Post-conflict reconstruction in the Middle East and North Africa region: A bidirectional parametric urban design approach

Introduction and background

Armed conflicts threaten many parts of the world and are increasing in speed of occurrence, frequency, scale and intensity.^1,2 Conflict and violence in the 21st century fundamentally differ from 20th century patterns of interstate conflict and ways of recovering from them. Poverty, unemployment, income shocks such as those sparked by volatility in food prices, rapid urbanisation and inequality between groups all increase the risks of violence. ³ Armed conflict causes major destruction of urban centres and changes cultures and societies by causing major loss to individuals, environments and resources. This has long-term, if not permanent, consequences on the built environment and the lives of local citizens. ⁴ Dealing with contemporary causes of armed conflict requires rethinking of the ways of alleviating them nothing short of a paradigm shift. This makes the case for inclusive, integrated urban reconstruction high, as social foundations hinge on the ability of urban space to support recovery of the social fabric.

Urban reconstruction following major conflicts is often a tricky business, as different international, national and local actors struggle to create a space for their interventions to support war-stricken communities (Dabaieh & Alwall, 2018). ⁵ Governments’ capacity and resources are often depleted, and communities are either returning from exile/internal displacement or struggling among the rubble of destruction, having lived trauma that affects their ability to return to normalcy. This is often difficult to addressing recovery planning. International agencies often struggle with lack of coordination between actors, with weakness of local capacities, with understanding the local communities, as well as with the limited budgets allocated for reconstruction. Time is yet another critical element that hinders efforts for equitable, inclusive and integrated urban reconstruction, due to the rapid proliferation of informality, complexity of land use and tenure and normalisation of practices that may inhibit an inclusive, resilient recovery. In addition, post-conflict reconstruction is commonly seen as an acute and single recovery action, separated from the process of constructing long-term sustainable settlements. ⁶ This need not be the case, however, as disruption caused by conflict has long-term implications and might provide opportunity to rethink architecture and urban practices, and a chance to escape from uninformed decisions of the past. ⁷

The aim of this article is to propose, and give examples of a parametric urban design approach to post-conflict reconstruction, allowing for a bottom-up participatory design process while maintaining a strategic and inter-scalar overview (from the city district scale over the neighbourhood scale to the street/building scale) of relevant planning parameters such as building densities, land use, public infrastructure and green space. The main token of this approach is a parametric urban model that supports three-dimensional (3D) visualisation of the intervention site at different scales and levels of abstraction while simultaneously being capable of reporting relevant urban data (housing units, street lengths, parks etc.). The purpose of this model is to provide a visualisation medium which may be comprehended by non-architects and modified instantaneously to reflect different scenarios and stakeholder interests, as well as a way to evaluate physical scenarios of the built environment against their associated urban data in real time. With a focus on post-conflict reconstruction in the Middle East and North Africa (MENA) region, the scope and functionality of the parametric urban model is demonstrated for an urban district in the Iraqi city of Fallujah. Fallujah has suffered severe damage from over a decade of armed conflict and the effects of war on the city are well documented by the World Bank. ⁸ It therefore constitutes a uniquely situated case for demonstrating the proposed approach.

Approach and methodology

In order to establish a realistic framework for the proposed bidirectional (top-down and bottom-up) parametric urban design approach to post-conflict reconstruction in the MENA region, three areas of investigation have been included into the study. First, the concept of collaborative urban design is presented and discussed. This represents the bottom-up direction, as it takes its point of departure in the needs and aspirations of local stakeholders. Second, the concept of sustainable urban development in the MENA region is presented and discussed. This represents the top-down direction, as it requires spatial coordination at different scales. By way of two examples, the combination of a bottom-up, collaborative approach and a top-down, sustainable approach, the bidirectional approach is demonstrated and summarised. Third, parametric urban design is presented and demonstrated as the medium which combines the two directions across different scales. In addition to these three areas of investigation, the demonstration site, an urban district in the city of Fallujah, is presented and motivated.

