Depending on the pressure and temperature of the thermal fluid, water can expand up to 1,100 times its initial volume when it converts to steam. No matter how it gets in your thermal fluid heater, it's your job to get it out.

Two samples from the same system: Sample A (top), from a low-point drain, shows a water layer below the heat transfer fluid, yet Sample B (bottom) show now signs of water infiltration in the fluid. There are a number of reasons that sampling may not show any water.

Maybe the first clue is the sudden appearance of hot thermal fluid in the overflow tank (or on the floor of the heater room). Or, it may be a drop in circulating-pump pressure. Whatever it is, water contamination in an operating thermal-fluid system is hard to ignore. What causes everybody to start scratching their heads is that the symptoms of water contamination don't always follow the rules. They can appear suddenly or over time, at an outlet temperature of 212oF (100oC) or at more than 500oF (260oC). The symptoms can occur when the system is already at operating temperature, or as it is warming up. No matter how or when it starts, it is not good news for anybody involved.

How much water can cause a “problem”? Water is almost completely insoluble in mineral-oil-based thermal fluids, and because it does not mix in, it behaves like water always does -- turning to steam at 212oF and 0 psig. Depending on the pressure and temperature of the thermal fluid, water can expand up to 1,100 times its initial volume when it converts to steam. Another way of looking at it is that a 7 oz glass of water can displace up to 55 gal of fluid when it flashes to steam. No matter how large the system volume, 55 gal of steam is going to cause a problem somewhere. Even if the expansion tank can handle the increase in volume, the resulting cavitation will accelerate pump wear.

Sometimes, it is obvious that water is going to be a problem. Even though every manufacturer warns against it, there are users who insist on using water to pressure test or to flush new systems. Drying the system with a nitrogen purge may look good on paper, but the water doesn't always evaporate within the time allowed.

Sometimes, it's not obvious that water will be a problem. For example, water can enter a system during construction and remain isolated in a cold dead leg for years without causing any problems. Vented systems located in areas with high humidity can become contaminated if the dewpoint is higher than the expansion tank temperature. Water can get into drums that have been stored outside. While a visible layer of water in a sample taken from a low-point drain is a sign of serious contamination, there are a number of reasons that sampling may not show any water. Because the water droplets do not flow readily along the bottom of the pipe, detecting the water in a sample from one loop (or even two or three) is more luck than skill. And, once the main circulating pump has run, the water will disperse into a “haze” which may not be visible if the fluid has darkened. While laboratory testing can detect water to the parts-per-million level, the bottom line is that no matter how scientific the results, you are still going to have to do something about it.

Proactive Approach

Every cold startup and every pump cavitation and/or pressure fluctuation problem should be approached with the assumption that there is water that must be removed if the system is ever going to operate properly in your lifetime. Simply running the pump while gradually increasing the fluid temperature will not completely remove the water -- it only means that the water will find a new place to hide. Many fired systems have some type of separate deaerator to allow gases to separate from the fluid. These are useful for non-condensibles (such as air or nitrogen) but are not effective on contaminants that can condense (such as water).

The vapor has to vent to atmosphere, and the only location in the system that can handle this is the expansion tank. The key to successfully removing all of the water in the shortest time is to make sure that the expansion tank temperature stays above 212oF to prevent the vapor from re-condensing. Every bit of steam that condenses and drops back into the fluid will increase the time required to boil-out the system. Throwing insulating blankets over the top of the expansion tank will keep the headspace walls hot. Adding a low-volume nitrogen purge will help sweep the water vapor from the tank as it is generated and also will help prevent fluid oxidation.

If your thermal fluid heater has one, open the valve on the warmup line that connects the heater outlet to the expansion tank, and also open the manual vent valve on the tank. If your heater doesn't have a warmup line, consider installing one. Open the control and block valves on every loop -- you don't want to go through all of this and miss something. Also make sure that the overflow drum is empty (just in case).

Bring the heater temperature up slowly to 225oF (107oC) and check for steam and pump noise. If all is well, increase the heater temperature 5oF (2.8oC) and again check for steam and allow it to stabilize. Repeat these steps at 5oF intervals until the system is stable with a pump-suction temperature of 250oF (121oC). It doesn't matter what the heater-outlet temperature is -- it's the pump suction temperature that is the most important.

Before you put the system back into service again, make sure that the warmup line is closed. If your heater has a deaerator, it will also have a thermal-blocking tank (or thermal-buffer tank) to prevent hot fluid from entering the expansion tank. Water will collect in this tank during the boil-out. If it doesn't have a drain valve on the very bottom, install one. PH