Narchal, R.M., "A Simulation Model for Corporate Planning in a Steel Plant", 1985

-590-

A SIMULATION MODEL FOR CORPORATE PLANNING
IN A STEEL PLANT

R.M. NARCHAL
Director (Corporate Planning)
National Productivity Council
Lodi Road, New Dethi-110003 (INDIA)

ABSTRACT

A Simulation Model for Corporate Planning has been désigned for
a Stee! Plant based on System Dynamics principles. This Model has been
designed for Material flow that takes place through a group of 12 production
shops arranged in six stages of production. The Model requires a time variant
input of Demand of 17 categories of finished steel products and 3 categories
of Raw materials. The Model generates behaviour of various objectives based
on the assumptions of the environment. The Mode! can be used for simulating
the impact of various strategic policy decisions on the corporate objectives.
The Model also guides the management in designing their long term invest-

ment policies related to expansion, modernisation and debott lenecking.
-591-

INTRODUCTION

The management of a steel plant in India has been very keen to develop
a quantitative model of Corporate Planning to achieve the corporate objec-
tives. With the help of such a model, the management is interested in simula-
ting the result of their various strategic decisions related to modernisation
and expansion investments, identification of bottleneck shop, and evaluation
of different project proposals before they are actually implemented. The
management realised the need of such a model based on the fact that in
the past a large number of investment decisions made by the company have
not shown the results as anticipated by the management. In the past the
criteria for investment decisions has been changing based on the environ-
mental conditions, the bias given by the top management and the other
conditions prevailing at that time. Based on the discussioris with the manage-
ment it was decided to develop a Simulation. Model based on the principles
of System Dynamics. The basic objective of this Model was that it should
assist the management of the company in designing their long term investment
and modernisation policies to achieve corporate objectives of the company.

This paper explains the principles of System Dynamics used in design-
ing such a model. The paper also explains the major applications of the
Simulation Model in the area of corporate planning. During the design of
the Simulation Model it was realised that the DYNAMO Compiler. available
with the steel plant has limited capacity. Therefore, it was decided to limit
the scope of the Simulation Model only to the production shops of the company
with an aggregated Financial Model.

SYSTEM BOUNDRIES MODELLED

This model has been designed for a material flow that taken place
through three main zones in the Steel Plant i.e. Iron making, Steel making
and Rolling. These three zones consist of group of twelve production shops
namely Coke Ovens, Sinter Plant, Blast Furnaces, Steel Melting Shops, Wheel
Tyre and Axle Plant, Blooming Mills, Sheet Bar and Billet Mills, Plate Mill,
Merchant Mills, Medium and Light Structural Mill, Strip Mills and Sheet
Mills. There ‘are only three categories of major Raw Materials i.e. Coal,
Ore and Scrap whose flow have been modelled in the six stages of production
-592-

as shown in Fig. (i). The material flow has been considered unidirectional
starting from Coal, Ore etc and resulting in finished Steel. The numerous
categories of finished steel products have been grouped into seventeen cate-
gories namely Black Plain Sheets, Galvanised Plain Sheets, Galvanised corruga-
ted Sheets, Merchant Mill Products, Skelp, Medium and Light structures,
Blooms, Sheet Bar, Billets, Plates, Strip Bar, Wheel Tyres, Ingot, Smal! Blooms,
Small Slabs, Rail Products and Forge Plant Prdducts. The Model does not
differentiate the finished product of Steel based on sizes and grades. They
are only represented by tonnages.

While designing the Model, a production shop consisting of main equip-
ment as well as auxilliaries is considered as a unified machine. The details
of the individual sections, machines and equipment are not considered while
modelling a production shop. In case, a number of production shops of diffe-
rent capacities are existing, in the Model these shops have been represented
by same number as hypothetical machines of average capacity. Similarly
there is no distinction made in the Model about the variety of sizes and
grades of material being fed to a shop.

A generic feed back structure of a production shop consists of eight
feed back loops. The same feed back structure exists in all the twelve
production shops. The integrated production flow, therefore, gets influenced
by 96 feed backs. These feed backs have been put in a Dynamo programme
consisting of about 1200 statements. The management have to interact with
the Dynamo programme with environmental scenarios inputs and its own
strategic policy inputs to see the behaviour of the various production outputs.
The Model provides an insight in the dynamics of production and assist the
management in changing the behaviour towards the desired end by allowing
them to simulate the impact of their strategic decisions.

