The Urban Decision Room: A multi actor design engineering simulation system

Third International Engineering Systems Symposium

CESUN 2012, Delft University of Technology, 18-20 June 2012

The Urban Decision Room

A multi actor design engineering simulation system

1

Delft University of Technology , Faculty of Architecture, Julianalaan 134 2628 BL Delft

2

Delft University of Technology , Faculty of Architecture, Julianalaan 134 2628 BL Delft

3

Planmaat, Mijnbouwstraat 106 2628 RX Delft

p.p.j.vanloon@tudelft.nl, p.barendse@tudelft.nl, sduerink@planmaat.nl

Abstract. This paper deals with the definition and construction of a decision based multi actor

urban design model which enables the integration of the allocation of a variety of urban land uses with the distribution of different urban functions: the Urban Decision Room. Urban design (and planning) is, among others things, about the spatial distribution of human activities and their physical facilities like buildings, roads, green areas etc. in amount, place and time over a well-defined area. In the Urban Decision Room the individual and specific visions and knowledge of the various participating parties with regard to the development area in question, are translated into individual and specific negotiable preferences for particular solutions for that area. By processing these preferences simultaneously and interactively in an Urban Decision Room, as opposed to dealing with them successively as in traditional design teams, the result is not a series of plan variants, but one common solution space.

Keywords. Multi Actor Design and Decision Making, Operational Goal Oriented Systems,

Urban Planning and Design

1 Introduction

In terms of structure, the Urban Decision Room (UDR) resembles a Group Decision Room. Both ‘rooms’ are interactive, with several people gathered together in a room each equipped with a computer. The computer network enables the participants to communicate with each other about the relevant topics. The network also enables to make calculations (using mathematical optimization models from Operation Research) of the ‘results’ of this communication and to represent it at each computer. The Urban Decision Room is based on the discipline of urban area design and planning. It is aimed specifically at decision-making processes in urban planning, particularly in complex, urban area development projects: hence the ‘U’ for ‘Urban’ in the name. This background in urban design also enables the UDR to support planning decisions that are made at concrete urban planning element level. This means that the UDR participants are not asked for idealised visions of policy themes. They are asked to come up with concrete solutions to an urban area planning problem (in terms of preferences for particular functions, number of plots, etc.).

The starting point in the UDR is that the individual and specific visions and knowledge of the various participating parties with regard to the development area in

question are translated into individual and specific negotiable preferences for particular solutions for the area. By entering these preferences simultaneously and interactively into a UDR, and not dealing with them successively as in traditional design teams, the result is not a series of plan variants, but one common solution space. A solution space in which alternative sets of preferences are possible and feasible. In other words, the UDR is a support instrument in the search for a final and common goal, rather than simply a provider of goals for individual parties. Experiments have been conducted with respect to the Laurenskwartier, an inner-city area of Rotterdam, to the Spoorzone, the central station area in Delft and to

Heijsehaven, an old city port area in Rotterdam.

Fig. 1. The Urban Decision Room

2 The Urban Decision Room as a Goal Oriented System

The Urban Decision Room is regarded as a particular type of ‘goal-oriented social system’. This is a system consisting of individuals and groups of individuals striving for their own goals and, at the same time, for collective goals. Within this system, the individuals therefore interact, take decisions in co-operation and in competition with each other (Van Loon et al., 2008).

A goal-oriented social system is a system which seeks to achieve a certain goal or goals, and consists of at least two makers. In general, it has n decision-makers (D1,...Di,...Dn). The fact that such a system contains decision-makers distinguishes it from empirical systems (systems in which processes are autonomous, natural and spontaneous). In the literature, a model of a goal-oriented system is often referred to as a normative model, and a model of an empirical system is a descriptive model (Van Loon, 1998).

The decision-makers in a goal-oriented social system are interconnected, in the sense that they exercise a certain influence over each other. It can be assumed that this influence can be exercised in two ways: through the primary system (PS) or through

the communication system (CS). The primary system is the actual object with which

the decision-makers are concerned. The communication system consists of all the communication channels through which information is passed from one decision-maker to another. See Figure 2.

Fig. 2. The goal-oriented social system

Hanken and Reuver (1976 pp. 49-55) have introduced two elementary units, which can be used to elaborate the model of a goal-oriented system: the system cell and the decision cell. The system cell gives a diagrammatic representation of a system: a set of objects which are interrelated in such a way that they form a more or less independent whole. A decision cell can be used to represent the decision-maker. The decision cell contains the goals and the choice procedure (algorithm). In other words, the contents of the decision cell represent what the decision-maker is seeking to achieve, and the way in which he intends to do this. In mathematical models, the choice procedure is represented in an algorithm, which is a mathematical procedure that allows a problem to be solved in an unambiguous way, in a finite number of steps.

