Sensitivity analysis of some critical factors affecting simulated intrusion volumes during a low pressure transient event in a full-scale water distribution system G. Ebacher a, *, M.C. Besner a , B. Cle ´ment b , M. Pre ´vost a a E ´ cole Polytechnique de Montre ´al, NSERC Industrial Chair on Drinking Water, Civil, Geological and Mining Engineering, C.P. 6079, Succ. centre-ville, Montre ´al, QC, H3C 3A7 Canada b E ´ cole Polytechnique de Montre ´al, Mathematics and Industrial Engineering, C.P. 6079, Succ. centre-ville, Montre ´al, QC, H3C 3A7 Canada article info Article history: Received 9 November 2011 Received in revised form 18 February 2012 Accepted 3 May 2012 Available online 14 May 2012 Keywords: Water distribution system Intrusion volumes Sensitivity analysis Transient analysis Downsurge event abstract Intrusion events caused by transient low pressures may result in the contamination of a water distribution system (DS). This work aims at estimating the range of potential intrusion volumes that could result from a real downsurge event caused by a momentary pump shutdown. A model calibrated with transient low pressure recordings was used to simulate total intrusion volumes through leakage orifices and submerged air vacuum valves (AVVs). Four critical factors influencing intrusion volumes were varied: the external head of (untreated) water on leakage orifices, the external head of (untreated) water on submerged air vacuum valves, the leakage rate, and the diameter of AVVs’ outlet orifice (represented by a multiplicative factor). Leakage orifices’ head and AVVs’ orifice head levels were assessed through fieldwork. Two sets of runs were generated as part of two statis- tically designed experiments. A first set of 81 runs was based on a complete factorial design in which each factor was varied over 3 levels. A second set of 40 runs was based on a latin hypercube design, better suited for experimental runs on a computer model. The simula- tions were conducted using commercially available transient analysis software. Responses, measured by total intrusion volumes, ranged from 10 to 366 L. A second degree polynomial was used to analyze the total intrusion volumes. Sensitivity analyses of both designs revealed that the relationship between the total intrusion volume and the four contributing factors is not monotonic, with the AVVs’ orifice head being the most influential factor. When intrusion through both pathways occurs concurrently, interactions between the intrusion flows through leakage orifices and submerged AVVs influence intrusion volumes. When only intrusion through leakage orifices is considered, the total intrusion volume is more largely influenced by the leakage rate than by the leakage orifices’ head. The latter mainly impacts the extent of the area affected by intrusion. ª 2012 Elsevier Ltd. All rights reserved. Abbreviations: AVV, Air vacuum valve; DS, Distribution system; FFD, Full factorial design (3 4 ); HAVV, External head on AVVs’ outlet orifice (m) (AVVs’ orifice head); HLO, External head on leakage orifices (m) (Leakage orifices’ head); LHD, Latin hypercube design; LR, Leakage rate (% of inflow); MFD, Multiplicative factor on AVVs’ outlet orifice diameter (AVVs’ diameter factor); WTP, Water treatment plant. * Corresponding author. Tel.: þ1 514 340 4711x2975. E-mail address: gabrielle.ebacher@polymtl.ca (G. Ebacher). Available online at www.sciencedirect.com journal homepage: www.elsevier.com/locate/watres water research 46 (2012) 4017 e4030 0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2012.05.006