-776-
An Evaluation of the Management
of Stocks in a Chemical Company
H.Sedehi, P.Verrecchia
TEMA S.p.A.
ABSTRACT
The present article describes an example of the application of the MDS
approach (Modelli Dinamici per Strategie - Dynamic Models for Strategies)
to the study of criteria for a supply. policy and for a consequent
economic evaluation of the different supply policies which may be
adopted. The aim of the study is limited to an evaluation of the direct
economic effects of the different hypothetical policies, excluding those
associated with the market and with different production steps.
The results of a series of simulations using the model are presented
along with an outline of the economic benefits deriving from the adoption
of a "tight-rein" stocks policy carrying a reasonable level of risk.
INTRODUCTION
This paper refers to the use of the MDS system (Gervasio, 1984,
pp.231-237) for evaluating the overall effects of alternative policies of
purchasing and managing raw material stocks on the economic results of a
company. :
The company considered runs chemical and petrochemical plant and produces
a wide range of products including both fine and heavy chemicals.
The production is located at several sites having a number of plants,
with a large variety of processes and raw materials.
The rigid schedules and the continuity required for production, typical
of process industries render raw material supplies ons of the most
delicate stages in the whole production cycle.
The supply policy thus takes on great importance and plays an important
role within the general company strategy.
-777-
In the past, with the emphasis focussed upon production, raw material
reserves policies were based upon the following criteria:
a) devolution of decision-making to the single factory level;
b) absolute pricrity given to the safeguarding of production.
This policy, in reality, tends to favour an overdimensioning of reserves
which leads to a freezing of liquidity and other related financial
disadvantages, which can not be ignored, given the high cost of money.
In the light of the growing importance of financial problems, with
respect to previous periods, the company felt the need for a close
examination of this policy with the following aims in mind:
i. To quantify the costs of running out of stocks and/or keeping an
excess of raw materials, as well as the relative financial costs
for the different supply policies, in order to find an "optimum"
policy, or at least a "satisfactory" one.
2. To succeed in formulating a set of uniform criteria as the basis
of supply policies, which could then be applied to each plant.
Point 1 has, as its central aspect, the evaluation of the main risks
connected with a policy of "tightening" reserves of raw materials with
respect to the current practice of the "abundance" of reserves, comparing
possible damage of the former with the known financial disadvantages of
the latter.
In this way it is possible to make certain choices which, while
respecting local plant autonomy, are based upon certain general criteria
(point 2).
The problem is a complex one due to a series of structural aspects.
a) The largely ’random nature of the system of supplies. The most
important raw materials generally come from faraway countries
and are usually transported by sea. Even with a careful
programming of supplies, the arrival dates of the lots purchased
as well as their quantities are often subject to great
uncertainty due to exogenous factors which are completely beyond
company control. These include, above all, “disturbances" which
lead to changes in transport plans and schedules (such as, for
example, weather conditions) and unpredictable commercial
policies of untrustworthy suppliers.
-778-
b) The difficulties in evaluating the economic damage resulting
from running out of raw material supplies. Often there may be
incidents which interrupt production, but not the regular flow
of sales of the finished product, or even incidents which cause
long down times leading to exhaustion of finished product
reserves thus, leading to direct effects on sales. The negative
effects of these two situations are serious, due to both the
high costs associated with interrupting the production cycle
(this requires a series of interventions and operations on the
plant) and also to possible consequent reductions in market
share and deterioration of company image.
c) Finally, a part of these products may be used as raw materials
for other productive processes of the company and this makes an
evaluation of the economic implications of running out of stocks
even more difficult.
The application described here has, for the sake of simplicity, been
limited to the direct economic effects of the hypothesised policies, thus
excluding those associated with the market and other production processes
which depend upon the particular product.
THE MDS SYSTEM
The structural characteristics of the problem require the use of a
simulation system, such as the MDS one, for the following reasons.
1. The MDS methodology (Bartezzaghi et al., 1982) allows one to
model the set of interrelations existing between the different
company subsystems involved (especially the supplies and
production circuits and the econcmic/financial subsystem), in
order to be able to measure correctly the costs relating to the
different supply and stock policies.
