MESH aims to provide a single interface bringing together parametric design tools in order to meet the operational needs of spatial planners and urban designers. It tackles the complexity of urban morphologies through the prism of environmental quality.
MESH explores new methods of parametric design to assess, compare, and improve urban forms and optimise them against performance indicators: energy consumption and requirements, comfort (thermal, visual, acoustic). It stands at the interface between researchers, designers and decision-makers, contributing to the methods used to analyse urban projects by the development of algorithms that generate, assess and optimise morphologies. Using a multiscale approach − neighbourhood, public space, block, building − MESH is able to adapt to the different phases of the design process and act as a scientific and technical resource for the actors responsible for implementing projects.
In the quest to control the process of urban sprawl and achieve European CO2 emission targets, new development operations seek to increase density and enhance connections. Their challeng is to combine density, comfort, attractiveness, mixity, proximity and energy efficiency.
Urban design choices dictate a neighbourhood’s capacity to reconcile all these factors, and govern the effective and sustainable use of resources. However, while certain instruments such as sunlight imaging are becoming standard, there are very few multi-criterion tools for the assessing the environmental performance – both quantitative and qualitative – of urban forms in the design phase. Moreover, projects receive an environmental assessment after completion, to certify a certain level of performance, and very few tools can help to directly modify and improve projects at the 3D modelling stage.
The recent development of parametric modelling, simulation and analysis tools constitutes a real opportunity to improve the operational processes of spatial development.
Ivry Confluences, one of the biggest development operations in the Paris region, is MESH’s first testing ground.
This ZAC (urban development zone) is set to be the arena for 1.3 million m² of programmes, consisting of 50% economic activities, 40% housing, and 10% amenities.
MESH ambitionne de développer un processus de conception itératif, permettant d'adapter les hypothèses morphologiques des formes urbaines, de manière rétroactive, en accord avec les données de programmation et le profil de qualité environnemental envisagés.
Tant la pertinence de l’évaluation que la réactivité de l’outil sont gages de son utilisation dans un cadre opérationnel. Le projet se pose alors les défis suivants :
MESH’s aim is to develop an iterative design process that will make it possible to adapt morphological hypotheses on urban forms, through a feedback process, based on the programme data and the intended environmental quality profile.
The utility of the tool in an operational framework depends both on the accuracy of its assessments and its responsiveness. The project thus poses the following challenges:
Iterative Genetic Process of Environmental Performance Design-Optimisation - ©MESH
In the MESH project, environmental quality assessment depends on a set of quantitative indicators. They are drawn from a state-of-the-art based on the cross-analysis of three inputs: geographical and climatic features, typo-morphology of the built fabric, and occupant comfort.
To move beyond this analytical approach, MESH undertake a systemic examination which aims to explore the assessments transversally, describe them, and subsequently quantify them. This may entail, for example, identifying and qualifying the interactions in the assessment methods, any domino effects in order to correct distortions, then incorporating the domino effects… essentially accounting for the complexity involved in calculating the environmental performance of a project’s urban forms.
The paradigm of multi-criterion decision support is used as a critical tool and as an alternative to the approach of optimising a single, global function. This paradigm, generated by decision theory, relates to the “dashboard” principle, in that it builds a system of performance indicators.
Representation of Environmental Indicators by Themes - ©MESH
This system of indicators acts as a basis for the development of assessment algorithms, beginning with the relative quantification of performance gains (energy, thermal, acoustic, etc.), before possibly moving towards a quantification of the intrinsic performance of an urban form.
Some of the assessment tools produced in this research are described in greater detail below.
NB: The assessment routines built into MESH partly include tools developed under OpenSource licence, in particular the functions library “Ladybug Tools”, an addition to the Grasshopper plugin on Rhinoceros 3D, developed by Mostapha Sadeghipour Roudsari’s team.
MESH is a single platform that combines generation, assessment and optimisation algorithms. Generation algorithms are used to produce variants on the initial urban form based on geometric rules defined with the project’s urbanists and architects, which incorporate the site’s constructibility and programming constraints. Each variant is automatically analysed against the selected criteria.
Evolutionary multi-criterion optimisation algorithms are used to activate a feedback loop: performance acts on form. In this way, the algorithm steers the process of generating variants, moving towards optimum solutions.
Mono-criterion optimisation on 5 indicators selected for Block 4G (Ivry Confluences) ©MESH
Multi-criterion optimisation on 6 indicators selected for Block 3I/3N (Ivry Confluences) Number of hours of sunlight, access to sky, Overlook, Façade irradiation level, Views over Confluence Park, views over the Seine. ©MESH
NB: The optimisation process at this stage users the evolutionary mono/multi-criterion algorithm libraries of the Galapagos (Mono criterion genetic algorithm, Rutten, 2010) and Octopus plugins (SPEA-2, elitist method based on Pareto, Ziztler 2001, and implemented in Grasshopper by Vierlinger, 2014).
The Goat 3.0 (implementation on Grasshopper of the opensource library NLOpt, Rechenraum & Feasible) and Opossum (optimization solver with surrogate models, Wortmann, 2017, based on the RBFOpt library) plugins are currently being explored.
The Design Explorer tool, developed by CORE Studio/ Thornton Tomasetti, makes it possible to visualise the individuals generated by the process. Using the selection cursors, one can quickly filter the urban morphologies that meet the performance criteria determined with the design team and the decision-makers.