During the rainy season of 1999, the spillway of the LA dam was about to overflow, due to high water levels in the reservoir. Since the dam’s construction in the 1960s the gates of the spillway had never been opened at the maximum water level. Because the system ‘gates-piles’ is so slim (each gate is 20m high and 8.3m wide (see Figure 1)), a physical model was studied to discover the hydraulic behaviour of the water flow near to the gates. The strategy of operating the spillway was to totally open the central gate with both laterals closed. The main interest was to detect the possible vortices formation in the neighbourhood of the ‘gates-piles’ system. In the bibliography, any study directly related to this case was not found. The most similar problem was reported by Bayazit [1], who was interested in possible resonance problems, although this work is related to vibration problems in partially open gates.

Due to the operation uncertainty of the system ‘gates-piles’, it was considered convenient to carry out a hydraulic study. In the preliminary tests of the physical model, vortices were detected near the closed gates. The natural frequency of the system ‘gates-piles’ is unknown; therefore to suppress the vortices was considered the best solution to prevent any potential hazard.

Problem description

As indicated by Knauss [4], the occurrence of vortices is accentuated when the approach channel has a pronounced curvature, as is the case of the LA dam spillway (Figure 1).

Naudascher [5] and Kisiolov [3] established that in the proximity of the discharging gates, vortices are formed. In the latter reference it is indicated that the vortices occur within a certain range of the ratio opening/head in such a way that when it is either very small or very big, vortices do not occur. In the physical model, which is described later, it can be observed that when discharging is only via the central gate, Mach waves did not cause problems along the discharge channel, therefore this policy was selected for the spillway discharge. It was thought that completely opening this gate would avoid the formation of vortices. However, as will also be seen later, this method did not completely eliminate vortices.

At the Institute of Engineering in Mexico, a physical model of the LA dam spillway (with a scale of 1:50) was built, which permitted a first analysis when the spillway operates. It could be seen that by fully opening the central gate of the spillway and leaving the two lateral gates closed, the vortex occurs in the vicinity of the piles forehead (in front of the closed gates) and comes out by the opened gate. It was also observed that if the central gate was opened completely and one of the lateral is partially opened, the vortex approached the partially open gate and left through it.

To attenuate the formation of vortices, initially it was thought to install a wood crosspiece which could float, upstream from the gates, to follow the evolution of the water level. With this device it was possible to attenuate the action of the vortex, but not totally eliminate it. Different devices for vortex elimination were tested; the best performance was obtained with a grid as shown in Figure 2. This grid could be made of metal (preferably stainless steel) or of wood, to avoid excessive weight, considering that the device ought to be fixed in place to avoid hitting the gate. As will be seen later, by putting the grid in place, an almost total calm was obtained.

It is important to clarify that, in spite of the device dampers installed and due the big dimensions of the model (located outdoors), it is not possible to achieve complete calm and eliminate the natural water oscillations. This occurs when the three spillway gates are closed and the water level is controlled using another discharge structure.

The reason for studying the spillway discharge in a physical model was to detect the possible vortices development near the floodgates. In this work, it is considered that scale modelling of vortices is not possible because there is no single similitude law that accurately correlates observed model vortices to those observed in prototype. The problem is that a Froude-based model does not accurately reproduce hydraulic phenomenon, such as vortices, which are affected by Reynolds and Weber numbers. Consequently, the vortex evaluation from the model study is qualitative.

However, according to Hecker [2], when a vortex occurs in a model it is very probable that it will also occur in the prototype. As the main objective of this work was to detect the possible formation of vortices and eliminate them, then the utility of the physical model was considered enough.


Five tests were carried out in all. Water level was maintained in the proximity of the elevation 539.7m (prototype), and the signals were captured by two pressure sensors (“Omega” mod PX541-003BGI). These sensors were set at two different levels placed in a lateral gate – one near the water level (above) and other near the floor (below) (see Figure 3). The sensors then sent the captured signals to a spectrum analyser (HP3582A). In each spectrum of Figures 3-7, frequency is in Hz and energy E in mV (1mV = 2252Pa, in the physical model).

First test

The three gates of the spillway were closed and discharge was made by another structure. In the corresponding spectrum, shown in Figure 3, it may be seen that there are no significant peaks in the sensor spectra. Hence the oscillations, due to the water level when there are no vortices, do not appear to cause important periodic fluctuations in pressure.

Second test

The two lateral gates of the spillway were closed leaving the central gate completely open. In the corresponding spectrum shown in Figure 4, it may be seen that the lower sensor begins to capture a peak, becoming significant towards the frequency of 0.75Hz, ie. with the central gate completely open the action of the vortex is not suppressed in those closed.

Third test

The operation conditions are almost the same as those of the second test, except that the gate where the sensors are installed has an opening of 2.5m (prototype). In Figure 5, the spectrum registered in this case is shown in which it may be seen that on partially opening the gate there is a pronounced peak towards the frequency of 0.75Hz.

Fourth test

This is similar to the previous test, but the gate has an opening of 4.25m (prototype). In this case the spectrum registered, Figure 6, shows a peak even better defined and a little larger than in the third test, towards the same frequency of 0.75Hz. So it may be concluded that, in the partially open gates, the vortices which are generated in their proximity produce periodic variations in pressure and of greater amplitude.

Fifth test

Once again, the two lateral gates of the spillway were closed and the central one was completely open, but now a grid similar to that shown in Figure 2 was placed in the opening of the gate were the sensors were installed. The spectrum registered in this case is shown in Figure 7. It may be seen that for neither of the two sensors are accentuated peaks registered. Even more, if this spectrum is compared with the one shown in Figure 3, in the total absence of vortices, not only can the similarity of these two spectra be seen, but also that the grid slightly attenuates the oscillations of the surface of the water.

According to the results of the five tests, it may be seen that when the vortex comes closer to the open floodgate, periodic fluctuations in pressure of larger amplitude take place, which are possibly due to a certain precedence movement of the vortex, or to rhythmic changes in the intensity.

When the central gate is totally open and the lateral ones closed, the vortex moves away from the floodgates and diminishes its action evidently but it does not disappear. On the other hand, when the grid is placed in front of the floodgate, practically total calm is obtained.

Finally, it should be remarked that the problems mentioned in the previous paragraph show the convenience of placing grids in front of each gate of this spillway, to ensure that they operate with an ample security margin.


In the physical model vortexes appear near closed gates when the central one is totally open. It is recommended vortexes be eliminated to prevent any type of problem in the piles-gates system. Different devices were tested and it was concluded that grids are the best system to eliminate vortexes appearing in front of the gates. The grids should be placed as shown in Figure 2, in the proximity of the three spillway gates in order to protect not only the lateral ones, but also the central one during the time in which it is opening or closing.

For the prototype, the principal recommendations should be to focus first on assuring that the central gate can be open and after be closed, because it can be blocked. When the grids are installed, it will be necessary to revise periodically the state of each grid to give adequate service when necessary.

At present, the grid is only a case of study. But the authors considered that the experience is interesting to report, in spite of the design method being unavailable. For now, trial and error is the method recommended to select grids.