Introduction of System Dynamics fa Urban Pe
Development Planning
a
es
5
Une
. fhy Dat
Wei Hongsen (Tsinghua University, vel
Abstract
This article presenting a suamary and analisis of the theory and nethed of Systez
Dynamics on the author,s experience in planning of the coordinated _develonent of
science, Technologe, economy and social advancement in such cities as Ceijing, larbin,
Anshan and Baotou, is ained to find out the advantage disadvantage of the $.) approach
to urban planning as well as to improve upon it.
System Dynamics which is considered in this article to craw on system theory,
information science ard cise actics, ally the feedback control principle ard
computer simulaticn, iencitl and aethad that can find an effective
application to impro a aulti-factur, son-linear, dynamic and/or
complex system, and also a lab pcratory that can verify the social economic policy on
strate, actual planning of cov gindted development of the Chinese cities and
regi i a §) points while left five(5) respects to be
a the idea of iestabtisiting * the S.D. “dominated coupretensive ugdel system is
thus developed to erhaice oJ stronsthen the above planning process.
I. General wat Planning
Ss: dD. is gainly applied to
the role program and edi
should be observed fa practice.
1
The és: eablishnert of obs
eed | in the regional pl
points must be stressed ani
Through analysis of which
providJ. The following principles
Rose ae Eeeble
to account urgent problea
tae E Wes Geudkeary SkosEtaae
ilitata modeling these necessary’ important
ita
awalitive anal
7 study while the
and
anc
quantitive cat
latter makes 4 2 former 3b
muc OTS
conforma to actual condit tions. Tey are dialictically
we must be fully aware that suantita
and correct than
qualitative analysis in expusiig ical Fre
s is characterised
1299
‘could be the better in space and
1300 System Dynamics '90
by intuition and completation and fussy knowledge when applying is a aulti- factor
dynamic and complex system. If so,subjectivism will prevail.Cn the contray S.D approach
which features problem-oriented gives various hypotheses and simulate the real process
through modeling by means of computer-aided facility. This method can make complex
system clear and reflect real conditions, also can change policy or parameter through
decision-maker to computer communication, thus having the qualitative anaysis a
qualitative study yell connected and combined to effectively sul the actual problen.
free of those errors caused by subjectism, intuition or sensational change. .
But we should not seid, the function of qualitative anaysis on modeling in regional
planning. On establishing S.D acdle, we aust Tully attach impor tance to the method of
holding the law of things by cence Intution and object.
correctly establish the better mocel of system analysis.cn mo Js] ine, should at first
have a rough dim and un-quan tative logical model in brain.Secondly we should make this
as referece-point and dynatic salce think ing so that we can the problens much clearer
and formal quantitative-studyiig wel.
3. System structure ceteraine Sustei fuietion
system constucture is in mutual velatious and mode in space tine aims every ele
of systemsit is a interna! ercund that makes system be a entirety and have sowe
functionsit is a essential mask that makes a system different form other system.
faxtion is system 's order gnc, cbitity that system exchanges matter energy and
infromation fora envirowent. I Tt is eter fluc waufestation of system constructure.So systen
construction is the order of f matagh effect of elements in system, is interior description
of system. Function is the order of s: an) ings,aiid is system
eri or description. A structure has a Funct tea will have a special
function,we mustdesign and coi i produce the fuction.
To some degree,regional planning is to find a optinum conctructure, and to obtain the
best effect.That is to Say ,accsi te re ional conditica, in order to achieve better
effect, better function, we shoul io ist new relations between ciery element so that it
state and could find the better econinic
constructure. It is regi tnat draws up concrete propramme for the optimum
constrcture that can give full play to the best function.
Ob tecttus poaltty evaten To ate bacte that tects tecth on mst tiepaees
4. Ubjective icality system is sain basis that tests truth and effectiveness of
oh vestablishing adel it shows agp
being, ub Parameter constant awd di
system.we must not preocuppy sure
not have the nature.On collecting aiid coal ing
reliable data.On establishing iioJei, we sould test ae va systea behavior and
devolop and perfect it in practice to make it solve correctly the real behavior of
objective system.
