Transferability of exfiltration rates from sewer systems

Background, Aim and Scope. Models describing in- and exfiltration of sewer systems require a large amount of data for calibration. These are often difficult and expensive to measure and to acquire. Therefore, this paper aims at clarifying whether results from various measurement campaigns might be joined to broaden the basis of an exfiltration model. Within this context, the transferability of exfiltration rates from one sewer (or catchment) to another is one of the crucial points. Materials and Methods. Exfiltration rates derived from field measurements and from literature (field and laboratory investigations) are compared with respect to - Catchment characteristics - Applied methods: tracer tests, blocking tests, and laboratory investigations - Experimental site: laboratory and field studies - Leakage area: Closed-circuit television (CCTV) serves as a substitute for sewer characteristics. From those records the leakage area is obtained calculating an exfiltration rate per day and cm(2) leakage area. Exfiltration was measured in two catchments and the findings were compared with published data for two catchments and two laboratory studies. As expected, exfiltration rates exhibit a wide range. We investigated whether the experimental design dominates the variance of measured data. The results are compared by means of statistical methods (Kruskal-Wallis analysis of ranks, Bootstrapping, and analysis of covariance-ANCOVA) to reveal significant differences in means. Results. The statistical analysis yields significant differences in exfiltration rates comparing the results obtained (i) in either field or laboratory and (ii) with different methods. Exfiltration rates measured in various catchments are not significantly different. Discussion. It can be shown that exfiltration rates obtained from field studies are affected by the measurement technique, whereas for laboratory investigations, the measurement technique does not influence the result in the first place. It is therefore difficult to jointly analyse laboratory and field experiments, i.e. a transfer of exfiltration rates from laboratory investigations to operational sewers is hardly feasible. It is also shown that results from different catchments are better suited for a joint assessment than results from differing methods. Conclusions. A joint assessment of exfiltration rates obtained in various studies is not feasible with the available datasets.