THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986
A STUDY OF SYSTEM DYNAMICS FROM THE VIEW OF SYSTEM THEORY
Qifan Wang Yang Xinnong
System Dynamics Group
Shanghai Institute of Mechanical Engineering
Shanghai, China
May 1986
bst:
Since L.V. Bertalanffy first brought’ forward the general
system theory, people have been paying more attention to
research work from the systems point of view, thus bringing
about the development of some system scientific fields such
as systems engineering, operations research and management
science, etc.
Now, in addition to general system theory, there are many new
subjects related to studying system dynamic behaviors and self-
organization such as synergetics, dissipation structure theory,
etc.
In this article, we study system dynamics from a philosophic
viewpoint and try to study important points and methods that
can be introduced by systems theory. We search into the
relationships between system dynamics and other fields.
INTRODUCTION
With the constant development in science and technology,
there has occured a developing tendency toward highly diversi-
fied and synthesized science and technology. Study methods,
once isolated from each other, have entered a new era --- the
general and systematical stage. In recent decades, new system
disciplines have been emerging one after another. For example,
the establishment and development of the general system theory
of Bertalanffy, the cybernetics of Wiener, and the information
theory of Shannon enable us to study various sorts of systems
theoretically and practically in an increasingly extensive
manner. The research work ranges from the concrete (e.g. techni-
cal, biological, or economical) study methods to he proposal
for establishing a system of completed or trans-disciplinary
system science. On the other hand, this trend attracts more and
more people's attention to the analysis and study of rising
disciplines.
System dynamics (S.D.), which studies complicated information
feedback systems, was created at the same time as those disci-
plines above. This systematic discipline, however, has so far
remained unfamiliar to scholars in many local areas and countries.
This article will, from the angle of system science, study the
characteristics of system dynamics and its particular view-
points and methodology and analyse the connections and distinc-
tions between this discipline and others.
439
440 THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 191
AS an new discipline, S.D. was first developed in 1956 by
day W. Forrester at MIT. In the 50's and 60's, Forrester con-
ducted in depth studies of industrial and economic systems
as well as systems combining human, financial, and technical
sectors. Through the analysis and study of the properties and
characteristics of these systems, he obtained many important
insights into systems with information feedbacks into and the
essential structure of systems. With the constant expansion of
the study range of S.D., its theories and methods are increa—
singly approaching perfection. For a long time, however, a great
number of people have been fascinated by its applications,
with little knowledge of its theories and basic ideas. Some of
them even have taken 8.D- to be merely a method of simulation or
a component of certain disciplines. In fact, S.D, since its
conception, has developed its own system. Being an independent
discipline, S.D. has its own theoretical system and scientific
methods. However, it has long been ignored as a result of its
close connection with application. Whether in complicated
systems study or in the establishment of large-scale, multi-
variable, nonlinear system models, the essential assumption and
simulations methods used in system dynamics are ath based on
certain system theories and methologies.
MAIN POINTS
With the help of computer simulation techniques, S.D. is an
interdisciplinary science which is based on feedback theory
combined with basic views of systems theory and application
of computer simulation techniques. S.D. defines a system as a
eomposition of different and interacting elements organically
linked together to accomplish a function for a common objective.
The prominent characteristic of S.D. is structure-function &
simulation, which differs from simple function simulation, such
as blackbox simulation. It constructs the basic structure of
a system based on the micro-structures within the system.
The micro-structure of a system is thought to produce the
macro-dynamic behavior of that system. Such models are much
more suitable to the study of problems varying with time.
Beginning with the relationships between the structure and
function of systems, we shall research further into the charac-—
teristics of structure-function simulation of S.D. in the
following paragraphs.
Structures and Functions
A system contains structures and functions. More exactly, a
system is an integration of structures and furictions. The struc-
ture of a system means the order of partss The structure of a
system depends on the characteristics of the parts and the
interactions or the relations between them. The function means
the order of process or the whole effect formed by the activities
of elements themselves or the interactions among elements.
