Smoke management systems can be complex and involve the operation of many interacting components, including detection systems, exhaust fans, natural ventilators, automatic smoke curtains, dampers, fresh air intakes, etc. Experience of actual installed systems in real buildings has led to concerns on the efficacy of some smoke management systems, especially over the lifetime of a building.
There is limited data on the reliability of smoke management systems. Klote and Milke  provide data based on five different system designs as shown in Table 1. The level of complexity was increased for each design.
System No of HVAC No of other Reliability of new Mean life of system fans components system before commissioned
Table 1: Reliability data given by Klote and Milke 
Table 1 shows that as the number of components used in the system increases, the reliability of the system decreases significantly, such that the mean life of the commissioned system with 5 fans and 54 other components is just 3 months. However, it must be stressed that the analysis assumed that failure of any one component would lead to failure of the complete system, which may not be the case in reality.
Moran  carried out a survey of smoke exhaust systems in shopping centres around the Brisbane region. 32 centres responded to the survey of which 15 had some form of smoke exhaust system. 5 out of the 15 centres reported problems, with various incidents of failure, such as: motor (1), contractor (1), fan bearing (3), fan motor blade (3), fire detection (2) and water ingress (1).
Moore and Timms  provided data on the efficacy of systems from less complex designs that are appropriate for low-rise shopping centres (see Table 2).
Probability of attaining the efficacy range: Design 1, Design 2 Efficacy range Maintenance, Installation and Commissioning quality.
Table 2: Reliability data from Moore and Timms 
These data demonstrate the benefits of a robust maintenance, installation and commissioning regime in increasing the probability of the efficacy of a smoke management system over the lifetime of a building.
Sources of failure
The efficiency of a smoke management system can be dependent on a number of factors.
Morgan  describes some common sources of error in the design, construction and implementation of a smoke management system. The following list of errors is not definitive :
Inadequate theories - All calculation methods have theories which have been developed for relatively idealised geometries. Real buildings can depart from those ideal conditions. Typical examples include: spill plumes rising past non-straight spill edges; where the smoke flow approaches the straight spill edge at other than a right angle; and plumes which are partially adhered and partially free.
Unknown input data - The building geometry is usually known when designing, but there are several necessary input parameters which are more subjective in many cases (i.e. specifying the design fire).
Poor communications - There can be poor communications between the system designer and the installer/s of the system; and between both and the architect; and between all of them and the regulators. Examples include:
a system which had smoke curtains installed 1.5 metres too short.
channelling screens fitted half-way across a shop’s open front, completely negating the reason for fitting them.
a hole in fixed smoke curtains at the reservoir boundary in order to allow a smoke detector’s light beam to pass through.
Problems can arise when late detail design changes are made without telling either the designer or the regulator because the changes are (wrongly) thought to be trivial.
Changes to the building structure over the lifetime of the building may also reduce the efficacy the smoke management system, if these changes are not brought to the attention of the designer.