Ethiopia’s 2,800-year-old walls could have carried 16 floors, a stress model finds

Grat Be’al Gibri, a palace at Yeha in northern Ethiopia built about 2,800 years ago, has walls that could carry as many as 16 floors, according to a structural analysis published in the journal Heritage in July 2026. The earlier virtual reconstruction of the building assumed eight.

The number describes what the masonry could bear in a computer model, not what the builders raised. The authors, Martin Drieschner of Brandenburg University of Technology Cottbus-Senftenberg and Mike Schnelle of the German Archaeological Institute, say so directly, and the caveat shapes how the result should be read.

Finite Element Models of Two Wall Sections

The palace dates to roughly 800 BC and is described in the HeritageDaily summary of the paper as the largest known palace-like structure of the early first millennium BC in South Arabia and East Africa. Only the ground floor and podium portions survive, so every claim about its height rests on reconstruction from comparisons such as rock and wall paintings, not on standing walls.

The study, titled as a numerical investigation of timber-reinforced walls with uncertain material parameters, is available on the MDPI Heritage page. Drieschner and Schnelle built three-dimensional finite element models of two representative pieces of the palace: an external wall corner and an internal wall containing a doorway. The walls are phonolite rubble set in clay mortar, laced with horizontal wooden beams of African olive and Cordia africana, and the ground-floor walls are 1.9 metres thick.

Because nobody knows the exact strength of 2,800-year-old stone, clay and timber, the team ran many combinations of material properties rather than one best guess. The result, as reported by HeritageDaily, is that variation in the timber mattered little, while the properties of the stone-and-clay masonry mattered most, with tensile failure in the clay-mortared rubble as the limiting factor.

Sixteen Storeys, Nineteen Storeys and the Safety Factor

Two numbers circulate for this result, and they do not conflict. The German Archaeological Institute’s press release of 9 September 2026 says the wall system would have allowed at least 19 floors, with the project’s leaders listed as Arnulf Hausleiter, Schnelle and Drieschner. The university’s own release gives the same at-least-19 figure and quotes Drieschner on the fascination of applying modern engineering methods one-to-one to a 2,800-year-old construction.

The 16-floor figure appears to be the cautious version. Archaeology Magazine’s summary reports that under worst-case combinations of heavy load and weak material the wall corner produced a load factor of 19 and the doorway wall a factor of 23, and that applying the study’s safety factor brought both down to the 16-floor estimate. It adds that in the best material case single wall sections tolerated loads equivalent to roughly 48 and 56 floors, figures the authors do not treat as a building height.

Against the earlier reconstruction, which proposed five regular floors plus three recessed upper floors, eight storeys sit comfortably inside the walls’ capacity, and sixteen is twice that.

A Fire, Not Structural Failure

The result also bears on how the palace ended. The MDPI paper concludes that an exceptional load must have caused the system to fail, and it points to a devastating fire in antiquity. In the authors’ reading, the walls had reserve capacity to spare, so ordinary overloading cannot explain the collapse.

The paper’s own conclusion is measured: the simulations show that a multi-storey building was feasible with the present wall constructions, and it adds that it is uncertain whether the ancient builders pushed the structural potential to its limits, while stressing that they were able to develop highly efficient building structures.

Missing Floors and Thinner Upper Walls

The analysis used the same wall thickness for upper floors as for the ground floor, according to the Archaeology Magazine summary, whereas a real palace would probably have thinned its upper walls, which would add capacity. In the other direction, the model has not yet included floor structures or the building’s full geometry. The DAI release says those will enter the structural calculations in later phases of the project, and HeritageDaily lists modelling the complete structure and testing alternative timber arrangements as future work.

Until that work appears, the safest reading is the narrow one the authors give: the surviving wall system was capable of carrying substantially more weight than the reconstructed eight-storey palace would have imposed. The paper does not say that Grat Be’al Gibri stood 16 floors high. What survives is the lower part of a complex about 60 metres by 60 metres, with ground-floor walls 1.9 metres thick, podium walls 2.2 metres thick and a podium about six metres high, and entrance pillars of roughly 10 metres that the paper says imply at least three storeys where the gateway met the main block.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview