Modern Structural Analysis      Modelling Process and Guidelines

A structual engineering textbook by Iain A MacLeod

What others say


Foreword to Modern Structural Analysis

By:  Michael Dickson FIStructE
Formerly Chairman, Buro Happold
President,  Institution of Structural Engineers  2005-06      01 09 05

This interesting book promotes a new way of looking at structural analysis.  It suggests that the ability to work with the model (as distinct from the solution process) is a primary issue which should be formally addressed in practice and in education.   The content is focussed on modelling issues and I know of no other text which does this so comprehensively.

The early chapters contain much advice necessary to help the reader establish how to formulate a numerical model which might be capable of simulating the performance of the actual structural system under investigation. The later chapters include a good outline of the issues involved in modelling of structures using finite elements. The two case studies given at the end of the book are a good device to put the excellent advice given in the earlier sections into some perspective for the reader.

I found it most useful to have in the same book a reminder of the theoretical basis of the full range of finite element types and a sound method as to how to employ analysis as a reflective tool towards a better understanding of structural behaviour.
The rigorous treatment for the process of validation of a model is most enlightening as is that outlined for verification of the results. After all, the iterative process of model validation and output verification are the main activities for gaining a true understanding of structural behaviour.

 My own experience working with Buro Happold tells me that robust structural design requires the willingness to develop an understanding of structural behaviour with a questioning mind. In most consulting offices, current practice is to undertake this using finite element models of increasing complexity as understanding of the problem at hand grows. Iain MacLeod describes clearly how to build up this understanding using sensitivity analysis and simplified loadings to test validity against expectations from parallel calculation and modelling experiences.  It is argued that risk will be reduced in practice if there is a rigorous analytical process that reflects the realities of current engineering practice in most offices.

 Most structures are of a reasonably conventional type and use well tried framing systems. Substantial experience already exists on their likely performance so hand calculations based on structural theory can be done to initiate formulation of the model or to act as a check on the results. However, even advanced classical methods struggle to model the sophistication of load paths in redundant or non linear structures where individual stiffness, material response and definition of restraint determines structural performance. In this case I have found that comparison of the output of simplified analytical results with physical models very useful as an addition to classical calculation  - as advocated in the second chapter..

The book is thus both a useful reference for the practitioner and a comprehensive learning guide for the student. It  builds on the publication by the Institution of Structural Engineers ‘Guidelines for the Use of Computers for Engineering Calculation’ published in 2002.  Its carefully constructed content successfully redresses the imbalance in risk between the finite element process based around generally determinate calculation output which has itself been derived from a possibly non-determinate understanding of the actual modelling process. In the Introduction, the author suggests that all structural engineers and all civil engineers who use structural analysis will find the contents of the book to be useful.  I think that he is right.




Comments from former students on the teaching of structural analysis as a modelling activity.

From: Dr M Y Rafiq, Senior Lecturer, School of Engineering, University of Plymouth, UK November 2005

I had the pleasure of knowing Iain MacLeod as my lecturer, supervisor and Head of Department at the University of Strathclyde, in the 1980s and I learned about his modelling approach to structural analysis at that time.  I teach modelling techniques to my students at Plymouth University and have got very positive feedback   practice about it.  I usually refer to Iain’s 1990 book ‘Analytical Modelling of Structuralfrom from  Systems’ in my lectures as an excellent source of useful information on analysis modelling, with greater emphasis on verification of computer results.  Iain’s new book - Modern Structural Analysis - Modelling process and Guidance is very important in relation to how we should teach structural analysis, which for many of us would be a step change that I undertook a long time ago.

From:  Matthew Petticrew,  Associate,  Ramboll UK,  November, 2010 

I was a student of Civil Engineering at Strathclyde University from 1994 to 1998, where I studied under Iain MacLeod.  At graduation I joined engineering consultancy Ramboll (formerly Whitbybird) as a graduate structural Engineer. I now hold the position of Associate.
I have found that the approach to engineering I learned at University has remained useful to me. I therefore teach the same approach to graduates under my supervision, and from time to time, students at the University. In particular I am of the view that critical validation and verification of modern analysis is more useful to the understanding of structural behaviour than a rigid adherence to the repetition of “by-hand” expressions, which students often perform by rote.

From:  David Mcwilliams,  Structural Engineer, Ramboll Uk,  November 2010

I spent the 5 years studying Architectural Engineering at the University of Strathclyde, with Prof. Macleod taking classes 3rd, 4th and 5th year (and advising on my dissertation throughout 4th year).  Where the majority of classes were calculation and theory based, the new approach which was introduced by Prof. MacLeod was, personally, very well received.  Where we had previously been getting bogged down with rigorous number crunching in structural analysis, we were now being taught a way of thinking, which gives far greater understanding of structural behaviour.  His use of computer modelling of structures as a learning tool was greatly advantageous to us, and gives a knowledge of how we model structures, and an understanding of how the structure behaves under various conditions.  Personally, this is an invaluable skill which he helped me develop, and I still use on a daily basis, from modelling a simple beam arrangement to a multi storey framed building.  Further to this, the checking procedures to computer modelling, which he enforced as a prerequisite to the modelling process, are a vital tool in using the structural modelling software in a design scenario for the company for which I work.

