ABSTRACT:
Michael D. Sohn and David M. Lorenzetti of Berkeley Lab have developed the first probabilistic algorithm for siting indoor samplers that detect airborne bio-agents. The algorithm, Probabilistic Approach to Sampler Siting (PASS), maximizes the probability of detecting a release. Specifically, the system chooses sampler locations in order to maximize the chance of detecting a release from a suite of relevant scenarios, subject to constraints such as the number and type of samplers, maintenance costs, and so on.
PASS provides sampler network designers with answers to questions such as: What sampler placements maximize the probability of detecting a biological release, given the uncertainties and variability in the building and release conditions? How does that probability improve with additional samplers? What sampler characteristics maximize the detection probability? and What sampler placements minimize occupant exposures from an undetected attack?
Unlike PASS, alternate sampler-siting techniques do not account for the relative likelihoods associated with uncertain, variable, and interdependent conditions such as the weather, release particulars (location, amount, timing, etc.), and mode of building operation. Therefore they cannot optimize the sampler system, in a probabilistic sense, against the expected threats it should detect. By accounting for the relative probability of each release, PASS avoids excessive sensitivity to highly unlikely scenarios, for instance, a release in a secure area of the building, if that leads to worse performance under more likely conditions, for instance, a release at a ground-level air intake. |