“ss
system Dynamics ‘20 1801
Features system dynamics on regicnal. planning
Cn regional planning,we find that because of various reasons the history statistics
data is brought about in fragments and because it is on the course of unceasing treade
varies and of industrial constructure adjestion, It can not make statictics meet the
requirenent. It has the more difficult in ovtaining specimen or sample ard data of the
longerterm. especially in ferfecture's eas science: tech lodge, Social well
coordiatating problem for it involves all trades aid professions aid vaviouse factors
that are conpelex and different in dimention and have many difficulty in converting
thes. In these fleids we wusi deal with a large nusiwer’ Of un-liniar problems. if it is.
by ecomicis and Input-Output analysis, there are many diffcities that could not
be exceeded.On dealing with this kind of multi-f tor , nabinies non-linear complex
systen problems,systex dynamics as tc the above metho
_ 1. Supporting system model is not only cata. It is at first its structure system
interrior causality.
Dann 1
on the princizal cf system constructure deciding or function, the behavior of
real system is all defined by tracture of mutual f Reaction and mutual connection
agony system elements.By mea; this, “sé can coveniently obtain the most mainly
pehiiennt form complex study
r system,and make study appraxiate to objective
eality. while the traditional quantitative mode! is supported by data.Tn Haerbin Econimic
ings Node research, apart form data +o be insufficient, unfull and
incorrect,we also fing that ata lasing | pheiomencn is not exc luded, especial ly high-grade
system such as economic and co heir mortage largely limits appliying Scope
& the applying of
lying S.D,we can obtain
tablish model and find
3 Seow ¢ rey hat v
quantitative econimic and Li,
nessary information form ex
out the constructure that reflects in Teri al system. This can largely
reduce the requirement for cata and can cor is quantitative study result or eractice
conclution that is obtained by the comination OF qualitative analysis inte quantitative
study, we call, this feature as structi lve. It_ is very important to city and resicnal
ae at the pre eseuut statics aiid dota information level of my country for it can
argly save tine and cost.
he
2. By simula tion me method, to better solve non-linear groblems
The man who establishs model to solve actu b 3 d the actual system
is mostly non-linear, especialy gigantic com} Social Econince
Ecosystem. Because we have not Found out the more etfecti mathematics that
cati deal with these non-linear problems eto make those Giles change into
linear problems to be solved. The concessioii of allel result in corresponding concession
of truth accuracy reliabity. This concession cai inadequately keep to the simple systen.
But if we should still apply this simple method, it should bring about the more harnful
result to the solving of gigantic _systen such as Social Econimic Ecosystem.We often
cosider coordination develar ing problem of gigantic systen such as science and technology.
vu
Coonimic, Social and Surruuiiigs, ww we ate faced by real task of coordination
1302 System Dynamics '90
development of Ecomic,Energy and Surrondings.e do not make a detour of non-linear stage
entirtity dealing with the complex non-linear system. This has led to much troudle to
gantitative econimic and Input-Output Anaysis.Sy means of S.D, the actual system can be
defined as the combining systen of multistage feek-back loop. With the combining of
qualitative analysis into quantitative study,we can easily handle this kind of non-linear
problems. Instead of being solved by mathemzatics calculating but by simulation we can
sal tte kind of problems more appropriate handling as effective to solve non-linear
problems.
3. Intuition and image when solving problen
With S.D,the problem to be solved can be formular.Acording to feature of problen, its
elements can be respectily represented by information flow and matter flow and currency
flow,and can be consisted to a oraganic entirety by system interior mutual conneetion.
This can avoid shortage in the past by deating with the problem with isolating these
systems. Wiile by the other middle methods, it can solve the difficulty of_ the
transfomation of different elements when applying genearal methods. Campared to Input
Output analysis, it has a pecular adv : Based on this advantage, it could make the
research man listen to suggests ly,make logical model appoxitate to real system.
