Structural Efficiency: current buildings implement 25 times the optimal amount of material

A recently published research paper demonstrates how current building structures hardly achieve a 4% structural efficiency, in other words they use 25 times more material than the minimum theoretically required.

Existing methods to assess structural design have been using indirect magnitudes -self-weight, embodied carbon- hence the structural capacity inherent to the material got lost in the process.

Instead, the novel Structural Efficiency Classification System -or S.E.C.S for short- measures “Stress Volume” i.e. the capacity of the material to translate forces over distances; hence it works consistently across materials.

Also, any building is compared to a theoretical optimal model resolving the very same structural problem, and the ratio between the stress volume of the theoretical optimal and the real design becomes the Structural Efficiency or SE.

The method is public and its theoretical foundations described in the paper:
Mena Q (2020) Towards a Structural Efficiency Classification System
(Structures, Volume 26, August 2020, Pages 298-310)

In that paper you’ll find:

  • how buildings designed following design guides do not achieve structural efficiencies over 5%
  • it is possible to design floor systems up to 50% structural efficiency: e.g. in steel/concrete, their self-weight is under 40 kg/m2

( As a reference: latest design derived from the direct application of the S.E.C.S. principles is a residential square floor, 4.2m side simply supported at corners and using existing standard materials. Its self-weight is under 15 kg/m2 and its embodied carbon also less than 15 kgCO2e/m2 )

1 Like

This is a very interesting paper @quisco, thanks very much for sharing! I like this: Among all the possible elastic layouts to solve a specific structural problem, the optimal is also the stiffest - if we can get cheap carbon fibres / nanotubes that sequester CO2 (see C2CNT) this could greatly improve structural optimization.

I have a germ of an idea to use Reinforcement Learning to train agents to grow to optimized structural forms applying random MC loading every iteration, and rewarding based on set objectives. S.E.C.S. looks like it provides a good objective for the model.

Hi @will.nash,
Material wise, in highly efficient systems, low working strains (i.e. the unit deformation of the material at its allowable stress) is paramount: light strong materials are not all the same, for example carbon fibres need to lengthen up to 14.3mm/m to reach their allowable stress; while graphene reaches its working capacity at only 0.75mm/m: the same structure manufactured in both materials at 100% efficiency (if that was possible) would have 19 times more deflection in carbon fibre, and could eventually need massive oversizing to control deflections.
For highly efficient building structures with typical serviceability constraints, working strain is more important than self-weight.
Anyway, efficient building structures will be so light I’m not sure it will be worth it as yet: check above the self-weight of our structural floor, made in normal (current) steels and concretes.