2. The behaviour of the real system, where the uncertainty in the
exogenous variables is added to that of the operating conditions
of production, can only be represented by computer simulation.
Only by using this instrument is it, in fact, possible to study
the economic effects produced by completely random processes,
which are, however, limited by the framework of hypotheses which
fix the maximum range of random variations of the significant
variables of the system.
3. The MDS system, in the present example, allows one to evaluate
the long term behaviour of suitable indicators of company
performance, given the adoption of a certain supply policy. The
need for representing the evolutionary dynamics of company
-779-
activity and for evaluating the long term performance are
related to two fundamental aspects:
- the necessity for ignoring transitory conditions and thus
describing the normal situation which effectively represents
the economic trends determined, "coeteris paribus", by the
simulated supply policy;
- the need for "filtering" anomalous results due to short term
random effects: the low frequency of annual purchases,
which characterises the supplies of strategic raw materials
for production, requires the extension of the simulation
over a long period.
The MDS system was used, first of all, for a single plant and for
supplies of a single raw material, sulphur.
This particular case is illustrated below, the concluding paragraph
showing the effects produced by the results on the overall policies of
the company along with some possible developments in the application of
the MDS system.
THE MODEL
The plant produces caprolactam (feedstock for synthetic fibre production)
and, as a by-product, ammonium sulphate. The entire production of the
plant is absorbed by group members and thus the relationships with the
market are limited to the economic transactions existing between the
Plant and the group to which it belongs.
The raw material examined is sulphur which, while not being the main one
on the basis of the quantity used, it is the simplest to deal with as it
involves fewer interactions with other plants. The main sources of
Primary supply are Canada and Poland. There are also other secondary
sources of supply (that is, sellers and not producers). A group of the
latter are called, for simplicity, the "Mediterranean area", while
another source is an Italian company which can provide small quantities
of sulphur for immediate delivery.
The availability of sulphur in Italy allows the possibility of
interventions whenever the reserves fall below the desired minimum
quantity. But these possibilities are diminished by the total amount
available, which imposes a maximum limit on the quantity of sulphur
available annually in Italy.
Since the "market" in which the company operates ensures the sale of all
~-780-
of its product, the supply plan may be defined annually on the basis of
the production plan (optimised for an efficient use of plant) and of the
predicted availability of the different sources of supply.
Once the production (and thus the production rates) and the initial
warehouse level of sulphur are known, it is possible to formulate a
purchasing plan. Such a plan defines: the desired arrival dates of the
lots purchased from the three commercial sources, the ordering dates
(having fixed the delivery terms for each supplier) and the quantities
for each order.
Having fixed certain exogenous conditions regarding the quantities which
may be periodically ordered (availability of the different suppliers),
the most significant decisional variable in the formulation of the
purchasing plan is the minimim stock level (MSL). The latter, given a
certain production plan, requires a certain scheduling of the purchasing
of lots, within the margins allowed by the Italian purchasing situation,
to resolve urgent problems of low reserves of the raw material.
The MSL value corresponds to the acceptance, by the company, of a policy
involving a certain risk, given the aleatory conditions of supplies and
production. These conditions can cause changes, even significant ones,
in previously formulated plans and thus breaks in stock.
The value of the minimum stock level used before the application of the
model was 5000 tons of sulphur.
Given this method of operation, the aim of the MDS model was to simulate
the dynamic behaviour of the supply and production functions and evaluate
the economic/financial effects in order to:
1. determine the minimum stock level for sulphur which statisfies
the criteria of economic efficiency;
2. determine, on this basis, the supply policy for sulphur.
Given the simplicity, in this specific case, of the market subsystem, the
subsystems represented using the MDS methodology are the
technology/resources and the economic/financial subsystems.
The overall model produced consisted of 23 levels, 40 flows and 30
parameters.
RESULTS OF THE SIMULATION
It can be seen that the decisional variable, which determines the
different purchasing policies for sulphur, is the minimum stock level.