From:  Graeme Quinn    5th Year MEng Student,  University of Strathclyde.  November 2010,

Throughout my education years, from school up until university, the methods of teaching subjects usually have a similar form. The framework of the learning process usually follows:

  1. Theory- the student if taught a background to a particular topic with emphasis on why it is important.
  2. Methodology- students are taught determinate method/s of solving problems and where these can be implemented.
  3.  Tasks- the students learn the methods and implement them in a determinate problem in the knowledge that there is only one solution.

This method of teaching provides the student with a basic knowledge of a topic and arms them with the confidence to tackle these problems to find a unique solution. It is also fair to say from a personal perspective that if a student learns the method then he/she can solve a problem in an exam and achieve a relatively high mark without having a deep understanding of the problem.
In my 3rd year at university, Prof. MacLeod taught a class called Structural Analysis. The class was of a different form to which is normally taught and concentrated on the modelling process rather than determining unique solutions by applying determinate methods. The framework of this class was in the form:

  1. Modelling- learning how to set up a model of a simple roof truss.
  2. Model Validation- to validate the model to ensure that it was implemented correctly.
  3. Results Verification- verifying the results output from the FEA software to ensure that they are what is expected.

By validating and verifying a model allowed us to understand the behaviour of a structure instead of just following a set out method. The form of learning provided me with the confidence to set up other models of a different form then validate it and verify the results. Understanding the modelling process required a degree of knowledge of the behaviour structures without which we would not know what was really going on.
I believe that the modelling process should not have been kept until my 3rd year of study but instead should have been part of structural mechanics from first year. Firstly, we should have learned how to calculate bending moments, shear forces and axial forces in a loaded simply-supported beam. Then we should have been taught how to model this configuration and verify the results output that we obtained from hand calculations. Then, the process of gradual-build-up could then take place and the understanding and modelling of increasingly more complex structural forms could be achieved.
In short, I believe that conventional learning of a determinate process to find a unique solution only provides the understanding of a problem up until a certain point. Understanding the modelling process of structural forms allows us students to develop a technique (not a method) of tackling unknown problems- a skill which we will need when working in practice after gradating.


From:  Andrew Rolland,   Project Engineer Civil,  Network Rail

When I studied Civil Engineering as an undergraduate at the University of Strathclyde I was fortunate to be introduced by Professor Iain MacLeod  to the idea that structural analysis is more about working with models than about doing calculations. 
Not long after I graduated I was asked by my project leader to look at the results of an analysis of a long span roof carried out by a third party.  Large trusses spanned onto columns.  The analysis showed excessively high moments at the tops of the columns.  I looked at the results and said to myself  “ What would Prof. Macleod say to do?  He would say “Check the restraints.”   The survey drawings and photographs showed that the trusses were connected to the columns by a pin at one end and a roller at the other.  There was no moment connection.  Problem solved.  While the error was rather obvious I believe that the discipline in working with models that Prof. Macleod helped us to develop got me to get to the result more quickly and has stood me in good stead in my career as a professional engineer.


A student experience

A student who had taken a class which included modelling came to see me with an analysis model of a building as part of a design project.  He had looked at the results and thought there was an error because the pattern of bending moments was not what he expected.   We examined the situation together and found a reason to explain the behaviour that he had observed.  Thus although the model was not wrong he (and I) had learned something about behaviour of that type of structure.  I said to the student "I am very pleased that you asked this question.  Was it because of what I had taught you in the modelling class?"  He replied that this was indeed so and that he now used the modelling process questions regularly in his project work.  He said that he had worked for a well known firm on a complex analysis of a building and remembered that the senior engineers working on the model tended not to look at the model and the results in this way.


Other comments

From:  Andrew Weir,  Associate Director,  Expedition,   January 2011  

Of all that I am hearing from 'academia', Iain MacLeod’s thoughts are closest to my own opinions (some of which are only just forming following finally getting to think about this stuff, despite going through the motions for the last 20 years.)

The approach that he is promoting is similar to that used in the nuclear and offshore industries although their thought processes are not as clearly laid out as his.

I'd suggest that Iain goes further to show that the ‘model>validate> verify’ approach is highly cyclical, particularly for more simple analyses. With fast computers people will always now keep running models, adding refinement and correcting errors as necessary. The essential skill being the continual validation, verification and interpretation of the results. Iain also suggests a 'QA' approach to analysis (more so in his book) which is always true, but importantly needs to be appropriate to the design stage and the importance of what is being modelled. For example, I'll run many cycles of models with a very limited check in place other than a quick look at the deflections (and BMs and reactions), but once at an appropriate stage the check become more thorough until finally there's a full QA check. (Sensitivity studies are also carried out at appropriate times).

Iain also talks about 'risk’ and ‘uncertainties'. I agree that an understanding of this is something that is missing from many engineers' understanding of analysis and is really a fundamental omission. I've seen too people analysing the hell out of a structure with 100mm x100mm elements in a flat slab to get a 'better' answer, missing the point that they've applied a blanket '4+1' live load, an E value from a code, section properties ignoring cracking and reinforcement, used elastic stresses etc. For example, if you ask a grad what the overall safety factor of a building is they normally say something between 1.4 and 1.6 (ie the load factors). It's always entertaining to see the long drawn out face of a grad when they tell you the utilisation factor of something already built is 1.03 and the relief when you tell them it 'may just be OK'.  I would go as far as saying that if you do not have a pretty good understanding of risk and how it affects the modelling process then you will never be a good structural engineer. Unfortunately, the design codes do not help develop any understanding whatsoever.