At the same time,when we have converted logical emi into the special purpose flow
model of S.D, the concept of the original large oxtension,elements is furture gistiact
and we have decided that every element is which kind of variable such as level variable
such as level variable, rate variable, assitant variable and so on. Up to now, the
relationship pattern between elements is further distinct.The limited factors of logical
model now could be considered and unnessary factors could be removed.When we have written
DYNAMO Lauguage procedure mou] based on flow-picture, this lauguage itself has the
function of looking for errior.so it can make us climinate every kind of error in form.
When the procedure model has Leen passed transtat ins and editing function we can most
intuitionly, imagely see how the model appoach to the property of the real system.
_ 4. Have policy-test and social-test and could give full play to the people's
subjective activity
Because system dynatics has adopted the methed of information and feed-back method, it
easily brings into desire valuce information of state avariable on establish model. the
different desire value has different result.This opens up the way to make science
experiment like seisnce on solving econimic and social problems.Campared to the input
of different policy to desired value.ve could adjust policy to achieve a certain aim.
The problems of being destrctive and risk that are inconvenlently tested can be tested
by system Dynatics model. We call this character as experiment. This is the most
important character campared to any other methods.New printed DYNANO soft ware (for
example PD-PLUS)also has the functions of two.paramenters of value-modifying and picture
-modifying and has provided as advantage to do lots of simutation tests and give play
to our subjective activity.For we can depend on the form of"if-then"by S.D method to
give full play to researcher's imagination to study the system behavior in different
states and can systematically attain different resutls to find out the better policy and
System Dynamics '90 : 13803
poliecy making prapramme and to forecast the result of the carrying of the propramnes.
5. S.D is deep and broad in analysis of middle- and long-term strategy
S.D is not only applied to solving the problems involved in science and technology,
economy and society ina macro way, e.g the general trend of macro-economy, the
proportional relation between accumulation and consumption relation between accumulation
and consumption, population, the changing trend of labour,but also employed to study the
production in an enterprise, inventory,sales in a micro way.
S.D is found to be much effective to analyse macro-economy. Science-technology and
society,espcially the general trend of the system behavior in the middle- or long-term
period. It can also be used for analysis of short-term behavior (3-5 years), middle-term
behavior (10-15 years) long-term behavior (more than 60 years).The longer the term is, the
more accurate the result is,e.g the Globe Strategy Study made by Dennis Meadows.
8. Per: rson-to-computer connection is easy for communication
fs $.D is a set of systematic nethods and equipped with a set of formative graphs,
specialised advanced language and other sof ftware, the real problem program is found to
well conform to the dialectiacally-meditated | law and very easy to use.This aethed also
featured intuition, free,of mat al formula or induct i e inference,it is easy to
conduct person-computer commun Ki i icating of
the researcher with the eiiduser,t because the wultural quality of the chinese users have
been well considered.This method is worth popularizing.
The limitations of S.D Method in Regional Planning
When practically applied,S.D is also found to be limited in some respects as other
methods will,so this approach is left to be inproved.Some limitations are given below.
1,Mathematical and Evonumic Theories are not provided in Regional Planning
If S.D is economy-science and technology- society-criented, mathematical or economic
theories must be provided to guide.But S.D is found to fail to give them. As we know,
theory varys according to different m i) there is sure to appear subjectivism.
Purther-nore, when study ing the co % of a nation, some economoic
policies must be involved.While a i “Sicence- technology, society model, the
nature of feed-back ring is difficult to deteralic. Ne must be aware that the special
asspect of a real fact should be well-analysed,otherwise there is no reliability.
