PTRC Summer Annual Meeting, 8-12 July 1974, university of warwick
§ETwgRK pESTGN AS AN JTERATIVE PBOCESS OF URBAN SIMULA?ION .A,ND EVALUATTON:
SOME APPLICATTONS OF THE POLIS SIMULATTON MODEL
Dipl. -Ing. !{ichael- Wegener
Battelle-Institut e.V. Frankfurt am Main
T. INTRODUCTION
More than twenty years of research and development have made urban transportation planning a mature discipline. There
exists a well-establ-ished eet of theory about trip genera- tion, traffic flows, choice of mode, and route selection. The methodology to apply thie theory to the solution of practLcal probl-ems is highly developed. There are numerous sophistica- ted techniques avail"abl-e to analyze, eLmuJ-ate, or forecast traffLe patterns or travel behavior. Nowhere in the planning profeseions are electronic eomputers used so extensively as in transportation planning models. Moreover, these models are indeed used for planning: ?hey are general-ly accepted as an indiepensable part of standard profesaional praetice, and their application Ls a prerequisite for major investment decisions.
However, the general acceptance of large computer nodels in urban transportation planning tends to conceal the fact that thLs advanced and expensive technique suffers from two grave inadequacieg:
(a) Even elaborate models are o11o-ehot traffic forecagtg baeed on usually shallorr land use projectians; they do not al"low for dynamic feedback between transpor- tation and the rest of the urban system.
(U) ComparLaons between network alternatives are ueually baeed on direct costs and performance of the network;
no comprehensive aseessment of social" costs, external effects is made.
In thig paper a pragmatic approach ie presented rhich, ln one integrated, computerized planning instnrment, combinea
dynamic simulation of spatial urban development, advanced tranaportation modelling techniquesr - and multi-attributed evaluation methodologY.
Network Design and Urban Simul-ation Wegener
The approach is based on POLIST Bil urban simulation model developed by Battelle-Frankfurt. POLIS is a multi-period, multi-region, dynamic, digital simul-atLon model of spatial- urban development. The model simulates the spatial distribu- tion of population, employment, construetion, and land use as it evolves in response to various stimuli (planning deci- sions and unplanned ?tmarketI developments) accruing over time. The spatial dimension of the urban system is represen- tecl in POLIS by a transportation sub-model eontaining the full range of elements of advanced transportatLon modelling.
The results of the urban simulafion, including transporta- tionr häy at any point in time be submitted to a formalized eval-uation procedure based on mu}ti-dimensional scaling of ut iJ. ity.
2. THE POLIS §IMULATION MODEL
The POLIS modeL is the first comprehensive simulation model
specificalJ-y designed for urban development planning in J.arge
cities of the Federal Republic of Germany. It is also the first such model practically tested with data of three cities
(Cologne, Vienna, Darmstadt).
2.1 ModeI Structure
The POLIS model uses elements of earlier models developed mainl-y in the US and adapts them to West German conditions.
The transportation sub-model, e.go, follows the classical-
scheme of.trip generation, trip distribution, modal eplit and trip assignment, however it takes into account specific traf- fic cond.itions in German cities by an extensive publie trans- it sector. In the developers I market sub-model the typical Lowry approach (f) which distributes housing as a function of the locations of basic employment has been replaced by a sequence of incremental al-location algorithms control-led Fy multi-dimeneional attractivity measures. Algo, the model is not a Forrester model (g), although it recognizes the ba-
sic dynamic feedback structure introduced into urban model-- ling by |tUrban Dynamics[.
In addition, the model contains some features which have not been present in most earl-ier models:
- POLIS allows to control spatial- development by zoning and land use regulatiorls.
- POLIS contains an extensive policy section that allows the ueer to introduce various kinds of time-sequenced and spatiall-y disaggregated action programs.
- POLIS also incorporates and exhibits eide effects of major physical changes.
-2
Network Design and Urban Si$ulation lfegener - POLIS has been designed for use in an interactive coß-
puter environment.
