Sage Journals: Discover world-class research

Abstract

The field of generative design strives to enable the automated generation of multiple suitable urban design alternatives. However, quantitative analysis is typically conducted in the later design stages, when options are limited, and modifications are costly. Generative methods enable the creating of designs that bridge traditional methods, through systematic examination of numerous options during the early design stages. Yet their utilization in architectural practices is limited, partly due to antagonism that stems from the perceived lack of involvement of urban designers in such methods (which often behave like black boxes). The objective of this study was to develop a generative design process that combines rule-based automated algorithmic processes with the active involvement of urban designers, thereby offering a wide range of design alternatives during the planning phase. Additionally, the study aims at ensuring the key involvement of urban designers throughout the design process. The methodology employed in this study is comprised of both quantitative and subjective evaluations that the urban designers can conduct themselves. This result is an interactive process that utilizes generative tools and computerized analytical measures for creating, evaluating, and screening multiple urban design options at the urban neighborhood level, with an emphasis on residents’ well-being, and based on the designers’ preferred parameters. This parametric workflow could assist designers in the early decision-making process and may be integrated into current urban design processes.

Keywords

Computer-generated parametric design urban design processes new urbanism theory urban well-being interactive design

Introduction

Motivation and scope

The rapid growth of cities and urbanization in recent decades has led to significant challenges in urban design and planning, including the creating of dense, potentially overcrowded living environments that can negatively impact the residents’ well-being and health (Fisher-Gewirtzman, 2017). These challenges, combined with issues such as global warming and dwindling natural resources, have led to the emergence of more sustainable and efficient approaches to urban design (Bardhan et al., 2015). A growing interest can also be seen in generative methods and performance-based designs—to create high-performing urban environments that also consider aspects such as the climate, energy, mobility and walkability, the economy, and perceived density among residents (Schnabel et al., 2017). The integration of advanced computerized tools into the design process has also played a role in this significant trend.

Parametric design methods utilize algorithms for generating multiple design alternatives, based on a given set of parameters, objectives, and limitations (Nagy et al., 2018). Such computerized methods and generative workflows may also streamline design processes, thereby saving time and resources (Wilson et al., 2019). Specifically, integrating the parametric design method into urban design processes allows designers to make quick and easy adjustments in all stages of the process (Çalışkan, 2017; Koenig et al., 2017; Nagy et al., 2018; Vanegas et al., 2012). Moreover, this could help decision makers address the complexities of urban space through a readily available tool.

As mentioned, the benefits of using generative design methods are considerable. While the solutions produced through generative design are not fundamentally distinct from those manually and traditionally crafted by designers, generative design methods facilitate the systematic evaluation of a vast array of alternatives. This evaluation includes measuring their performance upon creation and at the earliest design stages, when changes are still straightforward and cost-effective to implement. This process allows for the objective ranking of alternatives based on their performance. Furthermore, when an alternative is generated parametrically, it becomes easier and quicker to implement changes throughout the process, unlike traditional methods where each adjustment necessitates extensive planning/design and redrawing.

Yet, most architectural design practices only perform systematic quantitative analysis in the later stages. As such, traditional urban design practices tend to entail a team of architects who manually create a small number of design schemes (Austern et al., 2014; Çalışkan, 2017; Wilson et al., 2019).

Integrating parametric and generative methods into urban design practices can be a challenging task, due to the complexity of the urban environment, the need to consider numerous parameters and their interrelations, and the conflicting interests of the various stakeholders (al Qeisi and Al-Alwan, 2021; Stouffs and Janssen, 2017). This complexity is twofold, as it exists both at the macro and micro level (Çalışkan, 2017). The use of these methods also requires high levels of software programming skills and can be computationally expensive.

Some studies have utilized generative methods to create optimal alternatives (Calixto and Celani, 2015), yet this can be difficult in cases with multiple stakeholders and complex environments. It can also be challenging to assign weights to different metrics, as a means for reflecting the priorities of the various design aspects in optimization processes with multiple goals—thereby adding further complexity to the generative process (Fisher-Gewirtzman, 2018).

Designers may be deterred by the possible integration of computerized methods in their design processes, particularly in the early stages when their creative skills and experience are most valuable. There is also concern regarding the black box effect (Harding and Shepherd, 2017) of optimization-based generative processes that may result in only one solution for a given design issue. It is therefore important to create a process that involves the designers and their unique input for addressing these concerns.

As such, there is a clear need for a generative method that focuses on the specific process of urban designers, and that entails numerous factors and micro-decisions which may impact the outcome. Moreover, this method must also enable designers to conduct comprehensive assessments of the generated design alternatives—through both quantitative and subjective evaluation in an interactive manner.

Research objectives

The aim of this research was to develop a model that would enable urban designers to generate and evaluate multiple design alternatives at the neighborhood scale; in the early design stages when changes are easier to make; with an emphasis on the residents’ well-being; and in line with the Contemporary Urban Theory. In addition to being interactive, this proposed generative workflow would also entail automated and rule-based (i.e., parametric) algorithmic processes. In addition to developing this novel design tool, the study also strove to assess its efficacy and applicability via a case study and based on several indices for assessing the quality of the generated alternatives.

The research hypothesis posits that the alternatives generated through the interactive parametric model will exhibit comparable or superior performance to the existing neighborhood.

