Methods of Separation

Separation by Electrostatic

Figure 7–33 gives a schematic arrangement of a typical two-stage electrostatic system.

The principle behind any electrostatic system is that the oil/water separation process is induced through the application of an electric field. The electric field causes finely dispersed water droplets to coalesce, thereby increasing the droplet diameter. This phenomenon increases settling rates according to Stokes’ law, which states that “the velocity of a water droplet falling through oil is proportional to the square of the diameter of the droplet.”

There are two types of electrical separators presently used: direct current and alternating current. Direct-current electrostatic systems are devices employed with light refined fuels of low conductivity. Alternating-current electrostatic separators are applicable for heavier fuels having higher conductivity. Electrostatic separators are appealing because there are no high-speed rotating parts within the vessel.

Sludge removal is accomplished by the transport of solids in the effluent water from the vessel. This can be accelerated by a hydraulic system within the vessel that causes recirculation of the water.

The key parameter in sizing such a system is the selection of the number of stages. Generally, one can expect 85–90% sodium removal per stage. Three-stage systems are usually considered for initial sodium plus potassium inputs up to 100 ppm and two-stage systems are considered for initial sodium plus potassium inputs of up to 50 ppm. Although system startup times using electrostatic precipitators are somewhat long, usually taking three to four hours per stage, the operator is capable of detecting small changes in operating effectiveness of the system and can take corrective action with the system on line.

Electrostatic washing designed to gas turbine company specifications will remove sodium plus potassium concentrations to less than 1 ppm in fuels having specific gravities of 0.96 or less.

For efficient separation the temperature in electrostatic systems must be raised to 149°C (300°F), this requires a pressure vessel to prevent the water from boiling. This in turn can cause reduction of fuel viscosity and furthermore carbonizes the fuel oil.

Separation by Centrifugal Force

The mechanical separation process of washing fuel is shown schematically in Figure 7–34. This arrangement represents a typical two-stage centrifugal system. The principle behind this centrifugal process is that the oil and water are separated as a result of the centrifugal forces developed in the centrifuge. In contrast to electrostatic precipitators, which rely on increased particle diameters in a high voltage gradient field and an IG gravity field, the centrifuge employs a high gravity field of many thousands of Gs to accomplish the separation without the need of an electrostatic field. A variety of centrifuge models are currently available, having capacity ranges from 1.1–110 m/h.

Centrifugal fuel washing systems have been successfully applied to all gas turbines by various manufacturers using fuels with high density and high viscosity.

With correct system parameters, a typical two-stage centrifugal washing system can reduce sodium plus potassium from 150 ppm to a combined level of 1.0 or 0.5 part per million as specified by the turbine manufacturer.

Centrifuges are also effective in removing inorganic particles and dirt that are forced out with the effluent wash water. This added benefit minimizes the load on the downstream fuel filtration system.

Fuel Purification-Simplified Washing

The purification process is shown schematically in Figure 7–35.

Fuel purification is a one-stage extraction procedure that employs centrifuges to treat distillates and light crude oils without adding water to the fuel. This process relies on the fact that the salts are already dissolved in the entrained water; thus by removing the water the salts are removed as well.

Purification is practical with distillate fuel and light crude oils having a minimum 0.5% water in the fuel, with a salt concentration not exceeding 20 ppm of sodium plus potassium. Since the purification process eliminates the need for added water and demulsifiers, it also simplifies the overall waste disposal process.

Selection of Fuel Washing Equipment

Comparison of Equipment

The selection of whether centrifugal or electrostatic precipitators are used to reduce the soluble trace metals to acceptable limits is usually dictated by the customer or the turbine manufacturer; for example where space is at a premium or in case of marine use, the natural choice is centrifuges. For systems with continuous flows over 100 m³/hr, where space is not a major constraint, where the customer has at least 25 storage tanks with a capacity of 19,000 m³ each, and it can sit in the tanks for a minimum of two weeks, electrostatic desalters are acceptable.

The following gives a brief comparison of the two systems: centrifugal desalting system versus electrostatic desalting system. First, let us have a look into the technical part: mixing.

As was already pointed out, the centrifugal desalting system is using a multi-stage mixer, Figure 7–36, with multi-stages; the mixer is a low shear mixer. It is a similar process to hand mixing dough when making bread at home, a slow and gentle mix in comparison with the electrostatic desalting system that uses a high shear mixing valve (Figure 7–37), causing a very violent mixing method.

An electrostatic desalting system company in east Texas tested the two mixing methods (Figure 7–38) and concluded that if a multi-stage mixer is used in the electrostatic desalting system the salt in the finished oil is about 5 ppm. After five hours they switched to the mixing valve and the salt content in the finished oil increased within two hours to 20 ppm. And, when after six hours the system was switched back to the multi-stage mixer it dropped back to 5 ppm within two hours. The electrostatic desalting system company published this study and their conclusion was that they should supply multi-stage mixer with their electrostatic desalting systems and also supply additional filtration.

In addition to Figure 7–38, Table 7–14 gives a comparison of 12 items within the electrostatic and centrifugal fuel treatment systems.

TABLE 7–14

Comparison of Centrifuge/Electrostatic Features

Parameter	Electrostatic	Centrifuge
Time to reach startup stability	4 hours (per stage); 12 hours total	20 minutes total
Reliability	Good	Good
Maintenance	Low	Low
Desludging	By water effluent but can accumulate in vessels	By water effluent and by automatic bowl opening
	Approximately once a year vessels needs to be cleaned
Performance in sodium + potassium removal	Meets spec. if fuel is within system design spec.	Exceeds spec. even if raw fuel does not completely meets design spec.
Oil content in water effluent	500–2000 ppm typical, 50,000 ppm max	50 ppm typical, 500 ppm max
Switch positions
Response time to correct fuel out of spec. washed fuel	12 to 24 hours	20 to 40 minutes
Solid removal	Fair additional filtration needed	Excellent
Amount of heating required for fuel treatment	127–137°C (260–278°F)	98°C (208°F)
Fuel required to change system (cannot be used)	Approximately 1000–2000 m3	1–2 m3
Flexibility—servicing while operating	System cannot be partially shut down	Centrifuges can be removed from service while continuing to operate
Space required	Large area	Limited area

(Source: Westfalia.)

System Considerations

There is no one, basic, fuel-treatment system suitable for all applications, but a design can be tailored to the specific needs and requirements of each customer. The optimum approach results from an evaluation of operational requirements for the gas turbine, the expected fuel characteristics, the desired maintenance philosophy, the choice of equipment, the choice of additive, and the desired system flexibility. Generally, there is a tendency to request “worst case” design without giving due thought to the many constraints that may exist, which may result in an over-designed and expensive installation.