The structure of a system indicates the distinguishing features
of the form of a system while the function indicates the
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986 441
features of the behaviors of a system. The difference between
structure and function in a system is relative. They are mutually
causal and mutually prerequisite. Under certain conditions
the structure can be changed into function, and function changed
into structure. They cannot be separated from each others
Therefore, when analyzing and studying a system, we shall take
into account not only its functional behaviors, but also its
structure. Through the examination of the structure and function
of a system alternately, a model will be built which can better
reflect the actual system. This requires the collection and study
of not only various kinds of data and diagrams that reflect the
functions and the behaviors of a system, but also other informa-
tion about the system structure or about the interrelations and
the interactions between elements. This information is not merely
a great amount of data, for statistical data represents only
a@ small part of our knowledge of the world. Therefore, it reflects
merely a part of the functional phenomena of a system. To cons—
truct a model of a system really, we must investigate in detail
the causality of a system that the real world contains, and
connect visible dynamic variations and invisible causality
together.
With the use of logic and our abilities in analyzing and solving
problems, the right knowledge or conception of a system can be
developed, and then can be combined into the structures of
simulations. This process, however, requires knowledge of the
dynamic trends of a system and the interrelations between
structure and the function, i.e., the relations of the feedback
structure and the dynamic behavior of a system. In S.D.,
a feedback loop can be classified as positive or negative,
A system is the composition of those positive and negative feed-
back loops in a certain way.
Therefore, when a system grows exponentially, there exists a
positive feedback loop which plays the leading role in the system.
When a system, after being disturbed, returns to its initial
state, it indicates that there is at least one strong negative
feedback loop in the system. Oscillation shows the existence of
one time-delaying negative feedback loop or the existence of
more than two negative feedback loops. S shaped growth is produced
by connecting positive feedback loops and negative feedback
loops with nonlinear links.
The interrelation between structure and function is not only
helpful for us in analyzing the system model but also necessary
to the process of building a model. When we construct a model
of a system, it is helpful often to inspect the reference modes
of the system.
Basic Structure of a System
Another prominent characteristic, when we study a system using
S.De, is the essential structure of a system. S.D. supposes that
the feedback structures are the basic elements constitut:
a system. A system is made up of elements, activities of elements,
and information. Elements are the practical basis of the
existence of a system, while information plays a critical part
442 THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1981
in the system. Depending on information, the elements of
a system are constructed to be structure, and the activities of
the elements bring into the behaviors and functions of a system.
The concept of the composition of a system was formed gradually
with the development of science and technology. Especially
information theory, as a sort of fairly strict concept, was not
used until early this century. The information theory of Shannon
describes the reality and measurements of information
quantitatively.
Through studying the common law comparing the self-regulation
function in organic systems and the automatic control function
in machine systems, Wiener introduced the concept of information
feedback from organic systems into machine systems. Thus,
cybernetics was founded. In the meantime, Forrester was investi-
gating the dynamic functions existing in industrial and socio-
economic systems and introduced the concept of information feed-
back to the industrial and socio-economic systems. In fact,
there is an obvious phenomenon of information feedback in socio-
economic system. The problem is how to learn it. A structure
(or theory) is essential if we are to effectively interrelate
and interpret our observations in any field of knowledge. Without
an integrated structure, information remains a hodge-podge of
fragments. Without an organized structure, mowledge is a mere
collection of observations, practices and conflicting incidents.
‘But now the concepts of 'feedback" systems seem to be emerging
as the long-sought basis for structuring our observations of
social systems." (Jay W. Forrester, PRINCIPLES OF SYSTEMS)
In a system, the basic structure is. made of feedback loops.
A feedback loop is a loop coupling state, decision making
(executions) and information, which correspond to the three
components of a system, i.e., elements, activities and infor-
mation respectively. The changes of state variables depend on
the results of decision-making or actions, and the decision-
making @ctions) process can be classified into two kinds. One is
to depend on the self-regulation of information feedbacks
(such a phenomenon exists universally in the organic world,
society and machine systems); the other depends not on the
feedback of information but on an intrinsic particular law.