POLIS (:) is a balanced representatLon of major aepects of spatial urban development (fig. t). The urban area is spa- tially disaggregated into geographical sub-units (zonee) for which state variables representing stocks or activities are
collected (4). The zones are connected to each other and to the surrounding region by transportation networks (public transit, highway), which are coded by links.
Fig. 1
Aspectsof
urban developmentStarting
fromthe state of the
urban systemin the
base year'the
modeleimulates the
deveJ.opmentof the spatial distribu- tion of popuLation,
employment,buildingc
andland
use aswell as of transportation as it
evolvesover several
timeperiods until a planning
horLzon has beenarrived at (fig.
2)I{e.qener
Network Desi.en and Urban Simulation
Fig. 2 The POLIS simulation model
r
Iti
II.$r _l\
\§i tlr
$i\
5lI I I
-l F
gl3l
it
ü8r
it
B!
/nramütr
unaqe,S lnput
parärrct<r cha€es
fttle aae ti*,ol tones a}}reviatinns
qs
q pmgram§asunptions unsalenq €rrors
,ä- -/
lr
lr lr
ll
lr lt
I I I I I
{
I I I I I I I I I I I
)'
\..
ur\
z,z
pefEneter\
\/ data
lnput
patarretcr, data
/
\-/ Consistcrrov
A
of asstfiflUrß
;;;',-
anaMsis
l-- -
,. Cror-r{t
)
,it---___'_l]
lnttlaliw .f__z
I
pooulatisr,ünol'I H*f,1'?ä.1,,,,.,
I
tana uieI {ruo.airtr.of travet ti.
I
ün* tpadsI
o***try
I
I
attrocttut!I
I
maPsr
Iransporüation (/)
f,
AccestDüti'tY l-
l-
c.Attracti,vüty v)
H ISTAT HISIAI
\--/
Evaluatton
Restart
Constructlon Dernarü
\1
o
JJ
l
7__:r
atilicorurans ä serv'rirlec'
lnteruentions Constrrxtion atd
hM uJe
/i:\
Poblic Prognms Priuatc
@nstrucüton
F.f",gltfrt#a
lnteruontions
\ :)\J
r___z
summaru larü wC, tr€nsportatton cssh florus
hnettt *counü Sumrnary
CostS
M
üi/
Deneftts
-4
Network_Desi.cn and Urban Simulation Itegener The simulation of a period begins, with the analysisr de-
scription, and documentation of the state of the urban system.
The analysis starts with the sLmulation of traffic flows of the base year. Travel times computed in the traffic model are used to calculate accessibility indices of aII zoneB which are a measure of locational advantage with respect to various activitLes and infrastructure facilities of the urban area and the transportation system available. From accessibilities and other zonal attributes for each zone attractivity indices are computed which serve to express the market demand for land by various urban activities.
Next the allocatlon part of the model begins. First public ac- tion programs are executed. ?he model allows the introduction of time-seguenced and spatialJ-y disaggregated programs in the fields of housing construction, industrial development, educa- tional, social, recreational, and transport infrastructure.
Simultaneously with all construction programs necessary locaI roads and parkJ.ng facilities, with housing programs also
service facilities like kindergartens, elementary schools, neighborhood shopping ancl recreation areas are provided. The remaining construction activity is distributed over the urban area foJ-Iowing the market pattern of supply and demand by
private deve}opers within the restrictions indicated by the zoning plan. The likely distribution of private eonstruction for each type of activity is estimate«l as a function of attractivity and the available land in each zone. Displace- ment of one actLvity by more profitable ones is effected Ln
the model by demolition or by change of uge of buildings.
After simul-ation of. private construction, population and em- ployment projections are distributed across the available housing, commercial" and industrial building stock, including the updating of the respective demographJ-c, social, and €rt-
ployment distribution. Finall-y the availability of local- service facilities is checked against relevant standards.
Where service is severely sub-standard, the city administra- tion intervenee wlth a crash program.
