10. CONCLUSIONS

A summary of the failure loads predicted by ELFEN against the failure loads determined in the various tests that have formed the basis for the verification are included in Table 10.1. (Fuller details and discussion of the respective results, including comparisons with conventional arch assessment methods can be found in the respective parts of Sections 7 and 8 of this report).

The predicted failure loads determined by ELFEN correlate extremely well with the test results, for all the various forms of arch considered, including unstrengthened brick arches with and without partial ring separation, a random rubble stone arch, a multi-span arch and Archtec strengthened brick arches.

The results of the ELFEN analysis also correlate well with conventional methods of assessment in most instances, although the ELFEN predictions are generally closer to the test results than conventional methods. As has been noted by others, in some particular cases, conventional methods of assessment can give poor and non-conservative predictions of strength; for example it may not be possible to take account of multi-ring brick arch behaviour and so strength could be overestimated.

On the basis of the verification process documented in this report and subject to the use of the relevant material properties and parameters upon which the verification has been carried out, the use of ELFEN, in the way described, to determine the strength of both unstrengthened and Archtec strengthened masonry arch bridges (which are square or near square (up to the order of 20° skew)) is considered justified.

In addition, Finite/Discrete Element analysis of masonry structures provides several significant new capabilities over conventional methods of assessment, namely;

• Explicit representation of defects such as ring separation, local distortion and mortar loss is possible and the verification process using ELFEN has established that these can be reliably modelled; that is predictions of behaviour of both ring separated and fully bonded arches have been satisfactorily undertaken.
   
• Deflections and serviceability behaviour can be predicted. This has not before been possible. The verification process has established that deflections can be reliably predicted using ELFEN and this provides the basis for a significant and exciting new capability for modelling arch behaviour at the serviceability limit state.

Table 10.1 Summary of ELFEN Prediction Against Tests