MODEL OF THE STEEL PLANT

The integrated Model of the Steel Plant can be divided into two
parts as stated below:

i) Financial Model

ii) Production Model.
Ore

Scrap

Coal

-593-

Ore Fines
| Tron Making Zone I
Coke Ovens Sinter Plant
| Coke | Sinter
<< ¥.
9
Blast Furnaces
Steel Making Zone
_¥
Stee! Melting Shops.
Rolling Zone
x

Blooming Mills

¥

Sheet Bar & Billet Mill

|

Wheel Tyre & Plate Mill

Axle Plant

Sheet Mill Strip Mill Structural Mill} | Merchant
Mill

Fig. (i)

-594-

The Financial Model, in a limited manner simulates the financial!
consequences of the basic resources of production i.e. manpower, materials,
machines, etc. and largely depends on the results of the Production Model.
This Model simulates the behaviour of Return on Gross Block, Realisation,
* Works Cost and Undepreciated Gross Block. The feed back process governing
the financial flows could not be incorporated in the Financial Model due
to the limited capacity of the Dynam’o Compiler. The Financial Model has
no independent status and has to be run along with Production Model. |

The Production Model has been designed to simulate monthly Saleable
Steel production comprising of seventeen finished steel products. The material
flow takes place through six stages of production constituted of twelve produc-
tion shops. The Production Model has been designed by judiciously integrating
the production models of twelve production shops. A production shop has
» been the basic planning unit and a generic feed back structure has been
applied to model all the production shops. Inspite of the fact that underlying
feed back processes of all production models of all the production shops
are same, these models are unique to suit the typical material flow to and
from these production shops. The generic model of a production shop is
discussed below.

Generic Model of a Production Shop:

In this Model,production of a shop shows a dynamic behaviour from
month to month based on a number of influencing factors. These factors
depict interplay between the three main resources of production i.e. Machines,
Manpower and Materials. The machines have been the basis of computation
of capacity. To arrive at production, the utilisation of capacity has been
taken as a function of Demand for finished products, maintenance requirements
of machines, availability of raw materials and manpower. The impact of
power shortage has been put as an exogenous input in arriving at the capacity.
The influence of significant factors on production has been modelled through
feed back processes. There are a set of eight feed back loops existing in
each shop. The various feed back loops governing the production behaviour
in a Mode! of a shop are discussed below.
-595-

i) Demand Constraint Loop:

The influence of demand on the finished products of a shop can be
visulized as a feedback process. The shortages of Demand have been designated
as Demand Constraints. Higher is the Demand, lower will be the value of
Demand constraint. A. low demand indicates higher demand constraint meaning
throttling of production. As demand for finished products of a shop goes
up, the demand constraint on production reduces indicating a negative rela-
tionship of linkage between Demand constraint and production. The feed
back loop is shown in Fig. (ii).

a. Constraint

Fig. (ii)

Going round the loop, one finds that as production goes up, Demand
gets satisfied and therefore, goes down. The level of Demand decides the
Demand constraint. The value of Demand Constraint ranges between zero
and one depending on the level of Demand and Capacity for production.
All the three linkages round the feed back loop being negative, one finds
that the loop is negative in character.

ii) Material Constraint Loop:

The production of a shop is also considered as a function of the Raw
material availability. The shortages of materia! is represented as material
constraint and it indicates the extent of throttling of production on account
of material shortages. Its value ranges between zero and one. A zero value
of material constraint indicates complete production shut down due to non
availability of material, whereas the other extreme value of one indicates
that the production does not suffer in the month on account of material
shortages. The feed back loop is shown in Fig. (iii). When the materia!
-596-

availability goes up, the constraint of material on production goes down
indicating a negative relationship. As material contraint goes up, the thrott-
ling of production increases and the production goes down. This relationship
is also negative in character. The extent of feed back of production on material
availability indicates a negative relationship as when the production goes
up, the material availability goes down. The feed back loop is negative in
character as shown in Fig. (iii).

