The use of more elements than is thought to be necessary to offset the effects of unknown variables and prevent system failure.1

Design requires dealing with unknowns. No matter how knowledgeable the designer and how thoroughly researched the design specification, basic assumptions about unknowns of one kind or another are inevitable in every design process. Factors of safety are used to offset the potential effects of these unknowns. This is achieved by adding materials and components to the system in order to make the design exceed the specification that is believed to be necessary to meet the design requirements. For example, designing an Internet service that can support one thousand users is straightforward. However, to account for unanticipated uses of the service (e.g., downloading large files), the design specification can be multiplied by a safety factor (e.g., three). In this case, a safety factor of three would mean that the service would be rated to support one thousand users, but actually designed to support three times that many, or three thousand users.

The size of the safety factor corresponds directly to the level of ignorance of the design parameters. The greater the ignorance, the greater the safety factor. For example, structures that are well understood and made of materials of consistent quality, such as steel and concrete structures, typically use a safety factor ranging from two to four. Structures that are well understood and made of materials of varying quality, such as wood, may use a safety factor ranging from four to eight. When ignorance is combined with materials of varying quality, the safety factor can get quite large. For example, the designers of the Great Pyramid at Giza unknowingly applied a safety factor of over twenty.2

Increasing the safety factor in a design translates into the addition of elements (e.g., materials). More elements means more cost. New designs must typically have large factors of safety because the number of unknowns is great. If a design performs reliably over time, confidence that the unknowns in the system have been managed combines with the pressure to reduce costs, and typically leads to a “tuning” process to reduce elements and lower the safety factor. Unfortunately, this process usually continues until an accident or failure occurs, at which point cost considerations become secondary and safety factors are again increased.3

Use safety factors to minimize the probability of failure in a design. Apply them in proportion to the ignorance of the design parameters and the severity of the consequences of failure. Reduce safety factors with caution, especially when specifications extend beyond design precedents. Observe the rated capacity of a system when making decisions that stress system limits, and not the designed capacity (capacity including factors of safety), except in cases of emergency.