Long-lived Structures and Materials: Investigating Material and Construction Methods for Dry-stacked Corbelled Structural Systems

This research investigates dry-stacked, corbelled, compressive structures built up with elements that do not require mortar, fasteners, reinforcement or formwork. The relevance of these corbelled structures with regard to sustainability is twofold. Firstly, some of the oldest known human-made structures fall into this category, eliminating the need for frequent construction. Secondly, the lack of fasteners or mortar makes recycling the building materials much easier. 

Corbelled structures are also well suited for assembly by robots, since they require precise placement in complex geometries yet eliminate the need for handling multiple types of materials or fasteners. Corbelled structures are well suited for robotics response teams to disasters, for example, where a lot of loose rock material might be available. As a specific example of structures, we are focused on expanding Rafailidis’ work on corbelled structures (Rafailidis, 2014) that achieve maximum overhang (Paterson & Zwick, 2009) and have unexpected and complex geometries. These novel corbelled geometries have been tested so far with model scale materials like XPS foam, wood and plaster.

The SMART exploratory grant triggered an in-kind sponsorship by a global concrete manufacturer to fabricate concrete blocks at full scale to address three primary issues. 

How can these novel corbelled typologies found empirically, at model-scale, be scaled-up to full, architectural scale? 

The team constructed corbelled structures with a meaningful overhang and an increasing amount of courses. The optimum compromise between large block size, weight and handleability is therefore critical. The optimum size and weight was developed, defining the size of blocks dependent on weight suitable for a human and robotic arm respectively.

Different block geometries tend to lead to distinct corbelling typologies. Multiple block geometries were tested with hundreds of blocks for each test. Our research ultimately chose and focused on two block geometries:

• Rectangular block geometries allow for a maximum freedom in the overall arrangement of blocks
• Sawtooth block geometries are restricted in the aggregation possibilities, but allow for a self aligning feature. The self aligning feature is relevant for future research in robotic assembly. It allows for larger tolerances in the accuracy of the positioning arms. It also allows for more robust structures that could ‘fall back’ in place after an impact.

How can the novel structures and their respective stacking strategies be translated into construction-grade materials like concrete and brick? 

Multiple casting and forming techniques were explored to fabricate the two chosen block geometries. The challenge was to create formwork and fabrication techniques that are cost effective and that can create large number of blocks in short time with minimized manual labor. A successful technique was developed.

Another critical question that emerged from the transition into concrete was the maximum difference in density that could be achieved. Extensive test series were undertaken with novel concrete mixes and additives using the unique breath of knowledge from the industrial partner. The structural capacity of the mixes was tested partly at the our partner's research headquarters in Switzerland and at UB’s North campus at the department of Civil, Structural and Environmental Engineering. We explored aggregates, admixtures and fiber components of the mix designs.   We also achieved successful concrete mix designs with a structural performance in the range adequate for our assemblies, including mix design densities from 800kg/m3 to 4500kg/m3.

Another series of tests explored the potential to shift the center of gravity within a single concrete block. The bonding between the lightweight part and the heavy part within a block proved to be challenging. Ultimately we achieved a successful structural bonding between the distinct densities within a block and were able to shift the center of gravity to 0.64 of the overall length of the block. We produced over 250 units of different geometries and densities to be used for full scale testing. A first assembled prototype spans 12ft and provides a clear height of 7.5ft by strictly dry stacking.

How can the complex corbelled geometries accommodate unforeseen conditions on site and be made robust through robotic automation?

We implemented an optimal stacking strategy for robots that achieves the following criteria:

• Implements a vision algorithm to count blocks
• Computes the optimal simple stack
• Executes optimal plan.

The project also resulted in developed a program to generate new, stable 3D stacking plans that uses rigid body simulator and a structure search.

For additional information on this project, please contact either Dr. Georg Rafailidis (grafaili@buffalo.edu) or Dr. Nils Napp (nnapp@buffalo.edu).