Compressor washing, both on-line and off-line, is vital for maintaining gas turbine performance. But what is the best cleaning agent and how often should it be done? Meherwan Boyce, Boyce Consulting Group, and Francisco Gonzalez, Cheniere LNG O&M Services, review techniques and report on the results of recent tests in the field. It turns out that plain demineralised water performs just as well as soap solutions.
With ever higher pressure ratios being employed in gas turbine compressors, the cleaning of compressor blades by on-line water washing becomes an increasingly important operational requirement. In many plants this operational procedure has contributed literally hundreds of thousands of dollars to the bottom line of the plant.
Typically, the compressor used in gas turbines is of the axial flow type, with variable vanes in the early stages. In the case of the compressor shown in Figure 1, the IGV, followed by the next three rows, have variable vanes.
However, on-line water washing is not the full answer since after each wash full power is not regained, therefore a time comes when the unit needs to be cleaned off-line, as seen in Figure 2.
The effect of compressor fouling is also very important in terms of overall performance of the gas turbine since the compressor uses nearly 60% of the work generated by the gas turbine. Fouling of the compressor reduces compressor efficiency and leads to a reduction in the overall efficiency. The higher the pressure ratio the greater is the adverse effect of compressor fouling on thermal efficiency, as seen in the Figure 3.
The initial protection for the gas turbine is a good air filtration system that must be used before the air enters the turbine. However, a fraction of airborne salt always passes through the filter. The method recommended for determining whether or not the foulants have a substantial salt base, is to soap wash the turbine and collect the water from all drainage ports available. Dissolved salts in the water can then be analysed.
The optimal time at which to carry out off-line cleaning needs to be determined by taking into account loss of income due lost power sales, cost of labour for off-line cleaning, and additional revenue arising from improved performance.
Two approaches to compressor cleaning are abrasion and solvent cleaning. The use of abrasive cleaning has diminished due to erosion problems and liquid washing is now primarily being used. The new high pressure compressors are very susceptible to dirt on the blades which not only can lead to a reduction in performance but can also lead to compressor surge. Washing efficacy is site specific due to the different environmental conditions at each plant. There are many excellent techniques and systems for water washing. Operators must often determine the best approach for their gas turbines. This includes consideration of what solvents if any should be used, and the frequencies of wash. This is a complex technical-economic problem also depending on the service that the gas turbines are in and the plant surroundings.
Off-line water washing (with or without detergents) cleans by water impact and by removing the water-soluble salts. It is important that the water used should be demineralised water. The detergent/water ratio is also another important parameter. Water washing using a water-soap mixture is an efficient method of cleaning. This cleaning is most effective when carried out in several steps, which involve the application of a soap and water solution, followed by several rinse cycles. Each rinse cycle involves the acceleration of the machine to approximately 20%-50%of the starting speed, after which the machine is allowed to coast to a stop. A soaking period follows during which the soapy water solution may work on dissolving the salt.
On-line washing is being widely used as a means of controlling fouling by not allowing it to develop in the first place. Washing can be accomplished by using water, water based solvents, and petroleum based solvents or surfactants. The solvents work by dissolving the contaminants while surfactants work by chemically reacting with the foulants. Water based solvents are effective against salt, but fare poorly against oily deposits. Petroleum based solvents do not effectively remove salty deposits. With solvents, there is a chance of foulants being re-deposited in the later compressor stages.
Even with good filtration, salt can collect in the compressor section, as already noted.
During the collection process of both salt and other foulants, an equilibrium condition is quickly reached, after which re-ingestion of large particles occurs. This re-ingestion has to be prevented by the removal of salt from the compressor prior to saturation. The rate at which saturation occurs is highly dependent on filter quality. In general, salts can safely pass through the turbine when gas and metal temperatures are less than 1000°F. Aggressive attacks will occur if the temperatures are much higher. During cleaning, the actual instantaneous rates of salt passage are very high together with greatly increased particle size.
Some gas turbine performance deterioration can be recovered by engine cleaning, while performance deterioration due to internal engine component wear (Figure 4) can only be recovered by performing a shop inspection and engine overhaul.
The primary factors governing gas turbine performance deterioration that is recoverable by cleaning are the quantities of contaminants entering the turbine (via the inlet air filters and ducts, water from evaporative coolers, and from fuel) and the frequency as well as the thoroughness of engine water washing. At times unusual site conditions can accelerate gas turbine degradation. Unusual airborne contaminants from process mists, smoke (eg, from burning of sugar cane), oil, chemical releases, and dust storms, have for example been documented as causes of accelerated engine degradation.
A site-specific test programme should therefore be conducted in order to optimise the effectiveness of a turbine water wash programme. As the gas turbine performance deteriorates, the following conditions are usually exhibited: slow engine acceleration; tendency for compressor to surge; lowered power output; loss of engine compressor discharge pressure; increase in engine compressor discharge temperature; and decrease in compressor discharge pressure.
There are essentially three different types of gas turbine compressor wash systems: on- line wash system; off-line crank wash system; and manual hand held crank wash system.
