Answer: Air bubbles are commonly cited as sources of error, but with proper installation, both factors are easily controlled. By far, the most significant sources of error are barometric pressure (when using a non-vented VW sensor) and temperature variations.
Air bubbles: If air bubbles are present in the liquid, they can cause errors, because air is lighter and compresses more easily than liquid. However, the cell leaves the factory pre-filled with de-aired water. Special headers at the reservoir end of the tubing are overfilled, so the liquid is under pressure and remains under pressure until the final step when the tubing is connected to the reservoir. At connection time, the tubing squirts out liquid rather than sucking in air. Thus, air bubbles are not usually a problem.
Temperature: Temperature problems can be controlled by minimizing the length of tubing that is affected by the ambient temperature. Any tubing that is above grade can also be insulated to help reduce the temperature effects. Also the reservoir should be protected from the direct heat of the sun. It is also important to make sure that the reservoir is vented to the atmosphere. If the vent or the overflow tube is blocked, it is possible for ambient temperature changes to create either a vacuum or pressure within the reservoir.
Liquid Level Maintenance: It is important to maintain the level of liquid in the reservoir. Typical practice is to replenish evaporated liquid with water each time a reading is taken. To do this, fill the reservoir until the water flows out of the overflow tube. Ordinary water is sufficient. It does not need to be deaired, and it does not need to be mixed with ethylene glycol, since that component of the mix evaporates much slower than the water does.
Barometric pressure: The reservoir is open to atmosphere, so the pressure of the atmosphere acts on the surface of the liquid in the reservoir, and the transducer sees the combined pressure of the column of liquid and the atmosphere. This does not affect the vented VW settlement cell or the pneumatic cell (provided the pneumatic indicator is zeroed at reading time), but it does affect non-vented VW cells. What is the magnitude of these changes? They are many times larger than the resolution, accuracy, and precision of the system. A 1 millibar (0.75 mm Hg) change in atmospheric pressure is equivalent to 10 mm ( 0.4 inches) of water head. Weather fronts bring much greater changes.
The graph above shows 1.5 days of data from a settlement cell installed at the Isle of Wight. Atmospheric pressure was recorded by an electrical barometer. You can see how the settlement cell tracks changes in the atmosphere. The offset between settlement reading and the barometer reading is the head of water. As you can see, the settlement cell closely tracks the barometer.
The graph above shows a month of data from a long term test in the laboratory. Barometric pressure was recorded by a VW piezometer. During this month, atmospheric pressure varied over a 20 millibar range and was tracked by an equivalent change (300 mm or 11.8 inches) in head of water even though no settlement actually occurred. Again, the difference between settlement cell reading and the barometer reading is the head of water (about 2.5 meters).
As mentioned above, if you have a vented VW settlement cell or a pneumatic settlement cell, you do not need to concern yourself with barometric pressure. However, if you have a non-vented cell and wish to monitor small settlements, it is necessary to measure barometric pressure and to compensate settlement readings for changes in barometric pressure. Changes in barometric pressure can be very localized, so it is not sufficient to obtain barometer readings from the TV weather report or from the barometer on the office wall. Barometer readings must be obtained on site at the same time as the settlement reading. If you are using a data logger, then it should be programmed to read a barometer at the same time as the cell. If you are obtaining readings manually, you should use a high-quality, hand-held barometer to measure barometric pressure.