The system cell contains the variables and their interrelationships:

• The input variables

These are independent variables, which originate in the external environment or are output variables from other systems. They are independent because their value is determined outside the system, and not by the decision-maker.

• The decision variables/control variables

These are also independent variables. Their value is determined outside the system too, but this time by the decision-maker himself. The decision-maker (or management body) can ‘control’ the system using these variables.

• The auxiliary variables/intermediate variables

• The state variables

State variables are dependent variables, which serve as the system’s memory.

• The output variables

These are dependent variables, which are determined by the problem at hand. Their values are determined by the system. The output variables give the solution to the problem.

These elements all affect the workings of the system. See Figure 3.

Fig. 3. The system cell and the decision cell (After: Hanken and Reuver, 1976 pp. 27-32) When two or more decision-makers influence one system, the concept can be expanded by adding more decision cells. See Figure 4.

Fig. 4. A system cell with two decision cells

3 The Urban Decision Room‘s Basic Decision Making Model

We begin with an example for which a mathematical urban planning model can be constructed (Van Loon, 1998; Van Loon and De Graaf, 2011).

The decision-making problem of a housing association

A housing association wants to build a number of blocks of residential property and facility units (shops, school, social and cultural centre, etc.) on a particular site. The site covers 14,000 m2. The association hopes to complete the project within 16

months. A block (construction time 2 months) covers 1,000 m2, while a facility unit (construction time 1 month) covers 2,000 m2. A residential block costs 8.106 Euros, and a facility unit costs 5.106 Euros; the overall budget is 80.106 Euros. It is not necessary to cover the entire site. A survey has been conducted among the future residents. This has revealed that they value housing blocks and facilities at a ratio of 5:3. The aim is to ensure that the future residents are as pleased with their neighbourhood as possible. This problem can be represented mathematically in a LP model. X1 is the number of blocks of residential property and X2 is the number of facility units. Two decision-makers are involved in this problem: the housing association and the future residents.

MAX! 5 X1 + 3 X2 (appreciation) Sub: 1,000 X1 + 2,000 X2 ≤ 14,000 (site area) 2 X1 + X2 ≤ 16 (construction time) 8.106 X1 + 5.10 6 X2 ≤ 80.10 6 (budget) X1 ≥ 0 X2 ≥ 0

The simplex algorithm (a mathematical procedure which allows an LP model to be solved with 2 or more unknown variables) can be used to find the mathematical solution to this problem. Since the example has only two unknown variables, it can be solved using a simple drawing. See Figure 5.

Fig. 5. The solution space (shaded)

This modelling of the decision problem faced by the housing association is represented in diagrammatic form in Figure 6

The housing association and the future residents will undoubtedly continue negotiating their decisions and goals after this ‘initial’ solution has been found. Such negotiation is useful in order, for instance, to establish whether a change in the construction costs might better suit the preferences of the residents. Other, cheaper

building materials could lower the costs, which might lead to a better distribution of houses and facilities.

Fig. 6. The decision model for the problem faced by the housing association

The general structure of this model is as follows. This model is to select the values for the decision variables x1, x2, … ,xn so as to:

Maximize

Z

=

c

x

+

c

x

+

...

+

c

x

x

a

x

...

a

x

b

a

+

+

≤

x

a

x

...

a

x

b

a

+

+

≤

,

...

x

a

x

a

x

b

a

+

+

≤

.

0

,...,

0

,

0

≥

x

≥

x

≥

x

For the sake of brevity, we use Σ notation and write:

Maximise

∑

=

x

c

Z

i n j j ij

x

b

a

≤

∑

0

≥

x

This is adopted as the standard form for the linear programming problem. Any situation whose mathematical formulation fits this model is a linear programming model.

4 The Urban Decision Room‘s Basic Spatial Allocation Model

In urban planning not only the quantities of and the preferences for the resources (like land to be used, buildings, infrastructure) to be allocated play a role, but also the location of the resources in the urban space. (Binnekamp et al., 2006; Van Loon and De Graaf, 2011). With an extension of the basic multi actor urban decision model – the linear programming model with negotiable constraints – we are able to model the allocation of urban activities to space (urban land use). In urban design and planning, a dominant spatial dimension of resources is the position of resources in two- and three-dimensional space. This position is commonly expressed in floor plans, land use plans, and three dimensional models of buildings and their urban environments. In terms of allocation of resources, a floor plan is a proposal for allocation of architectural spaces to accommodate human activities such as living, shopping, eating, and office work: Which spatial layout of the resources fits the activities to be accommodated best, in accordance with stakeholders’ wishes, goals, and constraints, and with the architectural style chosen?

