Cement production is responsible for 8% of the anthropogenic CO2 emissions [1]. Reducing this share will be a strenuous task as the global population and urbanization rates are increasing and, therefore, the need for concrete will remain high [2], [3]. Simultaneously, despite the alarming statistic, a vast majority of demolished concrete buildings are structurally sound [4]. Therefore, this thesis aspires to push beyond the abundance of new concrete and outlines a design workflow for using reclaimed concrete. To enhance the material savings, the workflow focuses on double-curved, compression-only structures that benefit from the principles of structural geometry.

To implement the workflow, a novel stacking algorithm is designed using COMPAS [5], an open-source Python framework. It positions quadrilateral voussoirs on a funicular geometry’s thrust surface in a sequential one-by-one manner. The placement is guided by three global design principles. First, a pair of opposing voussoir sides are aligned parallel to the force flow while the remaining pair, which represents the load-transferring faces, is aligned perpendicular to it. This is done to prevent sliding failure. Second, the thrust surface is confined within the middle third of the voussoir to prevent the formation of hinges and cracks. Third, voussoirs are cut into hexagonal shapes and staggering is ensured among neighbouring voussoirs to create an interlocking geometry. Once a structure is completed, its stability is ensured by finding a compression-only equilibrium using the Rigid Block Equilibrium method [6].

The algorithm is then studied using various stocks with different voussoir dimensions of both regular and irregular side lengths, various deployment strategies that enable the user to customize the launch of the algorithm, and different thrust surface shapes. It is primarily assessed what fraction of the initial concrete volume can be retained in the structure. It was found that on average 47% of volume per voussoir is retained and that voussoirs of regular dimensions retain approximately 5% more of the initial volume. Minor differences exist in the results among different thrust surfaces with shapes of more uniform force flow retaining more concrete. Furthermore, smaller voussoirs mostly sized 0.7x0.7x0.2m required 20% less volume to construct a shell than a stock of 1.0x1.0x0.2m voussoirs. At the same time, the smaller stock required almost double the amount of voussoirs, thus, indicating competing goals that require the decision-making of the designer.

Next, high-level waste minimization was performed by checking if offcuts could be used to create other voussoirs from the remaining structure or if they could be returned to the stock for future placement. A pessimistic and an optimistic scenario were studied and it was found that 70-85% of the initial voussoir volume can be efficiently retained within the shell and the stock.

These results served as the basis for reflecting on how the stacking algorithm influences the suggested workflow by bringing design and fabrication-specific constraints to an earlier design phase. It is described that to improve the retained volume the designer can, first, alter their way of deploying the algorithm, second, modify the input geometry, third, custom select what stock is used for placement, and, fourth, inform the contractor of the level of detail to which a building should be deconstructed and post-processed. Thus, a designer is brought closer to the construction process, and incentives for a more collaborative cross-disciplinary workflow are outlined.

Finally, suggestions are made on how to improve the functionality and efficiency of the algorithm. The primary conclusion made is to expand the criteria based on which a voussoir is selected for placement. An emphasis is directed towards using criteria such as distance-to-staggering, local radius of curvature, local force flow, and dimensions of the surrounding voussoirs to limit the offcuts that will be created to ensure complete tessellation.

Overall, this study provides a workflow that is enabled by a stacking algorithm. It outlines how to design double-curved, compression-only structures and how their design can influence the entire workflow.