Skip to main content

CROSS Safety Report

Modelling of structures

Report ID: 994 Published: 2 August 2022 Region: CROSS-AUS


Overview

The reporter is concerned that some structural engineers assume without adequate verification that their computer modelling is correct and accurate; that they do not understand the limitations or basis for their modelling; or that the model chosen may not be appropriate for that structure.

Key Learning Outcomes

For civil and structural design engineers:

  • When correctly used, structural modelling can provide the engineer with a solid grasp of the upper and lower bounds to any design on which they are working.
  • Always ask the following (as a minimum) when assessing your own work or checking the work of others:
    • is the software selected for the analysis appropriate for the given application?
    • is the model correct and does it correctly and appropriately represent the structure under analysis?
    • how will the structure resist all loads and where are the load paths?
    • will the construction sequence have an effect on the design?
    • has independent verification and validation been undertaken based on the level of risk?
    • have alternatives been modelled to provide a “sense check” in case original assumptions are not correct?
  • It is important to verify the results from the model and for the more experienced engineer this may be by “gut feel” and a review of the deflected shape produced by the model. For the less experienced, the very minimum should be self-checking by simple hand calculations.

Full Report

Find out more about the Full Report

Our secure and confidential safety reporting system gives professionals the opportunity to share their experiences to help others. If you would like to know more, please visit the reporting to CROSS-US page.

The reporter is concerned that some structural engineers assume without adequate verification that their computer modelling is correct and accurate, that they do not understand the limitations or basis for their modelling, or that the model chosen may not be appropriate for that structure.

The reporter is also concerned that some engineers may not perform any internal checking and verification within their own office, and for larger projects, no adequate external checking and/or validation has been made of their model and resulting design. On several projects in recent years, the reporter is alarmed that the structural drawings provided to them have not been signed as checked internally.

Modelling, if correctly done can give very good indications of how structures might perform both under vertical and horizontal loads, especially using sophisticated computer graphics. However, common sense and experience are also needed as it is important to understand the background of the software and its limitations.

In a real structure, the behaviour of individual elements under load can be complex, depending on the materials used and many other factors. In analysis, idealised models of the frame or structure are developed to simulate how the real structure may behave. However, it is important to remember that they are just mathematical models and engineers (especially those less experienced) should ask themselves the following:

  • is the model correct and does it correctly and appropriately represent the structure under analysis?
  • how will the structure resist all loads and where are the load paths?
  • was a second opinion obtained within their office on the model they have adopted?
  • did they model some alternatives in case the original assumptions are not correct?
  • how did they model large penetrations in floors for stairs and the like?

As an example of incorrect modelling, the reporter has reviewed several projects where deflections of concrete floors were an issue, particularly for buildings designed in the 1970s and 1980s. Based on the modelling, these floors appeared to be satisfactory but based on real-life performance they were not. This highlights that there is no substitute for a critical review of the results.

There is no substitute for a critical review of the results

Deflection of concrete floors has been a vexed question for many years (particularly for reinforced concrete floors) and the reporter notes that deflection limitations in Australian Standard AS 3600 are minimal requirements and may not be appropriate for all projects. The reporter is also concerned that designers do not make use of available published information to establish initial sizes before inputting data into modelling programs. For example, the Guide to Long-Span Concrete Floors published by the Cement and Concrete Association of Australia (CCAA), which is free to download and should be taken into account. The reporter further recommends the Institution of Structural Engineers two-day online course on Understanding structural design which extends the principles developed in the earlier Understanding Structural Behaviour course to more complex, real structures and the all-important skills of approximate analysis for checking computer output and member sizing.

On a multi-storey project reviewed by the reporter, the concrete cores were assumed to be cantilevered from the basement and to resist all the lateral load from wind and earthquake. However, the cores were laterally restrained by the ground floor which in turn was acting as a diaphragm connected back to concrete retaining walls. It is more probable that the cores were cantilevering above the ground floor and not from the basement as assumed. Furthermore, in reality, significant additional horizontal loads would be applied to the retaining walls.

When modelling the effects of lateral load, it is important to understand the role of all vertical elements. For example, it is often not sufficient to ignore the effects of other load-bearing elements (such as precast panels or columns) as they may take some portion of lateral load depending on the relative stiffness of each element. 

The reporter points out that clause 5.2.3 of the Australian Earthquake Standard for buildings AS 1170.4 requires that all stiff components be considered as part of the seismic force-resisting system and designed accordingly unless they are separated such that no interaction can take place as it undergoes deflection due to earthquake effects. Yet the reporter has seen projects where load-bearing pre-fabricated concrete walls and columns have not been included in the analysis and all of the actions are assumed to be taken by the central core.

