Application of statistical experimental design for parametric simulation studies in building physics, with an emphasis on indoor air.

Parametric studies have often been used for sensitivity analyses in the field of the pharmaceuticaland agricultural sector. All such studies aims at bringing some kind of order out of complicatedrelationships between influencing factors and some response parameter(s). With experimentaldesigns and statistical analysis methods, it is possible to trace and quantify influencesof individual as well as combinations of input factors on the response parameter. Thisprocedure has so far very seldom been used within the building sector. The paper describes anexample within the area of indoor air quality, where a statistical experimental design procedurehas been used. The purpose of the example was to analyse the fraction of time of the totalheating season when the outdoor air ventilation flow rate was 24 litres/(s person) in bedrooms.Four different ventilation systems were examined and seven parameters were chosen,suspected to influence the air flow rate. This kind of investigation calls for a large number ofruns. In order to cover all combinations 4000 simulations have to be performed. With statisticalexpefiental design and analysis methods the procedure can been reduced to 400 runswithout significant loss in the quality of the results!