This cLoses the simuLation of the period. The state variabl-es of the model have received new vaf,ue§. The model starts, with
nevr parameters and assumptionsr the simulation of the next period. This cycle is reiterated until the planning horizon has been reached, i.e. the last period has been simulated.
For each simulated alternative the model gives spatial-Iy dis- aggregated information about the deveLopment of population, employment, physical structure, transportation, and the en- vironment. In addition, the costs of each alternative are accumulated and exhibited as cash fLows between various groups of the city.
Network Desi.qn and Urban Simulation hre.qsner
2.2 The Transportation Su!-Model
Let us now take a closer look at the transportation sector of the POLIS model. The transportation sub-model is proces- sed once at the beginning and end of each time period. It contains five distinct parts which are connected by feedback loops and operated in a cycle fashion: trip generationn net- work analysis, trip distribution, modal sp1it, trip assign- ment (rig. 3).
Two different trip .qeneration mqdels are used for trips with- in the urban area (internal zones) and for trips to or from the surroundi::g region (external zones). Trip production of internal zones is estimated as a function of various zonal activities for six trip purposes:
- work trips, - shopping trips, - recreation trips, - education trips, - business trips, - social trips.
Trips between the urban area and the surrounding region are estimated with a combined trip generation and distribution model.
Next for each interchange on either network (public transit, highway) minimum-path trees and origin-to-destination travel tj-mes are calculated. Transit travel times include walking and waiting at origin and destination as well as transfer waiting times estimated as a function of train or bus fre-
quency. A transfer penalty is added to account for transfer inconvenience. Car travel times include walking and feeder driving, congestion and parking times. Congestion times are a result of the capacity restraint applied to the assignment procedure ( see below) ; parking times are estimated as a func- tion of parking supply and demand in the destination zon@o Network analysis, trip distribution, modal sp1it, and trip assignment are executed three times in each traffic simula- tion. In each iteration only a portion of total traffic is cumulatively distributed to account for different loading leveIs of the highway network.
The trip distribution model is a bidirectional gravity model
which distributes trips originating from each zone according to attracting activities of other zones and exponentially weighted car travel times.
The resulting bidirectional interchange matrix then under- goes a trip-interchange modal split with the travel time ratio as the explaining variable.
-6
Network Desi.qn and Urban Silrrr"lation
Fig. I The transportation sub-mode1
We.qeneI
NET]^J0RK DAT#
ZONAL DATA
?o?ulati.on 'kiborforce enpIoyment
Servtce$
lqteau,on"
fiidents
.lchools
lLnrtvpe UnrllrJ}vtn lrnlctnueL tüne
traruit Unel
Une {re1vmcy
hteno naL tips,ltftLlaL tones ixtnaeonae f,n'ps, intcmaL ?onar
truaL tineJ
rni.nimun paths tove( tlmes
minirnun Psths
TEIP OI5IRIDUTION
nt*rzowL ttrps, t ntemat- e,xterneL ( trnptied {rip Seneretcon) intzrzpnat trips, 1p1a6oc- Ut€m§t
MOPAL SPLIT
IR,IP ASSI6NI.{ENT
Unk Loaas
frpqvenq distribtrtion of trovet ttmes
'
nt.4irlpl rnileeges/ täueL trnes coJts€nuiconmentat effects
Network Deeign and Urban Simulation hre.qener
In the assignment step traffic flows of each interchange are loaded to the respective network links. Public transit flows are always assigned to minimum-path routes (all or nothing).
On the highway network only fl-rst iteration flows are loaded to optimal routes, in succeeding iterations congestion times on overloaded links cause selection of sub-optimaI, less con- gested routes.
By merging of link data and link 1oads, performance measures
of both networks, like capacity utilj-zation, passenger or vehicle miles, travel times by mode or type of link, fre- quency distributions of travel times, walking, waiting, and
congestion times are calculated.
2.3. Feedbacks between Transportation and Land Use
As the transportation sub-model is but a part of the compre-
hensive urban simulation, feedbacks between the transporta- tion sector and other sectors of urban development can effec- tively be modelled.