Computational framework of parametric modelling

For purposes of this exercise, the parametric design modeller CityEngine (CE) was used in combination with ArcGIS online, as the main the parametric urban model. CE is based on the shape grammar-inspired scripting language computer-generated architecture (CGA). In CGA scripts, parameters and their variables are stored in attributes. Attributes are available for manual modification through an inspector window which is a part of the CE interface.

It is central to the approach of the present study that parameter values can be modified interactively, causing the parametric urban model to change dynamically. This prevents a complex computational framework as instant interactivity is typically not possible across different software platforms. Therefore, the computational framework is kept simple. Hence, it is limited to the following:

ArcGIS map data for terrain and streets;

CE for parametric modelling and using the CGA scripting language.

For CE, the inputs used are as follows:

ArcGIS map data in the form of street centrelines, terrain rasters and aerial photographs (accessed directly through the CE interface);

CGA scripts for model and data generation;

Constraint maps which inform the scripts with respect to desired variations to the design.

The output from CE is as follows:

3D models;

Two-dimensional (2D) maps;

Data (such as floor areas, floor area ratios, land use, number of dwellings, number of parking spaces etc.).

The CGA scripts were designed to cater to a number of different case uses. Depending on scale and/or focus, both level of detail (LoD) and mode of representation can be adjusted. While architectural detail such as facade openings, parapets and brise-soleil (examples used below) are irrelevant in large-scale representations and result in high polygon counts, they are indispensable for a qualitative evaluation at the street/building scale. Similarly, physical representation may be the most appropriate in small-scale 3D representations, but representation of land use, floor area ratios and so on may be more appropriate in 2D map representations at the district scale.

The functionality of scripts for the parametric urban model presented here must be a product of (a) relevant sustainability parameters for the specific place of application; (b) the social, cultural and economic specificities as defined by local stakeholders; and (c) the physical features (landscape and existing urban fabric) of the specific urban space. Thus, the examples given in this article are intrinsically generic, as they demonstrate only a subset of parameters defined by the authors, based on general sustainability parameters for the MENA region (hot, arid climate) and general liveability parameters (walkable cities, access to green space etc.).

Two sets of CGA scripts were developed for the present study. One set of scripts was developed to simulate the existing urban fabric. As no precise building data are available by lot, the script does not represent the exact physical reality. Instead, it simulates typical building structures, based on a typology study. The typology study was based on orthogonal (Figure 1) and oblique (Figure 2) aerial photographs. Four simplified building variations of a stepped, two-storey building with front yard and/or backyard were synthesised and represented in low detail. The first set of CGA scripts was designed to generate random variations of an existing building type. Most geometry was generated through simple setback and extrude operations. Elements shared with other types (facades, roofs, fences, gardens) were passed on to imported scripts. Variation was controlled through parameters with randomly generated values. The resulting geometry is shown in Figure 10(a).

OPEN IN VIEWER

Figure 1.

An edited Google map of Fallujah showing damaged roads (in red). The study area of the present study is a rectangular superblock located at the northern edge of the city (second from left/west), not to scale.

OPEN IN VIEWER

Figure 2.

Detail from orthogonal aerial photograph with building typology analysis.

Another set of scripts was developed to simulate possible building and green space typologies for reconstruction. While limited in variation for the generic demonstration of the present study, they comprise (a) a walled garden lot with randomly distributed trees and (b) four different building types, all with a form of courtyard/atrium, in different height distributions from 2 to 6 storeys. In a real-life scenario, both open and built-up spaces would be represented through more typologies, such as parks, plazas and sports fields and specialised building typologies, respectively. In the second set of CGA scripts, geometry generation for each lot (generated by a previous script) was passed on randomly to scripts generating four different building types or gardens. The distribution was weighed according to a building density attribute which was controlled by the greyscale levels of a constraint map (Figure 7) which was normalised. The lower the value of the building density attribute, the higher the ratio of gardens. The level of detail for geometry generation was controlled by a ‘LoD’ attribute. The resulting geometry is shown in Figure 10(c).

