Form-finding or form-improvement?

Questo dilemma nasce dalla sperimentazione di algoritmi genetici per l’ottimizzazione strutturale di forme libere e gusci. Essendo configurazioni spaziali resistenti per forma non è possibile, per questioni tipologiche, scindere la forma dal comportamento strutturale durante la concezione e lo sviluppo di un progetto.
L’algoritmo genetico è uno strumento versatile e robusto che è in grado di esplorare delle soluzioni ottimizzate all’interno di un dominio definito. Per dominio si intende il campo di casualità entro il quale il codice può cercare delle forme (individui) ed avviare la procedura evolutiva. A seconda dell’esigenza specifica del progettista deve risaltare o l’aspetto concettuale della gestualità formale oppure l’affinamento nel comportamento strutturale della configurazione spaziale di una forma non definita. In altre parole, si può aver pensato, studiato e disegnato una forma, una configurazione spaziale definita e considerata già valida e soddisfacente. Così da un limitato campo di esplorazione, definito appunto dalla forma, si vuole cercare un affinamento nel comportamento strutturale. Al contrario, possono essere noti dei limiti ben più flessibili per la ricerca di forma, come ad esempio la funzione (copertura), il volume disponibile (parallelepipedo), che però non prescindano mai dall’ottimo comportamento strutturale complessivo. Questa è la descrizione dei due approcci opposti ed estremi, ma che utilizzano lo stesso algoritmo genetico, perché l’obiettivo finale rimane comunque l’ottimizzazione e la ricerca del miglior compromesso tra forma e struttura. Solamente si da un diverso peso alle due componenti in base alle intenzioni e necessità del progettista.
Si può concludere la riflessione sottolineando che la compilazione delle righe dell’algoritmo che definiscono il suo dominio (campo d’azione, d’esplorazione), tecnicamente semplici e facili da scrivere, diventa invece una questione concettuale e metodologica fondamentale.
Lascio quindi al progettista l’ardua scelta, in quanto lo strumento è studiato apposta per migliorare la coordinazione e la sincronizzazione tra più software, ma come nessun mezzo progettuale può sostituirsi al compito proprio del progettista.

0 2688


  1. Djordje Spasic

    Very interesting topic.
    It would be very useful if this could be translated into English.
    Google translate, did not do the job correctly.