Fuel-Washing Equipment Design

Once the choice of a centrifuge or electrostatic desalter is made (see Figures 7–39 and 7–40), a support-equipment design approach must be selected commensurate with the system objectives. This equipment, which includes mixers, demulsifier injection pumps, heat exchangers, economizers, etc., dictates the actual performance of the system, the resulting maintenance philosophy, and the actual equipment cost. For example, units that are periodically shut down, such as peaking mid-range gas turbines, can have maintenance performed on a planned downtime basis. Continuously running systems require on-line redundancy of critical components such as heat exchangers and economizers. This permits scheduled cleaning without incurring downtime. Further, since the equipment can be maintained while operating, the fuel-washing size can be kept smaller, reducing initial costs. Likewise, redundancy of control air, critical pumps, and motors is recommended.

Fuel Oil Treatment

With two-stage counter-current washing systems (see Figure 7–41) the degree of purity specified by the gas turbine manufacturers is attained, and the trace elements are reduced to the specified limits.

The residual oil is first fed to a prestrainer where coarse impurities are separated out to protect the downstream pump.

After adding demulsifier to facilitate separation of the wash water in the purifier, the oil is conveyed to a heat exchanger, where heat recovery takes place. The residual oil is heated to the required separation temperatures in the downstream preheater. The separation temperature is determined by the viscosity.

The wash water separated from the oil in the second stage is added to the oil upstream of the first stage mixer. The residual oil is thoroughly mixed with the water in a multistage mixer. Salts in solution with the oil will be extracted into the water. The oil/water mixture flows to the purifier in the first washing stage. The solids are spun out due to the high centrifugal force; the dissolved salts, together with the wash water, are simultaneously separated out.

The salt content in the oil is substantially reduced and contaminants such as sand, rust, and cat fines are removed from the oil. Water is again added to the purified oil upstream of the second stage mixer. After intensive mixing, the oil/water mixture is sent to the purifier in the second washing stage, where further purification and desalting takes place.

The required wash water volume depends on:

• The quality of the residual oil

• The sodium and potassium content in the dirty oil (ppm)

• The specified sodium and potassium content in the clean oil (ppm)

• The water quality

The clean oil discharges from the separator in the second washing stage into the clean oil tank via a heat exchanger.

If required, analyzers can be fitted in the clean oil discharge. They are used for continuous monitoring of the water and salt content. If the specified limits are exceeded, the oil is fed back into the dirty oil tank via a three-way valve. As part of a research project, optimum purification efficiency was attained using treatment systems from Westfalia Separator, even when processing high-viscosity oils (1650 mm²/sec at 50°C).

When designing a system the following residual oil parameters are decisive for the capacity (see Figure 7–41):

• Viscosity (mm²/sec)

• Density (g/ml)

• Sodium and potassium content in the dirty oil (ppm)

• Specific sodium and potassium content in the clean oil (ppm)

Inhibition

Crude oils and heavy residual fuels usually contain organic vanadium, which is extremely corrosive to the hot gas path parts of the turbine. The vanadium is not present in liquid fuels in a water-soluble form; hence it cannot be removed through the washing process. Instead, its effects are inhibited by the addition of magnesium to the fuel in the ratio of three parts of magnesium by weight to each part vanadium. At this level of inhibition, the corrosive vanadium pentoxide that is formed during combustion reacts to form magnesium orthovanadate. This has a melting point sufficiently high to allow its passage downstream through the turbine without deposition on the hot gas path parts and resultant corrosion.

A variety of magnesium-containing inhibition agents are available, and their choice is determined largely by the economics of the system and the overall amount of vanadium present in the fuel. Three additives frequently employed are magnesium sulfonate, which is oil soluble and delivered ready for use, magnesium oxide, which is an oil-based slurry, and magnesium sulfate, commonly known as epsom salts, which must be mixed with water on the site. On-line inhibition systems are the preferred method of applying the magnesium agent. The inhibitor is injected into the fuel immediately ahead of its introduction to the gas turbine (Figure 7–42). The system consists of a small in-line metering pump and a flow switch that provides an alarm signal in the event flow of inhibition stops. Either constant flow systems or systems that provide a flow varying with the actual fuel use rate are available. Certification tanks have been employed in some installations. In this approach, batches of the magnesium inhibition agent are mixed directly with the fuel. Use of this system is not currently recommended, since there is a strong possibility that the additive will settle out.

Both oil soluble (magnesium sulfonate) and water-soluble inhibitors (magnesium sulfate) are used with an equal degree of success. The mixing ratio is three parts magnesium to one part vanadium and should not exceed 3.5:1.

For reasons of economy, the oil soluble inhibitor is primarily used for low vanadium concentrations (up to 20 ppm) and the water-soluble inhibitor for high vanadium concentrations (above 20 ppm). Figure 7–43 shows a water-soluble inhibitor.

Mixing Unit for Water-Soluble Inhibitor

A screw conveyor transports a certain amount of magnesium sulfate (MgSO + 7 H₂O) into the mixing tank. Prior to adding magnesium sulfate, the tank is filled to a preset level with demineralized water. An agitator ensures uniform distribution.

A conductivity meter in the circulation line monitors the concentration. Once the required concentration is reached, the solution is pumped into the storage tank. The entire process is automatically controlled and monitored.

Figure 7–44 shows the mixing ratio as a function of the magnesium concentration.

Fuel Analysis and System Controls

A modern, heavy-fuel gas turbine installation requires controls for the fuel treatment equipment as part of total system control. Actual philosophy is dictated by customer preference. Generally, each functional component is equipped with its own control system and has manual set points. Once the set points are preselected, the system can be initiated on-site (or at a remote area) after which the entire process is self-regulating. Monitoring of critical functions can be incorporated if desired.

Fail-safe and permissive features are provided as part of the control system. When a malfunction occurs in the fuel wash equipment, the fuel is automatically diverted to the raw fuel storage tank, rather than to the day tank, until the malfunction is cleared.

Additionally, the fuel quality must be monitored at various steps of the process: sodium and vanadium level after washing and total sodium and magnesium after inhibition.

For light fuels not requiring inhibition (vanadium less than 0.5 ppm) only the sodium level is monitored. This is most readily accomplished with a flame photometer, which is available as a manual instrument for spot checks.

Fuels requiring both washing and vanadium inhibition by a magnesium additive should be monitored with a direct-reading, automatic, emission spectrophotometer (shown in Figure 7–45). While this is a complex instrument, its operation is very simple and amenable to use by on-site labor. The operator collects samples, inserts them one-at-a-time in the excitation chamber, presses a button, and one minute later reads the analysis printed by a typewriter. About once every two weeks the instrument calibration is checked.

Operation on Refinery Crude

Light crude oil with a typical analysis shown in Table 7–15 is delivered by a pipeline to a power station located in central Saudi Arabia to operate six gas turbines. Sodium plus potassium level in “as-received” fuel runs from 15 to 4 ppm, only rarely does it come close to the value shown in Table 7–16 (21 ppm). Operation of this 54 m³/h fuel purification system (Figure 7–46) began at the end of 1981.

The centrifuges used at this location are of the automatic desludging type. Water and solids accumulating at the bowl perimeter are discharged at intervals of about four hours. A batch analyzer provides a spectrographic flame analysis of the trace metal in the purified fuel. Since this was the first purification system, it was equipped with the ability to add small amounts of water to the fuel if it was found to be necessary. This system was also equipped with demulsifier injection capability to prevent emulsions.