If we define the activities as demand (d) and the resources as supply (s) we can represent this problem (which is called in Operations Research literature the transportation problem or the distribution problem) in an LP model as follows:

Minimise 1 1 m n ij ij i j

Z

c x

=

∑∑

x

d

≥

∑

x

s

≤

∑

In this model xij is the representation of an activity i in space j. cij is the representation

of the cost (expressed in money, energy, appreciation, and the like) of the realisation of activity i in space j. This representation can be explained with two aspects of the

relationship between activities and spaces as follows: Since in buildings and urban areas human activities are not fixed to one unique space – or in other words activities are spread out over more spaces, like rooms, auditoria, corridors, zones, areas – a design expresses, among a lot of other things, a spatial pattern of different architectural and urban spaces to fit a set of different activities allocated to the designed spaces. In the remainder of this paper the index i refers to an activity, j to a space or zone, k to a lot, m to number of activities, n to number of zones and pj to the number of lot within zone j.

The second aspect concerns the fact that most of the urban spaces are suited for more than one activity, but of course not for all. This means that the designer can propose alternative arrangements of the activities required, for a given spatial arrangement of spaces. Also the other way around: for a given spatial arrangement of activities, alternative layouts of urban spaces may be proposed. By changing the input values of

cij, a representation of the design process on both aspects becomes available. With this

mechanism, a designer can represent his pattern of possible activities in such a way that he can see how well this pattern fits the activities required.

While urban spaces may be suited for more than one activity, they are not necessarily suited for all activities due to technical constraints such as daylight, noise hindrance, permitted location in the building, or conceptual constraints such as structure of spaces and patterns of connections.

The model for this design problem (the limited distribution problem) can be formulated as follows: Minimise 1 1 m n ij ij i j

Z

c x

=

_∑∑

a x

d

≥

∑

a x

s

≤

∑

{0,1}

a

=

value greater than or equal to zero if aij = 1. This means that if the designer decides

that space sj is not suited or otherwise not appropriate for activity i, he sets aij = 0 and

automatically xij becomes 0. In other words, using the zero and one value of aij, the

designer uses the model to calculate the best allocation of activities to the designed pattern of spaces.

In the representation of the space allocation described above, it is assumed that the total demanded space for activities equals the total supplied space for the activities. In the beginning of a design process this is often not the case. In architectural design and urban planning, demand and supply are independent of each other. They are not fixed at the start of a design process. Designers propose spatial arrangements of spaces based on their ideas, style, and concepts. Of course, these proposals are not that far from the required spaces, but they are not equal. So, a design can give ideas for activities one was not thinking of. Similarly, a designer can discover that he does not yet have space for an activity which certainly should be in the building. The designers have to find the best fit. With two extensions to the above model, it is possible to cope with this design question.

Minimise 1 1 m n ij ij i j

Z

c x

=

_∑∑

0

a x

S

−

=

∑

0

a x

D

−

=

∑

D

≥

d_min

D

≤

d_max

S

≥

s_min

S

≤

s_max

0

D

≥

0

S

≥

{0,1}

a

=

Sj Allocated space in zone j.

Dj Allocated space for activity

s_minj Minimum available space in zone j.

s_maxj Maximum available space in zone j

d_mini Minimum demand for activity i.

d_maxi Maximum demand for activity i.

5 The Urban Decision Room ‘s Decision Making Environment

society that experiences the problem, the group involved in the problem, and the organisation endeavouring to solve the problem. I shall refer to this below as the problem environment.

When a solution for a decision making problem is sought, the problem is removed from its environment, but is not completely dissociated from it. This would not be possible, since the problem ‘exists’ only in the context of its environment. The environment determines the social background, historical build-up and technical constraints of the problem. All parties involved in the problem (or wish to be) are also part of this environment.

On top of that there will be several decision making conceptions of the decision making problem environment, or parts thereof, which will lead to several design problems. A design situation soon becomes ambiguous and multi interpretable, as shown in Fig. 7.

Fig. 7. The ambiguous design situation

References

Binnekamp, R., Van Gunsteren, L.A., Van Loon, P.P. (2006), Open Design, A Stakeholder

Oriented approach in Architecture, Urban Planning and Project Management, IOS Press,

Amsterdam.

Hanken, A.F.G., Reuver, H.A. (1977), Sociale Systemen en Lerende Systemen, Stenfert Kroese, Leiden.

Van Loon, P.P. (1998), Interorganisational Design. A New Approach to Team Design in

Architecture and Urban Planning. TUDelft PublicatieBureau. Delft.

Van Loon, P.P., Heurkens, E., Bronkhorst, S. (2008), The Urban Decision Room, An Urban

Van Loon, P.P., De Graaf, R. (2011), Sustainability and Urban Density, A Decision Based Design Approach. In: Proceedings of Management and Innovation for a Sustainable Build