Skip to main content

CROSS Safety Report

Incorrect modelling of a cantilever

Report ID: 1144 Published: 17 November 2022 Region: CROSS-UK


A steel framed building was incorrectly modelled and designed resulting in excessive deflection of part of the frame even before the frame was loaded with the self-weight of the building. The modelling error occurred due to an incorrect support condition being applied.

Key Learning Outcomes

For clients:

  • Clients should check that they are appointing suitably qualified and experienced practitioners

For civil and structural design engineers:

  • Engineering analysis software should only be used by suitably trained and competent engineers
  • Understand how the structure should behave before modelling
  • Ensure the software models the structural elements as anticipated
  • Apply ‘sense checking’ through simplified traditional methods
  • Validate all analysis and design outputs
  • Single-person practitioners should consider arrangements for validation

Full Report

Find out more about the Full Report

The Full Report below has been submitted to CROSS and describes the reporter’s experience. The text has been edited for clarity and to ensure anonymity and confidentiality by removing any identifiable details. If you would like to know more about our secure reporting process or submit a report yourself, please visit the reporting to CROSS-UK page.


A reporter was asked to review the design of a steel frame structure which showed excessive deflection during construction. The structure formed a small extension of an existing building. The frame comprised beams and columns but with one column omitted in a corner due to site constraints. An edge beam into the corner was supported by a cantilever beam supported off a column. During the construction phase, this cantilever beam deflected excessively before even the full dead load had been applied to the structure.

A review of the design output revealed that the free end of the cantilever beam had been incorrectly modelled with a fixed nodal support condition applied. Thus, instead of acting as a cantilever, the member was in fact modelled as a fixed-ended beam. The nodal restraint condition had been user applied. Due to the restraint condition, neither the deflection of the beam nor the design forces (shear, bending moment) were correct. The software used provided a graphical indication of restraint positions and types so the error could have been detected by a review of the structure at the modelling stage. Furthermore, a review of the analysis results such as deflected shape, shear force or bending moment would have revealed an issue which could and should have been investigated. The steel member design was carried out using the software's integrated design module.

A design check based upon a correctly modelled frame showed that steelwork and associated connections were inadequate. The reporter confirms the error could have led to a structural failure with obvious safety concerns.

the error could have led to a structural failure with obvious safety concerns

The reporter considers that this problem arose due either to a lack of understanding of the software or of the implications of nodal restraint conditions, compounded by an apparent lack of review of the analysis outputs and steel design. At several stages in the process, the reporter says, the problem could have been identified and rectified before the structure was fabricated and erected. It was the reporter’s view, that the design engineer had conducted no validation of the model and being a single-person practitioner, there was likely no in-house review opportunity available.

The reporter goes on to share that they have encountered similar poor understanding of restraint conditions on other structural models reviewed. They consider this can happen when the engineer lacks experience in modelling. The reporter has seen similar mistakes made by new, graduate and very experienced engineers.

The reporter is concerned about the apparent lack of model validation undertaken. They contend that mistakes with restraint conditions can be detected before the design stage by carrying out a review of the analysis outputs. Even just reviewing the deflected shape of the structure can reveal modelling errors. It is also clear to the reporter that checking procedures are not being put in place or applied frequently enough.

reviewing the deflected shape of the structure can reveal modelling errors

The reporter concludes that analysis and design software is an invaluable tool, but it remains a tool that needs to be applied with caution and does not replace the need for the engineer to validate the results and design. The software can only analyse the structure as modelled, so if the model itself is wrong, the output from the software is also wrong, but not through the fault of the software.

Expert Panel Comments

Find out more about the Expert Panels

Expert Panels comment on the reports we receive. They use their experience to help you understand what can be learned from the reports. If you would like to know more, please visit the CROSS-UK Expert Panels page.

The apparent lack of understanding of both how the structure should behave and the need for design checks is a great concern as is any over-reliance on software. Furthermore, in this case, it was the modelling of a cantilever element that led to the error. Experienced engineers will understand that cantilevers are often safety critical because of their lack of redundancy.

A designer should not rely on software they are not competent to use. To be competent to use structural analysis and design software, designers should understand structural behaviours. Just as the reporter contends, observing deflected shapes, moment diagrams, support reactions and other data will readily lead a competent designer to spot any modelling errors. As with this example, zero deflection at the end of a cantilever really ought to flag to all designers a significant problem with the model. An understanding of structural behaviours is fundamental knowledge for any structural designer. Many argue that teaching graduates to understand intuitively how structures will behave enables them to validate their modelling and leaves them less likely to place inappropriate reliance on software. The Institution of Structural Engineers has published Text No.13 Behaviour of beams and two-dimensional frames in their Essential Knowledge series which provides an introduction to a qualitative approach to the understanding of structural behaviour.

leaves them less likely to place inappropriate reliance on software

If the model does not match what will be built, then it is probably useless but worst still may impact safety. Modelling experience does need to be built up so that the designer understands how for example, decisions concerning applied restraints and releases influence the outputs. Connections particularly need very careful consideration; there are usually no pins or fixed connections, just something in between and so there is an art to modelling to arrive at deflections, displacements and forces that are representative of the performance of the structure to be built. 

there is an art to modelling to arrive at deflections, displacements and forces

CROSS report 886 - Unconservative design of flat slab due to software modelling issues gives another example of an element that was modelled incorrectly leading to an insufficient structure being built. The designer must also be alert to the limitations of the software such that they do not use it outside of its validated capabilities. Designers should also be mindful that elements should be designed for all the forces that the model assumes for equilibrium, for example, the model may include torsional stiffness at connections which must therefore be followed through in the element design checks.

Engineers should always know how they expect the structure to act before creating the model, and then ensure that it does act so. If the engineer’s expectations do not match with the model, then they know that there is an error somewhere, either in the model, their expectations, or the software. The error is usually in the first two. As part of ‘sense checking’ simplified traditional methods should also be used to validate predictions to the right order of magnitude. A first check after running any analysis could be to ensure that the reactions agree with the applied loads and that the reaction locations are as expected. Some engineers will annotate graphical outputs from the analysis with reasoning as to why the results are as expected including commentary upon reactions, deflected shapes, moments and the like.

As part of ‘sense checking’ simplified traditional methods should also be used

Checking and validation

Outputs from analysis and design software must be validated; checking as per the firm’s quality management requirements should be followed. Checking independent from the design team will be required in some circumstances. Single-handed practitioners should consider appropriate arrangements for validation which will likely require that some checking is undertaken by others. Modelling is an activity which requires careful consideration by a single-handed practitioner as to how they can be assured of the adequacy of their design. The application of alternative hand methods could be considered as part of a validation regime. CROSS Report 1073 Concern over modelling of concrete frame building for construction stage similarly showed checking and validation of structural modelling that was insufficient.

Finally, it is important to note that clients should be aware of the opportunities and risks in design processes. Clients could consider the consultant’s validation proposals when appointing designers but should at least, check that they are appointing suitably qualified and experienced practitioners.

Submit a report

Your report will make a difference. It will help to create positive change and improve safety.

Our secure and confidential safety reporting system gives professionals the opportunity to share their experiences to help others.