During initial startup the water injection system was inadvertently activated, giving rise to erroneous indications that the light crude oil was being washed rather than purified. The interesting fact is that there was an improved performance of the system when the water injection was terminated. Water quality there is quite good; it is received from a local water desalination plant.

Purification tests have shown very easy formation of stable emulsions which can be troublesome due to their sedimentation at the oil/water interface and which cause a deleterious effect on centrifugal purification. The small amount of water used in the system (about 0.075 m³/h) is utilized in the operation of the centrifuge bowl.

Operational difficulties have been observed as fuel temperature dropped below 54°C, which appears to be the approximate “wax point” or the temperature at which waxes begin to precipitate from solution. Waxes are interfacial in nature, and on several occasions have completely filled the centrifuge bowl with this semi-solid. This in turn negates the separation effect of the microscopic water droplets containing must of the trace metals so damaging to the hot-gas-path components. Since crude oil temperature has to be maintained above the wax point prior to the purification system, the crude pour point (wax separation point) is therefore one of the most important criteria of system design.

It is interesting to know that none of the problems that sometimes plague the high firing temperature gas turbines have occurred: failure of flow dividers, fuel nozzle plungers, high-pressure pumps, fuel nozzle coking, etc., due to crude purification and turbine operation cycle. The first turbine water wash was after 3000 hours of turbine operation.

The simultaneous elimination of water and demulsifier injection had increased filter life.

Filter life is influenced by many factors: moisture, asphaltic sludge, inorganic particulates, bacteria (algae), wax out of solution, as well as chemical additives (demulsifier and vanadium inhibitor).

Increase in filter life was at first thought to be coincidence, and little investigation was given to their interrelation. The later lab studies on the oil field crude demonstrated conclusively that one oil-soluble magnesium inhibitor can have a major effect on filter life in the presence of only trace amounts of moisture. And that these chemicals also can cause fuel nozzle coking is proven by the extremely high magnesium content of the deposits.

Operation on Oil Field Crude

During 1981 and 1982 centrifugal lab investigations were conducted on an oil field crude shown in Table 7–15. Experience dictates any such investigation should be conducted on-site as exportation of drum quantities of light crude to a remote laboratory has proven to be a futile effort because of numerous factors. Exporting large samples has resulted in highly questionable results.

Test runs on oil field crude were made on a quantity of oil taken only hours prior to testing. Tests were run on the light crude from many points in its processing train, and a great deal of data were obtained concerning desalting as well as separation rates under various processing conditions. Figure 7–47 shows the actual physical nature of the oil field crude.

The variations are considerably higher than those seen in the refinery crude oil. During the peak periods of January to May 1981, washed crude from an electrostatic system was observed with 4.5 ppm of sodium plus potassium in the day tank. Passage through the test centrifuge gave a further reduction of sodium plus potassium to very acceptable levels as plotted in Figure 7–48. Note: Field experience showed a higher centrifuge efficiency.

Table 7–16 and Figure 7–49 show the extent to which sodium and potassium can be reduced in a centrifugal purification process on the oil field crude. Reduction of other trace metals is also given. Note: Field experience showed a higher centrifuge efficiency.

The above test also provided some answers to questions of current interest to gas turbine operators contemplating burning light crude oil. The conclusions reached and data presented are considered valid for the stations where studies have been made on refinery and oil field crude. Other crudes are under investigation, it is expected that most crudes are purifiable if their physical properties are similar.

Comparison of Centrifuge and Site Data

During 1981 a direct feed crude oil purification system was installed at the Gizan power station in order to purify crude oil for GE’s Fram 6 gas turbines. The system operated very successfully. At a later date the customer switched over to contaminated distillate and utilized the same purification system without any modifications. Westfalia Separator’s purification system can operate on crude oil or distillate fuel and obtain identical results.

Westfalia’s system reduced the sodium plus potassium level from an average of 2.92 ppm down to an average of 0.015 ppm (site data dated Sept. 26, 1989).

At a later date (1987) an additional system was installed (not Westfalia) for GE’s Fram 7. This system could not supply purified fuel directly to the gas turbine (it has been modified at present time) and its sodium plus potassium reduction is only 88%.

If we look into the purification system efficiency from the filter plugging data point of view, it can be noted (from site date) that the turbines that are being supplied with fuel that was purified in the Westfalia system have a filter life span of 1078–4103 hours on the low-pressure and high-pressure filters whereas the turbines that are being supplied with fuel that was purified by the other purification system have a filter life span of 95–1009 hours on the low-pressure filters and between 1578–1716 hours on the high-pressure filters.

From the above comparison of the two purification systems (summarized in Table 7–17) it is clearly noted that the Westfalia purification system performed as designed.

In summary, the data from centrifuge lab studies as well as from operations at various power stations, confirmed the viability of the process. A level of approximately 20 ppm of sodium plus potassium can be reduced to <0.5 ppm while operating at commercially acceptable flow rates for the standard centrifuges models available.

It has been found that purification operations on crude oil should be at 55–60°C to ensure that wax separation does not influence the operation. This appears to be true regardless of wax concentration, as proven by test runs on tank bottoms containing extremely high wax content (about 40% by volume).

Centrifugal purification of light crude oil can extend filter life on 5–10 μm paper cartridge filters to 1000 hours, as have been shown by both commercial and pilot plant operation. The operator must, however, be very careful in selection of additives to be used, as well as their injection point, as discussed previously.

Operating water on a typical light crude system of 54 m³/h has a total consumption of about 0.075 m³/h. This water is needed for the bowl hydraulic. For the same size plant (54 m³/h), a washing operation having a moderate 2% by volume water would consume about 1.0 m³/h of water.

Extremely high saltwater contamination level can be tolerated in fuel as received providing proper storage facilities are available and are rigorously maintained. Tests show sodium plus potassium level exceeding 1000 ppm can be reduced in 24 hours of quiet storage to about 40 ppm. Floating suctions and a floating pan with a fixed roof are also highly recommended to maintain clean fuel.

Special caution should be taken when planning for the successful desalting of certain crude oils by the centrifugal purification process without the addition of water to the fuel. The success or failure of this process is highly dependent upon both the design and configuration of the centrifuge and the processing systems.

Tung Hsiao Power Plant (OEM: Alstom)

An OEM, such as Alstom for a model such as their IIN, might ask for a fuel treatment system that can handle 158.4 m³/h (44 1/sec.) of residual fuel with a specific gravity of up to 0.986 at 16°C and a viscosity of up to 440 cSt at 50°C. Incoming fuel will have a sodium plus potassium level of 50 ppm and must be reduced to ≤0.5 ppm.

During the acceptance test the sodium plus potassium was reduced from 30 ppm to 0.08–0.10 ppm. It should be noted that this is the first (by any company) to achieve a Na + K reduction to 0.08–0.10 ppm. This fuel treatment system is the largest operating system in the world with the largest centrifuge in the world.

