Statistical Model for Live Floor Loads

Theoretical considerations and analysis of survey data suggest that live loads on floors of buildings can be represented by a two-dimensional stochastic process with independent increments, the prototype of which is the compound poisson process. An essential assumption, which is vindicated by analysis of the data, is the statistical independence of the loads on nonoverlapping areas, provided the areas are not too small. This is closely analysed. Rational probabilistic design is based on the upper tail of the load probability distribution, and it is shown that the tail can be well approximated by the Pareto formula Cx^−α. The fit to survey data obtained is compared with that of the alternatives of a normal and an exponential tail. From the representation of the tail distribution, rational load-reduction formulas are derived. A further consideration which has been so far neglected is the necessity to assure that none of the panels into which the floor area is divided collapse, and this leads to an increase of the design load with increasing floor area.