Research Reviews

Cellular Steel Frames: Why the Whole Frame Matters

/ Ali Eskandari

A Civil Insight review of peer-reviewed research, based on the full published paper.

The cellular steel portal-frame specimen S02 in its laboratory loading rig, with instruments, hinged bases and lateral restraints identified.

Circular openings in steel members can help coordinate structure and building services, but their structural behaviour depends on the surrounding frame. A study published on 28 September 2026 by Waleed M. Sayed, Ali Hammad and Sherif K. Hassan examines that interaction through two portal-frame tests and finite element analysis. Its reported capacity gain needs to be read with the specimen geometry and laboratory restraints.

Featured photograph: Cellular-frame specimen S02 in the laboratory loading rig. Source: Sayed, Hammad and Hassan, Scientific Reports (2026), Figure 8b (CC BY 4.0).

The research question and test arrangement

The Scientific Reports paper investigates how column stiffness influences a frame whose response is governed mainly by its beam. Researchers at Ain Shams University, Egypt, tested one solid-web frame, S01, and one cellular frame, S02, under a concentrated vertical load at beam mid-span. Both frames had a 2,000 mm span and 1,730 mm height, with hinged bases and rigid beam–column connections.

Both were fabricated from IPE120 parent sections. In the cellular specimen, cutting and rejoining the members increased the beam depth to 156 mm and column depth to approximately 166 mm. S02 also included additional local reinforcement. Openings were filled at selected critical locations, including the load application region. This was a comparison of two fabricated configurations, not simply a test of adding holes to an unchanged section.

The apparatus restrained out-of-plane movement, deliberately excluding lateral-torsional instability. Load cells, displacement instruments and strain gauges recorded the response. Ten material coupons supported the steel-property measurements; there was only one frame of each type.

What the measured capacity gain means

The solid-web frame reached an ultimate load of 61.80 kN at 17.48 mm mid-span deflection. The cellular frame reached 101.28 kN at 33.08 mm. The increase in peak load was approximately 64% for these specimens under this loading arrangement.

At the authors’ comparison deflection of 5.56 mm, equivalent to span/360, the corresponding loads were 31.89 kN and 47.30 kN: approximately a 48% increase. This is the study’s serviceability comparison criterion, drawn from its US-code reference; it is not a universal Australian project limit.

Failure changed as well as load capacity. S01 developed a plastic hinge at beam mid-span, while S02 failed through web-post buckling in the beam. Web posts are the strips of steel between adjacent openings. Higher frame resistance therefore remained bounded by local beam behaviour.

Two close-up study photographs show local distortion in the steel web between openings of the tested cellular beam.
Web-post buckling observed in cellular-frame specimen S02. Source: Sayed, Hammad and Hassan, Scientific Reports (2026), Figure 11 (CC BY 4.0).

What the numerical model adds

The authors calibrated nonlinear ANSYS models against the tests, then examined 16 configurations across four parent-section sizes. The simulated variations included column geometry, the relative stiffness of beams and columns, and full-depth stiffeners.

For the two calibration specimens, predicted peak loads were 66.88 kN and 109.16 kN, approximately 8% above the measured values. That agreement supports further investigation within the model’s scope; it does not independently validate every larger simulated frame.

The numerical work suggests that changing column stiffness can improve load distribution and overall response, while beam instability eventually governs. Civil Insight’s interpretation is that cellular members should be assessed as part of a frame and its restraints. An isolated member-capacity comparison cannot fully describe that system behaviour.

How reliable and transferable is the evidence?

The paper clearly documents specimen dimensions, material testing, supports and instrumentation. It also acknowledges that the two-frame programme was intended mainly to validate modelling, rather than provide statistically conclusive experimental evidence. There are no repeat frame tests from which to estimate construction variability.

Several changes separate S01 and S02: section depth, web geometry and reinforcement. The experimental gain cannot be allocated to openings alone. Some of the stiffeners were included to manage the laboratory point-load arrangement and are described by the authors as unrepresentative of standard practice.

The model used assumed geometric imperfections because the actual specimen imperfections were not measured. Its wider results therefore remain dependent on modelling assumptions as well as the tested evidence. The underlying datasets are available from the corresponding author on reasonable request.

The investigated load was monotonic and concentrated at mid-span. The results do not establish performance under wind, seismic cycles, fatigue, fire or unrestrained lateral instability. Different load distributions, support conditions and opening geometries need their own assessment.

Implications for architects and construction teams

For architects and services designers, cellular members offer a coordination question: where can service routes fit without compromising the intended structural behaviour? The study does not test installed services, penetrations added later or a complete floor system. Early agreement on opening locations and the restraint strategy is more useful than assuming that every opening is available for a future service change.

The research also contains no project cost, embodied-carbon or whole-life comparison. A higher laboratory peak load cannot, by itself, establish a cheaper or lower-carbon building. For Australian practice, the useful transfer is the need to check the complete structural arrangement, fabrication details and applicable design requirements, rather than adopt the reported percentage as a design allowance.

Original paper and publication checks

Sayed, W.M.; Hammad, A.; Hassan, S.K. (2026). “Experimental and numerical study of two-hinged steel frames with cellular members.” Scientific Reports, 16, article 30036. Published and version of record: 28 September 2026. DOI: 10.1038/s41598-026-67704-6. Original full text · Full published PDF.

The publisher identifies this as a research Article in a peer-reviewed journal. Publisher and DOI records were checked on 16 November 2026; no correction or retraction notice was found. This article is Civil Insight’s editorial review of the study.

Join the discussion

No sign-in required. Your email address will not be published. Comments may be held for moderation.