Optimising Gully Blocks to Reduce Flood Discharge

Flooding is the most frequent and socially disruptive natural hazard observed worldwide and is expected to increase in severity under climate change and due to urban expansion. This has prompted research in upland natural flood management (NFM) strategies and in using gully blocks as leaky barriers. Gully blocking is often implemented into degrading peatlands, primarily for restoration through water table recovery, erosion control, and soil restoration. However, they are not designed for flow attenuation and there have been relatively few attempts to test their capabilities to attenuate discharge peaks and reduce downstream flood risk. Past efforts to model gully block hydraulics are limited and those that exist have typically applied simple ‘weir’ and ‘orifice’ equations, sometimes tested against field observations of stage and discharge but never (to our knowledge) tested against detailed laboratory observations.

We collected 465 measurements through a series of 20 flume experiments in a 1 x 1 x 12.5 m flume under the range of discharges expected for timber gully blocks in UK gullies (i.e. 10 - 220 L/s). We examined the stage-discharge relationship under steady discharge for a timber barrier with a single configurable full channel width slot, 0.2 m above the bed and with slot height 10 - 100 mm. Mathematical modelling suggests that this design has the potential to considerably improve discharge attenuation relative to traditional gully block designs, but this has not been tested in the laboratory. This design functions in three phases, dependant on upstream pond height: 1) the slot functions as a weir from the point at which it overtops until the free surface reaches to the top of the slot; 2) thereafter it functions as an orifice with this as the only outflow point; until 3) the pond overtops the barrier when this is supplemented by weir flow over the top of the barrier. We find that the first phase weir flow is not well approximated by the classical weir equation, the more complete form accounting for upstream velocity improves the relationship, but resultant stage-discharge curves remain a poor fit to observations. However, both models (with and without upstream velocity) are a good fit to observations for phase 3, where the upstream pond depth is > 565.5 mm for a 10 mm slot barrier configuration. Taken together, these results suggest that weir equations are not appropriate for the shallow upstream depths associated with phase 1 but are appropriate for phase 3. The good news is that phase 1 will be short-lived in storms (early on the rising limb) thus the resulting error will have limited influence on modelling their hydraulic behaviour. In phase 2, orifice equations prove a good model, with both large and small orifice equations providing a good fit to observations and the large orifice equation providing a better fit at smaller upstream pond depths. These preliminary results are an encouraging step forward in pursuit of simple models for gully blocks to inform design optimisation and placement.