‘This is a wonderful book that should be on the desk of every architect and planner. It shows how

Box 20 Designing for Successful Failures
Dr Paul Bannister
Essentially all buildings are complex, one-off designs filled with a wide range of mechanical and 
electrical, automated and manual systems that act as valves on the resource consumption of 
the building. Irrespective of the best intent of all the parties concerned, it is inevitable that 
these systems break down. As a result, the resource consumption of the building depends 
critically on how the building responds to such failures.
This can be illustrated by the following simple example of a light switch. If a light switch is 
not turned off at the end of the day, the light will stay on until the following morning, wasting 
a great deal of energy. The light switch as a control system can be seen to be a poor solution 
as one simple (and probable) failure can cause much waste. Furthermore, the waste is po
-
tentially difficult to detect, in that it occurs overnight when no-one is in the building to notice 
the failure. No-one is negatively affected by the failure (except the accountant or owner, and 
they are unlikely to recognize the existence of the problem), and driving past the building many 
would assume that someone must be present to justify the lights being on.
Of course, there is a range of methods by which one could reduce the impact of this problem.
One could install automatic controls such as a timeclock or occupancy sensor; one could 
reduce the number of lights serviced by the light switch; one could improve the efficiency 
of the lighting system so that the energy impact of control failure is reduced; or one could 
provide an alarm that tells people that the lights were on overnight. Each of these methods has 
strengths and weaknesses but, together, they can produce a control that is robust to failure.
Given the range of potential failure scenarios associated with a simple light switch, it is small 
wonder that failures in more complex systems often go unnoticed. Each of these failures 
reduces the building’s efficiency and, in an extreme case, can lead to the performance of the 
building being worse than average rather than leading edge. To avoid this fate, designers need 
to consider the following techniques:
•
Identify the opportunities for failure. It is worth consulting with building maintenance 
specialists to establish this.
•
Reduce the number of moving parts. Fewer moving parts means fewer opportunities 
for failure.
•
Limit the domain of failure. Ensure that the total amount (kW) of the plant that would 
be incorrectly operated in the event of a failure is limited. This can be achieved by 
improving the basic efficiency of this plant, or by reducing the area or number of plant 
items attached to an individual failure point.
•
Remove compensatory mechanisms. Ensure that in the event of a failure there are no 
energy consuming systems that will operate to hide the failure.
•
Provide failure detection. For systems that are likely to fail with energy consequences, 
provide additional monitoring to alert the building operator to the failure.
•
Ensure that all moving components can be inspected and accessed readily. A plant that 
is difficult to inspect or access will be less likely to be detected and fixed.
•
Design buildings with ongoing ecological upgrading and simple passive adjustments in 
mind.
For any building, there will be a balance of each of these approaches required to achieve a 
building that is designed to fail successfully. In design, failure is successful if we in fact learn 
from it.

---

302
Positive Development