For the most part, on-line wash systems are intended as a supplement to off-line crank wash systems and not a substitute. In general, it is extremely important to implement a turbine crank wash programme in order to recover most of the recoverable performance deterioration.
At times when turbine crank washes are put off due to operating constraints, there may be the need to manually clean the turbine compressor blades to remove large build-ups of dirt on the blades.
Testing has shown that an additional five efficiency points were recovered in a Mars turbine when an off-line water wash was performed.
On-line washing is used while the turbine operating parameters are stable. The system can be used without disturbing the operation of a unit, and it does not matter if the turbine is operating at part or full load. On-line washing should be a routine and scheduled maintenance function.
On-line washing normally involves injecting of atomised cleaning fluid thus avoiding any problems that could be associated with abrasive cleaning methods, which could erode blades and damage component coatings.
For an off-line crank wash the turbine is manually cranked by the engine starter with the fuel and ignition system de-activated. As already noted, this type of wash is more effective in recovering performance degradation. Before performing an off-line crank wash, most of the low drain piping, together with igniter, torch and pilot gas systems, etc, are removed in order to avoid liquid pockets in the turbine fuel igniter piping. Manufacturers recommend different water and solvent-based washing fluids, and field technicians have tried many different combinations of relatively widely available household cleaning agents.
A preferred method for performing a crank wash is to employ a hand held wash wand. Several of the turbine inlet air inspection covers have to be removed in order to obtain access to the compressor inlet. A hand held wand is then rotated around the inlet turbine screen in order to spray the compressor air inlet evenly.
Hand scrubbing the compressor blades is another method of recovering the performance of an engine compressor. This is very labour-intensive but helps recover an additional level of performance. Not all gas turbines compressors can be inspected in the field, and the costs and benefits have to be evaluated carefully before this method is undertaken. Currently we perform this type of cleaning on a Solar Mars 15000 SoLoNox turbine that has to operate 11 to 12 months continuously without any shutdowns while minimising power loss. The resulting expected engine compressor polytropic efficiency improvement is about 0.5 to 1.0 efficiency points. In an engine that has to run for 11 months without a crank wash this extra efficiency provides a valuable increase in fuel savings and power delivered.
Water quality for both on-line and off-line water washes must be stringently managed, so as to ensure that impurities are not introduced. The table below gives a typical specification for the quality of the water required to complete a successful water wash:
Testing the quality of the water is critically essential before performing a routine on-line water wash. A common problem with demineralised water sources is the occasional fluid channeling in the catalyst beds. A demineralised water polishing skid is recommended for the turbine water wash in order to ensure good water quality.
The effectiveness of on-line water washing is greatly improved by paying attention to testing of the water quality. Detailed analysis of the water for every water wash is time consuming and expensive, but as an alternative a hand held meter can be used to measure conductivity, total dissolved solids and pH. The meter readings should indicate a water conductivity of <0.5 μmho, total dissolved solids of <1.0 ppmw and pH between 7 and 9.
The most popular water base solvents for on-line and off-line compressor washing have low metal contents, and are derived from highly active natural oils and surfactants. The water base agents are environmentally benign and contain no solvents that are capable of dissolving and removing engine deposits that accumulate with time. Other industrial cleaning agents such as Mr Clean have been known to work well in combination with water base agents and solvent base agents. Most turbine manufactures offer to the users an approved list of soaps that can be used with their turbines.
Many different types of solvent base agents are also used for on-line and off-line turbine washes. Most solvents are derived form hydrocarbon base stocks, and contain very little metal. Most of these solvents are considered to be not fully environmentally friendly, but with the proper procedures these agents can be used in a safe manner. These solvents are predominantly used during off line crank washes.
By far the most common agent used for water washing is demineralised water. Demineralised water systems are very effective if the filtration system effectively filters out the majority of airborne particles and oily substances.
To help determine optimal procedures for compressor washing, a study was carried out on three turbines in a facility employing 36 gas turbines for various purposes, 28 in mechanical drive applications and eight for power generation. The facility was chosen because the turbines were all similar, from the same manufacture, Solar, and, being on the same site, atmospheric and air pollution conditions were the same for each unit.
The tests were focused on identifying the most effective solvents and optimal frequency of washes.
The turbines are used in a hot oil combined cycle mode, with recovery of energy in the waste heat. Most of the gas turbines are in the 4000 hp size range, while the largest is a 15000 hp Mars machine driving a process gas axial compressor. Thirty-three of the thirty-six gas turbines were equipped with on-line water wash capability.
The turbines used for all of the on-line water wash tests were Solar T4702S Centaur machines. All of the turbines are configured identically, and all of the turbines have about the same number of operating hours.
The turbines are equipped with inlet air pre-filters and with primary filters made by Donaldson filters. The filters are configured with cylindrical/conical filter synthetic media. The filtration system is a “huff and puff” filter system. The filter efficiency is 99.5% for particles of 1-3 microns. The average pressure drop across the filters is 2.4 inches of H2O.
The Centaur engines coupled to a speed reducer gearbox and to a generator are capable of producing 3 MW. The turbine exhaust is routed to a heat recovery unit that heats up a hot oil media that is used for process purposes. The typical exhaust gas temperature is around 950oF before the hot oil coils and 200oF after the coils.
