIVES

Marketing Business Financial

Objectives Objectives Objectives

Sales &
Resources Demand
Requirements Operations
Planning Management
Planning

Master
Production MANAGEMENT
Schedule SYSTEM

Figure 14-1. S&OP and industrial management functions

The planning horizon will often be 18 months long with a monthly

planning frequency. These parameters are set to take account special events
such as promotions, special agreements, …
Several simulations could be performed in order to determine the optimal
strategy that will minimise the total cost while maximising the sales.

3. S&OP OPTIMISATION TECHNIQUES

Two approaches are frequently used: graphical methods and optimisation

by Linear Programming [4, 5, 6].
The spreadsheets and graphical methods are widespread because of the
easiness of use and understanding. The plans are obtained with few variables
settled at a time to let the manager compare the forecasted demand to the
existing capacity. These graphical methods work by iterations; they identify
different integrated and realisable plans but costs are not necessarily the
lowest. The manager must consequently use his feelings to determine the
appropriate plan.
Graphical methods generally proceed in 5 steps as follow:
1. Determination of the demand per month;
2. Capacity determination in normal work hours, in overtime and in
subtracting per month;
3. Identification of manpower costs, carrying cost, etc.;
4. Strategy evaluation changing workforce or inventory level;
4 Patrick Genin, Samir Lamouri and André Thomas

5. Setting of alternatives and balancing of total costs.

These management tools help to evaluate different strategies but do not
generate them. Whereas decision makers expect a systematic approach that
considers the whole costs and gives an efficient answer to that problem.
Mathematical models, using linear programming propose such an approach
[1]. In the following, an application in an industrial context is described.

4. VALLOUREC PRECISION ETIRAGE (VPE)

PROBLEMATIC

VPE produces steal tubes in parts or full length for automotive markets
(layer 1, 2 and 3 supplier) and for mechanical markets (heaters, boilers,
circuits). The Supply Chain initiative reengineers business process
(industrial and administrative) in order to reach 98 % of on-time delivery.
Settled in 4 production entities, The 10 flow-shops ensure the production
of the 70 commercial families by working 5 days out of 7 in 3x8. The
demand by products family "the load” is different each month. On the
contrary the capacity is relatively stable.
The S&OP is the monthly process for updating the tactical planning by
consolidating production and demand on a 12 months time-fence. The steps
are (Figure 14-2) [2]:
1. Demand Forecast calculation in the sales department in families and
production lines;
2. Load calculation and load and capacity balancing on each line by its
manager and parallel calculation by the supply chain manager at the firm
level;
3. Scenario construction and actions plan by line and for the whole
company;
4. Consolidation of resources requirements and availabilities and action
plans validation by the supply chain manager;
5. Monthly meeting to present the scenarios and choice of the strategy by the
steering comity.
SALES AND OPERATIONS PLANNING OPTIMISATION 5

Step°1 Step°2 Step°3&4 Step°5

Sales SC Load and capacity

department Department from other
production lines

Sales S&OP Tool Load and Capacity S&OP Meeting

Forecasts Balancing

Lines
Mangers Scenarios

Chosen S&OP
scenario
Production
parameters
evolution

Figure 14-2. S&OP Process of VPE

The firm adjusts a set of logistic variables to spread the workload during
its S&OP process:
– Seasonal inventories;
– Capacity adjustments (working during weekends and public holidays);
– Subcontracting (limited for strategic reasons);
– Backorders or inventories;
– Priorities by products family on production resources.
The determination of scenarios and associated total costs is difficult to
make by hands. Linear Programming makes it possible to find the optimum
for a whole set of given conditions.

5. THE LINEAR PROGRAMMING MODEL

PROPOSED FOR VPE

The model presented is simplified: we do only consider one production

line and only one products family. Moreover, data are truncated for
confidentiality reasons.
The S&OP determines for the considered products family:
– production level;
– inventory level;
– subcontracting level;
– number of additional working days or non-working days;
– S&OP over-cost engaged by the scenario.
The main over-cost is due to production. It is composed of the over-costs
caused by additional or non-working days and of subcontracting. In addition
to this production over-cost, VPE considers the carrying costs. The Business
Plan aims at a service rate of 100 %, that is to say any delay, and a level of
stock lower than 3 days, 195 T Consequently, algebraic stock cannot be
6 Patrick Genin, Samir Lamouri and André Thomas

negative. It thus lies between 0 and 195 T. That last constraint does not
consider the seasonal inventory.

