Sensitivity of hygrothermal analysis to uncertainty in rain data

This paper describes a small study carried out using a hygrothermal simulation tool to investigate the sensitivity of wall performance results to uncertainty in the amount of rain impinging on the wall. Design standards for hygrothermal analysis of proposed designs include methods for selecting appropriate moisture reference years and specify the amount of water that is assumed to intrude into the wall. Weather data used as input for modeling purposes is generally assumed to be reliable, but recent work has shown that there may be considerable uncertainty in the rainfall data. A small study was carried out to investigate the effect of uncertainty of rainfall data on the hygrothermal performance of a typical residential building envelope. Most hygrothermal models require fully populated hourly datasets, which include rain data. Many locations, however, do not have this kind of data although many have qualitative rain data. Ten locations with rain gage data were chosen as typical of most regions of Canada, except for the far north. Different methods in estimating rainfall were considered as well as variations on the amount of rain data were subsequently made. Several performance criteria, including total moisture content and a mold index, were compared. Although the choice of which method for deriving quantities of water from qualitative codes does cause differences in the hygrothermal response and consequently the performance criteria, these differences appear to be manageable. It is suggested that practitioners should show their awareness of the probable level of uncertainty by stating error bands for their predictions of performance. It should be emphasized that the sources of uncertainty dealt with in this small study are not the only ones, but that they do focus on water entry through leakage paths. The natures of the leakage paths likely introduce greater uncertainty, and should also be borne in mind.On d\ue9crit dans ce document une br\ue8ve \ue9tude qui a \ue9t\ue9 r\ue9alis\ue9e au moyen d?un outil de simulation hygrothermique dans le but d?\ue9valuer la sensibilit\ue9 des r\ue9sultats de la performance du mur par rapport \ue0 l?incertitude li\ue9e \ue0 la quantit\ue9 des pr\ue9cipitations de pluie qu?il re\ue7oit. Les normes de conception pour fin d?analyse hygrothermique des conceptions propos\ue9es incluent des m\ue9thodes de s\ue9lection des ann\ue9es de r\ue9f\ue9rence appropri\ue9es pour l?humidit\ue9 et pr\ue9cisent la quantit\ue9 d?eau que l?on suppose avoir p\ue9n\ue9tr\ue9 dans le mur. Bien que les donn\ue9es m\ue9t\ue9orologiques utilis\ue9es comme donn\ue9es d?entr\ue9e aux fins de la mod\ue9lisation soient suppos\ue9es g\ue9n\ue9ralement fiables, de r\ue9cents travaux ont permis de d\ue9montrer que les donn\ue9es de pluviosit\ue9 peuvent comporter un degr\ue9 consid\ue9rable d?incertitude. Une \ue9tude de faible envergure a \ue9t\ue9 men\ue9e dans le but d?investiguer l?incidence qu?a l?incertitude des donn\ue9es de pluviosit\ue9 sur la performance hygrothermique d?une enveloppe de b\ue2timent r\ue9sidentielle type. La plupart des mod\ue8les hygrothermiques requi\ue8rent des ensembles de donn\ue9es horaires pleinement aliment\ue9s, lesquels incluent les donn\ue9es de pluviosit\ue9. Or, de tels types de donn\ue9es font d\ue9faut en de nombreuses localit\ue9s, bien que plusieurs d?entre elles disposent toutefois de donn\ue9es de pluviosit\ue9 qualitatives. On a s\ue9lectionn\ue9 dix (10) localit\ue9s disposant de donn\ue9es issues de pluviom\ue8tres comme \ue9tant repr\ue9sentatives de la plupart des r\ue9gions du Canada, \ue0 l?exception du Grand Nord. Diff\ue9rentes m\ue9thodes d?estimation de la pluviosit\ue9 ont \ue9t\ue9 prises en compte, et des variations quant aux donn\ue9es de pluviosit\ue9 quantitatives ont \ue9t\ue9 apport\ue9es par la suite. Plusieurs crit\ue8res de performance, y compris la teneur en humidit\ue9 totale et un indice de moisissure, ont fait l?objet d?une comparaison. Bien que le choix de la m\ue9thode permettant de d\ue9river les donn\ue9es quantitatives d?eau \ue0 partir des codes qualitatifs influe effectivement sur la r\ue9ponse hygrothermique et, par cons\ue9quent, sur les crit\ue8res de performance, les diff\ue9rences qui d\ue9coulent de cette incidence semblent contr\uf4lables. Il est sugg\ue9r\ue9 que les intervenants fassent \ue9tat de leur reconnaissance du degr\ue9 probable d?incertitude en \ue9non\ue7ant des intervalles d?erreur pour leurs pr\ue9dictions de la performance. Nous soulignons ici que les sources d?incertitude prises en compte dans le cadre de cette \ue9tude ne sont pas les seules \ue0 consid\ue9rer, mais qu?elles visent essentiellement le probl\ue8me de l?eau dans les murs via des voies de p\ue9n\ue9tration. La nature m\ueame de ces voies de p\ue9n\ue9tration est susceptible d?introduire un degr\ue9 encore plus \ue9lev\ue9 d?incertitude dans les pr\ue9dictions et doit \ue9galement entrer en ligne de compte.Peer reviewed: YesNRC publication: Ye