University CrestEdinburgh Wave Power Group

Wide Tank 1977 - 2001




The Institute
The School
The University




The wide tank was housed in a temporary building that was attached to the James Clerk Maxwell Building (JCMB) at the Kings Buildings campus of the University of Edinburgh.

1977: Wide tank goes up
Bullseye wave
2001: Wide tank comes down


Two documents describe the Wide Tank and our approach to its design:

4th year report, Volume 1 of 3, The wide tank

Absorbing wave-makers and wide tanks


These photos show Wide Tank waves and sea-states

Many video sequences show the Wide Tank


We built the Wide Tank in 1977 for research at scales between 1/150 and 1/100 into the generation of electricity from ocean wave energy. It was designed to accurately represent multi-directional Atlantic wave activity in areas of up to 3km full scale width. It was the first multi-directional wave tank with absorbing wavemakers. In 2001 the Wide Tank was dismantled becasue the University needed the site for a long-postponed building project. 48 of the wavemakers were subsequently incorporated in to the new Curved Tank.

The nominal water depth was 1.2 metres, the internal length was 10 metres and the width was 27.5 metres. 'Width' referred to the dimension parallel to the wavemakers - hence the name 'wide tank'. The original waterproofing consisted of a loose lining made up from rolls of a polyurethane material which had a bonded-on terylene felt backing. This lining was contained within modular concrete retaining walls The terylene / polyurethane lining was later replaced by a light blue glass-reinforced polyester lining .

Along one width of the tank were 89 wavemaking paddles, of which approximately the first 75 were usually driven. (See 'wavemaker platform limit' line on the drawing). Along the opposite side were 2 metre long beach cages using expanded aluminium foil as an energy absorbtion medium. (See 'beach limit' line on the drawing). The upper part of most of one short side of the tank was glazed to provide an underwater view. The opposite short wall of the tank usually had a row of beaches along it.

The working area was up to 25m wide by 5m long. The wavemaker pitch was 12 inches (305mm).

A gusseted plastic membrane formed a water seal between the paddles and the tank. Water was displaced only by the moving front surface of the paddles. There were no rear waves to cause resonances and compromise wave quality. The paddles were driven by DC motors with force and velocity feedback to linearize response and maximize absorbtion of reflected waves arriving back at the paddles.

Wave fronts & seastates

Tank sea states were created as single wavefronts or as combinations of finite numbers of wavefronts.

Wavefronts were specified by:

frequency, period or wavelength (eg: 1Hz)
amplitude or height (eg: 1cm)
angle of travel (eg: -30°)
starting phase

Single wavefront sea states provided regular, monochromatic or 2-dimensional waves. Additional wavefronts gave 2 and 3-dimensional sea states with specified spectral content.

Wavefront frequencies were generally in the range 0.5 to 2Hz. Angles could be up to ±90°, although at extreme angles wavefronts tended not to propagate into the tank working area.

The 'bullseye' wave (shown here on the glass window of the wide tank) was composed of many wavefronts, each having the same period (0.8 seconds), but different angles.   The starting phases were chosen to bring each wavefront into phase at the same point on the glass.


The wavemaking software of 1977 had to run in real time as memory was very expensive. (Our best computer had 64 kB of RAM and a 20MB hard disk drive cost more than a year's salary). The Plessey Miproc was the fastest micro-computer that we could afford and it allowed us to generate seas with up to 70 'wavefronts'. There were 89 wavemakers and their commands were clocked in at 20 Hz. By the 1990's we were using the software from our spin-out company Edinburgh Designs Ltd. There was no practical limit to the number of wavefronts in a sea state because memory was now cheap and so the wavemaker command sequences could be produced off-line. The greater freedom allowed us to make two interesting new seas: 'sneak' and 'surf'.

Collections of sea states were defined by writing text files in a special compiling syntax not unlike C or Pascal. Standard ocean spectral algorithms such as Pierson-Moscowitz and Jonswap were included as high level functions, and wave directionality was usually specified with a Cos2n function.

Special waves

Wavefront starting phase was normally assigned randomly by the wavemaking compiler. However explicit manipulation of starting phase allowed statistically uncommon wave features (eg: 50 year wave) to be forced into sea states at specific times and places. Thus the following line: makewave focus (cosn(pm(1.0)*3,5),20,4,4) specifies a 1 second Pierson-Moscowitz sea, with cos5 spreading and includes a freak wave 20 seconds after sequence start at tank coordinate 4,4.

This technique also allowed easy implimentation of scarcely imaginable real world sea states such as repetitive deep water plunging breakers.

Repeat time

All sea states were pseudo random with their repeat times specified in the corresponding source files. A commonly used repeat times was 64 seconds, but this coul be extended indefinitely. All the features of complex sea states repeat with great precision which for most experimental work was greatly advantageous. During wavemaking a pseudo wave-gauge facility in the control computer could provide calculated surface elevation time-series for any point in the tank.