Table 14-1. S&OP data

Item Value
Unit inventory cost by period, cI 190 €
Unit backlog Cost by period, cB 2 300 €
Unit cost incurred per additional day, cOV 800 €
Unit cost incurred by non-production day, cNP 1 300 €
Unit cost for subcontracting, cSC 600 €
Beginning Inventory, I(0) 100 T
Available capacity expressed in unit per day, e 65 T/d

5.1 The variables definition

t is the period index. T is the number of periods (it is the horizon lenght :
12 in our case). D(t) represents the forecasted demand for period t. N(t)
corresponds to the standard working days in period t, N*(t) is the maximum
of working days per period t. u(t) is the standard capacity in period t. It is
determined by the formula (1). O*(t) is the maximum overtime capacity in
period t. Relation (2) gives it. The values used are presented in Table 14-1.

∀ t , u (t ) = e × N (t ) (1)
∀ t , O (t ) = e × [ N (t ) − N (t )]
x x
(2)

5.1.1 The decision variables

O(t) is the number of tons manufactured in additional days within period

t. S(t) is the number of tons manufactured in subcontracting during period t.
S*(t) is the upper limit to subcontracting. N(t) is the number of tons which
have not been produced during the non-working days in period t. I(t) is the
inventory level at the end of period t. B(t) is the backlog level at the end of
period t. P(t) is the total production carried out.

5.1.2 The objective function

The objective of the problem is represented by the minimization of the

sum of the different cost factors, i.e. the costs for production in overtime and
for non-production, subcontracting, inventory, backlogs. It determines the
over-cost of the determined scenario (3).
SALES AND OPERATIONS PLANNING OPTIMISATION 7

∑c
t =1
OV
×O(t) + cNP × N(t) + cSC × S(t) + cI × I (t) + cB × B(t) (3)

5.1.3 The constraints

The following describe the constraints in the model.

∀t, P(t) = u(t) +O(t) − N(t) + S(t) (4)

∀t, P(t) = D(t) + I (t) − I (t − 1) + B(t) − B(t − 1) (5)
∀t , I (t ) ≤ I * (t ) (6)
∀t, S (t ) ≤ S * (t ) (7)
∀ t , O (t ) ≤ O (t ) *
(8)
∀ t , N (t ) ≤ u (t ) (9)
∀t , 0 ≤ O (t ), N (t ), S (t ), I (t ), B (t ) (10)

Constraints (4) state that in each period the whole production is obtained
with the standard capacity plus or minus what is produced in overtime or not
produced, and subcontracting. The balance equations among the whole
production and inventories, total demand and backlogs are established
through constraints (5). Clearly, constraints (5) may easily be modified to
accommodate alternative assumptions concerning lost demand. Constraints
(7) are constraint conditions, stating that the amount stored must be less than
the storage capacity, for each time period. Similarly, constraints (8) stipulate
an analogous condition for subcontracted units limited by management.
Upper limits on overtime workforce capacity are given by inequalities (9).

Table 14-2. S&OP parameters

Period t 1 2 3 4 5 6 7 8 9 10 11 12
N(t) 19 9 21 22 21 16 22 20 22 20 20 20
N*(t) 24 17 30 31 30 24 30 28 31 30 20 30
D(t) 1475 510 1655 1320 1757 1210 1603 1475 1320 1685 1199 1782
S*(t) 300 300 300 300 300 300 300 300 300 300 300 300
I*(t) 195 195 195 195 195 195 195 195 195 195 195 195

5.2 S&OP optimisation

Working out the problem and using of EXCEL solver determine the
optimal solution according to the given conditions. The results are shown
below (Figure 14-3 and Figure 14-4).
8 Patrick Genin, Samir Lamouri and André Thomas
Sales & Operations Planning