The rel-ationship between land use and transportation is well known and constitutes one of the basic concepts of transpor-
tation planning: based on forecasts of zon.at.. Land use future traffic flows between zones are estimated. Subsequently, the transportation system is designed so as to serve this demand.
By far less investigated are the effects of the transport system, i.e. the transport supplyr orr land ll 8€r These very effects, however, are essential forces in shaping spatial urban development. In POLIS the accessibility variables are taken to account for these effects. If transport services of a zorae are improved, its attractiveness for development
increases. As a consequence, traffic flows to and from this zor:e will rise, the transportation system has to be expanded,
and so on .. .
This spiraling path of rrurban developmentrr is halted, when
technological possibilities, resources or the willingness to utilize them are exhausted. Many large cities al-ready have
reached this limit: Traffic flows exceed the capacity of vital traffic arteries I zotres with heaviest traffic loads experience marked reductions in accessibitity, even when in- vestments in the transport system are inereased. At the same
time, these investments consummate more and more of the scarce urban land (fig. 4).
The reproduction of this cycle in a computer model, however, poses no small technical problems. It requires that the com-
plete set of transportation model routines be executed not only oncer &s it would be with one-shot traffic forecasts, but once for each time increment of the simulation. As these
-B
Network Design and Urban Simulation We.eeneI routines are among the most time-consuming types of computer programs, the use of fast minimum-path and assignment qlgo- rithms is a prerequisite for this modelling techniquel).
:$
Elc,Fig. 4 Transportation and Land use
3. THE POLIS EVALUATION MODEL
The results of the urban simulation, incJ-ucling transporta- tion, may at any point in time be submitted to a formalized evaluation procedure which has recently been added to the simulation model. The procedure used is based on the Multi- attributive Utility Theory (MAUT), a decision aiding tech- nique relying on muLti-dimensional scaling of utility (6),
MAUT proceeds by decomposing a complex object of evaLuation (a plan) into its independent dimensions (attributes) by way
of a goal hierarchy. The attributes are i-ndividually evalua- ted by means of utility functionsl weighted, and aggregated by a formal additive composition model. On each level of the hierarchy the utility of the plan with respect to specific aspects, on its top level its total utility becomes apparent.
Thus it is possible to view attributes of one problem area within a larger framework, i.€. by relating them to attri- butes of other problem areas as well as to higher l-evel more general- goals.
1) .A Moore algorithm been developed for workg (5).
modified by l.ist storing techniques has minimum-path tree building in both net-
AüE$l6lilr
LAND U§E SYSTEH
TßAUS- POßTANOI\J
SYSIEI4
IßAFFIC
FLOhJS
Network Desi.en and Urban Simul-ation We.qener
Differences between the values structures of different groups invorved in the planning process are expressed in this model
by different weights and utility functions. rf not only one,
but several goal or preference stnrctures are used, it is possible not only to compare different plans n but also to show differences in the eval-uätion of the plans by clifferent groups. Thus, potentiar conflicts that may arise from a plan may become apparent (2, B).
For use with the Polrs simul-ation model a goal hierarchy has
been adopted the elements of which are implied by the aspects of urban deveropment contained in the pol,rs model (rig. 5).
The top level goal of the hierarchy is trthe cityrras it
changes during a simulation. The el-ements on the lowest level of the hierarchy are attributes, i.e. quantitative properties or indicators for intangibJ-e properti.es of the evaluation ob- iect; in this case they are data about the state of the plan- ning object rrcity" and its zones as provided by the simula- tion model.
The evaluation model receives these data from the simulation model and evaluates them by using one ore more goal struc-
tures. The goar or varue structures represent attitudes or interests of various groups of the community and differ in weights and utility functions. For each group the model €x- hibits utirity values for all leve1s of the hierarchy and for alr zones or any aggregate of them. Also, differences bet- ween the evaluations by the groupsl i.€. potential conflicts are shown.
