Overflow from combined sewer systems has been a hot topic in Denmark over the recent years. Danish utilities are therefore obliged to report the state of each overflow structure based on the yearly number of overflows and the calculated overflow volume. Vejle Spildevand has been doing this for the past 10 years. Yearly comparisons between observed and calculated overflows in Skibet have shown great differences, both numbers exceeding the permit each year. Vejle Spildevand needs to find solutions to reduce the overflows to protect the recipient Vejle Å. The unique situation in Skibet is that there are relatively many (and good) measurements in the sewer system. Other than that, a point rain gage has been situated in the center point of the catchment (covering a total area of approx. 60 ha) for 12 years, meaning that the quality of the rainfall input is good compared to other sites. Due to Vejle Spildevands recent transition from DHI-products to PCSWMM, the existing MOUSE-model was imported and corrected in PCSWMM in order to obtain better correlation between model results and observed values. The assumption at first was that the existing model was an adequate stating point, but needed a re-calibration. The first step however, was a thorough quality assurance (QA) of the model. The QA revealed several errors, both related to updated capacities in pumping stations, but also related to the subcatchment. For example larger roads with high imperviousness, which wrongly had been connected to the combined sewers. The QA-process has been applied systematically and attention was paid to details within each catchment, pump station and retention basin to reveal and correct errors, and afterwards observed overflows were nicely reproduced even with unchanged model parameters. To pursue this work, two specific tools has been developed within PCSWMM: (i) a specific Python-script to transfer the MOUSE model within PCSWMM (ii) an efficient event analysis tool to have a clear overview of the situation. This workflow has been applied to several different sites, and the conclusion is that a well-constructed, quality assured model, which is fed with representative rainfall data, does not necessarily rely on calibration of model parameters to replicate observations in the sewer system. This model was subsequently used to plan the re-build and upgrade of pump stations in order to increase the outgoing flow capacity. This is a much more inexpensive solution compared to a full sewer separation or upgrades to the retention volumes.
Click here to download a static PDF version of the presentation.
Click here to watch recorded presentation on YouTube.