Parameters
Beginning inv entory 100 T Carrying costs 190 €/T
Capacity 65 T/d Production over-costs in ov ertime 700 €/T
Production over-costs in non-working hours 1300 €/T
Backlog over-costs 2300 €/T
Subcontracting over-costs 600 €/T

Results
Period t 1 2 3 4 5 6 7 8 9 10 11 12
Demand D(t) 1 475 510 1 655 1 320 1 757 1 210 1 603 1 475 1 320 1 685 1 199 1 782 16 991 T
Production 1 235 585 1 365 1 430 1 365 1 040 1 430 1 300 1 430 1 300 1 300 1 381 15 161 T
Subcontracting ST(t) 140 0 215 0 282 170 173 175 0 275 0 300 1 730 T
Total prod 1 375 585 1 580 1 430 1 647 1 210 1 603 1 475 1 430 1 575 1 300 1 681 16 891 T
Inventory S(t) 100 0 75 0 110 0 0 0 0 110 0 101 0 T
Production in overtime HS(t) 0 0 0 0 0 0 0 0 0 0 0 81 81 T
Over-costs 84 14 129 21 169 102 104 105 21 165 19 237 1 170 K€

Figure 14-3. Optimised S&OP results

2 000

1 800

1 600

1 400

1 200
Tons

1 000 Demand
Inventory
800
Subcontracting
600 Production + Subcontracting)
400

200

0
1 2 3 4 5 6 7 8 9 10 11 12
Months

Figure 14-4. Optimised S&OP graph

The recommended strategy is inventory building during the under-load

periods and the use of subcontracting in overloaded periods. The demand
being strong during period 12, 81 tons are produced in overtime, less
expensive than subcontracting and storage during period 11. The over-cost
of this scenario is 1 170 K€. If the global approach is logical, the specific
choices for each month cannot be obvious [1].
Let us suppose now that various events occur in production few hours,
few days after the implementation of this S&OP scenario:

5.2.1 An exceptional order

It consumes 50% of the beginning inventory! How does the optimum

evolve?
SALES AND OPERATIONS PLANNING OPTIMISATION 9
Sales & Operations Planning

Parameters
Beginning inv entory 50 T Carrying costs 190 €/T
Capacity 65 T/d Production over-costs in ov ertime 700 €/T
Production over-costs in non-working hours 1300 €/T
Backlog over-costs 2300 €/T
Change Subcontracting over-costs 600 €/T

Results
Period t 1 2 3 4 5 6 7 8 9 10 11 12
Demand D(t) 1 475 510 1 655 1 320 1 757 1 210 1 603 1 475 1 320 1 685 1 199 1 782 16 991 T
Production 1 235 585 1 365 1 430 1 365 1 040 1 430 1 300 1 430 1 300 1 300 1 381 15 161 T
Subcontracting ST(t) 190 0 215 0 282 170 173 175 0 275 0 300 1 780 T
Total prod 1 425 585 1 580 1 430 1 647 1 210 1 603 1 475 1 430 1 575 1 300 1 681 16 941 T
Inventory S(t) 50 0 75 0 110 0 0 0 0 110 0 101 0 T
Production in overtime HS(t) 0 0 0 0 0 0 0 0 0 0 0 81 81 T
Over-costs 114 14 129 21 169 102 104 105 21 165 19 237 1 200 K€

Figure 14-5. Optimised S&OP results after the order integration

2 000

1 800

1 600

1 400

1 200
Tons

1 000 Demand
Inventory
800
Subcontracting
600 Production + Subcontracting)
400

200

0
1 2 3 4 5 6 7 8 9 10 11 12
Months

Figure 14-6. Optimised S&OP graph after order integration

Volume in subcontracting will be more significant involving an

additional over-cost of 30 K€. The other part of the scenario remains
identical (Figure 14-5 and Figure 14-6). This exceptional order is profitable
only if its margin is higher than these 30 K€. In addition, if this event is
anticipated, that makes it possible to warn the subcontractor to take the 50
additional tons. If this order is not anticipated, the 50 tons are made in
overtime instead of sub-contracted. 5 additional K€ (50 X (700-600)) have to
be added.

