Sewn fabric moulds used to cast concrete beams with far less material

Engineers at the Massachusetts Institute of Technology (MIT) have cast a 3-metre concrete beam in a sewn fabric mould, producing an optimised form that needs far less material than a standard beam. Compared with a conventional beam designed for the same load, it used about one-third of the concrete and half of the steel reinforcement. The study appears in the Journal of Building Engineering.

Why optimised shapes are hard to build

Topology optimisation is a computational method that places material only where it contributes to structural performance. The resulting shapes are often intricate, and pouring them into conventional rigid moulds is difficult or expensive. The few examples built so far have mostly depended on concrete 3D printers or machine-carved moulds, equipment that is scarce in regions where much new construction is expected. The world's building stock is projected to nearly double by 2050, with about 85 per cent of that growth in developing countries.

Fabric formwork, in which concrete is poured into a textile mould, has been studied for years. Its drawback is that the cloth deforms under the weight of wet concrete, so the final shape is hard to predict. Josephine Carstensen, senior author and Gilbert W. Winslow (1937) Career Development Professor in Civil Engineering, describes the team's approach as treating that flexibility as a route to optimised forms rather than only a problem to control.

Combining two design methods

The team paired two techniques. Graphic statics, which generates structural layouts almost instantly, was used to define the path of the steel reinforcement so that the bar carries an equal share of the load along its length, keeping it simple enough for builders to fabricate. Topology optimisation, which is slower but can place material freely, determined where concrete was needed and where it could be removed.

The resulting beam has a top section, narrow angled legs and a thin bottom section housing the steel. An automated process converted the 3D design into flat fabric pieces that can be cut and sewn, allowing for the way the cloth deforms under load. A simple timber frame supported the finished mould.

Casting, shape accuracy and testing

Lead author Mark Hellrich said the main practical challenge was the concrete mix. Woven cotton lets water seep out, and a runny mix presses harder on the fabric and distorts the shape. The team used a moderately wet, sand-rich mix with only small stones so it could flow through the narrow legs and thin base. It was packed by hand, without a vibrator, by prodding each layer and tapping and kneading the fabric.

All three cast beams were free of wrinkles and creases. The top surface stayed close to the design, while the bottom sagged slightly and the angled legs folded a little inward. Moulds made from 36, 56 and 76 fabric pieces were compared using 3D scans. More seams gave finer local control but did not necessarily improve overall accuracy, since small cutting and assembly errors can accumulate. The 56-piece mould gave the best balance.

One beam was load-tested with 36 water-filled buckets, up to about 30 per cent of its design load, to reflect everyday service conditions. Measured deflection at the centre was within about 2 per cent of the computer model's prediction.

Cost and where it could be used

The cost comparison depended on location. In the Global South, where materials are relatively expensive compared with labour, the fabric versions were about 10 to 20 per cent cheaper than a conventional beam. In the Global North, higher sewing labour costs made them about 16 to 55 per cent more expensive. The approach therefore looks most suitable for one-off or small-batch components, such as exposed beams, long-span roof framing, pedestrian structures and landscape elements, rather than repetitive rectangular beams.

Cement production accounts for roughly 8 per cent of human-caused carbon emissions, so reducing concrete demand is the main climate benefit the authors point to.

Limitations and next steps

The authors describe the 3-metre beam as a proof of concept, longer than the 1.2-metre beam from their earlier project. Scaling up is not a matter of enlarging the design, because self-weight and mould pressure change with length, so longer beams would need fresh designs. Only one beam was load-tested, and only to a fraction of its design load. The next step, according to Carstensen, is to test multiple beams to failure.

The paper's co-authors are Jackson Jewett, Abhinit Sati, Edward Chien and Emily Whiting, the latter three from Boston University's Department of Computer Science. The work was partly supported by the US National Science Foundation.

Source: Fabric molds could bring material-saving concrete beams within reach (MIT News: research, 7 November 2026).

Image: AI-generated illustration.