The fuel washing system consists of:

Naga Power Plant, the Philippines (Gas Turbine OEM: Alstom)

The OEM requested a fuel treatment system that can handle 13.8 m³/h of residual fuel with a specific gravity of up to 0.952 at 16°C and a viscosity of up to 220 cSt at 50°C.

Incoming fuel will have a sodium plus potassium level of 50 ppm and must be reduced to ≤1.0 ppm. During the acceptance test the sodium plus potassium was reduced from 70 ppm to less than 1 ppm.

The fuel washing system consists of:

Initial Power for Metro Manila in Record Time

To meet the very important third criteria and reduce the daily eight-hour brownouts, the project was split into two phases. The phase one gas turbines were completed in simple cycle mode by shipping and erecting preassembled blocks. It was possible to commence commercial operation only 8.5 months after the notice to proceed from National Power Corporation. The first block of three gas turbines was available after 13 months, the second block after 18 months.

To allow the energy to be provided to Metro Manila, the 230 kV HV lines to the nearest substation some 50 km away from the plant had to be built within the same short time period. This task was managed by NPC on time.

With the restricted choice of fuels the location was chosen adjacent to Petron Bataan Refinery to take advantage of the available heavy residual fuel oil.

Limay Bataan Combined Cycle and Extension with Block “B”

The plant is depicted in Figure 7–50.

Phase two of completion, in which a 114 MW steam turbine (ST) was added using the steam produced in the heat recovery steam generators (HRSG) and thus completing the combined cycle, was achieved 12 months later, with the commissioning of the ST of Block “A,” on 25 November 1994.

Financial constraints led the plant to be ordered in two separate blocks, to limit engineering efforts. Block “B” was to be identical to “A.” Block “B” was handed over on 25 February 1995.

The combined cycle process is depicted in Figure 7–51. Each combined cycle block comprises three Alstom Power GT 1 1N gas turbines, three unfired natural circulation heat recovery steam generators, the steam turbine, main transformers, and high voltage switchyards. Overall control for each block from a central control room is operated by use of Procontrol P system. Each gas turbine is rated at 70 MW and is fired on treated fuel oil (TFO) with startup on diesel fuel oil. Water injection is available to augment power output. The steam turbine is rated at 114 MW. Total site guaranteed output is 615.2 MW (dry) and 651.2 MW (with power augmentation) at an ambient temperature of 32°C. Sea water from Manila Bay is used for the main cooling system.

The heat recovery steam generators are supplied by KHI under a licensing arrangement. Each supplies 35.2 kg/s of steam at 50 bar/493°C to the steam turbine and 8.7 kg/s saturated steam for ancillary services in the plant.

The steam turbine has an integrated HP/IP casing and a single flow LP casing exhausting to a conventional titanium tube sea water-cooled condenser. For startup or ST trips each HRSG can bypass steam directly to the main condenser. Each power unit is coupled to an air-cooled generator producing power at 13.8 kV to be stepped up to 230 kV en route for the switchyards and then into the North Luzon Grid to feed Metro Manila.

The operating mode of the plant is normally in combined cycle. However, each gas turbine can be operated anytime on simple cycle as gas bypass stacks and diverter dampers are fitted before the HRSG. Figure 7–52 provides operational data.

The efficiency of the plant is—based on heavy fuel oil operation—at 46.6%.

Fuel and Fuel System

The residual fuel produced in the refinery is a very heavy product with varying quality, as can be seen in Figures 7–53 (specification) and 7–54 (actual supplies, demonstrated on the Na + K and heavy metals contents). The product would not meet the Philippines environmental standards with respect to SO_x emissions. Hence this “IFO” (intermediate fuel oil) must be blended with a low sulfur diesel oil (DFO) at the site to achieve a maximum level of 2.75% sulfur in the blended fuel oil (BFO). During the last year, due to some changes in the refining process, the refinery was able to provide fuel that met the requirements. This situation, however, was only temporary, and by July 2000, the existing fuel mixing system was again in operation.

The fuel system is therefore quite complex and requires attention as it is not only a multi-step process, but also has to feed the two blocks simultaneously and continuously with fuel that meets the gas turbine fuel specification and can achieve the environmental standards set by the Department of Environment and Natural Resources (DENR). The system is described in Figure 7–55. The raw fuel, IFO, is stored in two large 15,000 m³ tanks and is blended with diesel oil to meet the sulfur content criteria. As the sodium content is also quite high, fuel washing is required.

Each block therefore has its own fuel treatment plant (Figure 7–56) based on a two-stage washing process with electrostatic separation of the two phases, water and oil. This process adds water to the fuel to extract soluble salts and hydrophilic matter. This water is then separated by applying an electrostatic field that causes the water droplets to coagulate and separate out by gravity force. The separated oil must meet a sodium plus potassium content of <1 ppm and should therefore contain an absolute minimum of residual water. The extraction water must be virtually free of oily matter as it may only contain 10 ppm of residual oil. The fuel is then stored in four tanks of 5000 m³ at approx. 60°C and then preheated to approx. 110–130°C for the combustion process in the GT combustor. In order to minimize high temperature corrosion caused by vanadium pentoxide, an Mg-based additive is added immediately before combustion to the fuel.

Experience with the fuel handling has shown various problems that occurred as a function of time. One occurred after three years of operation when suddenly the separation of the water was extremely difficult to handle and multiple recirculations of the oil became necessary to meet the requirements. Investigation of the problem revealed that part of the water/oil mixture had partially become a quasi-stable emulsion that did take longer to separate, although different demulsifiers were added.

A detailed fuel analysis finally showed that the pH of the fuel, due to processes at the refinery, was in the higher alkaline range and varied between 8 and 9.5 instead of 7–8. In this pH range the alkalinity is causing ideal conditions for stable emulsions. The reason for this alkalinity was investigated and found in the refining process.

When processing crude oils, the normal practice is to reduce chloride salts to very low concentrations via electrostatic desalting techniques. Primarily this is done to control chlorides carry-over into the distillation unit and thereby minimize the formation of corrosive hydrochloric acid. In cases where efficient desalting is not employed and where unacceptable levels of chloride salts may enter the distillation column, many refineries adopt the less desirable use of sodium hydroxide (caustic soda, NaOH) for corrosion control.

The main purpose of the sodium hydroxide is to convert the more volatile calcium and magnesium chlorides in the crude oil to the more stable sodium salt. Under high temperature distillation conditions, sodium chloride hydrolyzes less than calcium and magnesium chlorides, and therefore the rate of dissociation to form corrosive hydrogen chloride (HCl) is reduced.

Also, naphthenic acids can be corrosive under certain high temperature conditions, and sodium hydroxide is sometimes utilized to control this type of corrosion in refinery equipment. Obviously, this may solve these refinery problems, but without looking at the effects in the power plant, where the presence of sodium is highly critical and treatment is carried out to remove sodium from the fuel.