Research background

Generative design in the urban scale: Related studies

Generative design principles stem from an extensive debate on urban planning “design methods,” which confronts the notion that cities are complex, organic systems and criticizes traditional, oversimplified, and predefined master planning in favor of a more dynamic approach (Mehaffy, 2008, 2011). Led by Christopher Alexander and influenced by thinkers like Jane Jacobs and the principles of New Urbanism (CNU, 1993; Alexander, 1964, 1987; Jacobs, 1961), this perspective asserts that cities inherently possess a “structural quality” enabling them to grow according to their own unique “laws of wholeness” (Alexander, 1979). This approach rejects conventional master planning for what they call “generative” methods (Batty, 2007). The generative strategy emphasizes adaptability and integrates a city’s unique character, moving away from fixed “diagrammatic” plans toward a more fluid, complex, rule-based, and organically evolving urban development.

The generative design method allows designers to explore complex alternatives through automation and algorithms. More specifically, it enables the exploration of large design spaces through the generation of novel, high-performing designs that meet specific objectives (Frazer, 1995). Procedural generation refers to a technique that allows designers to create a range of complex urban designs based on their defined set of rules and parameters (Fusero et al., 2013), and using computer-aided design (CAD) tools to explore large design spaces (Wilson et al., 2019). This approach is useful for creating designs that are responsive and adaptable to a given context (Schumacher, 2010), and differs from traditional design approaches, where designers refine solutions through manual iteration and experience.

Generative tools and computational design methods are often applied on small scales in architecture and engineering, to optimize specific issues such as building forms and facade geometries (Fisher-Gewirtzman, 2019; Rakha and Reinhart, 2012)—primarily as a means for addressing the demands and requirements of a single stakeholder (Brown and Mueller, 2016; Nagy et al., 2018). In the practice of urban design, however, these methods are less frequently adopted (Nagy et al., 2018; Çalışkan, 2017), with draft and illustration tools being more commonly used, with the help of CAD and manual 2D and 3D modeling. Yet these latter methods are dependent on teams of urban designers, who can create a small number of relevant schemes.

The complex process of urban design involves multiple disciplines and stakeholders, each with different interests and objectives (Nagy et al., 2018). It also involves dealing with intricate and detailed geometric and morphological designs at the urban scale, which can make it difficult to define input parameters without increasing computational costs. The urban design process can also be challenging for reconciling between the various interests and opinions to achieve optimal (or even satisfactory) results that meet the conflicting demands (Schnabel et al., 2017; Stouffs and Janssen, 2017; Wilson et al., 2019).

Several approaches have been suggested for the generation of urban fabrics. König and Bauriedel (2004) proposed a four-step process for generating urban designs (Ayaroğlu, 2007). Aldabagh and Alchalabi (2009) used geometrical shapes of individual buildings to create a regular grid, while Steinø and Obeling (2014) developed a six-step process for generating urban designs. Other researchers have used shape grammars (Beirão and Duarte, 2018; Beirão et al., 2010) or intera methods for the procedural generation of parcels within city blocks (Vanegas et al., 2012). These methods all involve the creating of a skeleton-based or oriented-bounding box subdivision. Koenig et al. (2017) developed a framework for generating urban design alternatives by synthesizing road networks, plots, and building volumes—using simple algorithms and an open-source toolset for the Rhino/Grasshopper environment called Decoding Spaces, which also includes optimization engines and evaluation tools. Wilson et al. (2019) created a computer-based methodology for analyzing and visualizing urban scale designs.

Studies on generative methods for urban design provide valuable insights and practical applications, yet they also entail a range of challenges. These challenges include lack of simplicity in the initial shapes; unrealistic or overly complex urban-network results; of scalability to the neighborhood level urban design; and lack of an end-to-end framework that analyzes the existing conditions of a given urban area while integrating programmatic, spatial, and geographical characteristics as input. In addition, these studies often focus on finding a single optimal design solution, rather offering a range of possible solutions that could be continuously addressed during the unpredictable design process. This proposed research strives to address these challenges, by creating a holistic process for generating, evaluating, and filtering design alternatives, while involving the designer in the selection process and offering a range of suitable planning alternatives based on the designer’s selection of relevant parameters and indices.

Urban well-being

Most of the world’s population currently lives in urban areas, with over 70% of the global population expected to live in cities over the next two decades (Fisher-Gewirtzman, 2017; United Nations, 2018). The well-being and quality of life of urban residents are therefore of great importance. Studies show a connection between urban designs and the residents’ well-being, with an emphasis on identifying design principles that meet the needs of city dwellers (Fisher-Gewirtzman, 2012, 2017; Fisher-Gewirtzman and Wagner, 2003; Gehl, 2013; Lynch, 1984; Marans and Stimson, 2011; Martin et al., 1972; Trossman-Haifler and Fisher-Gewirtzman, 2022). Research focus has been placed on principles of benefactor urban design in dense urban environments, to contribute to a sense of well-being among residents. Dense yet well-planned cities can offer a range of benefits and opportunities (Borukhov, 1978; Fisher-Gewirtzman, 2018). Poorly planned dense cities, however, could have significant negative impacts on their residents.