Material
Availability

Production _—

J

Materia! Constraint

Fig. (iii)

iii) | Down Time Loop:

This loop indicates that the maintenance needs of an equipment arrise
as a result of the usage. The underlying hypothesis is that if usage goes
up, the maintenance requirement will also go up in future and thus, reduces
the equipment availability for production. The phenomenon is called Down
Time Loop and is shown in Fig. (iv). It shows that equipment usage influences
the Down Time and there exists a positive linkage between the two variables.
The Down Time, in turn, influences usage i.e. when down time goes up,
the production usage goes down due to non. availability of equipment. This
linkage between Down Time and usage is negative in nature. The feedback
loop is negative in character.
-597-

Production/Usage _ Down Time

Fig. (iv)

iv) Capacity Expansion/Divestment Loop:

Demand for the finished products of a shop not only influences the
production but under certain conditions causes expansion/divestment of capa-
city. In Fig. (v) this phenomenon is explained with the help of a feed back
process. As demand goes up, management tends to increase the capacity
to meet the Demand. The increase in capacity is the cause of increased
production which satisfies the Demand. The feed back is thus negative in
character and guides the management in policy design for expansion of capa-
city. In case due to some reasons the Demand goes down, the feed back
process represents the phenomenon of divestment.

Production Demand
+ _

Capacity +

Fig. (v)
-598-

v) Manpower Constraint Loop:

This feed back process explains positive feed back phenomenon which
exists between four variables i.e. production, demand, capacity and :manpower
constraint. As the Demand rises the management tends to increase the capacity
indicating a positive linkage. With the increase of capacity the manpower
constraint increases. A higher manpower constraint indicates shortages of
manpower and this reduces production. The feed back loop, thus, is positive
in nature. The marpower constraint loop necessarily exists in conjuction
with the capacity expansion loop and therefore, has to be studied and analysed
along with capacity expansion/divestment loop. The Loop is shown in Fig. (vi).

3» Demand

V-
+ Capacity

Manpower
Constraint”

vi) Recruitment /Retrenchment Loop: Fig. (vi)

Demand for manpower fluctuates in response to the capacity. In
case of expansion of capacity, Demand for manpower rises. Till the recruit-
ment is effective, manpower constraint on production operates. Therefore,
any deficit on manpower, on one hand influences the manpower constraint
as discussed earlier and on the other hand it influences recruitment as shown
in Fig. (vii). When recruitment goes up, the level of manpower goes up,
thereby reducing the manpower constraint. The loop is negative in charac-
ter and operates on the management policy of recruitment/retrenchment.
-599-

Manpower
Constraint
a
Manpower. Recruitment
+
Fig. (vii)

vii) Machine Obsolesence Loop:

Based on the life of the machine, its capacity depletes. The rate of obso-
lesence effects the capacity, which in turn effects the obsolesence rate as shown

in Fig. (viii). A negative feed back process, therefore, explains the phenomenon
of machine obsolesence in the model of a production shop.

+
Capacity _ Obsoloscence

-

Fig. (viii)
-600-

viii) | Manpower Turnover Loop:

The manpower in a production shop is likely to go down due to turnover
and retirement of manpower. The turnover and retirement increases with the
increase of level of manpower. The feed back is, therefore, negative in character

as any increase in turnover rate decreases the level of manpower. The feed back
process in shown in Fig. (ix).

Manpower

Turnover Rate

+

Fig. (ix)

In addition to the feed back loops discussed above there were two more
feed back loops identified. One of these feed back loops was related to purchase
of the Raw material and the other related to power constraint. These feed back
loops could not be incorporated in the Dynamo programme of the Stee! Plant
model due to the limited capacity of the Dynamo Compiler.

The generic feed back model of a production shop has, therefore, eight
feed back loops as discussed above. Production is common variable lying on the
feed back loops of Down Time, Material Constraint, Demand Constraint, Capacity
expansion and Manpower constraint. Therefore, it provides a common focal point
for connecting various loops. The integration of the above loops results in a feed
back structure that has been the basis of design of the Production Model of a
shop. The integrated feed back structure is shown in Fig. (x).
-601-

FEED BACK STRUCTURE OF PHYSICAL FLOWS IN A PRODUCTION SHOP

PENDING
we ORDERS 4° ~ Ry INCOMING
i ay OROERS
.
/ . FINISHED
j \ 7 yo coos
y \ NET INVENTORY.
PRODUCTION
INCOMING MATERIAL + 4
RAWMATERIAL ORDERING DESPATCH
VATE _ RATE RATE |
A GROSS |
\ t PRODUCTION
+
\ . f = I
_ DEMAND
RAW —
SS MATERA gee,
~~ Ww i UTILIZED
CAPACITY
HOURS
AVAILABLE ones
ZCAPACITY HOURS
—— RAWMATERIAL +
“= CONSTRAINT