This phenomenon exists in the non-organic world, and information
in this case, however, is not non-existent at all, but is only
at a potential state and hence not utilized yet.
A feedback loop is a fundamental structure which is composed of
states, decision-making and information. A complex system,
therefore, is the composition of these interacting feedback loops,
and the general function is produced due to their mutual connec-
tions and interactions with each other.
Wholeness and Level of a System
The wholeness and level of a system are the theoretical basis of
S.D. to study and analyze systems by means of synthesis and
decomposition principles. "The whole is more than the sum of.
parts". This point, originating from the thesis of Aristotle,
has already become one of the important viewpoints in system
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986 443
theory. The characteristics of wholeness is different from that
of elements, because the structure and function of the whole are
different from those of the parts. This is known as the principle
of wholeness. It represents the fact that as the different parts
ina system interconnect and interact with each other to form
a particular structure, the characteristics of the system have
been changed. It is not merely a piling up of structures and the
accumulation of functions by which a new specific structure and
function are formed. A national system, for instance, is composed
of production systems, finance and trade systems, transportation
systems, and so forth. Obviously, the whole nation, as such a
big system, should not be considered as the simple summation of
these sub-systems. Therefore, when studying a system, we have to
consider, with the viewpoint of wholeness, the feedback mechanisms
and interrelations between the entire system and its sub-systems,
and those among the sub-systems. We should never just simply put
some sub-systems together to form the entire system.
Our emphasis on the wholeness of a system does not mean that a
system has no level and hierarchy. The relative independence of
elements and structures brings about a hierarchy of structure
and then the hierarchy in function results because of the relative
independance of processes and functions. The hierarchy of a system
results because a system is an integration of structures and
functions. When a system is being studied, it is of great signi-
ficance to clarify the level and hierarchy of the system for the
reason that a concrete natural law always implies the law of
certain systems hierarchy. Laws of elementary systems penetrate
naturally into advanced systems. However, the advanced systems
have their own specific laws. They can never completely return
to the laws of low level systems. Of course, we should not forget
that the objective world emerges not only in the form of continuous
development from elementary systems to advanced systems, but in
the form of the transfer between different levels in a systen
as well.
Thus, we can lead to three important principles in system study
or the establishment of a system model.
1) On the basis of the wholeness and the level of a system, the
principle of composition and decomposition can be applied to the
study of a system. We emphasize that a system and its whole effect
should be studied from the viewpOint of wholeness. On the other
hand, the concept of hierarchy and levels in‘a system implies that
a system is composed of sub-systems of different hierarchy and
levels. A complex system is composed of (positive or negative)
feedback loops of different forms, Thus, by using the decomposition
principle, we may dissect a system'and analyze the structures
and dynamic laws of a system step by step. First we analyze the
relations between a system and its environment and determine the
boundary of the system. Then we examine the levels-of the system,
analyze it and search for sectors of a system. Finally we gra-
dually reduce the examination scope, paying particular attention
to the feedback mechanisms and the structures of sectors. This
process is called gradual focus. Meanwhile, when we analyze and
study the mechanisms and the structures of a system, we should
rely on the roughness-to-detail principle, i.e., from rough causal-
loops to flow-diagrams which reflect the relation among variables,
and eventually to the writing of DYNAMO equations. The decomposi-
444 THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 19
tion process discussed above is of great importance to analvze
and understand the internal structures and the feedback mechanisms
of a system.