4. DESIGN AS AN ITERATIVE LtrARNING PROCESS
This combination of simulation and evaruation may be used as
a generar toor for supporting complex solution finding or design processes. First, a preliminary plan, or a set of alternative plans, is evaluated. The resurts of the evalua- tion suggest how the process may be continued. If all criteria are satisfied for all participants by a plan, it can be selec- ted for implementation. More frequentry, however, the plans will be not acceptable for one or more of the participating groups. rn this situation the process may be continued in three different ways:
(a) The planner proposes a changed plan which either con- tains new elements or modifies existing elements in the direction of a compromise.
(U) The participants agree to change their assumptions about future developments, i.e. they modify the simu- lation mode1.
-10
Nelwork Desi.e}r and Urban Simulation lf e.rr ene r.
IRAUTL TIMT/ITIP (MIIt) TAIIiiG IIi{/InIP (ilI)
PAR|(Iil6 siARcN il[t/IRIP
ITALl(UAY/IRIP (l(I',t)
PtRCil{IS StAIS A'/AItAEtt PRrvAcY(vttl./1oo PASs. ) sArrrY(iAI.Acc./4ro.PoP/Y )
ilEI POP. OTilSIIY (RTs/t|A) I ilousIR.JoBs/cAPI IA AcctssrB. ro JoBs (ritil) AcctssrB. ro REIATL (ätn)
BLUt cotuR/uflIIt c. RAIIo PtRC. AtItil RiSI0ürTS PoLLUTroil co(c/sotilYEAR ) TRAFilC ilotsE rdB(A))
,rr,ro-{
ntIGltB0RH000r stRv Icts
ll0t{s
-
IIt IGHEORtlOOD
Itli UlSAlr SYSIEI,I
ACIIliIII TS HiIHIT PLACTS
RtsI0tillrAr.
ACiIl/IIITS
EC0il0lilc ACI I', II I ES
PUELIC StRVICtS
ilil6H801ll000_
RtcRtAi Ioil
AGRICULIURT
-
It{0|,lsIRY-
EUSIiltSS_
sirvlcts
TACCISSIB. I0 P0P. (r,flil)
I
I--ACCISSIB. TO RTIAIL (HIil)
I
I nt lll L AcGtor,t. (titPL0yf,ttfl I )
I
l--ruoon lnu Rruo
RtGIonAL I
sloeltlc --l- pARxIrG suppLy/Du,t. RAII0
I
l-tr*c.*ooos or IoI. ARtA
Loc^L I
stt0ppIlG l- nfi.tr,rpLoyu.otils.(JoB/tlA)
I
L- RtIÄI L EMPLoYM./cAPI TA
*rr^rr{
to|JcAIIofi _
SOCIAL SERV. ,
I|EALIH CARE
RtcRtAiI0n
-
FiS. 5 The POLIS hierarchy (excerpt)
Network Desi.cn and Urban Simulation hle.gener
( c ) At least one of the participating groups agrees to change the weights of its goals or its satisfaction standards, i.e. it modifies the evaluation mode1.
If these three possible responses are seen within the frame- work of the planning process, the following five steps may be identified (fig. 6):
Fig.6 Simulation and evaluation and the planning process
Step
1:Step 2:
Step J:
Step 4z
Participants of the planning process define goals to be achieved by planning.
The planner is guided by these goals in for- mulating one or more plans in the process of design.
The consequences of the plans are predicted by the simulation model.
The consequences of the pl-ans are checked
against the.predetermined goals in the eva- luation model.
SOCIETAL GOALS
SI MU LATION
EVALUATION
I M PLEMENTATION
-12
Network Desian ancl Urban Simulation I{egener Step Jz A plan is adopted, if the goals of all Sroups
are satisfied. ff no such plan is found, the process is continued with one of the steps (a),
(U), or (c), until an acceptabl-e plan is found.
The motivation for these modifications comes from the growing information about the planning problem, the solution alterna- tives and their consequencesn and about the potential con- flicts arising from thern. This makes the solution fincling pro- cess an individual or collective learning process, in which through iterative application of. simulation and evaluation a plan that is acceptable to all participants is approached.
