Abstract:
The origin of this thesis was a long-standing interest in the performance of buildings in the years after completion, when the designers and builders have all moved onto the next new work. That interest grew as a result of conducting building surveys in the course of professional practice. The surveys often revealed incipient or actual building failures which required careful diagnosis to discover the cause, so that the failure could be prevented in future. For the knowledge gained from investigation and diagnosis to benefit the wider community, rather than merely the individuals concerned with one building, it became obvious that some system of objective and anonymous recording of the circumstances of each building failure was necessary. This thesis proposes a basis for identifying and evaluating building failures. Building failure is defined from the viewpoint of both the producer of the building and the user to ensure that it is the expectations of both that are considered when a building failure is being identified and evaluated. Identifying and evaluating building failures is a precursor to diagnosing the cause or causes of that failure. it is argued here that any evaluation of the causes of building failures must acknowledge the part played by natural causes as well as the part sometimes played by human error. It is also argued that placing emphasis on blame, and hence on legal liability, encourages universal denial of fault and works against the search for the truth. A system for classification of building failures by their causes is proposed as a means by which the knowledge gained from diagnosis of individual building failure events can be aggregated to reveal the pattern of failures in a sample of buildings. The results from applying the system of identifying, evaluating, and classifying building failures in a sample of New Zealand dwellings are presented. The main conclusion drawn from the work is that because natural causes are so difficult an influence to regulate, the best prospect for reducing the incidence of building failures is the avoidance of human error. Because human error can never be entirely discounted insurance against the risk of error is only wise. A second conclusion reached is that the proposed system of identifying, evaluating, and classifying building failures has been shown to produce useful results, even when the system has had only a written record from which to work.