This study applied several planning and geometrical principles, based on approaches to urban planning that prioritize the well-being of residents. These include the Theory of New Urbanism, the concepts of Jan Gehl, the concept of Perceived Density, and the Neighborhood 360 Index. First, the Theory of New Urbanism (Grant, 2005), an approach to urban planning and design, strives to create urban environments that promote residents’ well-being and quality of life—through principles of traditional neighborhood development, transit-oriented development, and walkable neighborhoods, that offer a range of housing and job types with a focus on creating a sense of community and sustainability. Danish architect and urban designer Jan Gehl (Gehl, 1987, 2013; Gehl et al., 2006), renowned for his people-oriented approach to urban design, suggests a range of principles, such as an interconnected network of short, logical routes, small spaces, and a clear city hierarchy, as well as vibrant “soft” edges and comfortable walking conditions (Matan, 2011).

The Perceived Density (Churchman, 1999), that is, how people experience the density of a given environment, based on factors such as the morphology and geometry of structures, the number of dwelling units, and the height of the buildings—all of which affect the residents’ well-being and quality of life (author, 2016). The Spatial Openness Index (SOI) is suggested to predict the perceived density, by calculating the 3D visibility, as observed from all indicated viewer positions (Fisher-Gewirtzman, 2003). Finally, this study is also based on the Neighborhood 360 Index, developed by the Israeli Green Building Council to encourage high-quality sustainable urban environments in new developments and urban renewal projects. The index strives to incorporate sustainability principles into the design process, while encouraging the consideration of various issues in the early planning stages.

Quantitative evaluation methods of urban environments

In this study, we applied a quantitative analysis methodology for evaluating urban environments using various computer-based metrics. Such methods can be classified by field of knowledge, such as solar analysis (Capeluto et al., 2006; Naboni et al., 2019; Natanian and Wortmann, 2021), network analysis (Brandes, 2001; Dogan et al., 2018; Hillier and Hanson, 1989; Song, 2005), airflow analysis and energy simulation (Naboni et al., 2017; Natanian and Wortmann, 2021), economic analysis (Nagy et al., 2018; Wilson et al., 2019), visibility and perceived density analysis (Batty, 2001; Fisher-Gewirtzman and Wagner, 2003), geometric and density analysis (Alexander et al., 1988; Fisher-Gewirtzman, 2017; Natanian and Wortmann, 2021), ecological analysis (Leskens et al., 2017), and mobility simulations (Branea et al., 2017).

Methodology and data

To create and evaluate the proposed model for generating and evaluating multiple urban design alternatives, the following four stages were conducted:

Stage 1. Review the literature as a means for identifying tools and theories in three relevant areas: generative design at the urban scale, evaluation methods for urban environments, and urban well-being parameters.

Stage 2. Develop a tool for generating parametric designs: selecting an existing neighborhood as a case study. Next, the urban designers participating in this experiment selected a range of fixed and variable planning principles, based on both the chosen case study and their own worldview. The authors then developed a parametric tool, by combining existing open-sourced tools and Python code in Rhino/Grasshopper environments.

Stage 3. The researchers generated a set of design alternatives, which were then evaluated through quantitative measures and manually filtered through an iterative and interactive process. The designers applied three methods that meet their worldview for evaluating the alternatives, that is, walking distance (Dogan et al., 2018), daylight hours (Roudsari et al., 2013; Natanian and Wortmann, 2021), and perceived density (Fisher-Gewirtzman, 2003, 2017b). The selection of these methods was based on their strong relationship to the well-being of urban residents and the availability of fast runtime calculation tools (additional analysis tools could be chosen for future research and practice). During this iterative stage, the urban designers assessed and rated the alternative designs, through the implementing of subjective decision-making.

Stage 4. The top ten alternatives were selected and compared to the performance of the chosen case study.

The interactive model for generating multiple urban design alternatives

Developing the workflow algorithm

The novel generative algorithm presented in this study was developed following previous research on parametric urban design and on the rule-based algorithmic creation of urban environments—with a focus on geometric and morphological complexities (Ayaroğlu, 2007; Beirão and Duarte, 2018; Beirão et al., 2010; Chowdhury and Schnabel, 2018; Koenig et al., 2017; König and Bauriedel, 2004; Steinø and Obeling, 2014; Vanegas et al., 2012; Vidmar and Koželj, 2015; Wilson et al., 2019). The purpose of the algorithm was to create a complex and realistic environment that could be used for detailed assessments, such as radiation analysis and perceived density. To create a design that enables degrees of freedom, the algorithm was written in Python, combined with open-source generative components for the Rhino/Grasshopper Platform, such as Decoding Spaces (Koenig et al., 2017).

The development process entailed the determining of the desired level of complexity and flexibility, while finding a balance between a large range of input parameters and reasonable calculation runtimes.

Selecting the case study

The Neot Peres Neighborhood in Haifa, Israel, was chosen as a case study for the present research due to its urban density, landscape views (the Mediterranean Sea in the West and the Carmel Mountains in the East), and widespread public criticism of its urban design (Merav Moran, 2018). The master plan for the neighborhood which includes about 2000 housing units. The neighborhood is relatively flat and enables the use of a simple grid for presenting the road network typology.

Choosing the leading design principles for the model

The generative algorithm developed in this research is based on design principles and on variable parameters, which are defined by the urban designers as the starting point. While the design principles dictate how the algorithm operates, and are expressed in all generated alternatives, the variable parameters enable the creation of a set of diverse alternatives for the designers to evaluate and choose from. The design principles are derived from the local master plan of the neighborhood, the New Urbanism Theory (Grant, 2005), and the designers’ spatial choices and preferences.

