Factors affecting the performance and cost-efficiency of sand storage dams in south-eastern Kenya

---

Josep de Trincheria, Walter Leal, Ralf Otterpohl, Erik Nissen-Petersen

Thursday 2 july 2015

8:30 - 8:45h at Oceania (level 0)

Themes: (T) Water engineering, (ST) River and coastal engineering

Parallel session: 10E. Engineering - Computational

---

More than 1500 sand dams have been built in Kenya since 1960. However, to implement sand dams showing satisfactory performance and cost-efficiency levels is a complex procedure because there are key biophysical and technical constraints which limit their real-life performance and cost-efficiency. Currently, these biophysical factors and technical constraints are not adequately understood. Therefore, the main goal of this study is to contribute to a better understanding of the factors affecting the performance and cost-efficiency of sand dams in south-eastern Kenya. This research discusses the implications of the results of an exhaustive on-the-ground evaluation of the performance and cost-efficiency of sand dams in Makueni County, south-eastern Kenya. A physical survey of the sand reservoirs was systematically carried out in 97% of the sand dams implemented in Kikumbulyu location. Relevant hydrogeological variables in 1800 sub-sections of 30 sand dams were systematically and extensively measured. The analysis of the data collected on-the-ground coupled with an extensive literature review facilitated the identification and analysis of the factors affecting the real-life performance and cost-efficiency of the sand dams evaluated. The actual performance and cost-efficiency of the 30 sand dams evaluated was very low. 83% of the sand dams accumulated volumes of sandy alluvium sediments lower than 1000 m3. 0% yielded volumes of water higher than 2.000 m3. At least 63% did not supply water to any household during the entire dry season. The average cost-efficiency for the water yield and supply capacity was 5.295 EUR/m3 and 6.382 EUR/household, respectively. The results are statistically significant in Kikumbulyu location and may also be highly representative of the real-life performance of most sand dams implemented in south-eastern Kenya since 1990. Applying the precautionary principle, sand dams should always be built by stages in catchment areas where there is a homogenous production of coarse and medium sandy alluvium sediments. The robustness of the structure to the intrinsic climate variability of arid and semi-arid areas may inversely decrease as compared to the height of the spillway and/or the height of the stages. The contribution from riverbanks should not be automatically assumed as an inherent characteristic of sand dams. In addition, potential evaporation and seepage losses, and the saturation of the sand reservoir, should be taken into account to estimate the water yield and supply capacity of sand dams. The specific yield of the sand reservoir and the seasonality of the rainfall distribution are also important parameters affecting the real-life performance of these structures. A clear effort is required to reconceptualise and change current technical guidelines and construction procedures. Specifically, a systematic adoption of a multi-stage construction procedure; a clear definition of the maximum stage height; lower construction volumes in order to reduce the costs; lower spillway heights in order to increase the resilience to floods and reduce constructions costs; consideration of the slope of the riverbed and maximum area of influence; siting in catchment areas producing coarse sandy alluvium sediments; siting in catchment areas producing fine sand only with spillway stages of reduced height, no siting where there is production of low volumes of sandy alluvium sediments; siting downstream of dry-season waterholes and water-indicating trees; construction of the sand dam onto underground dykes. In addition, it is suggested to use as maximum specific yield of coarse sand sediments 20-25%. The need to site the sand dams according to highly restrictive biophysical variables may reduce the replication and transfer potential of this type of technology.