A process contrary to the decomposition process is the composition
process, which is also necessary to system study. Although the
decomposition process is important, we should never only stay at
this stage. No matter whether it concerns the establishment of a
model the study of dynamic behaviors and trends of a system, the
final purpose of decomposition is composition. Only if every part
and every sector of a system are connected organically into an
integrated whole and the functions or behaviors of every part
conform to that of wholeness, can the model be made to represent
the entire effect and the entire structure of a system. Therefore,
in the establishment and the tests of a model, we must gradually
pass from the partial tests to the general tests, reform struc—
tures and behaviors of the system with the viewpoint of wholeness,
and inspect thoroughly and systematically the internal structures
and the feedback mechanisms of the system.
2) Various kinds of sectors of a system all have their natural
special laws. We should not press the laws of the system upon
the sectors. Neither should we mistake the characteristics of
sectors to be those of a system, nor based on narrow experiences
use characteristics of different sectors indiscriminately. The
pasie structure of a system is a positive or negative feedback
loop. Just because the positive feedback loops and the negative
feedback loops connect with each other, the structures of a system
are formed and the distinguishing features of the behaviors of a
system are produced.
3) To insure the adjustability of the entirety of a system, there
must exist one or several dominant parts which show the hierar-
chical orders. The dominant parts of a system are the core of a
system which will dominate the main structures and the functions
of the system. They decide the changes and the developing trends
of the system. Therefore, among the feedback loops within S.D.
systems, there exists one or more dominant loops. The characte-
ristics of these dominant loops and the interactions among them
largely determine the characteristics and behaviors of a system.
This is known as the principle of dominant parts function. The
dominant parts not only exist in stable systems, but also in the
evolution and the development of the system. In addition they
exist in the course of transition from a previous stable state
to a new one. Such a fundamental distinguishing feature, under
some conditions, can allow us to capture main factors, and hence
simplify the system model. It should be pointed out that the
emphasis on the principle of dominant parts does not mean the
abandonment of the functions of other parts in a system. The
dominant parts of a system can only be formed by the interactions
between parts. The behaviors of a system are not completely deter-
mined by the dominant parts only; they are the results of the
common actions of every part. In addition, under some space-time
conditions, the dominant parts and the non-dominant parts can
interchange. z
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986 445
Isomorphism of Systems
In the fieids of nature, human society, and human ideology, there
exist structural and functional similarities. These similarities
are the isomorphism of systems. Often several different structures
produce the same function, so that we can use one kind of equation
or law to describe different fields and different phenomena. This
is also the theoretical basis that we use in qualitative or
quantitative systems studies.
Isomorphism is based upon the similarities of structures and
functions instead of only similarity of functions. Therefore,
when we talk about the isomorphism of one system with another
system, it strictly means they are similar in structure and
function. When creating a model for a system, we can not simply
simulate the functions of the system. Rather should we construct
its structure which can truly represent the system. Such a model
is more objective and can more scientifically describe the real
world than the model which does not consider the structure
similarity.
For a long time, system dynamics was described in this way.
Its purpose was to link the laws and theories of the systems of
different fields. It was based only on the isomorphism of systems.
S.D. unifies the basic structures of systems by feedback loops
and simulates the different features of systems by functions of
SMOOTH, DELAY, SWITCH etc.. Thus, we build aS.D. model that can
represent the real system in both structure and function. When
building S.D. models of different systems, we find that there
exist many similarities among different systems in dynamic
behavior and internal feedback structures. For instance, birth
and death rates of populations and capital investment, disearding
rate of industry have the same interrelated positive and negative
feedback loops with a delay function in structure. Therefore,
from the view of their structures, they are isomorphic. So we
ean deduce that their dynamic modes may be exponential increase
or decrease or oscillation, but not the S shaped mode. As another
example, the exponential law is also called “natural increasing
law". It can represent the increase of capital or interest in
economics. In sociology, it is called the Malthus Law, whick
represents an unlimited increase in population. In scientiology,
it can represent the development of science and technology, and
so forth. We find that, in these different systems, their major
structures are always positive feedback loops. Through research
about the isomorphism in structure of these different systems,
it is helpful to raise up the analogous description of systems
to their logical homogeneous description. Then we can elevate
the special law of any system to a higher degree in order to
build a foundation for the general theory about structure and
function of systems. At the same time, it also provides a conve—
nient study of the specific structure and laws of a particular
system.
