51
Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

  • View
    215

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Computational Approaches to Space Layout Planning

Presented By Hoda HomayouniFinal project for ARCH 588

Page 2: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Introduction“What is Space Layout

Planning”Space layout planning is the

process of allocating a set of space elements according to certain design criteria.

It usually results in topological and/or geometrical relationships between elements.

Page 3: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Introduction“Why Computers?”

We humans are easily board, distracted, and tend to make mistakes when confronted with large and complex problems.

Our memory is vast, but our ability to recall these memories at will is limited.

Page 4: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Motivation“Why this is important?”

Kalay: ”If we could find a way to take advantage of the abilities of computers where ours fall short, and use our own abilities where computers’ fall short, we would create a very powerful symbiotic design system: computers will contribute their superb rational and search abilities, and we humans will contribute all the creativity and intuition needed to solve design problems.”

Page 5: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Introduction“Challenges”

□Solving ill defined problems.□Addressing qualitative constraints.□Evaluating Nonquantifiable

Qualities.□Having Creativity or using humans

creative abilities.□Convincing the architects to rely on

the program.

Page 6: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Computerized Space Layout Planning

“Solution Approaches”Kalay (2004) categorizes

computational design synthesis methods as:

1- Procedural Methods2- Heuristic Methods3- Evolutionary Methods

Page 7: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

1- Procedural Methods

Leverage our ability , as human designers, to specify local conditions and the ability of computer to apply or test these relationships over much larger sets of variables.

Page 8: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

1-Procedural Methods“Types”

□Complete Enumeration□Space Allocation

□Additive Space Allocation□Permutational Space Allocation

Page 9: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

1- Procedural Methods“Additive Space Allocation”

GRAMPA (GRAph Manipulating PAckage ) is an example of a program that uses additive space allocation.

The current GRAMPA was presented in 1971.

Page 10: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

GRAMPA“Structure of the program”

The methods of solution depend on a special linear graph representation for floor plans calledthe ‘dual graph’ representation.

Page 11: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

GRAMPA “Structure of the program”

□a “space” is defined to be either a room or one of the four outside spaces.

Page 12: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

GRAMPA “Floor Plan Graph with Dual

Graph”

Page 13: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

GRAMPA “Planar Realizations of a

Graph”

Page 14: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

GRAMPA “Final level”

□ Fill the region with edges representing those adjacencies not specifically requested by the design requirements.

□ Satisfying the physical dimension requirements.

Page 15: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

GRAMPA“Output”

Page 16: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

GRAMPA “Pros and Cons”

□The adjacency constraints are defined as either true or false.

□Long run time (23 min. for the example)□The program could handle only adjacency

objectives.□Finding all the realizable planar graphs not

necessarily would lead to a floor plan.□Exhaustive generation of the floor plan

could be replaced with a heuristic method.

Page 17: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

1. Procedural Methods“Permutational Space Allocation”

□ACTLOC is an example of a prototype that uses Permutational Space Allocation

□The first version was written in 1965□Current program has been developed

in 1992

Page 18: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

ACTLOC “Procedure of the Program”

The Program is based on compatibility between any pairs of activities.

Page 19: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

ACTLOC“Structure of the Program”

ACTLOC finds the optimum configuration by shifting activities around, keeping the arrangement which results in an improvement in the overall compatibility, then shifting activities again, starting from the current best.

Page 20: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

ACTLOC “The Output”

Page 21: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

ACTLOC “Pros and Cons”

□The program does not consider size of the activities.

□The program may be trapped in a local optimum

□There are many other factors that could not be simulated with the ACTLOC

Page 22: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

2. Heuristic Methods

Computational design methods that are inspired by analogies, just like the design synthesis methods that are typically inspired by analogies and guided by the architect’s own or another designer’s previous experiences.

Page 23: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

2. Heuristic Methods“Types”

□Analogical Methods□Case-Based Methods□Expert Systems□Shape Grammars

Page 24: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

2. Heuristic Methods“Analogical Methods”

□Borrow the idea of simulating space arrangements in layouts from the rules that has derived from other sciences

□Scott Arvin and Donald House (1999) were the first one who used physically based techniques to manipulate space layout planning.

Page 25: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Analogical Methods“Structure of the Program”

□The designer specifies and modifies the design objectives instead of directly manipulating primitive geometry.

□The plan adapts to the changing state of objectives by applying the physics of motion to its elements.