*
REQUIRED
EQUIPMENT
RS
FoSwn HOU
—MEN TIME
ne G, /

AYAILABLE +
MANHOURs \ monn , 7
REQUIRED es ce

~ \

+ AVAILABLE
EQUIPMENT 4 a se
RECRUITMENT ae HOURS SX EXPANSIUN
ete € N _ RATE
7 "
ee a \
! \ MAX +,

\ EQUIPMENT

i fons
POWER 1 HouRs + NO. OF
CONSTRAINT 1 a
+ * 7
eset. ile
POWER 7 ON DEMAND

ILABILITY = — . - .
AVAIL . =

OBSOLESCENCE
oe RAIC
-602-

Model of each of the twelve production shops is independently designed
and valedated before integrating them to arrive at a production Model of Steel
Plant. The integrated mode! requires input of demand of seventeen categories
of finished products and supplies of three categories of raw materials as exogenous
inputs from the environment. In addition, the production model requires the various
strategic parameter related to choice of technology, policy for expansion of capa-
city, recruitment etc. also as exogenous inputs.

MODEL VALIDATION

Before using the Mode! for generating future scenarious and undertaking
"What if" experiments, the Shop Models as well as the integrated Model have
been historically validated for a period of 36 months. The production and inventories
as generated by the Model were compared with the actual values taken from the
statistical records. Resemblence between the simulated results of the Model and

actual values indicate that the Simulation Model of the Steel Plant is a satisfactory
representation of reality.

The actual and simulated historical time series of Down Time, production
as well as Inventories have been super imposed on each other for the purpose
of comparison. In addition to this, the simulated annual values for the production
have been compared with the actual yearly totals. The two series i.e. actual and
simulated production have also been validated by computing the Root Mean Square
Error (RMSPE). The superimposed plots of simulated and actual production for
the Saleable Steel is given in Fig. (xi). It can be visualised from Fig. (xi) that
the monthly production behaviour simulated by the Model represents the actual
behaviour. By comparing the yearly totals of actual and simulated production,
it was found that the deviation was below + 2% of the actual results. The value
of Root Mean Square Error between actual and simulated series of Saleable Steel
production is 6.66% which was considered acceptable.

MODEL APPLICATIONS

The simulation Model can be used to carry out a variety of experiments
under controlled conditions to study the behaviour of various indicators of interest
such as Saleable Steel production, works profit, Return on Gross Block etc. These
-603-

ACTUAL PRODUCTION

MODEL PRODUCTION---------

Y= FG HSV watess

Fig. (xi)

VALIDATION PLOTS OF SALEABLE STEEL PRODUCTION
-604-

experiments tend to answer the question related to various strategies likely to
be adopted by the Management under a set of assumptions from the environment.
These experiments are grouped under four sets as given below.

i) Bottleneck Identification:

Investment decisions in a Steel plant has to be guided by understanding
of Bottleneck Shop in addition to various other. considerations of the management.
In Case, it is feasible to identify the bottleneck shop for the future, based on
a set of assumptions about environmental influences and decisions about Strategic
variables, the investments have necessarily to be made on priority bases in the
Bottleneck shop. The investments made in any unit other than the Bottleneck
unit are going to be futile till the Bottleneck unit is first debottlenecked. A wrong
bottleneck identification can result in very serious consequences leading to invest-
ments without any significant production increase of the Steel Plant. This Model
has the capability to compute Intensity of Bottleneck every month for every shop.
Over the whole simulation period it can identify in a dynamic manner the order
in which management decisions have to be taken for debottlenecking. In case,
for any length of time in future, behaviour of Demand, essential supplies and
power availability are specified, the simulation Model is in a position to arrange
all the production shops in the order of debottlenecking priority. The Model, thus,
guides the management in taking correct decisions related to investments for
debott lenecking purpose.

The dynamically changing conditions assumed during the simulation run
as well as the inter-dependencies taken into account amongst dynamic capacities,
demands, materials power availability of the twelve production shops in arriving
at priority of. debottlenecking is one of the major features of this model. The
intensity bottleneck graph as produced by the Model is given in Fig. xii).

ii) Generation of Future Scenarios:

The Model can be used to produce three sets of scenarios under the opti-
mistic, pessimistic and most probable assumptions of the environment related
to Demand of seventeen categories of finished steel products, supply of 3 essential
raw materials, availability of power and the likely inflation during the simulation
REGAN PLETE INS AT PLt83-8953-6 & JULY 1986

yPSUITIIOG Aysua] JOy yoTg

TeCCen, TERte re 1 Iggral, T2E4=Be TOetse+ Luter, TEMES. TaKbele TGTeT + TBweems TSS HSH
26.2 Cal 160 REAL
ec i a res sw ee nee ee ee eee RESTORE THe