5. APPLICATION IN TRANSPORTATION PLANNING
From the preceding paragraphs it has become clear that the proposed planning instrument is not a specific tool exclu- sively for transportation planning. However, the urban trans- portation planner may greatly benefit from it in as much as he views his activity in the context of comprehensive urban planning.
In its simplest application he may use the instrument to check the consequences of network clesign alternatives. For instance, he may introduce different highway configurations' changes of service levels in the public transit systemr new' public transit lines, or entirely new modes of transit. He may experiment with timing, sequence or financing of trans- portation programs. The model will give him information about the 1ikely develoPment of
- construction, maintenance, operatirg, and user costs, - network utLlization, inclu<ling link Ioads, frequency
distributions of travel times, walking, waiting' con- gestion, times, number of trips, mileages by mode or link type;
- comfort, €.g. car occupancy, percent seated;
safety, like nurnber of fatal accidentsl injuries, or property damages;
- environmental effectsr äs air pollution by CO and NO*
or traffic noise I
- aesthetic effects, e.g. space requirements for new
right-of-waysr or intrusion by traffic arteries.
In addition he may observe the likeJ-y consequences of his design alternative on the urban system at large. He may find out how following network or service improvements accessi- bilities locall-y change, but also how pollution and noise levels go up
"on..rr"entfy. He may look into the effects these changes have on the development of land prices and land de-
and the resulting shifts in the spatial distribution
Network Design and Urban Simulation lte.qener
of construction activity. He may watch how dispracement pro- cesses slowly change the land use, socj-al and age structure of certain areas. He may be interested to know if minority groups are affected by these changes, and whether their con- cerns are adequately accounted for.
rn a second step the transportation planner would join with planners of other planning departments or agencies, e.§o the land use planner, the school planner, the recreational plan- ner etc., to discuss design arternatives. Now each partici- pant contributes the opinions, i.deas, and constraints of his agency or discipline. In this way it is possible to combine
different transportation schemes with various concepts of land use, housing, industrial development, social, educa- tionalr or recreational planning. The model would show where
discrepancies between the concepts exist, where badly served areas or major diseconomies would result. This information may then be used to jointly search for more compatible p1ans, and eventually may lead to improved inter-agency coordina- tion.
In a third step of application also the evaluation model would be applied. Three levels of application are possible. on the first level only one goal stmcture is introduceci, €.,g. urban development goals as formulated by the municipal legislature.
rn this caser ähy future state of any plan alternative may be
checked against that goal structure. rf only one such state is evaJ-uated, the model shows spatial disparities in the dis- tribution of pubric services and other indicators of quality of lj-fe. rf more than one plan is evaluated, comparisons bet- ween plans may be made. On the second leve1 more than one
goal system is used. This allows not only to compare between
plans, but also to compare between different attitudes of different groups towards one single plan in any desired spa- tial, temporal, or sectoral cletail. In addition, it is pos- sible, to analyze the differences between group additudes and thus identify potential conflict zones or problem areas.
On a third leve1 groups or representatives of groups would themselves actively play a part in a participatory solution finding process. This would imply the ful"I implementation of the learning type planning process indicated above. The metho- dological, didactic and politi-ca1 problems connected with
such application are at present difficult to assess,
6.
S.AMPLDS oF IrroRKThe combination of simulation and evaluation has, in differ- ent stages of implementation, been applied to land use and
transportation planning problems of the cities of Cologne, Vienna, and Darmstadt. In Cologne and Vienna, only the simu- lation model has been used. For the testing of both models, Darmstadt (population 15o.ooo) was selectecl as an rexperi-
-14
Network Desi.qn and Urban Simulation Wegener
mentalrr city because of its closeness to Frankfurt, its
manageable size, and the availability of data. The Darmstaclt data, having been assemble<I and coded in the way required by
the POLIS mode1, haveserved as the experimental setting for a
series of workshops held during the last year with groups of different sj-ze and professional background. To give an im- pression of how the proposed planning instrument works, selec- ted results of these workshops are presented.