Principles inspired by the local master plan of the neighborhood

The design principles for the generative algorithm were based on the local master plan for the case study neighborhood, which includes programmatic, economic, and connectivity considerations at the metropolitan scale. These principles—such as zone areas and built areas, building volume and density, network typology, connectivity to the surrounding neighborhood and main traffic routes, and a central park axis with a cluster of public buildings—served as a geometrical baseline.

Principles inspired by the New Urbanism Theory

Additional design principles for the generative model were based on the New Urbanism Theory and measurable indices, such as the Neighborhood 360 Index. These principles included an orthogonal grid-based street network with a range of 60–180 m between junctions; walking distances with a range of 250–300 m from the residential homes to daily-used services; and lot depths to create shared courtyards. The model also included a set of predefined street sections, as per the New Urbanism Theory.

Principles inspired by the designers’ spatial choices and preferences

Based on their worldviews and intuition, the designers chose the following six free variable parameters, which when combined—enabled the creating of an initial set of alternatives (Figure 1):

1. Size of the grid.

2. Position of the park’s axis along the street network.

3. Distribution of the public buildings along the park.

4. Width of the residential plots.

5. Building heights’ delta factor

6. Building heights’ organizational rule in relation to their surroundings (i.e., mountains and sea).

Figure 1.

A diagram of leading spatial principles, chosen as quantitative input parameters for the generative algorithm.

The algorithm’s main steps for calculating the design alternatives

The first step of the generative design process is to create a design space algorithm model that can generate various design alternatives based on site conditions and constraints (Nagy et al., 2018). The parametric workflow algorithm in this study was developed through the following six stages (Figure 2):

1. Inserting the neighborhood borders as polygon input for the model.

2. Inserting number of grid divisions in the x and y directions of the polygon.

3. Determining one axis of the grid as a central park, and sorting the remaining roads based on a hierarchy.

4. Sorting the urban blocks by proximity to the park axis, while maintaining the land use mix and zoning areas for residential and public spaces.

5. Dividing residential blocks into plots and categorizing them into building typologies, based on their area size.

6. Generating building according to the typology of each lot. Determining the height of each building—based on distance from a given reference point, the building height rule type, and the building height ratio. A Python component calculates the number of floors required for each building in a recursive algorithm, ensuring that the required number and size of residential apartments and public buildings are maintained.

Figure 2.

Parametric workflow’s main steps.

Performing calculations for the initial set of design alternatives

The Colibri plugin (CORE Studio, 2016) was used to automatically iterate the generative algorithm and aggregate the data for each alternative in a comma-separated value type of database. The challenge at this stage was finding an optimal balance between the number of variable parameters, the runtime of the algorithm and analyses, and the designers’ ability to manually filter the generated alternatives.

Filtering out alternatives

After generating the initial set of design alternatives, the urban designers conducted an iterative process of computer-based analysis and manual filtering, to demonstrate the human-machine interaction. This iterative evaluation and filtering phase was comprised of the designers choosing a range of desired results for each metric, and then manually filtering out unrealistic alternatives using the Design Explorer Platform (CORE Studio, 2017). This platform could be perceived as a parametric space in which the performance of each option is presented in relation to all other ones. The order of the three analysis stages was determined in line with the designers’ worldview, values, and priorities, as well as the calculation time for each analysis. Below is a description of the reasons behind the designers’ analysis-related choices, as seen in Figure 3.

Figure 3.

Three analysis stages: (1) The shortest-path analysis for the walkability index. (2) Analysis of total daylight hours on December 21 (performed via Ladybug). (3) SOI analysis for the perceived density measure, demonstration via long (blue) and short (red) lines-of-sight.

Stage I: Walking distance

This study calculates walking distances based on the shortest-path algorithm of the space syntax (Zhan and Noon, 1998) and on three additional indices: (1) the average distance from each residential building to all public buildings; (2) the average distance from each residential building to the park; and (3) the average distance from each residential building to a particular public building. The starting point of the walking route is calculated from the center of the facade that is closest to the road of each building.

This stage of choosing the range of results for filtering the alternatives is a point in which the designer intervenes and influences the parametric process. In this study, alternatives with an average distance greater than 350 m to public buildings were removed, as were those with more than 500 m or less than 300 m to the park and the chosen public building. This was done to ensure that the neighborhood had a high walkability rate and that residents were not housed too close to potential sources of noise.

Stage II: Daylight hours

Sunlight is an important natural resource that is known to have a positive impact on the quality of urban environments, people’s health and well-being, and on energy consumption of lighting, heating, and cooling (Capeluto et al., 2006; Dogan and Park, 2019). Recent studies present a range of urban performance models for quick assessing urban designs in terms of daylight hours, energy loads, solar radiation, and thermal comfort (Naboni et al., 2017, 2019; Natanian et al., 2019).

This study focuses on a simple solar matrix and evaluation method that can be integrated into a holistic workflow using the Rhino/Grasshopper Platform; it is also suitable for urban designs, due to its scalability and compatibility with automated design processes. The LadyBug Sunlight Hours Analysis (Roudsari & Mackey, 2017), an open-source plugin for the Rhino/Grasshopper Platform, was used to measure the amount of sunlight in different areas of the neighborhood and during two different seasons (summer, June 21; and winter, December 21). The following three measurements were conducted:

1. Daylight hours in open park areas—based on the assumption that maximal sunlight hours in these areas create a pleasant outdoor space.