Essentially, the problems in the power plant result from the fact that it is difficult for refineries to add exactly the required concentration of sodium hydroxide for corrosion control, and the common tendency is to overtreat with an excess of NaOH. This results in high pH alkaline conditions in refinery effluent water streams, and the carry-over of “spent” sodium hydroxide into the residual oil fractions of the crude oil distillation process.

These high pH conditions, particularly if associated with oils containing naphthenic acids, result in the stabilization of oil-in-water emulsions and may lead to the generation of unacceptable oily effluent water streams. Experience shows that for pH of the effluent water higher than approximately 9.2, an effective effluent water treatment becomes difficult.

As the fuel spec in NPC’s Fuel Supply Agreement and in the GT fuel specification had pH requirements, the solution had to be found together with the refinery and the OEM of the fuel treatment plant. After thorough testing of various demulsifiers with the fuel/water emulsion in the laboratory and in the field, and the introduction of a more strict quality assurance program on the refinery side, the fuel treatment plant is now operating with higher tolerance on pH. Excursions on the pH are limited as a consequence of stricter quality control of the fuel in the refinery.

The pH problem also caused a consequent problem in the treated fuel oil tanks. Usually the fuel arrives with 0.1–0.3% of bottoms, sediment, and water (BS&W), which would partially settle out. The above-mentioned fuel instability caused formation of a very high amount of sludge that accumulated at the bottom of the tanks. It was noticed that the fuel quality started to deteriorate at the tank outlet and an investigation was carried out by intensive sampling of the tank. This investigation revealed that approximately 250 m³ of sludge type material was present in one tank; hence cleaning was the only way to restore the full tank capacity. The cleaning process was executed by a specialized company that separated fuel from the sludge such that approximately 90% could be recycled as fuel and the rest, cumulating in 66 drums of combustible material, was processed under control of the Department of Environment and Natural Resources (DENR). Tank bottom surveillance and regular tank sump drainage is now part of the daily operation program.

It can be summarized from this eight year experience in treating various fuels in Limay with the electrostatic equipment that this equipment is indeed maintenance friendly, as there is hardly any maintenance required and the process is very stable and runs reliably. Except for the case above and the wide range of basic fuels that was used, the electrostatic fuel oil treatment plant needed no further adjustment to achieve the requested fuel quality. The performance of the plant is depicted in Figure 7–57. Starting and stopping the plant has not been a problem. A restart of the process is easy and can be executed without difficulties. Within 4–5 hours after initiating the process, the plant is producing quality fuel at full throughput.

The fuel quality needs to be closely monitored for the following reasons: On one hand, the content of trace metals such as vanadium and sodium are critical for avoiding high temperature corrosion and on the other hand, physical and chemical parameters determine emissions and combustion quality. Therefore, a complete laboratory equipped for water and fuel analysis as well as emission monitoring is also part of the site.

Heavy Oil Operation vs. Diesel Oil Operation

It is often questioned whether the operation on a heavy fuel oil is attractive in light of the additional measures that have to be taken in order to reduce the immediate effects of the combustion of this fuel. Additional efforts are the fuel treatment itself and the additive dosing. Due to the corrosiveness of the environment hot gas path parts show higher wear and tear than comparative parts in clean fuel gas turbines. Comparing the effort on an annual basis, the additional cost for treatment, including all chemicals, maintenance, and operation of the FOTP is <5% of the annual fuel costs. Comparing the cost with the additional cost running on a fuel such as diesel oil, the savings are related to the annual diesel costs, in the range of approximately 10%. This takes into account that the diesel to heavy oil index has changed from >2 in 1996 to about 1.4 in 1999, which makes the difference smaller.

Taking into account the side effects on the turbine output and efficiency, these numbers are reduced to a benefit of roughly 7% for heavy oil operation. The fact that this fuel is available from the refinery next door and must be used in a cost-effective way makes the scenario quite attractive.

The inspection concept of A, B, and C inspections has proven to give enough information to be able to plan the major (C-) inspections in detail and hence be efficient. Nevertheless the gas turbine is the most delicate part of the plant. Incidents in the early days were fortunately limited to a second stage blade failure caused by high cycle fatigue, in combination with a resonance excitation from grid instabilities. The blade was modified for higher tolerance and has operated satisfactorily ever since. In fact, part of the blading is now reaching longer lifetimes than actually expected under these conditions.

In order to be able to carry out the major inspections quickly and efficiently, the individual steps of the inspection were critically analyzed. This analysis was then transferred into a work plan. At the same time it was also decided to invest in additional hardware on site, which can be preassembled to a higher degree and hence reduces the actual outage time. All these actions reduced the original outage time from roughly four weeks to only 12–15 days.

As an example, one of the decisions in O&M was to have fully bladed vane carriers on site that could be prepared for exchange while the unit was still in operation and then be exchanged as a whole. The blading and deblading process of the stationary blading is therefore outside the inspection period and also the parts removed from the machine can be moved out of the space where the turbine is being inspected. There is more room, which improves overview and planning. Similar solutions have been found for the combustion chamber, which is now overhauled with the same principle. Using a second combustor to replace the one going for inspection has also given a boost to quality in a sense that during the pre-assembly, there is enough time for quality assurance during assembly and training on the actual material.

As there are six units that need to be inspected, the local team has developed excellent skills in performing the inspections. The Limay Bataan team will now only require a few specialists, and the major portion of the work can be organized locally, in close cooperation with Swiss headquarters. At the time of writing this case study, the first two gas turbines of Block “A” were again inspected after roughly 50,000 equivalent operating hours. Both units were back in operation after 15 days.

Discussion

Where other published shock tube data are available, the agreement with the current data is good. However, it is acknowledged that there is a shortage of data in the region of interest to gas turbine combustion. The data tend to be at high temperature (greater than 1200 K), and pressures that are either too low (below 10 atm) or too high (over 50 atm).

The experimental data suggest a change in activation energy at low temperature (below ∼1200–1300 K, depending on the initial pressure conditions). This is depicted by a change in the slope of the data series. The change in slope could be due to other factors than a change in activation energy. Several possible explanations are briefly discussed below:

If the integrity of the lower temperature data is shown to be good, several significant implications arise. Firstly, the suggestion that the actual composition of a natural gas fuel has no impact on the autoignition delay time for temperatures below 1100 K would, if confirmed, offer the potential to greatly simplify the predictive schemes. Also, for practical systems, the ability to elevate the minimum autoignition temperature by the addition of water could signify additional benefits for humid air cycles.

Without exception, all of the predictive methods assessed here show good agreement with the high temperature experimental data.

Burner and Combustor Installation

Figure 7–69 shows a typical Alstom combustion system, showing the injector to combustor interface. The Tornado system is comprised of eight reverse flow “diffusion flame” tubular combustors. The Typhoon has six similar combustors. There is always one injector per combustion chamber.

In a tri-fuel LPG system, the fuel system, like the fuel injector, allows the use of fuels close to their vapor point. The system comprises a metering and shut-off valve train inside the turbine enclosure and a separate pumping skid outside of the enclosure.