2. lack of a set of scientific and all-round standared system
When considering the co-ordinated development of a nation in economic, science
~technoloay and society, there must be a set of scientifitic and all-round standard, by
now we have not worked cut such effcctive standerd system, and 8. D itself can't give
either.So the result obtained by S.Dlack of such assessment standard wiich means in this
two aspects! a) whether a region is in co-ordinated devel opaent in economy, science
-technology and soc correct. Thus, we have
seen that the me ordinated”, this is a
whether che measure given by 8.
ahi te
syStem established by $.D often lack”
rv)
1804 System Dynamics '90
vital weakness and difficulty.Under such conditi have to stay at the level of
“sbJective study"or"inasinary exper intent”.
te finns ve “e 5 is ui
ae ar
for $.D Is
ed to study the
and Lechnolugy “Society. I the
is destroyed
wnctioi Ly SS thee ecolluiiy"S Sciei
ii ig found to be wrong, thes i
fn two. ways? one is that differ cut
to D different model-maker, the other is that
on of parameter.
The subject
description of oie s
subjectivism may be involved in det
6 to the fact that the model-maker adopted different
Subjectivisn involved i i t : t
& problems.So we say sclentific feature still prevails
sci ntl ific method to diffe
the subjecti
4. gp tu ti-sta
368 feed-b
chars ‘acterised by multi-variable,
t,S.D approach by now have been
wlti-stage feed-back problem can't
TT foe
at te
the amount of profit turnec ty the state, this is because Une economy uit
dominantly a planned one, the sar ig sented accion is inalanced, the effect of le
is not tangable, and even sometimes Just on the contray, This means the investment
of "random"quality.So we think it may be feasible and suitable to well apply this S.D
approch in the esteri countries. while in a country with planned economy,such problems
can't be solved by considering the feed-back r
System Dynamics '90 13805
Total investment State investment ="
for production ——sself-col lected i? ft .
locality investment or local metal Tur
be a i industy |
i i
vA \ I
{ / \ ok te aye
H f ‘ Repay capability
: ~~
| Agriculture investment
\ Metallurgical product value
State expenditure Local fty, expenditure +
as
, a + Profit remained
| \ * in metallurgical
i \ Agriculture Industrial GNP industy
} \ any
State revenue :
\ Sa +
Profit turned to state Locality ievaiue Social GHP
5. S.D can't sive the optinal solution
Ona advantage of S.D is that it can simulate the soci
actually imagine the future development economy science-technslogy and society, ¢.8, if
we want to make _ simulation exper iment of the devel s Lio.
and society in 2000,we must before hand determine the tem structure in 2000,
nd decide on the input ints the system in 2000.To meet Uese must be some”pre-policy”
as a reference and some knewledge of structure parancter . These tow requirements mean
that S.D shall be applied in combination with other ¢ tatitative metho en though
me IO.
these two requirenents are met, S.D itself can't provide the optimal solution in the
sense cf theory.
6. S.D gives strategic guide instead cf specific advice
Regional f Planning is a general bekavior,S.D can only give strategic guide
advice.But macrocosmic behavior is never isolated.So in practice work,we sho
and policy
d not only
1306 System Dynamics ‘90
ra pee ee 1
the environmentas
oe dent me e a
precedent has besu set by jiow to study aid comect She asové three “facto ae We can
thee an
ed en at af
coinchide that $.0 theay is ab vai lance with practice.
1,
. ne tw meet
mien S.C is aves Th the analy
. 1
pulicy to aay 2
rar re
science- Cetin LObY on asset ) iv) She Lys
Stu, chev proportion in total nuxber
q
aid vesearch resuil dic Comere
TV sa
v7 the. pr ropor tion of the two elements is not
sive a proper Stancord _systen, if S0,a question Wi
raha T is S the, Tet Sea syotew?liuw ty make the Lest use of the local resoure:
we + v v +b. r4- ” tant.
We Shink that ap Red Pevuuticn ES Well G3 Gouindmic theory giice is importantly
Pai,
wuen S.D i and 2 developiis resources, it is found to be a weak
a
r te 7 :
wethod, but it cai, be cee ated by regisnal CDESS.
