Guidelines for the use of computers for engineering calculations (2002)

Errors in using computers for engineering design have led to drastic failures such as the sinking of the newly constructed Sleipner A offshore gravity platform in a Norwegian fjord in 1990. This structure was over 100m high, with a base area of some 16 000m2, and cost several million dollars.

These new guidelines advocate a logical and disciplined approach to computer-assisted engineering based around a process where each stage has built-in checks. The key components of this process are people, software and hardware. The Sleipner A case is examined and it is shown how the procedures described in the guidelines could have been mapped on to the platform design to identify the errors which caused the failure of the structure. A further case of the collapse of a civic centre roof in the USA is also analysed in a similar manner.

The guidelines also give examples of modelling reviews for a steel portal frame, a concrete headworks roof slab, and a slope stability analysis. Each is reported differently to emphasise that the format of the review document is not prescriptive and should be dictated by the needs of the particular project and office practice. Model development, acceptance criteria and model assurance are all examined.

The need for a systematic approach involving checks is emphasised, together with the need for adequate training and experience, stressing the fact that an experienced chartered engineer should have overall responsibility. The use of quality management systems is strongly recommended.

The guidelines have been prepared in response to growing concern regarding the appropriate use of computers for structural calculations and the recommendations of the tenth report of the Standing Committee on Structural Safety (SCOSS). Structural engineering practice has evolved radically in the last twenty years. Before 1980 the production of calculations was the most time-consuming aspect of design. The very slowness of the process provided time for the designer to think and rethink the basis of the model being used.

Computers now deliver the results of calculations very quickly and the problems of making a realistic representation of the structure have become more apparent as the time for subconscious reflection has reduced. As a result, more formal methods for developing and assessing the reliability of computer models have become necessary. These guidelines address this need.