As a common basis for a-r-l- workshops a number of basicly
different alternatives for land use and transportation plan- ning of Darmstadt were formulatecl together with staff members
of the Darmstadt city planning department (F.ig . ?\ z 1O/A New concentration of population and employment
in the northern part of the urban area, combined
with maximum investments for highway construc- tion; no improvement of public transportation.
20/B Incremental housing clevelopment added to olct
village cores i balanced transporation concept with moderate improvement of highway system and
transit service.
3O/C ttAnti-sprawltf concept with high density cor- ridor across the central city district I environ- ment-conscious transport scheme with cutbacks at highway construction; new lj-near (cabin type) transportation system along inner city corridor'
A fourth alternative is the hypothetical ttzerott or do-nothing alternative which assumes no public pJ-anning actions whatso- ever and represents pure market behavLor. These four planning alternatives were prepared for input into the simulation model and simulated in advance (fig. B).
The duration of the workshops was betweenthree and five days.
At the beginning the participants were asked to eval"uate the results of the four simulated alternatives. To ensure dif- ferent initial evaluations, the participants were divided into groups and asked to evaluate according to their group- specific interests. To facilitate group identification three typical representatives of social groups (trigtr, medium, low income) were sketched out in the form of written selfpor-
traits. Also in the workshop material, suggestions werde made
for the selection of weights and utility functions by each group in a first cycle of eval-uation (fig. 9).
Itle.qener
Ne twor']< De s i.eE__eng_ tIr!_An S imula t ion
+) .'-lb0
ti
..oU]\
ooIcn
+)tü^
dt{trO
+io+
Fl 0)
do
b0Ela
io
HN rldQ^+ + (+{
1rO6rl og<
a dori
f-
a b0
fr
,
g-"'--o s
n=
€<s
kl \-
§t
-.-Qo
-D-+I
I
rJ
_toER
= -.1
<\_
t
ts§oiaa
-§\o-S slL
.+3 3s\\
§äl
§P\\o
ö .E€:
s§
O
CO
O
m O e!
7,ä
*,-
/-4+,Eoa,
=<t
Eie
öU:O 6 E'\
'o-E
t\do
.E§af
INE §
y> z.
EÖ :E i.) oä <a s=-t6
E*b 6-3 5
=ö2 LF E
Jooo*.2- öutt i I
=z.il
*4 ä
Ie;$sz€
6 ;ö-9 ? id
5 >= >a rfl=
s35=83
ü_§
q8
dd ):
ß
.E 'Eodo or
=
o
o
I-
e>o
Eä§ a;-
Lc§
§BS z
-16
Network Desi.qn and Urban Sj-mulation Wggene,r
ot CiE
3
NV
L.t'
\ry/'
,
1
-9.
ooc, l.\
\o
sG{oo
ooOe
oo
q uo
1aq-
ooooO\O(3O
\.O ö lJ\
X\F
ooooo-fO(>
-t öoN?
ooooo§Joo(\t o Lr\
oooo
1970 1975 1980 1990 1995 1970 19?' 1980 1985 1 990 1995
1 970 19?5 1980 198' 1990
199'
1970 19951 transit
trips
L02 car
trips
DJ travel time/transit trip
(min)
Sf4 travel time/car trip (min)
5 travel time/transit CBD trlp
(nin)
l(G6 congestion time/car trip (min)
1975 1980 1985 1990
land required (ha) oemolition (ha) single family home construction kindergarten corr struction (places)
Fig. B Simulation results of
two Darmstadt plans:At left ilzerorr_alternative (NV), at right
30/c
L-'
3
5 4
1_
6
A
I
30 /c
I,
-rl-I
,
t
T(
,'i\\
\
\o
\
;/
N
\s
\ a\\l
\
\
Weg,ener
Network Design and Urban Simulation
.rl da
++.d
"(
.r{ Hb0.r{
;lv!