Die Gi
re ee , cmroary sete “
wie to ovlie Ue UhuvS Git wutend wabters in ecomomy Science and
D is ipo
tachi LoSa ain pciely Syobeu
c , xs t oe fat c n
Altiguah (3D iS effeuivia ope iicd ty fe ida Line? dyhaidic, iat ~ptage Peetu
it is found to be weak in sot
as
1GCN»
£ pe 18 a Bee es
ANG ViGCS GHG Wane thee ae Susie giane SySveyS.
sharp price-rise buying sprec,bank run railway accident, aud swial riot etu.ds ie
that there must exist some bi-coiditions for ch i.e nonclinear system, cnticg:
increase and fluctuations, etc.So $.D is ineffectual in determination of social entra
S.
Some complex phenominon may exist in economy-science and technology-socienty system,
but the complex reason may go beyond this system.Superficially, this problem is not
within the CDESS,in fact,we have to face a new challenge. Ve should see the problem in
a view of openness. In this circumstance,the result is only an approximation to the real
complex social giant system.
Some knowledge gained in comparison
1, Qualitative analysis must be combined with quatitative analysis
Regional planning is assoiated with the co-ordinated development of science and
technology,economy and society,and there are many sub-systems in the system concerned.
So there is no specified quatitative method that is available. Practically, we must
loy qualitative method which is the.basis of qualitative hethod to asertain the
actual condition.Because only to the bottom can the model reflect the system bshavior.
For the co-ordinated development of science-technology,ecnony and socirty, qualitative
analysis is considered as an very important way to accurately make the planning model
and to correctly refect the system features.fs is known,quatitative method is only a
mathematical approach in the abstract, when practically used,it has been simpified to
System Dynamics ‘90 . 1307
approximately reflect the object system so,this mettod is not sound or perfect.
As a matter of fact,qualtitative method can be a conplement.qualitative method which
is of intuition may be practised upon to make rough estimation and analysis. In the case
that there is much difficult to study by quatitative method, qualitative approach is
found to be effective.But quatitative method is systematic,comprehensive, accurate to
reflect the inter-relationship among all elements.And the modeling process can deeply
el the poteneial contradiction and system behavior to make people well understand
ie system.
As is stated above,we can conclude that quatitative and qualitative ie should
complement each othher.
IT. Select a proper model method according to the characters,conditions, necessasies
and goals of different regions
model method is a quantitative method applied more widely in regional planning today.
IT have analysed and compared several methods above,which have applying conditions and
advantages. We must select it according to the characters, conditions, necessaries and
goals. If we want to study the economic action in the near future, the data being
completed and accurate,we can selete a quantitative economic model.We have the knowledge
from studying Harbin economy-energy-enviroment mode].Harbin is a older industrial city.
Industry is of vital importance. Enterprises have been lead by goverment but industry.
Project study requires division of industry in Harbin according to the state standard.
But the obtained data is not fit to the state standard which have a great restriction
on selecting quantitative eccnmic model.System dynamic method, which has advantage of
representing system structure, is lower in accuracy of data and effectiveness of sample
term.So we select it.Practice demonstrates it avaible.We realing that most cities in
china,especialy heavy industry cities, are changing from goverment administration to
industry administration. It makes the requirement of production statistics chang greatly.
It is difficult to get statistic data conform to the state standard in long term. This
limits the application of Mathematic Programming and Quantitative Economy.So we select
system dynamic method in planning Anshan city and Baotou city.System dynamic must not
be selected in_any situation because it has some limitations. It is not accurate in short
term planning.It is no more accurate than othersin middle and long term planning.So we
can select it according to region, time,and condition.We think that we had better select
Input-Output Analysis to make planning of a province or a city.[t can help to collect
a lot of basic data which lay fondation of the planning in the future.Then,it is more
accurate and effective to use System Dynamic to study science and technology, eccnomy
and society coordinated development. Input-Output Analysis is widely applied in regional
planning in these aspects:
To analysis regional economy
2. To forecast and planning the jeoloment of Macroeconony
3. To simulate a complicated policy
1808 System Dynamics '90
III. Make the comprehensive mace! system to bring the function of every model method
inty full play in regional plan
Regioual e of actual issues. Different problems are
solved depeiids on the use of differeny quatitative and “juant itative method. Form the
content, it should include economic model, science and technology model,social developamnt
nodel , portation model, ecology model,enviroment quality model,education model, labour
force model, Livi waual ity model, etc. In a word, it is a social system engineering related
to the region. It is required to reveal the mechanism and rule of development of social
ant systen.én effectire cul y aud method has not been given. Different model nethod
is usec ts ) satisfy the reat ifferent regional planning in the practice. As
ove is used to make sure the economic proportion in
all wae of the region. antitative chic model is used to represent the
evelopment quantitative womic sectors.A optimal coutrol model
uscd to optinize some economic problens.A system dynamic model i is used to solve middle
long term development tendency and to do the test of science and technology, economy
socisty coggeling development. & systen evaluation mcdel is used to evaluate
tion, ete. In ordei to overcome the short coming of different method, we should use
he advantages of all methods into full play. So we
Sudy the mutual relatiowanong Scicie and technolvay, economy and society and
nt their internal cope: sna use these kinds of msthods to solve
three kinds of _problens i in Har “bin economy, enersy and civironent coor dinates deve lopiaent.