Expert Panel Comments

Find out more about the Expert Panel

An Expert Panel 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-US Expert Panel page.

Today’s structural engineers have a wide range of tools available that allow fast and detailed analysis of different structural forms. However, it is important to clearly understand the assumptions, simplifications and the basis of any software being used for modelling the structure. As the reporter points out, the software used for structural analysis is based on an idealised mathematical model of the structure. There are limits to how accurately the models can represent the real structure in its real environment.

When correctly used, structural modelling can provide the engineer with a solid grasp of the upper and lower bounds to any design on which they are working. Simple models can be used to understand the results of more complex models and all modelling should be considerate of likely construction tolerances, secondary effects, and serviceability implications (vibrations, accelerations, line of sight, etc.). It is also important to consider the effect on structural behaviour from temperature changes, concrete shrinkage, and the possibility of unexpected loads (robustness). If the material response is ductile the consequences of overlooking such effects may be limited, but if the material response is brittle, these effects may lead to failure. 

Inherent inaccuracy of modelling deflections in materials such as concrete

As noted by the reporter, engineers should be aware of the inherent inaccuracy of modelling deflections in materials such as concrete.  Informing a client that a reinforced concrete beam will deflect 21.7mm in the long term would provide a false degree of precision, whereas a more realistic statement would be that the long-term deflection of the beam is likely to be in the range 10 to 30mm.

It is important to verify the results from the model and for the more experienced engineer this may be by “gut feel” and a review of the deflected shape produced by the model. For the less experienced, the very minimum should be self-checking by simple hand calculations to check that the results (e.g. span/depth ratios) are of the right order, and that the total support reactions are equal to the overall loads applied. For complex structures, a very simplified model may be used as a comparison to check the expected load path and to provide a “sense check” for the more complicated model with all elements included.

It is well established that human errors are the cause of most structural failures and a European research project - The human factor in structural engineering: A source of uncertainty and reduced structural safety - has studied how subjective decisions, individual knowledge and the use of advanced tools and codes affect structural safety and structural design. This research revealed a large variation (around 300%) in results when a common design task was carried out by different engineers and introduces the term Engineering Modelling Uncertainty (EMU).

Further examples of incorrect modelling of structures that may go undetected are outlined in two investigations of existing bridge capacities that make sobering reading:

  • The UK Highways Agency Contract 2/419 Technical Audit of the Application of BA79 A Review of Bridge Assessment Failures on the Motorway and Trunk Road Network. This audit of 249 desk-top bridge assessments that had been carried out by a range of consultants was undertaken to determine why there was an unexpectedly large number of assessment failures. Disturbingly, two of the eight reasons for assessment failures were failures in the desk-top assessment process itself through “inappropriate or too conservative analysis for assessment” and “misinterpretation or inappropriate application of the assessment code”.
  • The Queensland Department of Transport and Main Roads (TMR) had a similar experience as presented in the paper by Shaw, Pritchard and Heywood: Bridge Analysis: are we data managers or engineers? TMR conducted an audit of around 500 desk-top bridge assessments that had been carried out by different consultants. This found a range of errors including errors in the use of 3D models, inappropriate software defaults and incorrect load application.

The CROSS Topic Paper Reflective thinking discusses similar concerns. Reflective thinking is a constant drive to ask questions and to make appropriate responses to them. It is characterised by a healthy scepticism about all inputs to processes, the processes themselves and about the outcomes from processes.

To summarise, the risks associated with inadequate modelling include:

  • models with assumptions about boundary conditions, element interaction and unrealistic structural behaviour;
  • designers not understanding the limitations of software;
  • inadequate time assigned to checking of models and independent rudimentary first principle checking for logical output; and
  • standards and guidelines may not be suitable to cover the developments in software or communications between designer, detailer, fabricator or builder etc. 

These risks can be mitigated as follows:

  • ensure the competency level of the modelling and analysis software operator is adequate for the task;
  • undertake suitable quality control based on level of risk;
  • undertake independent verification and validation based on the level of risk; and
  • having a quality control system in place that incorporates the modelling, analysis and model transfers.

The CROSS database contains other examples of the problems associated with incorrect modelling. Attention is drawn to the recent report 1073 Concern over modelling of concrete frame building for construction stage. This report stresses the importance of considering the effect that the construction sequence may have on the design and the need to design the structure through all stages of its life. Failure to ensure that the design and proposed construction methodology are compatible may lead to a structure which is unsafe to build or indeed unsafe in use.

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.