Coal, Gas

Water Contamination in Hydrogen-Cooled Generators Lurks as Serious Operational Threat

Issue 8 and Volume 107.

By: Ed Borkey, Fluid Energy and Troy Reynolds, Progress Energy

Water contamination in hydrogen gas can lead to significant problems in hydrogen-cooled generators, including electrical arcing and winding failures. The phenomenon of how water is introduced into generators, however, is not well understood, nor is the proper instrumentation normally in place to alert operators to a potential problem. The focus of this paper is to share facts and field observations from more than 80 hydrogen coolant dryers and associated instrumentation, which shows that water contamination may be more prevalent than is commonly thought.

The Need for Pure Hydrogen

Hydrogen gas is used as a coolant for electric generators for two reasons. First, it has the best heat transfer properties of any gas, with a specific heat of 3.4 Btu/lb-F at standard conditions. On a mass basis, this makes it more than 14 times more efficient than dry air for removing heat. Second, it has the lowest atomic weight of any known gas, which keeps wind resistance losses within generators to a minimum. In other words, the wind resistance to generator rotors turning at 3600 rpm is reduced when using hydrogen gas rather than other gases as a cooling medium.

Most generators in use at coal-fired and combustion turbine plants are two-pole units, meaning that they turn at 3600 rpm. Nuclear plant generators typically are four-pole units, which means they turn at 1800 rpm. In either case, it is important to keep pure hydrogen in the generator case to keep overall efficiency high.

Click here to enlarge image

Because air is 14.4 times more dense than hydrogen and water is 8.9 times more dense, it is important to keep these impurities out of generators to keep wind resistance losses to a minimum. For example, on an 800 MW generator, a 2 percent decrease in purity due to water contamination can cost a utility more than $300,000 per year in power sales. This is because frictional losses of the hydrogen coolant within generators increase as the density of the coolant increases (Figure 1).

It is also commonly understood that excessive water contamination in hydrogen-cooled generators negatively affects component life. High relative humidity induces stress corrosion cracking on generator retaining rings manufactured from the steel alloy, 18Mn-5Cr. There is also evidence that newer retaining rings made of 18Mn-18Cr are susceptible to stress corrosion cracking. Replacing these generator retaining rings can cost more than $1 million on larger generators.

Water contamination in hydrogen gas also contributes to lead carbonate deposits, which form as a result of water and carbon dioxide reacting with lead and acid flux within generators. Lead carbonate is a conductive and porous material that can trap moisture, leading to electrical arcing and failure of generator windings.

Hydrogen Dryers

Drying systems based on internally heated regenerative dryers are commonly used to remove moisture from hydrogen gas. Such systems typically consist of a coalescing pre-filter, a loose bed carbon adsorber, a regenerative dryer, an afterfilter, and a blower (Figure 2).

Click here to enlarge image

As with all desiccant dryers, it is important to keep oil and liquid water out of the desiccant beds. Liquid water can disable a desiccant bed; however, water can be regenerated so that the desiccant can be used again. Conversely, turbine oil contamination from the generator will coat the desiccant and ruin it so that it cannot be regenerated. When turbine oil contamination occurs, the only solution is to replace the desiccant.

A desiccant dryer is only as effective as its ability to regenerate itself. As one vessel is drying the hydrogen gas, the other vessel regenerates itself by energizing the heaters within each desiccant chamber and by purging a small amount of hydrogen gas across the desiccant to carry the moisture, and potentially other contaminants, out of the system. Depending on ambient temperature and the amount of water being removed from the generator, varying amounts of liquid water can condense in the purge exhaust drain line. This drain location is an excellent place to gauge the effectiveness of the hydrogen dryer. By measuring the amount of condensate at this location over time, it is possible to accurately calculate the amount of water contamination within a generator. Moreover, this calculation is conservative because the purge gas carries additional water out of the system in a vapor state that does not condense into the purge exhaust drain location.

All desiccant materials are abrasive and create dust over time. To keep desiccant fines from being introduced into the generator, a particulate afterfilter is used. The final component in the hydrogen drying system is a positive displacement, rotary lobe blower that continually circulates about 10 actual cubic feet per minute through the dryer system. This blower is necessary to overcome pressure drop associated with the dryer, filtration system, and piping.

Water Entry

The water content in turbine oil has a huge effect on generator dew point. The alarm point for many utilities is 500 ppmw of water in turbine oil. When water content in turbine oil reaches these high limits, it also creates high dew points in generators, and it does not all occur by oil infiltration into the generator.

Turbine oil pressure is usually kept 10 psi lower than hydrogen gas pressure to keep oil carryover at a minimum. As hydrogen leaks out of the seals, a flow path is created for water vapor to enter the generator. This can be explained by water vapor seeking equilibrium. In other words, the partial pressure of water vapor in the turbine oil is greater than in the generator. As water vapor seeks equilibrium, water vapor enters the generator, even though the hydrogen gas is at a higher pressure. This can be further understood by studying Fick’s Law of thermodynamics that addresses diffusion of gas molecules.

Experience has shown that when water in turbine oil reaches 100-150 ppmw, one to two quarts of water are routinely purged out of the dryer vent line in a 24-hour period. In other words, this is the amount of water that existed in the generator in a vapor state. As discussed above, there is an additional amount of water vapor that is carried out in a vapor state that is not part of this measurement. When water content of the turbine oil approaches 500 ppmw, one to two gallons of water are routinely drained from the dryer vent line each day.

New 150 MW GE generators that have been recently supplied for gas turbine power plants have an internal volume of 2250 ft3. One gallon of water removed from this generator each day equates to 8.3 lb of water. With a specific volume of water of 21.06 ft3/lb at standard conditions, 8.3 lb of water consumes 174.8 ft3 of space in the generator. If this is the only contaminant, the purity is 92.2 percent by volume.