Page 26: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Analogical Methods“Structure of the Program”

□The spaces and walls are modeled as point masses .

□Adjacencies between spaces are modeled as springs that connects the masses

□Objectives are translated into forces applied to the Masses.

Page 27: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Analogical Methods “Other Objectives”

□Orientation Objectives□Interior

Objectives□Exterior

Objectives□Separation Objectives

Page 28: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Analogical Methods “Geometric Design

Objectives”□Alignment

Objectives□Offset Objectives□Area Objectives

Page 29: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

“Analogical Methods”Geometric Design Objectives

□Gravity Objectives□Proportion Objectives

Page 30: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Analogical Methods “The Output”

Page 31: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Analogical Methods“Pros and Cons”

□It help architects to feel the nature of design problem

□The behavior of spring May not be the same as what we need for manipulating the design in all situations

□There are other design criteria that the program does not address

□Unable to handle multistory buildings

Page 32: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

3. Evolutionary Methods

Potential to Produce Novel Design Solutions.

Superior to other search algorithms for problem consisting of large unstructured search spaces.

□Genetic Algorithm□Neural Networks

Page 33: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

3. Evolutionary Methods “Genetic Algorithm”

□Based on Survival of the fittest□Genetic codes: Genotype□Population of solutions: Phenotypes □Fittest phenotypes are chosen by the

means of Fitness Function.□Three operators for manipulating the

representation of genotypes: Selection, Crossover, and Mutation

Page 34: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

3. Evolutionary Methods “Genetic Algorithm”

□By Rosenman and Gero in 1999 □two examples of work for evolving

designs by generating useful complex gene structures

□The first example uses a genetic engineering approach whereas the other uses a growth hierarchical model

Page 35: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Genetic Algorithm “Genetic Engineering Approach”

□Making the complex genes from basic genes and using them in next generations as evolved genes

□The individuals are evolved through a number of generations

□An additional operation identifies particularly successful combinations of genes

Page 36: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Genetic Engineering Approach “Evolving Representation”

Page 37: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Genetic Engineering Approach Design of Architectural Floor Plan

□The Fitness Function:

Page 38: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Design of Architectural Floor Plan

“Evolved Representation”

Page 39: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Design of Architectural Floor Plan

“New Requirements”

1- Minimal overall wall length 2- No walls with open ends, that is, no

walls that do not build a closed room;3- 6 rooms; 4- Room sizes 300, 300, 200, 200, 100

and 100 units.

Page 40: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Genetic Engineering Approach “The Output”

Page 41: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Genetic AlgorithmHierarchical Growth Approach

□ A multi-level approach.□Each level has its own definition

and requirements.□ at each level, a component is

generated from a combination of components from the level immediately below.

Page 42: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Hierarchical Growth Approach“The Fitness Functions”

□At room level: Minimizing the perimeter to area ratio and the number of angles.

□At zone level: Minimizing a sum of adjacency requirements between rooms.

□ At house level: Minimizing a sum of adjacency requirements between rooms in one zone and rooms in other zones.

Page 43: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Hierarchical Growth Approach“The output- Living Room

Generation”

Page 44: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Hierarchical Growth Approach“The output- Living Zone

Generation”

Page 45: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Hierarchical Growth Approach“The output- House

Generation”

Page 46: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Hierarchical Growth Approach“Pros and Cons”

□Organizing the generating process□Making the design process more

meaningful and manageable to the architects.

□Having multiple fitness function for addressing the same issue may make a system biased toward the one that is more likely to happen.

Page 47: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Hierarchical Growth Approach“Pros and Cons”

□The fitness function can not adapt itself to the design process.

□The program is not capable to recognize the novel solutions by itself.

□The program can not learn from its mistakes.

Page 48: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Conclusion“Comparison Chart”

Page 49: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Conclusion

□Despite the increasing need of architects to have a computational assistant in the design tasks, still they don’t show much enthusiasm to use the available programs

□Most of the programs are still at a research prototype stage

□The runtime of these programs increases exponentially by increasing the objectives so that the program can not handle calculating complexities.

Page 50: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Conclusion

□Architects want to design themselves and let the computer do the redundant jobs.

□They don’t trust these programs.□They do not want to loose their job to

a software.

Page 51: Computational Approaches to Space Layout Planning Presented By Hoda Homayouni Final project for ARCH 588

Future Works

□Producing a software that benefits from combinations of the advantages of the current systems.

□Finding the right user of the program and design the program based on the users’ need.