2. Daylight hours on the building facades from all directions—based on the assumption that maximal sunlight on building facades ensures maximal penetration of natural light via the building’s interiors, thereby saving on lighting energy and enhancing a sense of well-being.

3. Daylight hours on the building rooftops—based on the assumption that this will maximize the potential for solar energy production or green rooftops.

The results were not normalized for the variance of the areas measured, to give priority to alternatives that offer more daylight hours.

Stage III: Perceived density

Perceived density indices have been found to impact the sense of well-being and behavior of urban residents (Fisher-Gewirtzman, 2012, 2022). SOI measure perceived density (Fisher-Gewirtzman, 2017) via the volume of visible open spaces that can be potentially seen from a given viewpoint (Fisher-Gewirtzman, 2003). In this study, The SOI was calculated by measuring visibility at the junctions of a street network, based on the assumption that junctions are places that people tend to pass by very frequently. The average total lines-of-sight lengths at all junctions, and the sum of the total lines-of-sight lengths for each alternative, were measured.

Results: Leading generated alternatives via the interactive algorithmic model

The ten alternatives that were selected, based on the urban designer’s criteria and the filtering process, are presented with ID cards that display the input indices used in the generative algorithm and the results of the measurements and filtering iterations. These ID cards allow for easy comparison of the tradeoffs between different indices, as a higher score in one aspect may result in a lower score in another. Moreover, the use of ID cards and bar charts enabled the designers to comparatively monitor the performance of the various alternatives, while making informed decisions based on the designers’ priorities and worldviews.

As mentioned, the article outlines a filtering process that merges an objective, quantitative assessment of alternatives’ performance—where the researcher selects the value ranges for criteria from a pool of alternatives—with a subjective, manual selection. This approach is designed to address the “black box” issue prevalent in generative design processes. From an initial pool of 1910 alternatives, after three rounds of quantitative filtering, the designer narrowed the selection down to 25 alternatives for manual/subjective review. This quantity allows for an in-depth evaluation of each option and the ability to manually choose based on the content. The subjective criterion guiding the selection in this this study was the aim to ensure a diverse range of alternatives. It is important to note that all alternatives have the exact same local master plan definitions regarding land use mixes, built volumes, density levels, public programs, and overall typology (Figure 4).

Figure 4.

Best alternatives ID cards.

The case study: Comparative analysis

Using the Design Explorer Platform (CORE Studio, 2017), the case study was compared to the top ten alternatives created via the proposed algorithmic parametric model (Figure 5). The findings indicate that the ten alternatives that were created via the parametric model outperformed the existing neighborhood in the average walking distance between the residential buildings and the public ones, and in some indices relating to daylight hours. On the other hand, the existing neighborhood outperformed the generated alternatives in the average walking distance between the residential buildings and the park, and daylight hours in the open spaces during the winter season. In the SOI indices, the generated alternatives outperformed the existing neighborhood by about 30% in relation to lower perceived density among residents. Overall, these results support the research hypothesis whereby the alternatives that were generated by the interactive model performed at least as well as the existing neighborhood, if not better for some indices.

Figure 5.

Comparison of leading parameters and analysis results between the case study (the bold red line) and the leading ten alternatives.

Discussion

The aim of this research was to create an interactive algorithmic model that generates and assesses urban design options at the neighborhood scale—using a combination of urban design parameters and computational tools. The proposed workflow allows urban designers to create multiple design alternatives in the early design stages, based on their priorities and preferences—with the goal of enhancing residents’ well-being in urban neighborhoods, despite ever-increasing density requirements.

The workflow presented in this paper enables a balance between an algorithmic generative tool and the important input of the urban designers, based on their knowledge, experience, and perspectives. In this manner, the designers are not replaced by automated processes; instead, they can leverage such processes for generating and evaluating a much larger number of urban design alternatives that would not otherwise be possible.

This study builds on previous research that explores the development of parametric urban environments using a range of generative methods (e.g., Al-Sayed and Penn, 2016; Aldabagh and Alchalabi, 2009; Beirão and Duarte, 2018; Beirão et al., 2010; König and Bauriedel, 2004; Nagy et al., 2018; Nickerson, 2008; Popov, 2010; Rakha and Reinhart, 2012b; Schaffranek and Vasku, 2013). Contrary to prior studies, however, the primary objective of this research was to establish an interactive generative workflow that prominently involves the urban designer. By incorporating the computational and analytical advantages of computerized tools, as well as the designers’ subjective worldviews and irreplaceable intuition, the planning process benefits from the best of both worlds.

The proposed model also enables flexibility, as different sets of parameters, indices, and measurements can be applied. Indeed, the subjective preferences of the urban designers play a key role in this selection process, which is based on tradeoffs between metrics. The process outlined in the article differs from traditional optimization processes, which aim to find few optimal solution, by seeking to develop several strong alternatives and understanding their performance strengths and weaknesses through measurable criteria.

Creating a generative tool allows architects and urban designers to conduct in-depth examination of many design alternatives that they would not otherwise be able to analyze. The demonstrated parametric model may allow for more freedom, including randomness, which could result in alternatives that the urban designer may not have considered. This helps enrich the design process, by encouraging new design directions and thought processes.