Due to the widely different thermodynamics properties between diesel and LPG, it was considered less risky to add a separate LPG pump stream to a proven dual fuel system, than to risk the consequences of modifying a single diesel pump stream to cater for LPG. There are also a number of advantages in using separate diesel and LPG pump streams, instead of a single pump stream. These are (a) no risk of cross-contamination between LPG and diesel tanks, (b) able to deliver turbine as standard dual fuel and add LPG later, (c) no requirement to purge the complete LPG system before restarting the turbine on either diesel or natural gas fuels.

Fuel Metering Device

The injector’s fuel metering device comprises a metering valve surrounded by cooling circuitry. This is designed to ensure metering of liquid LPG fuel during all turbine-operating conditions. The valve’s piston operates under the action of differential pressure between fuel supply and combustion chamber pressures. Three metering ports are progressively opened as the valve piston moves. The valve crack point is set to give a minimum fuel system pressure of 20 Bara to prevent pure propane from phase changing in the fuel pump, manifold and injector.

As the LPG and diesel supplies are separate, it is necessary to pre-cool the tertiary valve before introducing LPG. For example, if the injector is started on diesel pilot and main and run at part load until steady state, the valve reaches temperatures of 100°C. Before valve crack, expanding a small fraction of the fuel into the separate cooling circuit at the lower combustor pressure actively cools the valve port. The latent heat of vaporization that results from the expansion and phase change, of this small fraction, is utilized to cool the injector valve, keeping the main flow of fuel in liquid phase.

The valve body will always be cooled to a temperature below the saturation temperature of the fuel at the fuel supply temperature. This is the result of the pressure differential and phase change. The lower absolute pressure cooling flow will always be lower in temperature than the saturation temperature of the fuel at supply pressure. This is always true for any composition of LPG even for extreme compositions containing large amounts of ethane, pentane and hexane.

Fuel Transfer Circuits

The fuel is then transferred to the injector tip via three fuel transfer circuits. The cooling fuel flow is kept separate from the main fuel until the fuel is vaporized at the injector tip. The low mass flow, but intermittent high volumetric flow rate of the cooling flow, would result in slug flow if the vapor cooling flow were to be mixed with the main fuel. Slug flow resulting in pulsations of liquid and vapor would be very detrimental to turbine operation.

The fuel transfer passages for the cooling flow have conflicting design requirements. The passages must be sufficiently large in flow area to accommodate a wide variation in volumetric flow rate while being sufficiently small to maintain sufficient high fuel velocity to maintain a dispersed two-phase flow. The volumetric fuel flow rate of the cooling flow varies significantly as the heat load varies. Initially when the LPG flow is initiated the LPG valve can be 100°C or greater. At these conditions the LPG cooling flow would be mostly vapor at the low fuel pressure at which this injector is designed to operate (20–30 Bara). Under prolonged steady state operations on LPG the cooling flow will be mostly liquid, as the steady state heat load becomes small. The ratio of vapor to liquid in the cooling flow will also vary as a function of fuel temperature, fuel composition, engine combustor pressure, and fuel supply pressure.

The coolant passages were designed to have a relatively large flow area in order to accommodate high volumetric flow, when the cooling flow is mostly vapor, while also having relatively small distances between opposing walls in order to limit the size of the bubbles that can be formed. Thin passages with high volumes will only form small bubbles, compared to other passages such as tubes that have minimized wall surface area to flow area. Using thin flow passages for the cooling flow allows large cooling flow areas to accommodate a wide range in volumetric flow rates without resulting in the formation of large bubbles that would cause unstable combustion.

The bubbles that do form in the fuel injector are after the metering device and are limited to the separate cooling flow, which is less than 10% of the fuel flow. Engines in the field have been shown to accept this amount of two-phase flow in the fuel system without resulting in unstable combustion.

Fuel Vaporizer

The active cooling concept of the fuel metering device combined with the design of the fuel transfer passages has provided controlled fuel flow of the mostly liquid main flow and at times mostly vapor cooling flow to the injector tip near to the combustion zone. The fuel must now be injected in a spatially controlled manner into the combustion chamber.

Most conventional fuel injectors would use a pressure atomization or air spray device to accomplish this task.

However, two-phase flow makes the use of any device that creates a pressure drop after the fuel is metered problematic. Only one metering device can be used per single stream of flow and the inlet of this device must have liquid only in order to have consistent predictable flow rates. Others have attempted to address the above problems with liquid LPG injection in two ways.

One way is to use a pressure atomizer at the injector tip with no other fuel-metering device such as a valve. However, this approach requires high minimum fuel pressure to maintain liquid phase (typically 100 Bara). For example, well-insulated fuel injector circuits would require a minimum pressure for propane to be 18–20 Bara (reference Figure 7–68). The difference between the minimum fuel pressure and the lowest operational combustor pressure could be 8 Bara for typical industrial gas turbine engines. This combined with these engines needing a 4:1 fuel range between maximum load and load reject results in very high maximum fuel pressures. The pressure atomizer would therefore require about a 16:1 pressure ratio to accommodate a 4:1 mass flow ratio. For engines that have at least an 8 Bar difference between minimum combustor pressure and the saturation line pressure, the fuel pressure could therefore exceed 100 Bara.

Another way is by placing a variable area metering device, such as a valve, near an air spray atomization device at the tip. This reduces the maximum fuel pressure. The close proximity of these components also prevents two-phase flow from separating into liquid and vapor components by minimizing the time and volume available for separation. However, these devices are problematic because of the difficulties of placing tightly toleranced moving components such as a spool valve in regions of high thermal gradients.

The tri-fuel injector developed jointly by Alstom Power and Delavan has been designed to overcome the reported difficulties with LPG injection, described below.

The fuel-metering device is placed in the back of the injector (see Figure 7–70) to limit heat conduction to the valve, thus minimizing mechanical problems. The injector’s fuel transfer circuits, designed to keep the main and coolant flows separate, also allows the injector’s fuel vaporizer to be positioned close to the injector tip where there is sufficient heat and air for both vaporization and atomization processes.

For the fuel vaporizer, the initial conceptual design was to use an air blast type of prefilming atomizer. The three cooling circuit flows and the three main fuel flows are introduced into an annular vaporizer in an evenly spaced symmetrical pattern. The idea was to use air to spin the LPG fuel into a film where it could be atomized by additional air outside of the injector.

Using air to add momentum to the fuel and spin the LPG fuel into a thin film had two advantages: it induces no pressure drop into the fuel circuit and the heat from the hot air partly expands the fuel thus increasing its volume. Also keeping the existing dual fuel design of the pilot, main, water, and gas circuits meant that the LPG filmer had to be at a larger diameter (see Figure 7–70). This would make the production of a continuous film difficult in this application without somehow increasing the volumetric flow rate of the fuel. The idea of adding more hot air to increase the volume of the fuel in order to create a film therefore developed. The philosophy was therefore just to add hot air and vaporize as much of the fuel as possible and then treat LPG as a gas when injecting it into the combustion chamber.