5.2.2 Capacity restriction of our supplier

Because of a contract for a new strategic market, our subcontractor can

treat only 80 % of our needs! What becomes of the optimum?
10 Patrick Genin, Samir Lamouri and André Thomas
Sales & Operations Planning

Parameters
Beginning inv entory 100 T Carrying costs 190 €/T
Capacity 65 T/d Production over-costs in ov ertime 700 €/T
Production over-costs in non-working hours 1300 €/T
Backlog over-costs 2300 €/T
Subcontracting over-costs 600 €/T

Results
Period t 1 2 3 4 5 6 7 8 9 10 11 12
Demand D(t) 1 475 510 1 655 1 320 1 757 1 210 1 603 1 475 1 320 1 685 1 199 1 782 16 991 T
Production 1 235 585 1 365 1 430 1 407 1 040 1 430 1 300 1 430 1 335 1 300 1 441 15 298 T
Subcontracting ST(t) 140 0 215 0 240 170 173 175 0 240 0 240 1 593 T
Total prod 1 375 585 1 580 1 430 1 647 1 210 1 603 1 475 1 430 1 575 1 300 1 681 16 891 T
Inventory S(t) 100 0 75 0 110 0 0 0 0 110 0 101 0 T
Production in overtime HS(t) 0 0 0 0 42 0 0 0 0 35 0 141 218 T
Over-costs 84 14 129 21 173 102 104 105 21 169 19 243 1 184 K€

Figure 14-7. Optimised S&OP results after restriction

300
Inventory Subcontracting
250

200
Tons

150

100

50

0
1 2 3 4 5 6 7 8 9 10 11 12
Months

Figure 14-8. Optimised S&OP graph after restriction

Subcontracting is now limited to 240 tons per month (Figure 14-7 and
Figure 14-8). Additional days are to be envisaged for the overload periods.
The resulting over-cost is of 14 K€.

5.2.3 Reduced capacity

An event in production constraints VPE to produce during the first two

months with a reduced capacity of 20 %! Where is the new optimum?
SALES AND OPERATIONS PLANNING OPTIMISATION 11
Sales & Operations Planning

Parameters
Beginning inv entory 100 T Carrying costs 190 €/T
Capacity 65 T/d Production over-costs in ov ertime 700 €/T
Production over-costs in non-working hours 1300 €/T
Backlog over-costs 2300 €/T
Subcontracting over-costs 600 €/T

Results
Period t 1 2 3 4 5 6 7 8 9 10 11 12
Demand D(t) 1 475 510 1 655 1 320 1 757 1 210 1 603 1 475 1 320 1 685 1 199 1 782 16 991 T
Production 988 468 1 365 1 430 1 365 1 040 1 430 1 300 1 430 1 300 1 300 1 381 14 797 T
Subcontracting ST(t) 300 42 290 0 282 170 173 175 0 275 0 300 2 007 T
Total prod 1 288 510 1 655 1 430 1 647 1 210 1 603 1 475 1 430 1 575 1 300 1 681 16 804 T
Inventory S(t) 100 0 0 0 110 0 0 0 0 110 0 101 0 T
Production in overtime HS(t) 87 0 0 0 0 0 0 0 0 0 0 81 168 T
Over-costs 241 25 174 21 169 102 104 105 21 165 19 237 1 383 K€

Figure 14-9. Optimised S&OP results after event

300
Inventory Subcontracting
250

200
Tons

150

100

50

0
1 2 3 4 5 6 7 8 9 10 11 12
Months

Figure 14-10. Optimised S&OP graph after event

Overtime are carried out in period 1. Subcontracting makes it possible to

fulfil demand during months 2 and 3. Then the scenario remains identical
(Figure 14-9 and Figure 14-10).

5.2.4 Forbidden Storage

Storage becomes impossible – Imax(t) = 0. What becomes the optimum?