. 7 fr . 2 FSERPL Te
: . Yeu foe » FSLRPL Tew
‘i : “YT ro. LFS + SISPTome
. . » Yo te R FLIES .
. . » ' id a FES .
. . * He ON TSTSP .
S : » bow ris F + RESTOLT
: E . “. Mon ITS P 2 MPA TBL
: é > Yr oR 4 FRITS @ + 1SUT

we ee ee wpe ee te tc cente He = FS eRe ee ee SILT
. . * Yo. HM FURS + OLae NPT
. + * Yo. KA PFURL + FSB OLT
. ; » t.o4n Pru Rt 2 PROP Te ed
* a ™ Yo. MP Tr RS. 2 STe ROFL !
* . » rt. r T FOR 2 RSTO L, Pen s
. . » Yo. PM 1 Fin . Tief be Pa a
: E ts pu T Fan a RETR L eat i
. . Ve “Pe Tourn oN S Pa
. . Yo. P _ iT LERS ao Rbehde Pmt

a i i ees Ter carr s ett Bey
. . Yr. ew T FAK a TEP tees
: ‘ » Ws Pom your a RSTro
. . * Yo Point LF oe RSTefa
. . * re. PoONK T tran . Pet
. . . Ye eoom iran SSERL!
. . Ye. Po FM ure 2 ONSTAR
z . » Ye Po fm Care aNd
: . » Ye. Po f 4 Lb orsK = Stem
‘ . oY Ye Po oF oe Lori oS

. be mee ee we ey le PT om ote Ler ine © 6 KSef Den

i ee : ’ ° te. ° “ L oFR 2.
. . . vs. tr “WHOL BFK .
. . > t. Liu * Lorre "
‘ : . te te hOLAPR a
. z » te. e rn
. . » ve P en a

-606-

period. In each of these scenarios of the steel plant based on set of assumptions,
a priority for debottlenecking of various shops is also listed down by the Model.
This priority is applicable for the future period for which the scenarios are prepared.
The Model, thus, guides the management to plan for their future strategies based
on the optimistic, pessimistic and Most probable scenarios.

iii) Listing of Strategic Variables

The Model lists down the various Strategic policy variables which are
within the control of the management. The behaviour of the production is caused
based on the dynamics ‘created due to the feed backs existing in the Model as
well as due to the strategic policy decisions taken by the management. The Model
tends to answer the various "What if" questions related to choice of the strategies
by the Management. The Model can be used to simulate the impact of various
strategies on the Corporate objectives before the strategies are actually implemented.

iv) Project Evaluation

Whenever a Strategic Policy variable is to be changed by management,
it may require a modernisaion project to be undertaken. Often the situation arrises
when more than one project can be thought to change value of a strategic policy
variable. In such cases the Simulation Model can also be applied to carry out
evaluation of projects. This experiment can be done by running the Modelon computer
with two sets of values of the parameters. One set of values. assums that the
project is not implemented and the other set assume the project has been imple-
mented. To illustrate this application of the Model, a project of putting additional
sinter plant was taken for evaluation. The results obtained with and without addi-
tional sinter plant are given in Fig. (xiii). Based on the results achieved the manage-
ment can take a decision about the project. In case more than one projects are
to be compared, the Model needs to be run for all the projects and comparison
made. In each case the simulation Model also lists down the new priority for
debottlenecking. The Model can, therefore, be very effectively used for making
investment decisions related to various projects in future.
-607-

—
Without Sinter Plant WithSinter Plant
S.No. | Variable Year Year
1 2 3 1 2 3
1. Works Profit 1808 2006 | 2292 1833 | 2088 | 2380
(Mi Ilion Rs.)
2. Realisat ion 4918 5552 | 6301 4940 | 5552] 6301
(Mi Ilion Rs.)
3. Works Cost 3110 3546°| 4010 | 3108 | 3464) 3921
(Million Rs.)
4
4 Retun % 45.5 50.52] 57.73] 46.16) 52.64 59.96
Fig. (xiv)
Comparison of Result with and Without Sinter Plant
ACKNOWLEDGEMENTS

The above work was carried out jointly by a team consisting of two NPC
consultants and six executives of the Stee! Plant. The work done by Mr. Rakesh
Kumar, Director, National Productivity Council on this project is deeply acknow-
ledged. The auther is also thankful to the team of Stee! Plant for providing all

the relevant information in designing the Model.