The study area of Fallujah lends itself well to simple parameterisation, as it is typical of many urban developments in the MENA region from the past 3–4 decades. Consisting almost entirely of long ribbons of back-to-back single-family housing lots of near identical sizes in an orthogonal layout, scripts did not have to cater for very different nor complex lot geometries. While street centrelines were available through ArcGIS online, block subdivisions were scripted to approximately match actual conditions (see Figure 3).

OPEN IN VIEWER

Figure 3.

Oblique aerial photograph with typical buildings for cross reference and validation.

Collaborative urban design and parametric modelling

Collaborative urban design refers to a set of iterative and participatory methods that stakeholders, professionals and policy makers can use to reach consensus on the planning of neighbourhoods and cities. There is an array of collaborative design methods that have emerged over the years and that have proved to democratise and spread city and neighbourhood design decision-making, such as charrettes, workshops, Q&A meetings and so on^17,18. By means of such methods for citizen engagement, the field can be levelled when it comes to issues of power and interest, and more transparent negotiations with a wider array of stakeholders may be allowed. Parametric urban design has obvious potentials for collaborative design processes. This has been addressed by Jacobi et al., ¹⁷ Kunze and Schmitt, Kunze et al., Kunze et al., Steinø et al. ¹⁸ and Steinø. ¹⁹ If combined with parametric design tools, collaborative urban design can offer fast, detailed, flexible options for urban planning and development that can be easily explained to a variety of audiences, allowing them to take more engaged action in the planning of their built environments.

Parametric design as a method of choice facilitates the generation of different urban design scenarios through parametric variation. It allows to quickly generate large-scale generic designs (district or neighbourhood scale master plans), while at the same time adding detail (street and building scale 3D models). ²⁰ Rather than making one-off plans for urban areas, parametric design may support iterative and step-by-step decision-making among stakeholders(Figure 4). This may not only reduce the communication gap between urban specialists and urban dwellers, but also between governments and constituencies, and between different groups in a city. It may also provide the opportunity to rethink the rebuilding of cities and lives at the early stages of post-conflict reconstruction. Early involvement of different stakeholders (residents, builders, planners etc.) may be an opportunity to rethink post-conflict reconstruction through an integrated approach to the design of housing and public places, and to engage with citizens in processes of peace-building through the active physical reclaiming of space.

OPEN IN VIEWER

Figure 4.

Parametric urban design for collaborative urban design processes. At this participatory design workshop in Denmark, a parametric urban model (on display) was used interactively to visualise and discuss different design scenarios.

While the authors have only tested the parametric urban design approach theoretically in a MENA context, it has been tested in real-life planning at the masterplan scale in Denmark. ¹⁹ Apparently, the MENA region and Denmark represent very different socio-cultural contexts with radically different traditions in urban design and planning. However, the ability of laypeople and non-designers to understand urban planning and design documents and visualisations in these settings are comparable. Hence, the added value of using parametric models is likely to be transferable, given their dynamic and spatial nature, which promotes ease of visualisation, discussion and modification of design in a collaborative manner.

A bidirectional planning and design approach

Top-down approaches to planning at the larger scale are not necessarily adversary to a bottom-up approach to home building. In his well-known Belapur incremental housing scheme, Indian architect Charles Correa devised a simple, yet very regular fractal principle for the distribution of housing lots for low-cost residential buildings that fit both approaches in one design. At the smallest level, seven lots are clustered in a 3 × 3 unit square, leaving the centre and one corner units open(Figure 5, left).

OPEN IN VIEWER

Figure 5.

Schematic representation of Correa’s Belapur housing scheme, showing the fractal nature of the site plan. Drawing by the authors, based on original drawing by Charles Correa.