Carbon formation in the vaporizer is not an issue as long as the temperature of vaporization at the combustion pressure is less than the temperature at which carbon will form. For most industrial gas turbine engines this stays true at least until the fuel contains mostly C₈ and higher hydrocarbons. For this reason most naphthas can also be mixed internal to the fuel injector with hot air. As long as the fuel stays partly wet the temperature will stay low. This prevents auto-ignition and carbon formation from small fractions of C₁₆ and higher hydrocarbons in naphtha.

General Operation

The described tri-fuel system entered commercial operation in March 1999 on two Tornado gas turbines (units 1 and 2), both 6.45 MWe rating. Prototype 2 tri-fuel injector became the production standard. The lead engine has accumulated about 1029 hours, while the second has about 361 hours on LPG. Most of the above LPG running has occurred during winter, between November 1999 and March 2000. Including natural gas and some diesel operation total running has now exceeded about 8000 hours.

Fuel Pressures

The reported winter LPG running presented the coldest ambient conditions and therefore the highest power conditions. This operation resulted in the highest LPG fuel pumping pressures. Table 7–22 shows that fuel pumping pressures are within the original design intent (30 Bara).

Combustor Exit Temperature Spread

Figure 7–71 shows comparisons of combustor exit temperature spread between natural gas and LPG fuels. This is measured by 16 off thermocouples positioned circumferentially in the inter duct between compressor and power turbines. This spread level has shown no deterioration for LPG running, confirming that all eight injector LPG valves are operating consistently.

Transient Response

The tri-fuel engines can change over to and from LPG from a reasonable low load, say 1 MW, up to full load with complete confidence, although changeovers to LPG at FSNL have been locked out because of uncontrolled speed rises during the LPG cooling cycle. Changeovers from LPG at FSNL are possible but an uncontrolled speed rise (approximately 250 rpm) does occur during the 20-second nitrogen purge of the LPG manifold. No further work was conducted to optimize low load changeover, as it was not essential. Unit 2 was proven to LPG load shed from a variety of loads including full load; however, Unit 1 could only load shed from 2 MWe. The load-shed abilities are therefore not consistent and further work is required. It may be necessary to further optimize the LPG valves port shape to improve load shed. There are no issues with standard dual fuel operation.

Combustion Dynamics

During initial commissioning in March 1999, a combustion system dynamics survey was conducted for all fuels. An increased level of combustor dynamics, with a center frequency of about 100 Hz, was measured for LPG operation relative to both natural gas and diesel running. The amplitude of this dynamic increased in magnitude as the load increased; although maximum levels never exceeded 300 mpsi RMS. As these dynamics were within levels of experience (< 500 mpsi RMS), no further action was taken to reduce their levels.

Later in November 1999, a low frequency audible rumble in the exhaust became apparent on both turbines when running on LPG. There was no audible exhaust rumble for diesel and natural gas fuels. Combustion system dynamics showed no increase since the initial commissioning. The exhaust noise was reported to change with ambient conditions. It is probable that the exhaust noise is generated by the higher dynamic levels measured for LPG running, which only becomes audible at certain ambient conditions.

Fuels

A large portion of the commercially available naphtha and fuel oil is derived from crude oil by atmospheric distillation. In contrast to naphtha and fuel oil, the term “condensates” does not apply to a specific fuel, but is rather used as a synonym for hydrocarbons that exist in a liquid state at standard temperature and pressure, and that occur as significant byproducts of the natural gas extraction process. Specific properties (Table 7–25) and their validation are of central importance in characterizing liquid fuels. Characteristic of condensates and naphtha are a high calorific value in excess of 42 MJ/kg, low density, low viscosity, a low flash point, and a comparatively high vapor pressure.

System Description

Gas-Turbine Liquid Fuel System

Whereas the fuel delivery system is designed around those characteristics specific to the power plant, for the gas turbine standard modular skids are used for natural gas and liquid fuels, so that use of a particular fuel is independent of the existing infrastructure.

The annular combustion chamber of Siemens 3A-Series gas turbines is equipped with 24 identical hybrid burners for natural gas and liquid fuels. Each hybrid burner in turn has a diffusion burner and a premix burner for liquid-fuel operation, as well as a diffusion burner, a premix burner and a pilot burner for operation using natural gas. In addition, a separate lance can be used to inject water directly into the flames to reduce NO_x formation.

The three natural-gas burners and the premix burner for liquid fuels have controlled feeds, i.e., the entire quantity of fuel supplied by the feed control valve in each case is injected into the combustion chamber. The liquid-fuel diffusion burner has a controlled feed and return, which allows control of not only the quantity of injected fuel but also the feed quantity and injection pressure, and thus permits control of the directly linked atomization quality. The liquid-fuel diffusion burner can operate in the load range from startup to base load (approx. 12 kg/s fuel), without the addition of water. In power operation the return mass flow from the diffusion burners is directly fed to the injection pump again (recirculation).

Compared to water injection via the lances, the direct mixing of water with the fuel and the operation of the premix burner achieve a greater reduction in NO_x emissions using the same quantity of water. In the first development phase, as represented by the Santa Rita plant, water is admixed via a jet mixer upstream of the injection pump (Figure 7–73). Moderate water pressure is required, allowing use of the water skid originally designed to be used with lance injection. The jet mixer increases the feed pressure of the liquid fuel/water mixture upstream of the injection pump by up to 10 bar, which also allows correspondingly greater injection pressures to be attained at the burners.

The density and viscosity of the liquid fuel/water mixture vary according to the mixture ratio and the composition of the fuel, which places heavy demands on the injection pump. The mixture ratio is defined as the ratio of the water mass flow to the fuel mass flow. The gas turbines at the Santa Rita plant have been successfully tested with mixture ratio over 100%. A major milestone in the development of the multi-fuel gas turbine was therefore the replacement of screw pumps with centrifugal pumps, as the latter are well suited to low-viscosity liquids (<1 mm²/s for naphtha and condensates) and also have a characteristic favorable to the combined supply of diffusion and premix burners. The zero delivery head of the centrifugal pumps is approx. 1600 m.

Because the water is added prior to branching the mixture to the diffusion and premix feed lines (see Figure 7–73), both have already been implemented for the newest gas turbines via a standard dual-fuel system, water is added separately to the diffusion and premix feed lines, so that the mixture ratio can be varied independently for the two burners.

In gas operation, a switch off/switch on of burners, e.g., at changeover from diffusion to premix operation and back, can be implemented rather easily. No cooling or purging procedures are required. In liquid-fuel operation special precautions must be taken during startup and shutdown of the gas turbine, at changeover to liquid fuel premix operation and when changing over from gas to liquid fuel operation and back. If a liquid-fuel burner is not in operation, there is a risk of circulation flow among the individual burners as a result of slight local pressure differences in the combustion chamber, which would soon lead to overheating or the formation of coke at the nozzles. To prevent the ingress of hot gas when the liquid fuel burners are shut down, water is used to flush the residual fuel from the lines and nozzles, which are then supplied with cooled compressor outlet air, so-called seal air. Before a liquid fuel burner starts to operate, the empty sections of the line are filled with water. This not only allows the atomization pressure to build up rapidly, but also cools the nozzles so that the fuel cannot ignite prematurely.

