A great deal of the work with which Origen is involved, is concerned with failure analysis and fracture mechanics. Both of which require realistic and accurate assessment of loads and hence stresses to which the component or system is subjected. These loads/stresses may be obtained in various ways, and a good starting point is often to refer to the original design documentation. Typically the engineer/designer would have selected (presumably) realistic loads to which the component or system was subjected, and designed the component in such a way that the stresses (and deflections) were always less than the design limit, by an appropriate ‘load factor’ or ‘safety factor’. In this conventional design approach, the component would not yield or fail by other mechanisms, including inter alia fatigue, where the number of cycles, or design life, is also considered.
This ‘design’ approach generally works well, as long the loading conditions and cycles of repetition have all been considered and appropriately accommodated, allowing new systems and structures to be safely (and economically) designed for their planned operational conditions and lifetimes. This is borne out by operational experience, as long as failures do not occur. The system can be re-designed if loading conditions change, to ensure safe operation under the new loading conditions. This is all very well, but is heavily dependent on the skill/experience of the designer and his/her assumptions in modelling and assessing the loads (and particularly cyclic loading) correctly and accurately. Difficulties arise, however, when the loading conditions are not adequately understood or modelled, and assumptions about the magnitude of the loads, or their line of action, during operation have to be made. When the component degrades through, for example, wear or corrosion, its section size, or the line of action, may vary and bending can occur if the loading becomes eccentric. The ‘ideal’ loading conditions assumed in design are frankly seldom truly accurate, and this deviation can lead to additional unexpected loading scenarios. Wear and corrosion, if not monitored over time can rapidly lead to higher stress and degradation of the whole structure, often through fatigue cracking. Thus for meaningful understanding of failure and true assessment of safety of industrial systems it is important to re-iterate the need for truly realistic and accurate estimates of the operational stresses, in practice.
As long as the base assumptions truly reflect the loading conditions and the consequent behaviour of the materials, failure can typically be prevented. However, in some cases poor design and inaccurate assessment of the stresses stems directly from the ‘false sense of security’ obtained by the modelling process itself. During design, and in cases of failure, it is essential that true representative estimates of the actual operational stresses be obtained. This is often achieved by Finite Element Analysis Modelling but these models require accurate understanding/modelling of the loads and boundary conditions. Failure to represent the loads correctly will lead to incorrect interpretation of the stresses. As FEA modelling capability becomes more easily available it is often undertaken by inexperienced and inadequately qualified ‘experts’. This introduces a conundrum in that the coloured contour plots generated by FEA modelling are very attractive and beguiling and can be misleading if loading and boundary conditions have been applied incorrectly.
Although the results of strain gauging measurements of the ACTUAL as built structure under load are often less easy to visualise, they provide a more reliable measure of the stresses introduced by actual loading. However, the results of any strain gauge measurement exercise are limited to the gauge position, can only really give the surface strains, the loads that can realistically/safely applied during testing and are limited to the change in strain from a base load condition. In addition strain gauging can only be used once the structure or prototype component has been built and not during the initial design process.
Unfortunately given the elegance of FEA contour plots and the relative ease with which these plots can be generated, there is a tendency to believe FEA results implicitly, often in preference to the strain gauging results, with potentially disastrous consequences. In essence the two techniques are complementary and should be used together, with the strain gauging results complementing the FEM models and providing input as to actual loading. Where ever possible one needs to assess the model, using actual stress measurements, as obtained from strain gauging and measurement of in service loading conditions and use this data to revise the design models used in prototyping.
The proponent of each technique (FEM versus strain gauging) knows the limitations of their own stress analysis evaluation. The true FEM expert knows the limitations of the model (e.g. neglecting stress concentration features such as welds, the contact model employed, the limitations of how loading and boundary conditions are applied, etc.), but his users are beguiled by the pretty illustration of resultant stress contour plots. Conversely the strain gauge analyst can ‘see’ and truly determine the effect of loading of a particular type (from bending, say, or whether the loading is axial or eccentric) on the measured strains, and hence, stresses, and gets an extremely good ‘feel’ for the actual stress encountered. His audience, however, tends to regard strain gauging as less elegant and more prosaic, whereas in reality it is the more reliable method of calibrating stress analyses. In truth one needs both techniques and in the ideal cases these would be used together to complement each other.
The message is simple, when used by specialists, numerical/FEA modelling is an irreplaceable design tool but has limitations that can best be addressed using experimental stress measurements techniques such as stain gauging which should be used to validate the numerical models and define actual load cases imposed by service conditions. It is for this reason many organisations such as ASME require strain gauge verification of the design process.
A future tech tip will include examples where strain gauging has made a powerful contribution to the understanding of the failure modes in real structures.