12 Patrick Genin, Samir Lamouri and André Thomas
Sales & Operations Planning

Parameters
Beginning inv entory 100 T Carrying costs 190 €/T
Capacity 65 T/d Production over-costs in ov ertime 700 €/T
Production over-costs in non-working hours 1300 €/T
Backlog over-costs 2300 €/T
Subcontracting over-costs 600 €/T

Results
Period t 1 2 3 4 5 6 7 8 9 10 11 12
Demand D(t) 1 475 510 1 655 1 320 1 757 1 210 1 603 1 475 1 320 1 685 1 199 1 782 16 991 T
Production 1 235 510 1 365 1 320 1 457 1 040 1 430 1 300 1 320 1 385 1 199 1 482 15 043 T
Subcontracting ST(t) 140 0 290 0 300 170 173 175 0 300 0 300 1 848 T
Total prod 1 375 510 1 655 1 320 1 757 1 210 1 603 1 475 1 320 1 685 1 199 1 782 16 891 T
Inventory S(t) 100 0 0 0 0 0 0 0 0 0 0 0 0 T
Production in overtime HS(t) 0 -75 0 -110 92 0 0 0 -110 85 -101 182 -37 T
Over-costs 84 97 174 143 244 102 104 105 143 240 131 307 1 875 K€

Figure 14-11. Optimised S&OP results with 0 storage capacity

350
Inventory Subcontracting
300

250

200
Tons

150

100

50

0
1 2 3 4 5 6 7 8 9 10 11 12
-50
Months

Figure 14-12. Optimised S&OP graph with 0 storage capacity

The under-load periods cause unemployed hours and the overload

periods imply the subcontracting and overtime saturation. The over-cost of
the strategy is then 1 875 K€, that is to say a variation compared to the
optimum of 60 % (Figure 14-11 and Figure 14-12)!

6. LIMITS OF LINEAR PROGRAMMING

What do we have to conclude from preceding simulations? That the level

of beginning inventory is not very significant? Certainly, but it is more
interesting to note than the optimum is not stable!
Linear Programming makes it possible the Supply Chain Manager to
generate the optimal scenario for a given set of parameters. It is more
adapted than the graphic techniques to exploit problems with multiple
constraints by providing a technique that leads to an optimal mathematical
solution, for a given set of conditions.
SALES AND OPERATIONS PLANNING OPTIMISATION 13

However, the scenarios developed previously show that an event in

production can make strongly diverge the optimum and thus to change the
optimal scenario. In a dynamic mode, nothing prevents the model to give
highly different solutions at only two days intervals. On a given date, the
firm directs its strategy towards the optimum, implements heavy actions (an
investment for example), the following day, the conditions are different and
the actions differ widely! The dispersion of the mathematical optimum given
by the linear programming model is caused by the parameters variability in
time whereas they are regarded as static in the model. This assumption is not
always exactly satisfied in practice on the whole time-fence of the S&OP:
for example, subcontracting capacity can be punctually limited.
Use of such tools is inseparable from attentive check of the significance
and validity of these assumptions, which in practice are unfortunately called
into question. Indeed, the coefficients of resources consumption or the costs
generally depend on the quantities: a subcontractor reduces his unit price if
the quantities are more significant. Linear programming cannot treat these
cases. “Linear Programming allows convex structures of production and
storage costs (non-decreasing marginal costs). What is awkward in this type
of constraint is the impossibility of introducing a launching cost, because in
general, at on a few months term, the production is realised at non-
decreasing marginal cost, once production of the first unit released” [3]. A
problem formulation in linear programming language requires lots of
assumptions (linearity and independences of the variables).

7. FUTURE PROSPECTS OF S&OP

How can the models that use Linear Programming be extended? The
answer will be to seek models using the non-linear and dynamic
programming. Certain authors suggested, in particular cases, other
approaches (stochastic optimisation models, Monte Carlo) [7, 10]. How can
the S&OP be approached in the industrial case of multiple production lines?
A deep research in the scheduling techniques must provide a suitable
answer. How can the scenario suggested by the linear program be
reinforced? The answer is by using the S&OP …
The S&OP is the process that drives the capacity level and actions to be
implemented so that Master Production Schedules can be completed. It must
thus be relatively stable, because of the weight of the taken decisions. For
example, the inventory level is limited by the storage capacity. In case of
events, the scenario of replacement can become then extremely expensive
for the firm whereas a simple contract with a storage partner could be signed
to provide a more robust scenario.
14 Patrick Genin, Samir Lamouri and André Thomas

