Draft GT2006-90154 Operation on Process Off-gas of a 24MW SGT-600 gas turbine on an LNG plant in China Torsten Strand, Siemens Industrial Turbomachinery, Sweden A privately owned LNG plant was started up at the Tuha Oil Fields in western China during 2004. The plant is running on associated gas from the oil field and the LNG is delivered to central China by trucks. The plant was delivered by Tractebel with Linde AG being responsible for the LNG process design. The compression set of the refrigeration cycle consist of a three-stage Ebara compressor driven by a 24 MW Siemens SGT-600 gas turbine operating on the off-gas from the LNG plant. The operation of the gas turbine integrated in this plant is associated with some special challenges: - the ambient conditions out in the desert. - the fuel varies from natural gas to a process gas consisting of methane diluted with up to 28% nitrogen - the refrigeration medium circulated by the driven compressor is changing in composition with load - the starting procedure with the compressor in the refrigeration loop A combustion test was performed to verify that the DLE combustion system could accept the variations in gas composition. The control system was modified to handle the variable gas qualities in the fuel and in the refrigeration loop. The commissioning of the gas turbine/compressor set was a long process governed by the LNG plant commissioning. It included some unexpected events. In the end all is working well. It is shown that a standard SGT-600 DLE unit can start and operate reliably and with low emissions on very much diluted natural gases. The LNG plant This very special LNG plant is situated at the Tuha oil fields in western China out in the desert some 300 kilometer south-east of the city of Urumqi, not far from the ancient Silk road. The plant is operating on associated gas from the oil field. The liquefied gas is transported by trucks to main China. The LNG plant is working like a huge freezer were the compressor for the cooling medium is driven by a gas turbine. The temperature necessary for liquefying natural gas is -180°C. The refrigeration medium is composed of hydrocarbons. The process is producing a boil-off gas that consists of methane and nitrogen. The content of nitrogen is increasing with plant load, up to 28 volume % at full load. The off-gas is used as fuel for the gas turbine. The requirements on the gas turbine Ambient conditions The plant is situated in position as far a possible from any sea, some 600m above see level. The ambient conditions are thus somewhat special • the pressure level is a little below normal, 93 kPa • the air temperature goes from + 45 to - 20°C • the humidity is in the range 4 – 15 % which generally means that the air mass flow in the gas turbine is on the low side and with a low specific heat. Figure 1: Schematic of the plant The gas turbine fuel The fuel is natural gas at start of the plant, but then mixed with an increasing amount of process off-gas in proportion to plant load. Also the process off-gas changes with load. It is consisting of methane diluted by nitrogen, the amount of which is increasing with load. At full load the fuel is supposed to be 95% process gas with a nitrogen content of 28 volume % and 5% natural gas: • the gas turbine combustion system has to coop with the changing fuel composition A general requirement was that the gas turbine combustion system had to be of dry low NOx type. © Siemens AG 2006. All rights reserved. The cooling medium The cooling medium is also changing with load, which means that the three-stage refrigeration compressor is operating at different conditions. It is normally natural gas at plant start up. With increasing plant load heavier hydrocarbons are added to decrease the temperature in the LNG tank. This has some implication on the gas turbine: • the variable speed compressor, which is driven by the power turbine of the gas turbine, is always in the cooling loop • the heavier the medium the higher the shaft torque (also at start up) • the compressor surge control, which is included in the gas turbine controller, has to be automatically adjusted for the change in cooling medium composition • the nominal compressor power demand is 17 MW at to = 32°C, but could on a cold day reach 22MW The waste heat recovery unit The exhaust system includes a waste heat recovery unit for oil preheating with a rather unusual configuration, which causes a somewhat high and variable pressure drop. GT10B Tuha exhaust duct Stack Butterfly valve GT Silencer Oil heater Figure 2. The exhaust duct with the heat recovery unit and it´s controlling butter fly valve The gas turbine/compressor set The gas turbine SGT-600 with a nominal shaft power of 25 MW at ISO conditions was chosen for driving an Ebara three stage compressor. The compressor has a maximum speed of 7400 rpm which fairly well matches the two stage power turbine, which has a nominal speed of 7700 rpm. The gas turbine has • a 10 stage compressor with a pressure ratio of 14:1 • a DLE combustion system consisting of a film cooled annular combustor with 18 burner of the lean premix type • a two stage internally cooled compressor turbine • a two stage uncooled power turbine Figure 3: Cross section of the SGT-600 gas turbine The main initial questions for this application were • will the combustion system handle the diluted fuel gas • how to modify the fuel control system for operation on variable nitrogen content in the fuel • could ignition be achieved on diluted gas • how to arrange surge control on the three stages of the driven compressor, considering the changing cooling medium A gas analyser in the LNG plant was installed for continuously supplying data on the fuel and the cooling medium to the plant and gas turbine controllers. The combustion system The DLE