System dynamics has been widely used in many fields such as
socio-economic systems, population, energy, transportation, etc.,
ranging from enterprises to professions and from cities to the
whole country and the world. Various kinds of S.D. models have
appeared continuously. This provides a good base on which to
build a "standard model" with standard units. We should never
446 THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 15
build a model simply for building it, but should study its
special behaviors, laws and structures in order to make the
research on systems more and more standardized and generalized.
Viewpoint of Interrelation
Interrelation represents the general relationship between the
whole and the part; the part and the part; and the system and
the environment. We must think in terms of relationships
between element motion and information in mutual interaction.
S.D. studies the interrelations between elements within systems,
or in other word, the causality. Whether in socio-economic sys-
tems or in engineering systems, the phenomenon of mono-cause—
multi-effects, and even sometimes the phenomenon of the crossing
of causal links exist. With the expansion of our study range,
the past single causal relations are becoming more and more
unreasonable. Consequently, there arises the need to have a more
reasonable causality which can better represent real relation-
ships. S.D., as a discipline that studies general system dynamics
behaviors, also requires this.
With the development of knowledge of system structure, the know-
ledge of causality has undergone several stages ranging from
mechanical causality to statistical causality and now to feedback
causality. The basic structure of a system dynamics model is the
feedback structure; in other words, the system structure that
S.D. studies is dependent upon the causality of feedbacks, which
is one of the important stages in the study of systems. If we
further dissect the feedback loops of systems, we can find that
the causality among elements in the feedback loops are all in
mechanical and statistical forms. They are inthe forms of tradi-
tional statistical relations or traditional logical relations.
These relations are applicable only in the description of clear-
cut or predictable interrelations. With the increasing expansion
and depth ofthe study range of S.D., the phenomena and problems
which we are facing are becoming more and more complicated because
of the involvement of the human element, the multiplicity of
objectives, and the diversity of behaviors. These problems cause
the interrelationships among elements to be such more vague and
complicated. This makes the methods for describing current
systems somewhat outdated.
In an economic system, for instance, the capital allocation
among different econimic departments, the determination of the
proportion of accumulation and consumption, and the requirements
for education and technology development through economic growth
all include a process of thinking, selecting, identifying and
decision-making. This requires us to have a method which is able
to describe this kind of causality. In fact, the process of
thinking stated above is set up on the basis of causality of
feedback. Therefore, by means of S.D. theory and the combination
of other theories (e.g. the disciplines applied in the description
of vague causality), a method that can represent the vague causal
relations and describe the process of thinking and decision—
meking needs to be established. This will be helpful in expanding
the study range of S.D. and in increasing the validity of the
S.D. models
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986 447
CONNECTIONS AND DISPARITIES WITH OTHER DISCIPLINES
In recent years, with the endless increase of study research
topics, and the continuous emergence of transversal disciplines
and overlappinz disciplines, it is necessary for S.D., as a
systematical discipline, to study the characteristics and the
methods of other disciplines. It is necessary to study the
relationship of S.D. with these disciplines, and to draw some-
thing worthy for reference from them in order to complete and
improve the theories and the methods of S.D.
Synergetics and System Dynamics
At present, there is a new notable discipline in the realm of
systems study--synergetics, which is a discipline concerning a
systems’ self-regulation. A popular example of self-regulation
occurs, when the workers in a factory behave in a certain way
according to the instructions given by the president. We say
that this system has an organization and possesses the functions
of an organization. When the workers behave in an actively and
coordinated manner in accordance with some mutually tacit
regulations, the process is called self-organization. Generally
speaking, if a particular structure and functions are actively
formed on the basis of some rules without continuous external
instructions, it is termed as a phenomenon of self-organization.