have thr ree
rment
ne
t
Q
a)
OLD MD rece
nae
ge
o
Ap
2
can
vi mole of ‘economic system,
lve the problem of eneray,
usé almesphe
We do the first attempt to link the system dyna
Harbin economy-energy-environent coordinated dove lopent.P rofessra! DYN Plus gives
the necessasy technical means to comlete linking, The sketch for model linking
indicated as below:
System Dynamics '90 1309
ee | cee
i economic development j energy
tT SD model ~ forecasting | data of energy
i
“deve lopment * plan fea renent
oars
WeStRENS Padi
fe oi tine mn
f Greimigauicn moac.
to obtain the function of stable state au
can, be fourd through the operation cf model.
2. we can raise efficiency of research by tal
calculating function of computer.
tee
ix
3. Learning form other model's advantage to make up for a deficit of system dynamic
in dealing with some problens,we can make the whole research be more scientific and
accurate.
Meny scholars in China have done a lot in this field to give. many good experiences
which are worth spreading.Typical examples are three mode! conbination which are belong
to the comprehensive integrated models, i.e.Zhuzhou mode! combination made by Yang Yuji
and Xiang Yuanwang,Xinjiang model combination made by ‘Wu Jianzhong, Hua Luogen model
i
combination made by Li Baojian. The radical difference among the three model is the
different: mode! selected by leading model .Zhuznou model make quantitive economic model
as its leading model and indudes SD model as its system simulate model Xinjiang model
makes SD model as its leading model.tua Lucgcng model makes itself as leading model.
This model system should be gra
to suit. the need of regional Planning research.Th
i. Model base .
Nodel base is established with the input-output model, quantitative economic model,
system dynamic nodel economic contre] model,optinizotion control model as its main past
to make all models’ software lunguage chaige each other. The model base also should
1310 System Dynamics ‘80
. aal=] tn masance wnd-)
include Some OSS model, dats process model,
rae Por {Cee Le
decision model Gault re
2. Stratesy base
fs to a region. it should make strategy and plan
regional strategy base Sigite be formed afte:
strategy is made. Tt is a dynamic devel
on a See.
HAUS Sr tes ane SECCorS. a
~ ae
oN semen
C GCC ORR
the regiunal comprehensi
3. Data base
we Should establist' a data base which can be used &: by all kinds of models. It indudes
i aiid present in the fields of ;, science and technology, society,
resources, geograrhy.
e Includes the newest economy, science
Kind of region.
S
ai Guede Las Ree Fepoees x, Sexe
Guatyois base Toi importait Stem
cee e 0 wok 5
he bdcuitow ita k every pens Of the important items to
iis ° ut Yert 7
devetoged be form a detwan WIth duléiple levels.
ae go Plilices ls yop
ive OF these Loses wn We
eats Goa Sr 20 cee Vaaues vn cate ute
in Pesitn, Tt should be a ieeosar J vv Compieca uy Coicurauicin
L.
and feasible to do that.
CC BY-NC-SA 4.0