The S&OP becomes the decision tool to make robust the scenario
suggested by the model according to the given set of conditions
(parameters). The Decision Support System must give an average answer
that will be the best possible response for several sets of conditions. The
S&OP will not be then mathematically optimised but will vary little when a
change in parameters will occur.
Design plans are the traditional tool used to establish the robustness of
the system answer (i.e. reduce the variability of the answers) by influencing
the control parameters (of adjustment) [9].
In the case of our S&OP models, the input parameters represent the Sales
forecasts. The control parameters are those that make possible to control the
system: costs of overtime, subcontracting, maximum inventory … They are
the action levers for the planner. The parameters of disturbance are all those
which intervene on the system independently of the will of the planner. It is
the case for example for the beginning inventory, the capacity of the line …
Taguchi calls the not-controllable parameters “noises” [11]. The more
“robust” a system will be, the lower the variability in scenario. A scenario
will be robust if it is not called in question by non-controlled external factors
(noises).
Use of Design plans allows to test, with a restricted number of trials, the
average scenarios that optimise the S&OP while limiting their variability.
The planner must fix, during the process, the nominal values of the control
parameters according to a double optimisation:
– the optimum operation of the system,
– the resulting robustness.
We neglect too often the second optimisation. We then work out
strategies on paper that cannot to be implemented or lead to the sub-optima
in a “disturbed” environment.

8. CONCLUSION

Within the framework of recent work, we have initiated this way of

research on the robustness of the S&OP thanks to simulations carried out by
taking into account controllable and not-controllable variables for the
industrial system and the manager [12]. Our survey of the literature made
clear to us that others mathematical tools can be interesting to test these
scenarios. We want to show that an optimum research in this field must be
obtained by various levels of simulations: a first step to define an optimal
target, a second to define a tolerable range of variation without degradation
of cost and finally a law defining the marginal loss according to the
difference to the optimum.
SALES AND OPERATIONS PLANNING OPTIMISATION 15

REFERENCES
1. Crandall, R.E., 1998, Production Planning in a variable Demand Environment,
Production and Inventory Management Journal, 4th Quarter, APICS, p. 41-48.
2. Genin P., 2000, Le Plan Industriel et Commercial : Recherche des pratiques optimales
du processus d’élaboration, Mémoire de DEA, LATTS, Ecole Nationale des Ponts &
Chaussées, Paris, France.
3. Giard V., 1988, Gestion de la Production, Economica, Coll. Gestion, 2ème édition.
4. Giard V., 1998, Processus productifs et Programmation linéaire, Economica.
5. Heizer and Render, 1995, Production and Operations management, Prentice hall.
6. Lamouri S and Thomas A., 1999, Optimization of the process of development of the
SOP, In: Proceedings of the International Conference on Integrated Design and
Production (CPI' 99), Tanger, Maroc, p. 328-338.
7. Moutaz K., 1998, An aggregate production planning framework for the evaluation of
volume flexibility, Production Planning and Control, Vol: 9 Nbr 2, p. 127-137.
8. Nollet, Kelada and Dioro, 1992, La gestion des opérations et de la production, Editions
Morin Gaetan.
9. Pillet M., 1992, Introduction aux plans d’expériences par la méthode Taguchi, Editions
d’organisation Université, Paris.
10. Silva Filho O. S., 1999, An aggregate production planning model with demand under
uncertainty, Production Planning and Control, Vol 10 Nbr 8, p. 745-756.
11. Taguchi G., 1987, Orthogonal arrays and linear graph, American Supplier Institute
press.
12. Thomas A. and Lamouri S., 2000, The new problem with Sales, Inventories and
Operations planning in a Supply Chain environment, In: Proceedings of the
International Society for Optical Engineering on Intelligent Systems in Design and
Manufacturing, Boston, USA, p 321-329.
13. Vollmann, Berry and Whybark, 1992, Manufacturing, Planning and Systems Control,
The Business One Irwin.