According to self-organization theory, a system is able to evolve
from the state of disorder to a state of stable order and is
also able to move from a structure in good order through a state
of disorder to a structure in a new stable order. For a system
to change from disorder to order and keep the stable orderly
structure, it is necessary that the system exchange energy and
material with the external world. This is known as an “open
system". It is also required that the "open system" lie far away
from the equilibrium state and that non-linear relationships
between the elements exist within the system. Disorder means
that the independence of elements in a system is in the dominant
position. In this case, there is no significant evidence to show
the connected structure of the system entirely. When a system
is in a disordered state, a stable pattern and an unstable
pattern mode must exist. The stable pattern, which corresponds
+o the fast relaxing variable, restrains against stimulations
or disturbances. Synergetics believes that the influence of the
stable pattern in the course of changes of a system can be
neglected.
The unstable pattern, which corresponds to the slow relaxing
variable, however, dominates the process of the change of a system
and determines the macro-behaviors of the system. Synergetics
calls it the order parameter. The order parameter responds
strongly to the external influences and the internal fluctuations
of systems. Under certain conditions, it will bring the system
into a new stable structure.
At present, most of the systems that S.D. studies are non-linear
and open systems which are far away from the equilibrium point.
448 THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 19
In these systems, a stable structure is formed due to the
interactions between the elements and develops and evolves
according to certain laws. Consequently, S.D. models are
established on the basis of orderly stable systems; they do not
include the evolution of systems from disorder to order.
A stable orderly system, however, can possibly change into
an unstable system. Thus, the system becomes more chaotic and
disorderly, or even collapes. A nation, for instance, when
threstened by an energy crisis, may experience business going
bankrupt, increasing unemployment, and skyrocketing prices,
which make the economic system change from a stable orderly state
to an unstable disorderly state. This may in the end, cause the
economy to collapse entirely. Therefore, as a discipline that
studies the long-term dynamic behaviors and the phonomena of
systems, 5.D. should consider structural changes in systems,
and causes and factors which bring about great changes, especiall;
those factors which cause systems to change from orderly states
to disorderly and chaotic states.
In a S.D. model, one or more dominant loops or sensitive para—
meters exist. A negative feedback is always able to resist
external disturbances and internal fluctuations to a certain
degree, while a positive feedback loop or sensitive parameter
is easily affected by such disturbances and may even magnify
the effects of disturbances and fluctuations. This may lead a
system to disorder. Because of nonlinearity, a stable system,
that is, a system in which the negative feedback loops play the
dominant role o=” be changed into a unstable system in which the
positive feedback loops play the dominant role in some period.
In synergetics, fluctuations are considered to be the crucial
cause of systemic changes. In a system, many parameters which
possibly give rise to fluctuations must exist. Once a system
is in the critical state, fluctuations may result in great
Yesponses that effect the system. These may be only in a small
range at the beginning, but in the whole system at the end.
The responses will carry the system from order to disorder or
from an old structure to a new one. Accordingly, when analyzing
and studying a model, we should not only seek good plans and
policies but also devote work to those unfavorable factors that
may cause changes in a system. In other words, we are required
+o pay special attention to those positive feedbacks and
sensitive parameters, as well as fluctuating points, which may
bring about the great changes in a system. It is necessary to
study the degrees of their effects, to find out the critical
factors that result in the changes in a system, and to try
to find a method which is able to control and identify the
changes in a system.
In self-organization theory, the slow relaxing variable, playing
the dominant role in the activities of a system, is used as the
order parameter to describe the activities of the system. Order
parameters are applied to describe the kind and degree of order
of systems. The Liapunov function is employed to judge whether
a stable structure has formed or not, and catastrophe theory
is used to determine the forms of a system structure according
to the kinds of order parameters. Many S.D. scholars, however,
think that the principal factors determining the activities
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986 449
of a system are the dominant loops which represent the dominant
structure of a system. The dominant loops are frequently replaced
due to the variations of the nonlinear factors and the parameters
within the system and the influence of the exogenous variables.