The safety concept has given rise to the liquid multi-fuel requirement that all liquid fuel lines which are pressurized during startup of the gas turbine (including startup using natural gas) must not contain hazardous fuels such as naphtha and condensates, as these fuels would pose a significant risk to the gas turbine, such as in the event of a faulted start. For this reason Siemens multi-fuel gas turbines generally use fuel oil for startup and shutdown. Moreover, use of fuel oil means that in the event of a leak, which in many cases is not evident until there is pressure in the line, the escaping fuel is initially the much less hazardous startup fuel. Changeover to the primary operating fuel (and back) following gas turbine synchronization takes place within the fuel delivery system, which means that no noticeable changes occur in the operation of the gas turbine. In the event of a gas turbine trip, the lines downstream of the stop valves are completely evacuated and partially refilled with water. The liquid-fuel skid is then flushed.

Operating Experience

The gas turbines at the Santa Rita plant have been successfully tested using naphtha, condensates and fuel oil. When water is added to the fuel in a ratio of 1:1, NO_x emissions are well under the World Bank standard of 72 ppm.

In the part-load range the characteristics of the different fuels (primarily the density) led to variable turbine operation, which made it necessary to modify the structure of the gas turbine controller such that the operating characteristic of the gas turbine can be adapted to the fuel in use.

Base load is reached without restriction regardless of the fuel used.

Acceptance measurements were taken in Block 1 in May 2000, using condensates as the fuel, and in August 2000 in Block 2, with distillate oil. It should be emphasized that the expected gas turbine output and efficiency values were either met or exceeded.

(1) Startup Characteristics

This power plant adopts a steam turbine starting system. After reaching the rated speed, it switches from the existing steam to HRSG steam once the conditions of the HRSG steam become acceptable for the steam turbine. The plant then increases its load automatically up to the minimum operational load of 150 MW. Once reaching 150 MW, the load can be changed between 150 and 300 MW depending on the power demand. The power plant also shuts down automatically. During load down, the cooling down process is designed to make the mismatch between the steam and the steam turbine metal temperature small so that the time to the next hot start is shortened. The expected time for starting (from ignition up to full-load operation) is approximately 90 minutes for a hot start, 150 minutes for a warm start and 200 minutes for a cold start.

(2) Plant Performance

Table 7–27 shows the actually measured performance. The plant output and plant thermal efficiency after compensation to the standard conditions (ambient temperature at 15°C and LHV base) sufficiently satisfy the planned values. The plant thermal efficiency reaches 51.2% compared to the planned 48.5%, and achieves the highest level in the world for BFG firing CCPPs.

(3) Environmental Characteristics

By effectively transducing the low calorific blast furnace gas to electric power, this power plant contributes greatly to environmental preservation as follows. Firstly, this is a combined cycle power plant that effectively utilizes the blast furnace gas instead of the coal fired thermal power plants popular in China. It is estimated that the plant emits approximately 1,900,000 tons less CO₂ a year (assumed operation of 7000 hours per year) than that emitted by burning coal in a coal-fired thermal power plant of the same output. Secondly, although the environmental characteristics depend significantly on the composition of the blast furnace gas, as it contains nitride and sulfur compounds, its combustion produces less thermal NO_x than burning natural gas because of its low calorific value. Thus, this plant indicates the significantly low NO_x value of 15 ppm (15% O₂ base, without a de-NO_x system).

Operation Records

Table 7–28 shows the availability of the plant for about five months from May 2007, when the plant was taken over, to the end of September. While the plant availability depended on the scheduled shutdown due to maintenance of the blast furnace shortly after taking over the plant we experienced the degradation of plant availability because of initial failures and other incidents. After that, the plant reliability improved gradually, finally reaching 100% in September 2007.

Conclusions

Extensive ignition, emissions and combustion dynamics testing of biodiesel on standard DLE SGT-100 combustion hardware has been conducted to investigate the feasibility of using biodiesel as a fuel to power industrial gas turbines.

In ignition testing, two series of tests were performed to identify key properties of the proposed fuel. The first set of tests identified the likelihood of the engine lighting based on the measured probability of the ignition of a single combustor.

This demonstrated that a high likelihood of ignition would be possible depending on the choice of air assist pressure. The second series of tests investigated the lean ignition and extinction limits at various air temperatures. This showed that the ignition limit was different to that measured for mineral diesel, with the biodiesel being easier to ignite at high air mass flow rates and harder at lower flow rates. It was also demonstrated that air temperature did not have significant impact on the lean ignition limit. It was further shown that the ignition window for the SGT-100 lay well within the predicted curves for the lean ignition limit and therefore that good ignition should be possible. The lean extinction limits were found to be very similar to the ignition limits.

Emissions and combustion dynamics were measured in a high-pressure test rig. Performance of the burner on biodiesel was mapped by variation in operating conditions such as the air inlet pressure, temperature, the combustor pressure drop and flame temperature. The results were compared with those of diesel at the same conditions. For all the operating conditions tested, biodiesel had lower NO_x emission than diesel. For biodiesel at an air inlet pressure of 14 bar, there was a strong link between NO_x and the combustor pressure drop, where the NO_x level was reduced as the combustor pressure drop was increased. It was not the case for biodiesel at 9 bar and for mineral diesel at all pressures investigated.

However, after FBN correction, comparable NO_x emissions were obtained, although diesel showed slightly higher NOx emission than biodiesel at 9 bar inlet pressure.

Analysis of the high-pressure rig testing results showed that for biodiesel at low flame temperature and low air inlet pressure, a high degree of premixing was achieved, so NO_x level was independent of inlet pressure and combustor pressure drop. At high pressure and high flame temperature, the optimal level of mixing could not be achieved due to the amount of fuel supplied. Where mixing is not optimal the thermal NO_x formation is higher and dependent on inlet pressure and combustor pressure drop. However, for mineral diesel, good premixing was achieved at all pressures investigated and little dependence on inlet pressure and combustor pressure drop was observed.

At lower flame temperature, biodiesel had a significantly higher level of CO than diesel, particularly at a lower inlet pressure of 9 bar. For both fuels, the combustor pressure drop had apparent influence on CO at lower flame temperature with CO level decreasing with the combustor pressure drop decrease, and this influence diminished as flame temperature was increased. Both fuels had comparable levels of UHC emission at higher inlet pressure and biodiesel had slightly higher UHC emissions at lower inlet pressure.

For combustion dynamics, both fuels had comparable frequencies, but biodiesel had lower rms dynamic pressures at lower flame temperatures and higher when flame temperatures exceeded full load value.

Fuel droplet sizes were calculated and biodiesel had bigger droplet size, particularly at lower pressure, lower flame temperature and lower combustor pressure drop. The dominant contribution of big droplet size for biodiesel came from its much higher viscosity.