In conclusion, by using the proposed workflow in multiple iterations, designers can explore a large range of design parameters and indices to create more accurate and suitable design alternatives that meet their goals and desired outcomes.

Limitations and future research

This study contributes to the field of urban design and residents’ well-being in both theory and practice. However, several research limitations should be addressed.

The experiment centered on a site-specific case study, potentially limiting generalizability and reliability. Input parameters and analysis iterations were restricted to demonstrate the iterative process. Only three quick analyses were conducted due to workstation constraints. The initial set of alternatives was reduced to 1.910, possibly excluding valuable alternatives.

Future research could optimize the iterative process for broader alternatives.

Further research could focus on improving the iterative process to generate even more diverse and impactful parametric alternatives. At the neighborhood scale, the emerging field of generative design requires more exploration, especially in handling the complexities of real urban environments and reducing abstraction levels. Comparing different generative methods and integrating algorithms from machine learning and deep learning could be valuable.

Future research in generative design could involve computerized models in the decision-making of prominent urban designers. Incorporating virtual reality experiments for assessing design options, with both quantitative and subjective evaluations through user responses and interviews, could be beneficial. Expert evaluations might also provide insights into urban design practitioners’ perspectives.

Despite architectural design optimization advancements, this study did not utilize optimization tools. While this research strives to overcome fundamental problems that are related to optimization processes, optimization algorithms provide an effective solution for large-scale parametric studies. Future research could integrate such algorithms to expand the parameter set, ensuring valuable alternatives aren’t overlooked.

Supplemental Material

Supplemental Material - An interactive method for generating and evaluating urban design alternatives in early design stages

Supplemental Material for An interactive method for generating and evaluating urban design alternatives in early design stages by Anat Talmor Blaistain and Dafna Fisher-Gewirtzman in Environment and Planning B: Urban Analytics and City Science.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) received no financial support for the research,authorship,and/or publication of this article.

Data availability statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.

Supplemental Material

Supplemental material for this article is available online.

Architect Anat Talmor Blaistain (MSc) is an accomplished architect and researcher with a rich background in architecture and urban planning. She earned her B.Arch. and M.Arch degrees with distinctions from the Technion Institute of Technology's Faculty of Architecture and Urban Planning. After her studies,Anat contributed to diverse projects in several architectural firms. She then joined a startup focused on developing automated economic analyses for urban-scale projects. Currently,Anat is a member of the AVP lab research group at the Technion,where her work emphasizes computational design generation and optimization,aiming to innovate urban planning methodologies on a substantial scale.

Professor Dafna Fisher-Gewirtzman (PhD),Head of the Architecture Visual Perception Lab,head of the research and development centre and previously head of the Architecture program at the Faculty of Architecture& Town Planning at the Technion-IIT;Her research focus is in the field of visual analysis and simulation and development of novel,automated design tools,directed toward the development of sustainable built environments;Environments for the Aging Brain;Interdisciplinary research incorporating Architecture and Neuroscience involving experiments in virtual reality. Her research is supported by the ISF and JoyVentures. She is a UNESCO fellowship recipient,and a laureate of the Yanai Prize and the Henri Gutwirth Fund.

References

Al Qeisi

Al-Alwan

(2021) Generative urban design concepts and methods: a research review. IOP Conference Series: Materials Science and Engineering 1090(1): 012085.

Al-Sayed

Penn

(2016) Translating analytical descriptions of cities into synthetic urban design models. In: Proceedings of the seventh international conference on design computing and cognition (DCC’16), Chicago, IL.

Aldabagh

Alchalabi

(2009) A mechanism for creating new shape and syntaxes in old Mosul city. In: International generative art conference, Milan.

Alexander

(1964) Notes on the Synthesis of Form. Cambridge, MA: Harvard University Press, Vol. 5.

Alexander

(1979) The Timeless Way of Building. New york, NY: Oxford University Press, Vol. 1.

Alexander

(1987) A new theory of urban design. In Center for Environmental Structure Series. Oxford: Oxford University Press, Vol. 6.

Alexander

Reed

Murphy

(1988) Density Measures and Their Relation to Urban Form. Madison, WI: Center for Architecture and Urban Planning Research, University of Wisconsin.

Austern

Moral

, et al. (2014) “The urban Genome” published as conference contribution at the 19th International Conference on Computer-Aided Architectural Design Research in Asia - Rethinking Comprehensive Design: Speculative Counterculture, CAADRIA 2014 - Kyoto, Japan.

Ayaroğlu

(2007) Urban Complexity and Connectivity: Emergence of Generative Models in Urban Design. Ankara: Middle East Technical University.

10.

Bardhan

Kurisu

Hanaki

(2015) Does compact urban forms relate to good quality of life in high density cities of India? case of Kolkata. Cities 48: 55–65.

11.

Batty

(2001) Exploring isovist fields: space and shape in architectural and urban morphology. Environment and Planning B: Planning and Design 28(1): 123–150.

12.

Batty

(2007) Cities and Complexity: Understanding Cities With Cellular Automata, Agent-Based Models, and Fractals. Cambridge, MA: MIT Press.

13.

Beirão

Duarte

(2018) Generic grammars for design domains. Artificial Intelligence for Engineering Design, Analysis and Manufacturing 32(2): 225–239.

14.

