An Experimental and Numerical Study of Infiltration Trenches in Urban Runoff Control

The rapid growth of Brazilian cities during the last decades is causing serious problems in many aspects of urban infrastructure. Urban drainage is not an exception: the usual drainage systems, that simply try to quickly transport water away, by modifying natural water courses or by using pipes and galleries, have proved to be inefficient and excessively expensive. In order to solve this problem, new solutions have been created and studied, trying to favour source control. This paper presents an experimental study with two infiltration trenches performed at IPH-UFRGS, in Porto Alegre, Brazil. The first trench, of 10 m length, 0.8 m width and 1.0 m depth, is being monitored since July 1999. The second trench was built as three separate but connected modules, each one of them of 3.0 m length, 0.8 m width and 1.0 m depth, and it is being monitored since September 2001. After two years of continuous modelling, both trenches are still able to control excessive runoff volumes. During the whole period, all runoff volumes were infiltrated into the soil through the trenches. Trench I presented higher infiltration rates, suggesting the occurrence of preferential flow. Also, the analysis of subsequent events in both trenches shows that the recession becomes faster as soil saturation increases, raising the hypothesis of creation of soil pore connectivity due to soil water saturation. The Bouwer Model (1969) was selected to represent the hydraulic functioning of the trenches, taking into account the typical characteristics of the regional soil (with high percentage of clay). This paper discusses the first results of the model simulations, claiming that they were not completely satisfying, probably due to: 1) preferential flow; 2) lack of a sufficiently clogged restriction layer; 3) simplifications of the model. The selected model considers that, for granular soils, the variation of the hydraulic conductivity ( K ), respect to the capillary tension ( P ), occurs in a relatively narrow zone. Consequently, it considers P and K as constants. This simplification cannot be applied to clayey soils, such as that where this trench is located. A modification of the model, taking into account P and K variations as a function of the infiltrated water volume, so that these values are computed at each time step, is proposed in this paper.