All of the gas turbines are equipped with an on-line water wash system, see Figure 5. The on-line wash system consists of a wash ring located outside the turbine inlet air plenum, with several pigtails connected to eight atomising type nozzles, which are aimed at the inlet guide vanes, as shown in Figure 6.
Figure 7 shows an inside view of the turbine inlet air plenum showing the off- line water wash ring and the on-line water wash nozzles. The wash nozzles are rated to discharge 0.24 gpm of wash fluid at 100 psig. The wash fluid is stored in a 26 gallon SS holding tank. The tank is equipped with a pressure connection used to pressure up the tank in order to inject the wash fluid into the wash ring through stainless steel piping. The system is portable and is moved from turbine to turbine.
The turbines are also configured with a crank wash ring with nozzles that do not atomise the water.
Four tests were conducted on three identical gas turbine generator sets sitting next to each other, as shown in Figure 8.
Test one: demineralised water with soap from different vendors
Figure 9 shows the results of a water wash only and soap washes, using soap from two different vendors, once a week on the three identical adjacent gas turbines. In this plot, the crank wash was performed on 19 July. On 23 July, the water quality was not being monitored and the turbines were washed erroneously with off spec water. A large drop in compressor performance was noted, indicating the fouling of the compressor by contaminants in the water. The turbines were then washed once again using water meeting the specifications set out in the table on p39, and with a water-soap solution using a soap from two different vendors. On 25 July, the turbine performance improved on all three units. As can be seen the wash with water alone was just as good as soap solutions.
Test two: demineralised water, water base soap, and solvent base soap
The object of this test was to determine which soap was better, solvent base or water base soap. Demineralised water was used as a third agent. In Figure 10 notice the vertical gap between the water and the soap solution results in the earlier part of the test. As time went on the vertical gap between the soaps and water was reduced to almost nothing. This suggests the solvents were more effective during the first week but after some time the efficacy of the solvents diminished to the point where they were no more effective than plain demineralised water.
Test three: different water base solvents
The object of this test was to determine if there was any difference between the water base soaps supplied by different vendors.
Water base soaps are considered more environmentally friendly than solvent based options, but each vendor claims that their water base soap has a different chemistry and different “philosophy” in terms of how the soap cleans the compressor blades.
In the tests the soaps were used twice a week, with a demineralised water rinse.
Figure 11 shows the results of these tests and indicates that all of the soaps performed in exactly the same way.
Test four: different wash frequencies with demineralised water and water based solvents
The object of this test was to determine the optimum frequency for performing on-line water wash tests. The tests consisted of a daily demineralised water wash on one gas turbine, and a demineralised water wash twice per week on a second gas turbine. The third gas turbine was washed daily using demineralised water mixed with a water based solvent.
Figure 12 shows the results and indicates that over time a demineralised water wash twice a week is most effective.
The average fuel consumption for a typical Centaur turbine is about 36 MMBtu/h. It is not uncommon to find users that perform crank washes based on power loss, and not on overall thermal efficiency loss, nor loss of compressor efficiency, which is rarely measured. Power loss in a simple cycle gas turbine is caused by many factors, such as an increase in pressure drop in the air filtration system, compressor fouling, fuel heating value changes, combustor fouling, turbine expander fouling and increase in turbine back pressure. Thus a 10% loss in output power, which translates to a 10% loss in thermal efficiency, can have as part of its losses a 6%-8% compressor efficiency loss, which contributes a 3%-4% loss in overall thermal efficiency.
In a Centaur turbine, a 3% saving in fuel for one year is worth $29 171 for one turbine only based on a cost of $5/MMBtu for natural gas. In a fleet of 30 turbines this equates to $875 124 per year.
Planning an effective on-line water wash programme needs to take into account fuel losses, materials and labour required as well as lost production. Typically, an on-line water wash of the kind considered here can be done in an hour by one person. Since each turbine application is different, it is for the user to compare the costs associated with lost production to the fuel savings associated with an off-line water wash programme.
Water alone as effective as soap
What constitutes the best combination of on-line water wash and crank wash will vary from location to location. By monitoring the performance and performing several water wash tests, any site can determine the best water wash combination. The results of the tests described above indicated that, under most operating conditions, a demineralised water wash done twice weekly is as effective as using water-soap mixtures. A few other observations relating to compressor cleaning:
- Good air filtration is the key to prolonging life and avoiding compressor fouling.
- The water used should be demineralised. The use of non-demineralised water would harm the turbine.
- In the tests, demineralised water performed as well as soap solutions.
- All water–solvent washes should be followed by a water rinse.
- After numerous water washes, compressor performance will deteriorate and an off- line crank wash will be necessary.
- An off-line wash should be done whenever compressor performance diminishes by 3%-4%.
- It is imprudent to let foulants build-up before commencing a water wash regime since the foulants will be washed down stream causing blockages in the last stages.
- Low carbon stainless steel should be used for tanks, nozzles and manifold to reduce corrosion problems.
- Spray nozzles should be placed where adequate misting of the water occurs, minimising downstream disturbance of the flow.