Thus, S.D. scholars are now attempting to look for and evoluate
the dominant loops by means of eigenvalue analysis and the prin-
ciples of Liapunov. These are helpful in the study of the .
structure and the dynamic behaviors of a system, as well as for
the simplification of models.
Econometrics, Economic Control Theory and System Dynamics
In the realm of economic systems, several schools of thought
have developed which quantitatively study economic laws. Two
typical types are econometrics and economic control theory.
The former is a field that uses the tools and concepts of the
two disciplines--statistics and economics; the latter is a
discipline that introduces control theory into the realm of
economics. Both of them pay special attention to the accuracy
between the model and the statistical observed values of real
systems and establish a model on the basis of statistical methods
and differential equations. Most of the economic theories are
static rather than dynamic, and traditional mathematical tools
have great difficulty in analyzing and studying nonlinear rela-
tions. Therefore, it is rather difficult for them to represent
complex and nonlinear dynamic systems.
At present, control theory is only successfully applied in
linear systems. In nonlinear systems, its theory and application
meet with difficulties, which limit its applied range and fields.
Control Theory mainly studies the chdracteristics around the
equilibrium point or the operating point of a system, so that
it is difficult to use for long-term study. Econometrics and
Control Theory, however, have many skills with respect to the
definition of the precision of a model and the determination
of the parameters, and are suitable for short-term and precise
predictions.
In contrast to the above fields, 8.D. seems to focus more atten-
tion on the internal mechanisms and the structure of a system
and more emphasizes on the information feedbacks and the inter—
relations between elements. S.D. supposes that the data reflect
only one side of a system. It also supposes that in a socio—
economic system there are not only definable and precise relations,
but also vague and random factors. These relations and factors
are very difficult to solve by traditional mathematical methods.
CONCLUSIONS
Nowadays there is no doubt that system dynamics ia a branck of
Systems Science.
If the science systems in the past have been classified into
two major categories, social science and natural science, then
systems science, a frontier discipline which combines natural
science and social science, should be added to the current
science systems.
450 THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 198
Systems science studies mainly the types, the general patterns,
and the laws of activities of systems. It consists of two
aspects -- theoretical methods and practical engineering appli-
cations. Currently, among those which have been successfully
applied in engineering are management engineering, engineering
control theory, economic control theory, econometrics, etc.;
and in theory are general system theory, cybernetics, information
theory, operations research, etc.
Now, as a systematical discipline which studies the dynamic
behavior of general ystem theory, S.D. should become an important
branch in systems science. From the aspect of its development
and study range, it has gone far beyond mere applications. lts
theories and methods can not only be directly applied in sim—
lation and analysis of actual systems but also can provide the
study of the structures and the functions of general ystems with
a theoretical basis. S.D. therefore is not only a method of
analyzing systems by means of simulation techniques, but also a
systematical discipline which studies dynamic behaviors, internal
structure, and the functions of general systems.
S.D. has been developing for 30 years. During much of this time,
S.D. has mistakenly been viewed as only a means of stimulation
or an equivalent of DYNAMO by people being strangers to the
field. And its position in system science has not been fully
discussed and studied. With the development and popularization
of S.D., it becomes increasingly important field. Nowadays,
system dynamics has been developing as a bridge connecting
social science uJ natural science. Moreover, it is expected to
be a powerful tool for promoting the creation of a complete
hierarchy of system science.
REFERENCES
Bertalanffy, L.V., (1980). General System Theory. New York,
George Braziller.
Forrester, Jay W-, (1968). Principles of Systems. Cambridge, Ma:
MIT Press.
Haken, H., (1978). Synergetics ~- An Introduction. New York,
Springer-Verlag.
Qian, Xuesen, (1982). On Systems Engineering, Peking, China.
Wang, Qifan, (1984). Methodology and Development of System
Dynamics. Future and Development, Vol. 19, No.4, Peking, China.
Zhang, zongjun et al., (1982), Proceedings of Economic
Cybernatics. Sangxi, China.