€
'Fl ü
BF-o )','tr
t+oo t{Hd5
L h.l
'.{ q) h
t+.r{
(/lFl H.dJ+
o5Ao)lt
§6+)
E
oA
u)z
ti
oB Ilq A
c!ro
§
E+Oqr q{ LoF{
+oo
A+'ti A- obDO'r{
a!Xq)
o\
a
..{bD
tu
(o§\
N
LO
\$
r\
(\LO
Nc\
NLO
§-E\
c!rr)
!oc\
c{Ir)
(Y)c\
Nt.r) a
c\N
tJ)
RR RR RR
8-
f, =
\
lo dt cl31 2t
EICICINN
.:l -t s
\
RR
I ol . o
I !l > o
| -l >t
l-lcrrE
I El +l o 6
l-lpr--
IDIFI66
I EI iI RR I -l
| 'r
I I
\
3P
55
I Br
R
\
d.i
ol o'
tldlaec )tI -'
=l r' ! i
ell
-lEl a; HH§{Xfrfifr_t it
5l 5
-18
Network Desi.qn an$ Urban Sipul?tion \tre.q elr e r As may be expected, in all workshops the first evaluation showed considerable differences between the attituctres of the groups towards different plans and between the gl:oups I satis- faction levels (Fig. 10). After the first cycle the results were discussed within the groups. In a gaming discussion rvith all groups the conflicts that had trecome apparent were ver- balized, ancl the conflicting positions were stated; also first possibilities for a compromise were indicatecl. After the gaming session a trplanning commissionrf sat down to develop a compromise p1an, while sirnultaneously the groups did their second evaluation. The result of. the seconcl cycle of simula- tion and evaluation showed in all workshops significantly
re<luced conflicts between the groups and a convergence of satisfaction 1eve1s. Ttris tendency continued in the third
c j-rc1e , where it was executecl .
groups 1-2 groups 1-)
Fig. 10 Evaluation results:
potential conflict
,::...:
groups Z_1
Conflict maps show zones.
Other remarkable results were: satisfactions of all groups tended to develop negatively, regarrlless of the simulated plan. This tendency was most obvious for the low income group and less apparent with the high income group. hlith each cycle conflicts between low and high income groups lessened; while the low income group more and more adoptecl the values anrl
standards of medium and high income people, its dissatisfac- tion with its conditions of life increased.
Network Design and Urban Simulation We.qener
7. ACKNO}TLEDGEMENTS
The research reported in this paper has been conducted by
the urban planning group of Battelle-Frankfurt under research contracts by the Bundesministerium für Raumordnung, Bauwesen
und Städtebau, the City of Vienna and the Science and Human
Affairs Program of Battelle Institute.
REFERENCE§
( f ) f .S. Lowry, ^A, Model of Metropolis, Rand Corporation,
Memorandum RM-ttO35-nC, Santa Monica, CaI., 1"964
(Z) J.W. Forrester, Urban Dynamics, Cambridge, Mass., 7969
(3) SimulationsmodeLl POLIS, Benutzerhandbuch, Vorläufige Ausgabe, Stand August 1972, Vo1. 12 of the publication series ItStädtebaulictre Forschungtr of the Bundesministe- rium für Raumordnung, Bauwesen und Städtebau, Bonn 7973
(4) POLfS, Simulationsmodell für die Stadtentwicklungspla- nung, Inventardaten, Battelle-Frankfurt,, 197)
(5) POLIS, Simulationsmodell für die Stadtentwj-cklungspla- nung, Routensuchalgorithmen, Battelle-Frankfurt, L974
(6) H. Raiffa, Preferences for Multi-attributed Alternati- ves, Memorandum RM-5858-DOT/P.C,, Rand Corporation, Santa Monica, Cal., 1969
(7) V. Bauer, J. Meise, M. hiegener, Urban Systems Studies, Part One, Evaluation in Urban Planning, Approach and Project PLan, Battell-e-Frankfurt, 1972
(B) V. Bauer, A. Gebert, J. Meise, Urban Systems Studies, Part One, Evaluation in Urban Planning, Report,
Battel-1e-Frankfurt, 797 3
20