For a float to drift at a certain depth two requirements must be met. First, its weight must be adjusted to match the weight of the water it displaces – what is known as Archimedes’ principle. Second, it must be less compressible than water. This, so that if it is displaced from its equilibrium depth, downward say, it will not compress as much as the surrounding water resulting in a restoring buoyancy force. If the float is to follow an isopycnal surface, we must add a further requirement, that its compressibility match that of the surrounding water. We do that with a compressee (see blog on the compressee). For a float to remain on the same isopycnal we depend upon the fact that the ocean is thermally stratified such that water gets denser with depth due to decreasing temperature. But the resulting restoring forces are smaller, so it becomes far more challenging to target it to the desired density surface.
The reason is that while the mass of the float is well-known, its volume is not. What I mean is that only at pressure will the various parts of the float be fully pressed against each other and any air pockets hiding in o-ring grooves and crevices be fully compressed or dissolved. Thus, floats must be ballasted at pressure. In short, it is easy to build an isopycnal float. Putting it on a desired isopycnal, i.e., density surface requires a pressure vessel. Fortunately, we had a huge one here at GSO, a bit oversized, but otherwise well-suited to our needs. Jim Fontaine, engineer and master of all skills needed in a modern laboratory, developed the routine of ballasting into a fine art.
The challenge is to measure the weight of the float with its compressee at pressure in fresh water. We know water density quite accurately given pressure and temperature. Before closing the tank, we adjust the float’s mass and hang a light chain below it so that only the lowest part of the chain rests on the bottom of the tank. When we apply pressure, the float may sink slightly at first as the remaining air pockets are dissolved. After that the float will start to rise with increasing pressure since fresh water is more compressible than the float which by design is matched to that of salt water. With a camera in front of a window in the tank we can see how much the float has risen and thus chain it is lifting and hence the mass needed to match the weight of the water it is displacing. Given the difference in density between the water in the tank and the target density we can determine the additional mass needed to achieve neutral buoyancy at the desired isopycnal in the ocean. While this routine worked well, there were two shortcomings. First, the tank cover was big and heavy with umpteen massive bolts to tighten each time we closed it. Second, the chain had a tendency to pile up on the bottom such that the float wasn’t lifting quite as much as we might think. Also, the tank water had a tendency to stratify – leading to uncertainties in estimating water density.
Jim solved both issues beautifully when we got a 12’ long cylindrical stainless steel pressure vessel. It was a beauty. First, he excavated a hole in the ground for the tank so that its upper end was at elbow height making it easy to service the tank: to insert and remove a float, and to lower the tank cover into place. Second, the pressure vessel sat inside a larger diameter PVC tank in which thermostated water was circulated enabling us to control the temperature of the tank thereby preventing the water from stratifying. With a simple pulley system, the heavy stainless-steel lid of the tank could be lowered into place and lifted to the side with little effort. The pulley served also to lower the nearly 2 m long float assembly into the tank. But, most importantly, he replaced the chain with a titanium beam from which the float would hang. The change in float weight with pressure would be measured with strain gauges measuring the bend in the beam. I was a bit skeptical that this would work given the enormous pressures bearing down on the strain gauges. But the four sensors were wired to cancel out the effects of both temperature and pressure. It was surprisingly effective, stable, and accurate. We know this because at the start and end of each day of float ballasting, Jim would hang a standard float to calibrate the strain gauge beam. I think he used the same beam for the entire history of tank usage.
To adjust the weight the float so that it was just a few grams heavy, he placed a metal basket of known mass on top of the glass pipe and added small, calibrated weights as necessary. The basket had a magnet which he coupled to a magnet hanging on the beam. The small magnets served as a mechanical ‘fuse’ that would part if there was risk of the beam being overloaded. The entire cycle of pressurizing the tank, cycling pressure up and down several times and recording all data was down automatically under computer control. Mark Prater write the code for determining the exact compressibility of the float and the exact weight needed for neutral buoyancy at sea. They did a fantastic job. To give you an idea of this, consider that the float weighted typically about 10 kg yet with the strain gauge beam they could determine the weight of the float to within a few tenths of a gram. In other words, the float’s weight was known to a few parts in ten to the 5th power!