A probabilistic approach to risk analysis : a comparison between undesirable indoor events and human sensitivity

Sammanfattning: The main objective of this thesis is to develop a probabilistic procedure similar to the one used in modern design codes for building structures. The purpose of the procedure is to estimate the risk of an unhealthy indoor environment to occur, i.e. the risk for humans to become unhealthy indoors because of the design and construction of the building. The developed risk analysis procedure is based on the IEC standard of risk analysis combined with fault tree analysis (FTA) to evaluate the risk of an unhealthy indoor environment both qualitatively and quantitatively. Structural reliability analysis (SRA) is used in the quantitative evaluation, since several random variables can be handled using functions to express the relationship between the basic events in the fault tree. The use of SRA reveals the risk to be defined as the violation of the limit state function. The objective of the risk analysis is to compare the undesirable indoor event with the consequence to humans, i.e. to estimate the risk by comparing the random variables as in SRA by using either first-order second-moment method (FOSM) or simulation to find the probability of limit state violation, exceeding a threshold value or to estimate the reliability index. Environmental impacts with the potential to cause an undesirable indoor event of concern to human health are identified to be microorganisms and substances from microorganisms, emissions, and ionising radiation. The proposed risk analysis process is applied to a single-family dwelling founded on a concrete slab on the ground built in an area with high levels of radon concentrations in soil. The undesirable indoor event Yenv "Radon concentrations in indoor air", is evaluated and compared with the available dose-response relationship Xenv. The probability of limit state violation living in the single-family dwelling analysed is pf = 0,002 per year. The undesirable indoor event is also compared with the threshold value 200 Bq/m3 stipulated in the Swedish Building Regulations to not be exceeded. The probability pf = 0,0001 to exceed the threshold value. The estimated safety index is compared with the safety indices in the Swedish Design Regulations where structures are designed according to different safety classes. The estimated safety index is in approximate accordance with safety class 1, which is valid for design of structures where risk for serious injuries or death to humans as a failure consequence is minimal. The risk analysis process is also applied to a tenant-owned dwelling with the undesirable indoor event Yenv defined as “Legionella contaminated aerosols in indoor air”. The random variables of interest in the analysis are the initiating amount of bacteria, and the time of stagnant water. The constant of specific growth rate was estimated from tests made in the dwelling shower and considered to be deterministic, though it depends on several variables, such as the amount of nutrients, biofilm and water temperature. In the risk estimation, the undesirable indoor event Yenv is compared with guiding values, since no dose-response relationship is available describing the sensitivity to Legionella bacteria in the population. The probability of exceeding, for example, the guiding value 10000 cfu Legionella bacteria per 100 ml in the shower living in the tenant- owned dwelling is pf = 0,006. The conclusion from the work is that the undesirable indoor event Yenv, similar to the load effect S in SRA, and the dose-response relationship Xenv, similar to the resistance R in SRA, can be compared as in SRA.