At the next level, three such larger units are clustered in a 2 × 2 unit square with the unbuilt corners facing towards each other and leaving one larger unit with only 2–3 lots, thus also providing an opening at this level(Figure 5, centre). This last organisation principle is repeated once again at yet a higher level. In total, this produces a fractal system of individual housing lots organised around shared courtyards of growing sizes at three different levels (Figure 5, right). Nonetheless, everything is distributed on a strictly Cartesian grid.

While three different basic housing unit types were originally provided by the architect, the idea was that – over time – the units would be expanded and adapted by the residents to their individual needs and likings. The flexibility and variation achieved by this approach is remarkable. Each fractal cluster of lots is self-contained and constitutes entities which may stand alone without appearing to be unfinished or incomplete. Yet, the system allows for the seamless addition of more clusters to build organic wholes. As such, it represents an emblematic example of the combination of a structurally varied, yet simple top-down master grid and a bottom-up self-build approach to the construction of the individual housing units. In a similar fashion, the Chilean architect Alejandro Aravena and his company Elemental develop housing projects offering ‘half houses’ designed uniformly by architects. ²¹ A void is left for residents to subsequently fill out the other halves of the houses. By this approach, a rational and efficient overall layout of the housing scheme is facilitated while safeguarding variation and user participation through individual completion by the residents.

Correa’s Belapur housing scheme was originally planned and designed entirely by the architect, leaving it to chance and unsolicited processes of adaptation for the scheme to mutate into subsequent variation and densification. Elemental, on the other hand, makes a virtue of including residents into collaborative design processes through series of workshops throughout the different phases of design, bidding, construction and subsequent workshops for expansion and collective spaces. ²¹ Both Correa and Elemental are occupied with the use value and adaptation of their designs to the cultural, social and economic needs of residents, as well as to local climatic conditions, building materials and technologies. As such, they are proponents of a process of urban (re-)development which differs from traditional public and market-driven modes of production. While public social housing typically focuses on keeping construction costs down with little regard for the use quality, commercial home building, on the other hand, treats dwellings as products and therefore aims for the mainstream ‘taste’ of the anonymous consumer.

In the context of post-conflict reconstruction in the MENA region, approaches similar to those of Correa and Elemental make a lot of sense. As Syrian architect Marwa ²² contends in her vision for the reconstruction of Syria after the civil war, the post-independence mode of urban development in her home country has failed to respond to the needs of urban dwellers, culturally, socially and economically, as well as environmentally. Or, as Elemental puts it: ‘The housing problem in the world will only be solved if we are able to combine top-down public policies with bottom-up self-construction capacity’ (p. 19). ²¹

The urban spatial production adapts and changes drastically after the shock of a conflict and under the stress of reconstruction, to the risk of it being considered a deviance if one only considers the pre-conflict conditions. This change in the architectural and urban production can be directly linked to the shock of the destruction, and to the various stressors for those who practice architecture and make decisions related to space-making in the years during and following a conflict (Benbih, 2016). Therefore, capturing these changes through an in-depth coordination and collaboration between the dwellers and the decision makers at various levels, and using cutting edge architectural and urban advances. This participatory involvement can help translate these changes and can support reflecting the post-conflict urban-scape.

Urban structure and scale

Creating sustainable and liveable urban environments requires different measures at different scales. While some measures must be taken on the urban and district level, others relate to the neighbourhood and building levels. Measures for creating sustainable urban environments include to minimise carbon emissions and pollution, both during construction and operation of urban space, which in turn includes the consumption of materials and fuels, as well as the introduction of natural and low-energy techniques and technology. Measures for creating liveable urban environments include socially and economically sustainable land-use distribution and infrastructure as well as an urban design which facilitates physically pleasurable and convenient urban spaces.

Different regions in the world feature different environmental, social, cultural and economic conditions which require different responses in order to successfully create sustainable and liveable urban environments. The requirements which are geographically specific to the MENA region and even sub-regions thereof are thus partially different from those of other regions. Through post-conflict reconstruction, the pre-existing structures, physical as well as socio-cultural and economic, must be taken into consideration, while accommodating potential new socio-economic changes.