  2. Hi,
    Thanks for your comment. This is a very old post that I wrote when I first developed a simple Genetic Algorithm in Rhinoscript, in March 2007. At that moment, I was at the beginning of my PhD and I was using this blog to write a kind of research diary or notes. It was called PROG. In Italian, this could refer to programming, but also to project (progetto). All the posts were in Italian as it was just a way to save some thoughts and quick notes.
    However, I have decided to leave this post because it stresses a fundamental issue. I am talking about the difference between experimental tools and numerical tools in structural design. Experimental tools are in fact physical models. While numerical tools can either simulate a form-finding procedure made with models or be used for mathematical optimization. The latter could already be redefined as a process of form-improvement. It does not have to reach any optimum; it could also be terminated to suboptimal solutions. Furthermore, the designer is allowed to follow the optimization process from the beginning to the end, interacting with the algorithm and reasoning on the problem throughout the all procedure.
    In some cases, for instance when we use Genetic Algorithms and other kind of meta-heuristic optimization tools, the optimization process in itself can become more important, from the design point of view, that reaching an optimal solution. It becomes a tool of form-exploration, a computational process of morphogenesis (Computational Morphogenesis).
    Everything starts from a parametric definition of a problem, with boundary conditions, design variables and a solution domain. Then an objective or fitness function is defined. These are fundamental design phases, which will affect the optimization process and allow the designer to focus on the variations of the system (in my opinion the title of the last book by Spuybroek, The architecture of variation, is really stressing the point).
    Let me quote here some parts of more recent texts that I wrote in the last year. They could be of interest on the topic but more mature than the original post. They are just a few paragraphs taken from some articles.
    *** Texts from personal publications
    The approach of architects and engineers after the Second World War was completely different. Works characterized by an elevated spatial complexity, such as the Kresge Auditorium by Saarinen, the BP service station on the Bern-Zurich motorway by Isler, or the bridge over the Basento River by Musmeci, were, in that period, the result of a creative-generative process that indissolubly welded the structural contribution to that of form exploration. At the beginning of the century, not even Gaudì could draw the steeples of his Sagrada Familia without first having studied their mechanical behavior – he had to simulate the basic properties of the stonework through the use of hanging models, therefore shifting the design to the resolution of a “form-finding” problem, which had the aim of searching for the structural optimum through catenary arches.
    It was in fact impossible to separate the representative component of architecture from its conformative core.
    Digital technologies are also radically modifying the work of civil engineers. Numerical calculation techniques like FEM (Finite Element Method) as a whole are replacing experimental structural design and analysis methods. In the same way, physical models for the form-finding of RC shells or tensile structures are no longer being used. The way now is to use mathematical optimization which, on the basis of one or more chosen criteria, takes advantage of the computation power of the computer to interactively search for optimal solutions to a problem from among a series of possible candidates.
    This change is relevant, from the architectural design point of view, for at least three reasons.
    (1) Unlike in classical form-finding, the topology of a structural system no longer needs to be fixed. It can therefore become the object itself of the optimization process, as in the case of the design of the new TAV station in Florence, which was developed by Isozaki and Sasaki on occasion of an international competition in 2003. An immense flat roof is here suspended in the sky from an organic structure, from which both the topology and the final tree-like shape are derived through the use of an extended version of the ESO (Evolutionary Structural Optimization) technique.
    Given an initial spatial configuration, and calculating the Von Mises stresses through FEM analysis, this algorithm interactively removes the inefficient parts of the structure, in this way minimizing the waste of material. In this case, it was also able to add new ones where needed, thus guaranteeing an optimal mechanical behavior to the overall system.
    The façade of the Akutagawa West Side office building, designed by the architect Hiroyuki Futai and the Hiroshi Ohmori research group from Nagoya University, was conceived using ESO. The façade of the Sagrada Familia was also curiously re-designed in this manner as part of a research project coordinated by Jane Burry of the RMIT University – this work had the aim of studying any possible analogies between the results of the topological optimization and the natural forms originally conceived by Gaudì with the hanging models.
    (2) Compared to the projects by Heinz Isler and Frei Otto, optimization also allows the original form-finding concept, literally aimed at the search of the optimal form, to be changed into what can be defined as “form-improvement” – this new process is instead aimed at improving the performances of an already existing spatial configuration, which does not necessarily mean reaching the structural optimum.
    For example, as far as the Kagamigahara crematorium is concerned, no physical model used to obtain the inverse of the tension-only hanging membrane would have been able to translate the idea of the architect Toyo Ito into a structure. Instead, through optimization, the floating RC roof, figuratively inspired by a cloud, was first freely modeled as if it were a sculpture and was then structurally honed through a Sensitivity Analysis (SA).
    Adopting this technique, Mutsuro Sasaki reduced the total strain energy of the roof shell and iteratively modified its curvature. Based on gradient calculation, an SA automates the traditional “trial and error” design method – in this way, Sasaki avoided a slow and repetitive drawing/verification process of the form, which would have required several manual FEM analysis runs.
    This is the strategy that was also used for the Grin Grin Park in Fukuoka and the Kitagata Community Centre in Gifu – another two cases in which the designer was able to consider free-form spatial configurations, sub-optimal from a structural point of view, only by means of an SA.
    From simple resolution instruments, this and other numerical optimization techniques become efficient “form-exploration” tools to support the conceptual phases of architectural design. For this reason, they are also often referred to in scientific literature as “Computational Morphogenesis” strategies.
    One always starts from a well-defined design problem, i.e. one which can clearly be formulated in a parametric way. An optimization strategy is then added. It takes on the role of guide in the study and evaluation process of the architectural form. GAs, for instance, are meta-heuristic optimization techniques which, being inspired by the principle of natural evolution, generate entire “populations” of design solutions (in this case, free-form spatial configurations) among which only the best are “selected” and “recombined” iteratively among each other. In our example, the GA metaphorically allowed those NURBS surfaces which structurally presented a low mean value of vertical displacements to “survive”.
    (3) The last fundamental aspect of optimization is that it is not just limited to resolving questions of a static nature, which instead is an intrinsic characteristic of form-finding based on physical models. Techniques like GAs can be used in all those cases in which an architectural performance can be formulated through a mathematical function and, technically speaking, it can therefore be “minimized”.

Trackbacks for this post

  1. […] tutti è il problema del ‘dominio’, per il quale faccio riferimento ad un articolo precedente “Form finding or form improving?”, nel quale avevo già riflettuto a riguardo. Per capire di che questione si tratta è bene spiegare […]

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.