Dam-break project - Data and MATLAB codes

This website features the dataset from flume experiments on dam-breaks conducted at Politecnico of Turin.  The experiments were conducted at the Giorgio Bidone hydraulics Laboratory in Politecnico di Torino, Italy. The flood wave channel, the dam and water release mechanism, the rod canopy comprising the vegetation, the water level imaging system and data acquisition, and the test runs conducted are now described. 


The flood wave channel

The 11.6m long prismatic channel used here has a rectangular cross-section that is 0.5m (=B) wide and sides that are 0.6m in height. The smooth concrete channel bottom is elevated 1.27m from the ground floor. The channel sides are made of glass to permit optical access. The glass sides are further enforced using a steel structure. This steel structure does not allow optical access of the 0.035m nearest to the channel bottom. A mechanical wheel allows the channel to rotate around a pin that can be adjusted so as to vary So from 0% to 3%. The channel is filled directly with water from below by a pipe and the outflow from the channel discharges into a tank after passing over a rectangular weir.

The dam break

A wooden cofferdam with an instantaneous opening is used to model a dam-break. The wood is waterproofed as this treatment allows the wood not to deteriorate during the experimental duration. The cofferdam is fixed on an aluminum double T-support and is free to move up and down through a vertical railing structure attached to the steel body of the facility. A pneumatic cylinder is fixed on top of the vertical structure and powered by a compressor located on the floor. The compressor directs an 11bar pressure to the pneumatic cylinder forcing a disc to move rapidly upwards. The disc is connected to the piston rod, which in turn is fixed to the cofferdam frame. This system uplifts the cofferdam at a speed of 0.86 m/s thereby mimicking the instantaneous release of water into the flume following dam break.

The vegetation

The vegetation immediately downstream from the dam is composed of an array of a polymeric resin cylinders. The cylinders are fixed onto six plastic boards each 0.15m wide and 1.75m long. To cover the entire cross-section, the boards are positioned side-by-side three at a time for a total length of 3.5m. The panels are attached to the channel bottom using silicon. This attachment allows the rods not to move during the test runs. The cylinders comprising the rod canopy are rigid with uniform diameter D=0.006m and height hc=0.10m. The rods are arranged in a staggered configuration with a spacing of 0.035m transversely and longitudinally, while the distance to the diagonal is 0.0175m. This arrangement resulted in a density m=1206 rods m^(-2). The no-vegetated case in the same facility is also tested.

Water level measuring equipment and data acquisition

The main variable measured here is water level h(x,t) variations along the channel at regular temporal intervals. To obtain h(x,t) without flow interferences, three Sony Handycam HDR-XR500 cameras are used to image the water surface profile. Each camera is equipped with a 3-3/16" wide-screen touch-panel LCD, a Sony's premium G Lens, and a remote control to start all cameras concurrently. This camera model is able to record high-definition AVCHD video and store it in a 120GB hard disk. The space-time resolution used in the experiment is the best available from such a camera model (1920 x 1080 pixels at 29.97 frames per second). The cameras are situated on a horizontal bar at a distance of 1m from each other. They are aligned with the bottom of the channel when the slope is 0%. The distance between the cameras and the side-glass is 1.5m thereby allowing each camera to record a movie of the full glass in its field of view. The three cameras cover a total length of 3m starting from 0.5m upstream of the dam. To avoid reflections from windows, two black cloths have been placed behind the cameras and behind the flood wave channel. Since water is transparent, it is difficult to automate the detection of the water surface profile from images without additional markers. For this reason, water was mixed with a Rhodamine dye that becomes fluorescent and emits red light when being excited with light at different wavelength (green light is used here). The green light is emitted by two laser generators with 200 mW power fixed over the channel on two supports welded to the metallic frame of the facility. Each laser emits a narrow beam of green light that crosses a glass cylinder with a diameter of 3mm. When the light crosses the cylinder, it is refracted and generates a plane of light perpendicular to the bottom of the channel with the same direction as the flow. The addition of such a dye enhances the imaging and automated detection of the water surface.

Test runs and slope/dam water level configurations

The test runs were performed using four differing static water levels behind the dam (Ho= 0.15m, 0.20m, 0.25m, 0.30m) and four differing bed slopes (So= 0\%, 1%, 2%, 3%) resulting in a total of 16 configurations. The 0% slope configuration was repeated ten times for each Ho thereby allowing the acquisition of statistically robust water level data not affected by outliers. The outcome of the analysis showed a low standard deviation between different water profiles after 5 replicas. This led to a decision of performing only 5 replicas per Ho and S_o configuration. Hence, water level data for each of the 16 configurations are presented as averages of the 5 water level replicas. For each test run, the channel slope is first configured to one of the four So values. Prior to commencing a test run, the gate is closed so that a water reservoir is established behind the dam. The reservoir is filled until the desired Ho is reached. The remaining portion downstream from the dam is initially dry. The Ho is measured by a hydrometer fixed to the glass panel of the flume facility. The water behind the dam is then mixed with a precise amount of Rhodamine calculated in relation to the volume of water stored. The goal is to reach a color that has the same shade of red for each experiment. Once the wave channel is set, the next step is to prepare the water level imaging equipment. The two lasers are started by turning their activation key. The compressor connected to the hydraulic piston is turned on with a switch that allows it to acquire 11bar pressure rapidly. The three cameras are turned on simultaneously with a remote controller. The test run is initiated when compressed air is pumped into the piston through a rubber pipe pulling the wooden gate of the dam up and ends when all the water is discharged. The acquired movies are converted to images and then analyzed using MATLAB (Mathworks, Natick, Massachusetts, USA). The analysis transforms the detected water level from pixel coordinates to metric coordinates thereby providing h(x,t) for each run and all 16 configurations. Each run lasted from 7-10s with the flood wave passing the entire imaged sections by the 3 cameras in 4-5s. Measurements for the non-vegetated case were conducted for Ho=0.15m, 0.20m, 0.25m, and 0.3m but for a flat slope. 

Processed Data

The data files can be download from Dropbox at this address




The file is structured in 4 subdirectories (one subdirectory per slope). Each file is a matrix that contains water level h (cm) and position x (cm) at fixed time intervals (deltat=0.09s).