Beirão

Mendes

Duarte

, et al. (2010) Implementing a generative urban design model: grammar-based design patterns for urban design. In: FUTURE CITIES [28th eCAADe Conference Proceedings/ISBN 978-0-9541183-9-6], Zurich.

15.

Borukhov

(1978) The trade-off between density and other objectives: a re-examination of planning norms. Geojournal 2(1): 71–80.

16.

Brandes

(2001) A faster algorithm for betweenness centrality. Journal of Mathematical Sociology 25(2): 163–177.

17.

Branea

A-M

Gaman

Badescu

(2017) Urban mobility analysis on efficiency and sustainability by means of transportation. IOP Conference Series: Materials Science and Engineering 245(8): 082020.

18.

Brown

Mueller

(2016) Design for structural and energy performance of long span buildings using geometric multi-objective optimization. Energy and Buildings 127: 748–761.

19.

Çalışkan

(2017) Parametric design in urbanism: a critical reflection. Planning Practice and Research 32(4): 417–443.

20.

Calixto

Celani

(2015) A literature review for space planning optimization using an evolutionary algorithm approach. In: XIX Congresso da Sociedade Ibero-Americana de Gráfica Digital 2015, Brazil, pp. 1992–2014.

21.

Capeluto

Yezioro

Bleiberg

, et al. (2006) Solar rights in the design of urban spaces. In: International conference on passive and low energy architecture, Santiago, Vol. 23.

22.

Chowdhury

Schnabel

(2018) An algorithmic methodology to predict urban form-an instrument for urban design. In: Learning, Adapting and Prototyping - 23rd International Conference of the Association for Computer-Aided Architectural Design Research in Asia (CAADRIA), Beijing.

23.

Churchman

(1999) Disentangling the concept of density. Journal of Planning Literature 13(4): 389–411.

24.

CNU (1993) Congress for New Urbanism. Newsport News, VA: CNU. https://www.cnu.org/.

25.

CORE Studio (2016, January 16) Colibri. Half Moon Bay, CA: CORE Studio. http://core.thorntontomasetti.com/colibri-release/.

26.

CORE Studio (2017, January 3) Design Explorer. Half Moon Bay, CA: CORE Studio. http://core.thorntontomasetti.com/design-explorer/.

27.

Dogan

Park

(2019) A critical review of daylighting metrics for residential architecture and a new metric for cold and temperate climates. Lighting Research and Technology 51(2): 206–230.

28.

Dogan

Samaranayake

Saraf

(2018) Urbano: a new tool to promote mobility-aware urban design, active transportation modeling and access analysis for amenities and public transport. In: Proceedings of the symposium on simulation for architecture and urban design, Delft, 1–8.

29.

Fisher-Gewirtzman

(2012) 3D models as a platform for urban analysis and studies on human perception of space. Usage, Usability, and Utility of 3D City Models–European COST Action TU0801, 01001.

30.

Fisher-Gewirtzman

(2017) The association between perceived density in minimum apartments and spatial openness index three-dimensional visual analysis. Environment and Planning B: Urban Analytics and City Science 44(4): 764–795, On line: http://epb.sagepub.com/cgi/reprint/0265813516657828v1.pdf?ijkey=VL1KjRmmLddXwRi&keytype=finite.

31.

Fisher-Gewirtzman

(2018) Integrating weighted views to quantitative 3D visibility analysis as a predictive tool for perception of space. Environment and Planning B: Urban Analytics and City Science 45(2): 345–366, (First Published 15 Nov 2016) http://epb.sagepub.com/cgi/reprint/0265813516676486v1.pdf?ijkey=OpOVzF7C3MR5mbm&keytype=finite.

32.

Fisher-Gewirtzman

Wagner

(2003) Spatial openness as a practical metric for evaluating built-up environments. Environment and Planning B: Planning and Design 30: 37–49.

33.

Frazer

(1995) An Evolutionary Architecture. Chennai: Architectural Association.

34.

Fusero

Massimiano

Tedeschi

, et al. (2013) Parametric Urbanism: a new frontier for smart cities. Planum the Journal of Urbanism 2(27): 1–13.

35.

Gehl

(1987) Life Between Buildings. New York, NY: Van Nostrand Reinhold, Vol. 23.

36.

Gehl

(2013) Cities for People. Washington, DC: Island Press.

37.

Gehl

Gemzøe

Kirknaes

, et al. (2006) New City Life. København: The Danish Architectural Press.

38.

Grant

(2005) Planning the Good Community: New Urbanism in Theory and Practice. London: Routledge.

39.

Harding

Shepherd

(2017) Meta-parametric design. Design Studies 52: 73–95.

40.

Hillier

Hanson

(1989) The Social Logic of Space. Cambridge, MA: Cambridge University Press.

41.

Jacobs

(1961) Jane jacobs. In The Death and Life of Great American Cities. New York, NY: Random House.

42.

Koenig

Miao

Knecht

, et al. (2017) Interactive urban synthesis. In: International Conference on Computer-Aided Architectural Design Futures, Berlin. 23–41.

43.

König

Bauriedel

(2004) Computer-Generated Urban Structures. Aachen: Juniorprofessur Computational Architecture.

44.

Leskens

Kehl

Tutenel

, et al. (2017) An interactive simulation and visualization tool for flood analysis usable for practitioners. Mitigation and Adaptation Strategies for Global Change 22(2): 307–324.