combustion system of SGT-600 was developed in 1989-90 and has a long experience on natural gas. The annular combustor is film cooled. About 60% of the compressor air is used as combustion air in 18 conical burners. The main fuel is injected along the air slots of the burner and the pilot fuel in the centre of the cone. e d c b a Figure 4 Cross section of the combustor and burner a) combustor, b) one of 18 burner, c) main gas manifold, d) pilot gas manifold, e) igniter and flame detector © Siemens AG 2006. All rights reserved. The experience of operation on diluted gas was limited to lower dilution rates. Several units are since long in operation on diluted natural gas in the Netherlands. The Dutch gas contains 14% nitrogen. Combustion air Vortex breakdown Main Gas Pilot Gas Main Gas Gas injection holes the burner, which was equipped with a glass slot along the cone. The fuel system was rebuilt for separate injection of nitrogen into the main and pilot fuel supply lines. The objectives of the test were in addition to the standard measurements of combustion testing (flows, pressure drops, temperatures etc) to • watch the flame for changing position and shape with variation of nitrogen content • measure combustion pulsations and emissions • check ignition performance Ignition Flame front Main&pilot fuel pfr mfuel, p, T mburner m tot, T, p Figure 5. The DLE burner of SGT-600 Air Main&pilot N2 Video cameras p The combustion system had also been operated on propane, thus with lower fuel injection velocities. ∆p mcool The general philosophy for this DLE burner is that due to the high combustion air flow (which is 80% nitrogen) there is a very limited influence on the combustion kinetics from a small amount of nitrogen in the fuel. However the nitrogen in the fuel will increase the fuel injection velocities, which could change the distribution of the fuel in the burners. The question was than if the injection geometry (main hole sizes or number of holes, pilot injector slots) had to be modified to accept the addition of 28% nitrogen. It was decided to perform a single burner test in order to find out if the standard geometry would work or is it had to be modified. The combustion test The single burner test was performed at the Combustion Test Centre in Whetstone outside Leicester, UK. A cylindrical test combustor for one burner was used. The geometry and wall cooling configuration was designed to resemble the annular combustor as closely as possible. It is always difficult to model the effect of the neighbouring burners of an annular combustor in a single burner test. Test burners with standard and enlarged main injection holes were fabricated. Several pilot injectors with enlarged and modified injection geometries were tested in a hydraulic rig and the best were chosen for the combustion test. It was believed that the increased pilot injection momentum could have a destabilising effect on the burner flow. The test combustor was installed in a combustion test cell, which had been modified for flame visualisation. One video camera was viewing the flame in the combustor and one was looking into N2 m, p, T P´ NOx, CO, TOC, O2, CO2 Figure 6. Schematic test set up The first test object was the standard burner and pilot fuel injector. The test program covered a number of points on the gas turbine operating line at different ambient conditions. At each point the pilot fuel ratio was varied. The testing started with operation on natural gas. Nitrogen was first added in steps to the pilot gas and then to the main gas. Test result It was found that the nitrogen injection in the pilot had little influence on the flame position. The flame position was however quite strongly influenced of nitrogen injection in the main gas. The flame was creeping more and more inside the burner, as an effect of the stronger fuel injection towards the centre of the burner At very high nitrogen rates in the main gas an unstable situation was reached: a repeated process of • flame slowly retracting into the burner • instant flame back (the flame moving upstream on the centre line of the burner, not heating the burner walls) • flame “spitted out” The range for operation on diluted gas was shown to increase by increasing the pilot fuel ratio, mainly as an effect of reduced main gas penetration. The standard burner could operate without coming close to flame back on 28% nitrogen, so the test on © Siemens AG 2006. All rights reserved. ©Siemens Power Generation 2003. All Rights Reserved the modified burner was found unnecessary. A test with a modified pilot injector was performed but without any measurable influence on the performance. Power Generation Figure 7. Flame back sequence The results of the test were very positive since it was found that • the standard burner and pilot injector could accept the high nitrogen content with margin • the nitrogen dilution range could be extended by increasing the pilot fuel ratio • no combustion pulsations were found • flame back conditions could be indicated in advance by the increasing pressure drop variations across the burner • ignition on diluted gas could be achieved with a slight increase in fuel flow, basically to keep the combustible gas content constant SGT-600 Single Burner Test at 100% load 400 N2 in pilot NOx 60 300 N2 in main fuel 40 200 Pilot fuel ratio 20 100 CO 0 70 75 80 85 90 CO ppm v Pfr% , N2 ratio % , NO x ppm v 500 80 95 0 100 Test reading no Figure 8. General emission test results at 100% load The emission performance looks basically the same at all loads as shown in figure 9. NOx is strongly depending on pilot fuel ratio. NOx is decreasing with increasing N2 in the pilot fuel, but not affected by the N2 in the main fuel. CO is not affected by the N2 in the pilot fuel but increased by the N2 in the main fuel. The general conclusions are that • the SGT-600 cone burner has a very wide capability to accept diluted fuels • • the dilution limit resulting in flame back can be detected in advance by a standard pressure drop measurement on the combustor ignition can be achieved on diluted fuel dilution of natural gas results in lower NOx emissions Plant commissioning Since the refrigeration compressor is in the cooling loop of the LNG plant, the commissioning of the gas turbine and the plant had to go hand in hand, which proved to be a comparatively long process. Normally the SGT-600 is commissioned in six weeks, but in Tuha it took several months. 