Results and discussion for the parametric urban model

The presented parametric urban model is organised so that it may display an existing urban fabric, a totally redeveloped urban fabric or combinations thereof. Computationally, the starting point is shapes representing lots in a subdivision(Figure 6). Depending on the set ratio of redevelopment, lots will either generate existing building types randomly (Figure 6, A–D) or new building types based on the grey value of a constraint map for each lot(Figure 6, E–H). The constraint map triggers different ratios for green space, as well as different combinations of new building types, some thereof with different building heights as a function of the constraint map.

OPEN IN VIEWER

Figure 6.

Flow diagram for the parametric urban model. In addition to buildings, the script also generates gardens.

In the following, three examples are presented to illustrate the versatility of the parametric urban model with regard to scale, transitional stages of urban reconstruction and detailed architectural studies. The model is designed to visualise the existing urban fabric as well as an imagined future urban fabric as a result of urban reconstruction. The examples take their point of departure at the district – or superblock – level, which has defined the Fallujah study area. The existing urban fabric within this area (as in neighbouring areas) consists of an almost undifferentiated repetition of a single typology (two-storey single-family homes), interspersed mainly with mosques, resulting in a monotonous urban structure(Figure 7).

OPEN IN VIEWER

Figure 7.

Aerial map of the existing urban fabric. The study area is bounded by the main streets visible along the edges of the image, not to scale.

In the offered examples, spatial, organisational and economic differences are introduced. To this end, a constraint map(Figure 8) was used for redeveloped lots to trigger the script to differentiate the frequency of built-up and open spaces, as well as the building heights within the area. This map was created to illustrate an example of how, over time, neighbourhoods may develop within the superblock, separated by more green and open areas. In the model, brighter values trigger predominantly green spaces interspersed with low buildings, and darker values trigger fewer green spaces and taller buildings. The constraint map is also used to control street widths, so that brighter values will trigger narrow streets and darker values will trigger broad streets. Otherwise, the street network is left unchanged in the presented examples.

OPEN IN VIEWER

Figure 8.

Constraint map for the study area. In the parametric urban model, the constraint map is overlaid with the aerial map and referenced by the script for each lot.

In the first example, the district scale is visualised at three stages of redevelopment ranging from no redevelopment to total redevelopment (Figure 9). For comparison, existing buildings are rendered in darker shades while new buildings are rendered in lighter shades. The imprint of the constraint map becomes gradually more visible in the green structures. The example illustrates how the parametric urban model can be used to visualise structural aspects of different redevelopment scenarios at the district scale.

OPEN IN VIEWER

Figures 9.

Plan views of the study area showing a gradual transition from the existing situation to a future state of redevelopment, not to scale.

In the second example, the neighbourhood scale is visualised at similar stages of redevelopment to the first example (Figure 10(a)–(c)). Different building typologies are detectable, ranging from existing typologies in Figure 10(a) to new typologies in Figure 10(b). For areas with low densities (as controlled by the constraint map), the script generates courtyard-type single-family houses in a contemporary interpretation on redeveloped lots. For areas with higher densities, the script gradually generates more multi-storey, multi-family houses with different types of courtyards for natural ventilation and daylight. The example illustrates how the parametric urban model can be used to visualise morphological aspects of different redevelopment scenarios at the neighbourhood scale.

OPEN IN VIEWER

Figure 10.

3D representation of the parametric urban model at neighbourhood scale: (a) existing two-storey building typology, (b) intermediary scenario with existing and new building typologies and (c) differentiation of densified development (right) and green areas (left).

