Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates*

Transcription

1 Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates* Cylinder wear and cylinder lubrication oil consumption are essential parameters for the operating economy of large 2-stroke diesel engines. Therefore it is of great economic importance to minimise the cylinder wear and/or the consumption of cylinder oil. The new HJL SIP lubrication concept meets these demands. Tests in service have shown a potential of up to 50% cylinder oil savings without trade-off in liner wear. INTRODUCTION During the last 25 years a continuous increase in specific cylinder oil dosage has taken place for 2- stroke engines. This development has been triggered by super long stroke engines where actual oil. service results have proved this increase particularly necessary. Similarly, results from shorter stroke engines have led to engine designers' recommendations of increased lubrication. One may wonder why this development never has been thoroughly questioned, as a) the sulphur content has by and large remained unchanged and consequently the TBN of the lubrication oil has remained more or less firmly at the level of 70 irrespective of lubrication brand. b) the pure lubrication demand has never reached the dosages actually applied. The cost split of keeping a cylinder unit in service, as shown in fig. 1, may indicate why. The distribution shown is based on a rather low specific cylinder lubrication consumption of 0.65 g/bhph(mcr) when liner and rings are run-in. It is obvious that the apparent gcure h for poor cylinder condition, the increased lubrication oil consumption, has been accomplished on the users' account, being the only party to pay for the cylinder The high lubrication oil dosage is not just a matter of increased costs it is just as well a matter of environmental considerations as engine smoke formation and oil sludge production naturally is increased with the cylinder oil consumption. In the light of the above considerations it is quite obvious that the cylinder lubrication system as such needed thorough scrutinisation. The final devel-

2 Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates the cylinder oil supply tubes is comparatively low, (typically a few bar). The cylinder oil can only enter the inner liner surface, if the cylinder pressure is lower than the pressure Consequently, in the oil supply tubes. the cylinder oil is not necessarily dosed at the optimal position of the piston, in relation to the position of the oil quill. The oil compliance due to dissolved air in the oil may even imply that more lubricator piston strokes are needed before oil is supplied to the cylinder surface. In the HJ SIP lubrication system, the dosing takes Fig.1 Typical Cylinder Cost Distribution, K90MC-engines opment and testing of a new system is based on a cooperation and development between the authors' companies. The result of the endeavours is now available in the shape of a patent pending system, HJ SIP lubrication(swirl Injection Principle)[1&2]. Here a new technology has been brought about to enable a possible lower lubrication dosage. The gas swirl in the cylinders of 2-stroke uniflow engines is used to carry fine atomised oil droplets to the entire periphery of the cylinder liner prior to the passing of the piston rings. The basic idea of bringing the lubricant close to the top of the cylinder liner where the highest wear takes place is not new and has been known in several 2-stroke engines with 2-level lubrication, however, the oil distribution on the liner sliding surface has previously not been dealt with. The results obtained show a potential for most users to reduce the lubrication dosage by about 50%. DESCRIPTION OF THE SYSTEM The main criterion for the project has, as mentioned, been the need for a better initial distribution of the cylinder oil on the liner surface. Furthermore, it was desirable to improve the timing of the lubrication and, if possible, make it less crucial, based on the following conditions: In most traditional systems the pressure level in place before the piston top reaches the oil quills. Due to their special design the injection valves do not open until a well defined relatively high pressure in the oil supply tubes has been reached. The lubricator pump stroke is effected shortly after the scavenging air ports close, during the upward movement of the piston. The pressure in the oil supply tube will always be higher than the pressure in the cylinder at the time of dosing. It ensures that dosing is taking place at every lubricator pump stroke. Because of the early timing, the timing point is not as crucial as it is in systems where the oil has to be supplied during the very short period when the piston rings are off the oil quill. The delay due to the ultimate velocity of the pressure wave is in this connection of less importance, but it can of course be taken into consideration determined/optimized. when the timing point is During the upward movement of the piston the scavenging air in 2-stroke uniflow diesel engines is subjected to a powerful rotation at the same time as the gas is dislocated upwards in the cylinder. Thus, the gas follows a helical path or a swirl on its way from the scavenging air ports to the exhaust valve. An oil drop in this swirl will due to the centrifugal force be forced against the cylinder wall and eventually settle there. This effect is utilised by introducing the oil portions in the cylinder through nozzles as a cone of oil droplets of suitable size. By adjusting the nozzle dimensions, the oil outflow speed and the pressure before the nozzle, it is possible to control the system parameters in such a way, that practically all the oil settles on the cylinder Journal of the JIME Vol.37, No.2 42