-781-
In order to determine its "optimum" value it will thus be necessary to
carry out, for each alternative minimum stock level, a number of
simulations covering all the possible configurations of the external
environment. For each possible range of uncertainty there will then be
different values of the accumulated result, which summarise tke . 10-year
evolution of the company, as described using the MDS model.
The results of the simulations are given in Figure 1, which shows the
plots of the accumulated result E as a function of the chosen Minimum
Stock Level (MLS) and of the variations in the parameter r, relating to
the delivery dates. In the figure the continuous lines interpolate the
set of discrete points corresponding to the same value of r.
«|
wy
Figure 1. The Accumulated Result
-782-
The amount by which the minimum stock level can vary is between a minimum
of 1,000 and a maximum of 5,000 tons of sulphur (the latter is the level
used so far by the company).
Within this interval, the plots of the accumulated results show a clear
minimum (it should. be remembered that the MDS model shows a positive
result with a level of negative sign).
The position of the minimum, on varying r, is to be found within an
interval of between 2,000 - 2,500 tons of minimum stock.
One fundamental result of the simulation is thus an indication that the
"abundant" stocks policy (MSL = 5,000 tons) does not satisfy the economic
criteria of the accumulated result, and does, in fact, worsen the
situation.
Entering into the merits of the overall behaviour of the accumulated
result, the following points can be made.
That part of the curve corresponding to low minimum stock levels is the
zone in which breaks in stock occur. The gradient of the curve is very
steep due to the very high fixed costs of breaks in stock. On going from
MSL values of 2,000 tons (breaks in stock are then practically absent) to
values of 1,000 tons (with 5-6 breaks in stock over 10 years,
corresponding to 15 days of uncertainty in the arrival of the sulphur)
there is a worsening of the accumulated result of the order of 2.5 - 3
billion lire (US$ 1.3 - 1.5 million).
The part of the curve corresponding to high MSL values is a region where
breaks in stoch are absent, but where there are often excess quantities
of stock.
The gradient of this part is less steep and is determined on the basis of
the costs of excess raw material and the increased financial costs.
Figure 1 shows that' the minima of the set of accumulated result curves
are concentrated in the region of 2,000 - 2,500 tons of minimum stock.
In particular, when the behaviour of the minima is compared with the
range of uncertainty, r, it can be seen that the shift Amin. between
the "absolute" minimum of the accumulated result (for r = 5 days) and the
"relative" minima (r > 5 days) is fairly small when the delays/early
arrivals are within % 15 days (see Figure 2).
-783-
1250 La
\
|
60} — —- — — — y
\ |
1 |
a |
1 i | ——
5 10 15 20 '
Figure 2. Shifts in the minimum values of the accumulated result
Given the hypothesis that this time interval will not be exceeded, this
means that operating with a MSL of between 2,000 and 2,500 tons will not
lead to breaks in stock, or penalties for excess of stock, and will
produce a relative "stability" of the accumulated resuit.
Comparing such 2 policy with an "abundant" stock policy (MSL = 5,000
tons) it can be seen, for example, that for an uncertainty of 10 days in
the arrival of the sulphur, there will be, other things being equal, ‘an
improvement in stock management of about 250 million lire
p.a.(US$125,000), of which 60 million are due to reduced financial costs
and the remaining 190 to the reduction in costs of excess stock.
REFINING THE METHOD FOR MSL DETERMINATION
If greater precision is required in the determination of the minimum
stock level a more detailed analysis is required regarding the
uncertainty of the arrivals. It should be noted that the parameter, r,
utilised (which was given the values of 5, 10, 15, 20, ..., days) is the
estimate of the maximum delay/early arrival considered plausible for a
-784~
given supply situation (an automatic mechanism included in the model
gives a "random" arrival date within the interval t% zr where ty is
the programmed arrival date).
It should be possible for a purchasing expert, using his own estimates
and/or using past data, to construct a graph of the type shown in Figure
3, where p is the probability that Q, the arrival date, does not lie
within the interval ty t r at least once a year).
In the graph it is assumed that the probability of a delay/early arrival
of more than 20 days with respect to the planned arrival date, te > is
-very low (almost insignificant) and, vice versa, the probability of
delays/early arrivals of more than 5 days is very high.