45.

Lynch

(1984) Good City Form. Cambridge, MA: MIT Press.

46.

Marans

Stimson

(2011) An overview of quality of urban life. In Investigating Quality of Urban Life. Berlin: Springer, 1–29.

47.

Martin

March

(1972) Urban Space and Structures (Issue 1). Cambridge: Cambridge University Press.

48.

Matan

(2011) Rediscovering Urban Design Through Walkability: An Assessment of the Contribution of Jan Gehl. Perth: Curtin University.

49.

Mehaffy

(2008) Generative methods in urban design: a progress assessment. Journal of Urbanism: International Research on Placemaking and Urban Sustainability 1(1): 57–75.

50.

Mehaffy

(2011) A city is not a rhinoceros: on the aims and opportunities of morphogenetic urban design. Built Environment 37(4): 479–496.

51.

Merav

(2018, June 26) How to kill a neighborhood: this is what happened in Haifa. https://www.themarker.com/realestate/.premium-1.6212987.

52.

Naboni

Meloni

Coccolo

, et al. (2017) An overview of simulation tools for predicting the mean radiant temperature in an outdoor space. Energy Procedia 122: 1111–1116.

53.

Naboni

Natanian

Brizzi

, et al. (2019) A digital workflow to quantify regenerative urban design in the context of a changing climate. Renewable and Sustainable Energy Reviews 113: 109255.

54.

Nagy

Villaggi

Benjamin

(2018) Generative urban design: integrating financial and energy goals for automated neighborhood layout. In: Proceedings of the Symposium for Architecture and Urban Design Design. Delft, The Netherlands, 265–274.

55.

Nagy

Zhao

Benjamin

(2018) Nature-based hybrid computational geometry system for optimizing component structure. In Humanizing Digital Reality. Berlin: Springer, 167–176.

56.

Natanian

Wortmann

(2021) Simplified evaluation metrics for generative energy-driven urban design: a morphological study of residential blocks in Tel Aviv. Energy and Buildings 240: 110916.

57.

Natanian

Aleksandrowicz

Auer

(2019) A parametric approach to optimizing urban form, energy balance and environmental quality: the case of Mediterranean districts. Applied Energy 254: 113637.

58.

Nickerson

(2008, June 23) Generating networks. In: Third International Conference on Design Computing and Cognition (Dcc08), Atlanta, GA.

59.

Popov

(2010) Generative urban design with cellular automata and agent based modelling. In: Conference Contribution - Paper in Published Proceedings, Australia.

60.

Rakha

Reinhart

(2012) Generative urban modeling: a design work flow for walkability-optimized cities. Proceedings of SimBuild 5(1): 255–262.

61.

Roudsari

Mackey

(2017) Ladybug. Fairfax, VA: Ladybug Tools LLC.

62.

Roudsari

Pak

Smith

(2013) Ladybug: a parametric environmental plugin for grasshopper to help designers create an environmentally-conscious design. In: Proceedings of the 13th International IBPSA Conference Held in Lyon. France, 3128–3135.

63.

Schaffranek

Vasku

(2013) Space syntax for generative design: on the application of. In: Proceedings of the Ninth International Space Syntax Symposium, Seoul.

64.

Schnabel

Zhang

Aydin

(2017) Using parametric modelling in form-based code design for high-dense cities. Procedia Engineering 180: 1379–1387.

65.

Schumacher

(2010) The parametric city. In: Zaha Hadid Recent Projects, GA. Tokyo: ADA Edita Tokyo, 42.

66.

Song

(2005) Smart growth and urban development pattern: a comparative study. International Regional Science Review 28(2): 239–265.

67.

Steinø

Obeling

(2014) Developing a parametric urban design tool. Some structural challenges and possible ways to overcome them. Architecturae et Artibus 6(1): 51–57.

68.

Stouffs

Janssen

(2017) A rule-based generative analysis approach for urban planning. In Morphological Analysis of Cultural DNA. Berlin: Springer, 125–136.

69.

Trossman-Haifler

Fisher-Gewirtzman

(2022) How urban wellbeing is influenced by spatial urban parameters (density, morphology, vegetation & commerce), as examined in a VR framework. Architectural Science Review 65: 370–384. DOI: 10.1080/00038628.2022.2091510.

70.

UN (2018, May 16) 68% of the World Population Projected to Live in Urban Areas by 2050. San Francisco, CA: UN. https://www.un.org/development/desa/en/news/population/2018-revision-of-world-urbanization-prospects.html.

71.

Vanegas

Kelly

Weber

, et al. (2012) Procedural generation of parcels in urban modeling. Computer Graphics Forum 31(2pt3): 681–690.

72.

Vidmar

Koželj

(2015) Adaptive urbanism: a parametric maps approach. IGRA USTVARJALNOSTI (IU)/CREATIVITY GAME (CG)–Theory and Practice of Spatial Planning 3: 44–52.

73.

Wilson

Danforth

Davila

, et al. (2019) How to generate a thousand master plans: a framework for computational urban design. In: Proceedings on the 10th Symposium on Simulation for Architecture and Urban Design SimAUD, Atlanta, GA.

74.

Zhan

Noon

(1998) Shortest path algorithms: an evaluation using real road networks. Transportation Science 32(1): 65–73.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.39 MB

0.00 MB