In the third example, the parametric urban model is used to make simple visual daylight evaluations at the street/building scale (Figure 11(a)–(d)). To this end, the model is rendered at an architectural LoD with balconies, parapets and facade openings, allowing for a sense of scale and depth. In the example, brises-soleil are turned on/off, different building typologies are tested for neighbouring buildings and the street width is altered. The example illustrates how the parametric urban model can be used to visualise architectural aspects of different redevelopment scenarios at the street/building scale.

OPEN IN VIEWER

Figures 11.

Street view of the parametric urban model at street/building scale. Examples of the effects of (a) arcades, (b) brises-soleil, (c) building mass and (d) street width on shading.

In the given examples, the parametric model is used to make different scenarios for a geometrically simple urban environment. The scenarios are shown at different scales and examine temporal changes (district and neighbourhood scales) as well as a visual examination of the effect of different architectural and morphological configurations on shading (street/building scale). Yet, basically, any scenario which could be expressed in a 3D form is conceivable. It is only a matter of adapting the underlying CGA scripts to the given purpose. While the examples only demonstrate aspects of physical form (albeit with colour codes for old and new buildings at the first two examples), the model might also be rendered to visualise other concepts of urban design, such as land use, floor area ratios, figure/ground and so on. As for enumerations, lengths, areas and volumes may be reported as data outputs, such data might be reported quantitatively along with the 3D model, possibly for parallel scenarios for comparison.

Ultimately, the power of the approach is that everything is kept in one single parametric model which is used across all scales, levels of detail and forms of representation. This allows for substantial testing, variation and scenario building, all of which may be evaluated with respect to all of the above. As collaborative urban design processes are iterative and must negotiate multiple planning and stakeholder concerns, this represents a powerful tool for the analysis, visualisation and reporting of practically any space-related question pertaining to this process.

In practice, variations to the parametric model may happen in a number of ways, some of which require preparation, while others may happen interactively, or live, that is, on request from stakeholders in participatory design workshops like the one shown in Figure 4. First, the scripts themselves may be altered or expanded in order to generate different or more diverse design scenarios. Second, the data input, that is, the constraint map, may be altered in order for the scripts to change their behaviour. Changing the scripts as well as the input data is likely to take more time than is feasible in the course of a live session and should, apart from very minor adjustments, be prepared before workshops or between multiple workshops.

Third, scripts may be modified by changing attribute values (visualised in the software as sliders). And finally, different sets of attribute value settings may be saved as scenarios for subsequent comparison and evaluation. Setting attribute values and saving different scenarios may be done live to visualise the effects of different design changes. While operating the sliders for complex scripts (with many attributes) may require a professional operator, this may nonetheless happen on the immediate request of non-professionals. If needed, simpler scripts (with less versatility) may be operated by non-professionals, based on a brief introduction. Detailed script is provided in the supplementary material provided as an annex to this manuscript see ( supplemental material).

Conclusion

Post-conflict reconstruction in the MENA region, one that addresses aspects of sustainability and liveability, may be guided by a bidirectional planning and design approach, combining top-down concerns for sustainable (re-)development and bottom-up concerns for stakeholder involvement. Such an approach may be facilitated through parametric urban design in the form of a parametric urban model. With such a model, both architectural aspects of physical form and space as well as urban planning aspects of land use, density and so on may be tested and visualised.

As the parametric urban model is a single model within a single software, it can be dynamic and reflect all relevant scenarios across all relevant scales and levels of detail in real time. As such, it represents a unique medium for the deliberation and negotiation of multiple planning and stakeholder concerns throughout the urban development process. Because it can output both 3D visualisations and data reports, both types of output are mutually enriched and may address the needs of both professionals and non-architect stakeholders.

In this article, an example was given of the application of the parametric urban model to an urban district in the Iraqi city of Fallujah. The model was demonstrated at three scales, the district, neighbourhood and street/building scale. Temporal scenarios of urban redevelopment as well as an example of the visual examination of shading were demonstrated through parametric variation of the model. In the examples, also different levels of detail were demonstrated, displayed through parametric variation.

Supplemental Material