Given this, the graph of the accumulated result can be re-examined (see
Figure 1), where the line (a) represents the locus of the minima for the
stock levels.
It is assumed that working points along that line are chosen. Remaining
at the intersection with the curve r = 20 days it is practically certain
that values outside of the range will not occur and, consequently, every
extra penalty for breaks in stock due to the arrival of sulphur outside
of the predicted range may be excluded. In this case, however, there is
a loss of "margin' equal to the segment DA (estimated over 10 years) with
respect to the improbable case corresponding to r = 5, with no values of
arrival dates outside of the range te. t 5 days.
By taking progressive intersections between the line (a) and the curves
with the parameter r = 15, 10 days, working points with progressively
lower theoretical losses of margin are obtained (segments CA, BA) which
include, however, growing probabilities of values falling outside of the
ge and thus of extra costs.
10 1s
Figure 3. Probability associated with the "range of uncertainty"
-785-
For a quantitative evaluation of these extra costs the following
assumptions can be made:
- for low values of the probability p one may ignore the event of
a value falling outside of the range more than once a year;
- given operation along the line (a), locus of the minima, it is
assumed that every value falling outside of the range causes a
break in stock (this assumption is a very cautious one because
of. the serious effects of breaks in stock).
Given these assumptions, the extra penalty p.a. can be taken, for low
values of p, to be equal to the product p(r) for the fixed costs of
breaks in stock (500 million lire, US$250,000).
This is shown in Figure 4, where an unbroken line has been used for the_
meaningful portion of the graph. .
Figure 4. Extra costs due to breaks in stock
The loss of theoretical margins p.a. must be added to these costs, and
these are: evaluated as a function of decreasing values of r and can be
quantified using the intervals EA, DA, CA, BA (reduced by a factor of 10,
equal to thesnumber of years' simulation). On adding the two trends, the
trend shown in Figure 5 is obtained.
ae
-786-
Figure 5. Extra costs due to breaks in stock and loss of margins
CONCLUSIONS
The methodology and the method developed allow the extension of
simulations to situations involving a higher level of complexity, with a
simultaneous consideration of the ranges. of uncertainty relating to
several variables. The model thus allows one to define efficient
policies for the management of stocks over wideranging conditions of
variability of the environment and of the economic criteria relating to
the long term performance of the company.
It is thus possible to objectively and quantitatively <neasure the
economic benefits deriving from the adoption of a stock policy which pays
more attention toe the global effects which it has in the company, rather
than a policy having a strictly "functional" orientation (especially
towards production, as in past company experience).
The MDS application has thus led to the beginning of a general process of
revision of the supply policies, following the lines and the criteria
developed in this example.
There is the prospect of applying the MDS methodology to more complex
situations. One cen develop a simulation model, extended to include
several plants of the same company, which takes into account the
interactions existing between the different operational units and their
production. For example, the already mentioned utilisation in series of
products from some plants which are used as raw materiais or partly
prepared products for others.
Also, systems with a high structural flexibility can be developed in
-787-
response ta the different requirements of the management. Regarding the
latter, it is possible, in particular, to introduce procedures capable of
modifying in a simple and rapid way the interactions which link the
different “operational states", each of which is characterised by a
suitable description. In this way it is possible to carry out
simulations given the hypothesis that the operational context which is
being analysed. changes, i.e. that the macrostructure of the model
changes. This allows the study, for example, of supply policies for
single plants and for integrated groups of them.
REFERENCES
E. Bartezzaghi, S. Mariotti and H. Sedehi. “A Dynamic Simulation
Modelling for Firm-System Strategies", 7th Symposium liber Operation
Research, St. Gallen, 1982.
E. Bartezzaghi and S. Mariotti (Eds.) "Modelli per le Decisioni
Strategiche Aziendali", Franco Angeli, Milan (1983)
V. Gervasio, "A System Dynamics Approach to Corporate Modelling",
Proceedings of International System Dynamics Conference, Oslo, Norway,
(1984) pp.231-237.