11 100 • Start procedure The start up of the compressor set showed to be somewhat different, since the compressor required quite a lot of power already at low speed in spite of fully open surge control valves. Normally the gas generator is accelerated to an idle = no load condition. But in this case high load was reached already during that acceleration so the acceleration sequence had to be modified to a slower continuous loading process, governed by the time needed by the plant itself. The commissioning was mainly done during the summer with ambient temperatures in the range of 35-45°C, somewhat low ambient pressure and a humidity of around 4 %. The air mass flow into the compressor could be characterised as “low and thin”. Too high engine exhaust temperatures were measured during the initial gas generator acceleration, often tripping out the gas turbine. The start setting values (cross ignition gas level, acceleration ramps) were adjusted with some improvements. After some time it was found out that the explosion proof thermo-couple wiring between the gas turbine and the control room was of different standard, resulting in too high readings at high ambient conditions. The temperature limits could with this knowledge be compensated for this effect. In due time the wiring was exchanged. The gas turbine was supposed always to start on natural gas. However after operation on process gas there is diluted gas in the 55 m long supply piping, not natural gas. The problem was solved by a control program modification, so that the start control sequence is for some time using gas data from the last stop to calculate the fuel valve values. A small gas pipe for natural gas was installed for the torch burners instead of using the standard set up with propane bottles, which were scare in supply in this part of China. Cross ignition on the diluted gas showed to be of no problem, as expected after the combustion test. © Siemens AG 2006. All rights reserved. The increased and varying pressure drop in the exhaust duct was at first a problem, but after some time it was found that the controlling butterfly valve had a 90° position error. It was closed when supposedly open. After correcting the valve the exit pressure drop caused no problem. The tuning of the compressor surge control system showed to be rather tricky. It was no problem to tune each bypass valve on the three stage compressor, but it took some time to get all three stages working smoothly together. The difficulty seemed to be related to the changing compressor stage characteristics with changing composition of the operating medium. Performance test At the end of the commissioning a performance test was run with a special set up of measuring equipment including registration of combustion pulsation, dynamic combustor pressure drop and emissions. The load variation on the gas turbine was governed by the plant operation and was rather limited, between 13 and 15 MW. Also the operating conditions was set by the plant. A systematic test program could not be followed. However, increasing plant load means basically lower refrigeration temperature and changing fuel conditions as described above • increasing amount of nitrogen in the fuel gas • increasing amount of heavier hydrocarbons in the cooling medium During the test series at different plant loads, which took several days, pilot fuel ratio was varied up and down to find out if there were any risks for combustor pulsations or flame back (indicated by increasing amplitude of the dynamic combustor pressure drop). The N2 content in the fuel gas also varied due to changing plant load. Some additional N2 was injected in the fuel in order to extend the dilution rate somewhat. It was however not possible to reach the absolute full load conditions during the test. The results of the engine test were very good • no combustor pulsations or flame back indications even at extreme pilot fuel ratios, so the pre-set pilot fuel ratio schedule could be used • the emissions with nitrogen dilution was following the same trends as seen at the single burner test. NOx is increasing with pilot fuel ratio, but decreasing with increasing N2. CO is decreasing with pilot fuel ratio but increasing with increasing N2. Conclusions The integration of the gas turbine in the LNG process was not without problems, but they were sorted out before and during the commissioning. The combustion test showed that the standard combustion system could handle up to 35 vol-% N2 in the fuel. With a modified fuel injection still higher rates would be possible. The site test confirmed that the SGT-600 gas turbine could be started and operated on both natural gas and a process gas consisting of methane diluted with a varying amount of nitrogen. The NOx emissions were slightly lowered by the dilution, no combustor pulsations or instabilities were detected. Reference Baseload LNG production in Xin-Jiang – a remote source of clean energy for gas consumers in China. Xiang Dong, Xin Jiang LNG Development Co, Eginhard Berger, Linde Ag, Albert Meffert, Tractebel, Li Wei Bin, SSEC . 2005 AIChE Spring National Meeting in Atlanta, GA, USA © Siemens AG 2006. All rights reserved.