Continuing with this example, 8.3 pounds of water equals 58,100 grains of water. Spread uniformly through this same generator, this corresponds to 25.8 grains/ft3. The resulting dew point would be approximately 106 F. This would undoubtedly create a saturated environment including liquid water in the generator. However, based on extensive analysis of more than 80 hydrogen drying systems, the authors have observed a generator dew point of greater than 100 F on only one occasion. In that case, the excessive dew point was primarily due to a cooler leak on a water-cooled generator. Therefore, when more than one gallon of water is removed from a generator in a given day, water must be continually added to the generator through the seals as described above.

Hydrogen Purity Analyzers Can Mislead

Because hydrogen has the potential to be explosive when mixed with oxygen, it is important to know the purity of the hydrogen gas in air during generator operation or in a stand-by mode. Hydrogen gas analyzers are calibrated to display the purity of air in hydrogen during generator operation. If a given purity analyzer displays 98 percent pure hydrogen, the analyzer assumes that 2 percent of the gas is air. In fact, the impurity can be a combination of air, carbon dioxide, water, and oil vapor. When the impurities are other than air, there is an element of inaccuracy introduced to the analyzer output.

If the carbon dioxide has been thoroughly purged from the generator prior to operation, the likely contaminants then include air, water, and oil vapor. Many hydrogen purity analyzers also include dryers upstream to remove water and oil vapor prior to entering the hydrogen purity analyzer. In these instances, it is more likely that the impurities measured by the analyzer are indeed air.

Operators of hydrogen-cooled generators need to be aware that just because hydrogen purity may display a high reading, this does not mean that the generator is dry. The dryers installed upstream of generator gas analyzers eliminate water as a contaminant and a potential signal of impurity.

As an example, dew point measurements taken at many gas turbine sites over the last several years revealed high purity readings on the purity analyzers that did not coincide with the measured dew points. Typical dew points ranged from +20°F to +70°F, while the purity analyzers usually showed 99 percent and higher. Further investigation of the purity analyzers, and discovery of the upstream dryers, explained the discrepancy.

This can be a particular problem on generators used for peak load conditions. Some generators are equipped with internal electric heaters to keep relative humidity of the hydrogen coolant low, but many utilities are reluctant to operate these heaters for safety reasons.

Generators that are not running and exposed to ambient temperatures are more susceptible to high humidity environments than generators in operation. For example, if a stand-by generator has a hydrogen coolant dew point of 40 F and the ambient temperature is 80 F, the relative humidity is 25 percent within the generator. As ambient temperature drops at night or during cooler seasons, relative humidity increases. In this example, if ambient temperature falls to 40 F, saturated conditions prevail. As cooling continues below 40 F, liquid water condenses into a liquid and may potentially freeze if temperatures fall below 32 F.

Mathematically, a 63 F dew point in a generator corresponds to 2 percent impurity by volume. As discussed above, it is unlikely that this impurity would be accurately displayed on a generator gas analyzer or purity meter. For a new 150 MW GE generator, this 63 F dew point corresponds to one quart of water in the generator. The interesting fact to note is that, in this example, it takes only one quart of water in the generator to create 2 percent of impurity and a high humidity environment.

Possible Solutions

One alternative to installing a hydrogen drying and circulation system is to continue the common practice of “bleed and feed” of hydrogen gas into generators. When hydrogen purity decreases, contaminated hydrogen is vented from generators and replaced with pure cryogenic hydrogen. The problem with this approach is that operators typically are unaware when high humidity environments exist within generators. Remember that to ensure accuracy, hydrogen gas analyzers need to be kept free of contaminants like oil and water. Drying and filtration devices used upstream of these instruments help ensure that the impurity being displayed is air.

The “bleed and feed” approach could also be used if dew point or humidity were known within the generators. However, utilities rarely have hygrometers installed on generators, and are often reluctant to install hygrometers because of the potential for sensor contamination. When operators question the accuracy of instruments, action is rarely taken based on the output provided.

Click here to enlarge image

The optimum solution is to install hydrogen-drying systems as shown in Figure 3. With this approach, water is removed from generators 24 hours per day. Because of the filtration used on these systems, it is now possible to measure both generator and dryer dew point, and alert operators to potential problems.

From a financial perspective, the cost to install hydrogen drying systems ranges from $60,000 to $100,000. The variance depends somewhat on equipment options selected on the drying system, but mostly on the varying costs of installation. Consideration must be given to the hydrogen piping into and out of generators, the availability of power, and location of the systems. In any case, the relative low cost of installing these systems versus the high cost of generator downtime, replacing expensive generator components, and increased wind resistance losses, makes installing hydrogen drying systems a sound investment.


Edwin C. Borkey, P.E., is General Manager for a distribution and service company called Fluid Energy, and a separate compressed air consulting business called N2O2. He has a BS degree in mechanical engineering from Ohio Northern University and is a registered Professional Engineer in North Carolina.

Troy Reynolds is an electrical engineer with Progress Energy. He has 30 years of experience in the power generation industry and holds a bachelor’s degree in electrical engineering from North Carolina State University.


N. L. Kilpatrick, Experience with In-Service Examination of Nonmagnetic Rings on Generator Rotors, Westinghouse Electric Corporation,

Westinghouse data from Drawing Number KHS 831207

General Electric data from Drawing Number 334 HA923

Troy Reynolds, Generator Hydrogen Dew Point Final Test Report, Carolina Power and Light, 1993

H. Feichtinger, G. Stein, I. Hucklenbroich, Case History of a 18Cr18Mn Retaining Ring Affected by Stress Corrosion Cracking, 1997