3 Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates Fig.2 HJ SIP Lubrication System wall before it is caught up by the piston top. The directions of the nozzles as compared to the flow in the cylinder are arranged in such a way that the interaction between the individual oil droplets and the gas flow in the cylinder ensures that the oil droplets hit the cylinder wall periphery over an area corresponding to 1-2 times the distance between two neighbouring oil quills. Within this area the thickness of the oil layer varies, however with a good coverage over at least an area corresponding to the distance between two neighbouring quills. In this way the oil is distributed nearly evenly over the cylinder periphery already before the piston ring passage. Furthermore, the upward direction of the nozzle will cause the oil to hit the cylinder wall higher up than the nozzles. If the nozzles are placed at the same height in the cylinder as the existing oil quills the oil will then, already when introduced in the cylinder, not alone be better distributed over the cylinder surface but also be gdelivered h to the cylinder surface closer to the cylinder top where the need for lubrication is higher. Both of these conditions cause a better utilisation of the oil with an expected improved ratio between cylinder life and lube oil consumption. The supply of oil to the cylinder surface is effected gtimed h and in volumetric measured portions by means of a modified HJ timed lubricator. The design of the system is shown in fig. 2. At suitable intervals a number of valves (3) are placed in the cylinder liner (5), which are characterised by being adjusted to open at a certain pressure in the supply tube (2) leading from the lubricator (1) to the individual valves (3). At the end of the valves (3), a little recessed from the inner liner surface, a nozzle (4) is mounted, through which the oil is transformed to droplets when the pressure in the supply tube (2) reaches a certain pre-set value. The valves (3) are designed according to the same principles as traditional fuel valves. The oil leak occurring due to the design is lead through a return tube (6) back to the lubricator supply tank (7). (1) or the The valve design appears from fig. 3: The pressure tube is connected to the inlet (2) and the oil is lead through a strainer unit (3) to the space (5) between the valve housing (4) and the inner part (7). From there further on to the inclined holes (9) in the inner part which also contains the seat for the valve needle (8). The valve is kept closed by the spring (6). When the axial force on the valve needle from the oil pressure exceeds the spring force the valve opens for the oil to enter the space (12) in front of the nozzle opening. The spring force and thereby the opening pressure of the valve can be adjusted using the adjusting screw (1). The leak oil is carried off through the leak oil connection (10). The valve housing is mounted in a bore in the cylinder liner (11) and fastened with two screws as shown in fig. 4. Fig. 2 shows the valves mounted radially in the Journal of the JIME Vo1.37, No.2

4 Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates Fig.4 HJ SIP Lubrication Valves, Mounting Possibilities model of the gas flow in an engine cylinder in which the oil is introduced through nozzles, as described above. Fig.3 HJ SIP Lubrication Valve cylinder liner. This mounting is clearly the most simple, however, outer conditions may necessitate a non-radial mounting. Such possibilities are shown in fig. 4. The system may be mounted on new engines or retrofitted on engines already in operation. 2. Design of valves etc. for the system. 3. With the purpose of demonstrating the oil settling on the cylinder wall and the function of the system on the whole, a test rig in the form of a helical gwind tunnel h, was designed. It simulates the scavenging air flow in the cylinder at the expected injection time of the cylinder oil. The results from the tests were in good conformity with the computer simulations and the expectations to the system. Computer simulations of the air flow in the wind tunnel formed the basis for the construction. COMPUTER SIMULATIONS Three major areas for the utilisation of CFD simulation tools within the project were defined. The PLANNING By November 2001, the system has for test purposes been operating on two cylinders on a MAN B&W 12K90MC in a sailing vessel for approximately 14,000 operation hours. As mistakes in connection with such a test can be extremely costly, an extensive 3-step design and verification programme preceded the installation: 1. Before designing the system a series of computer simulations were carried out with the purpose of making the expected effect of the system probable. The simulations are based on a dynamic first part was to determine the gas flow conditions inside the cylinder to obtain knowledge about the environment into which the oil spray was to be injected. The second part was to determine a spray type which is able to produce an appropriate oil film at the cylinder liner surface. Finally, the geometry of a wind tunnel for testing different valves and prototypes was developed. Gas flow conditions inside the cylinder The simulation results showed a fairly strong and stable, reproducible swirl flow inside the cylinder for a significant period of time between end of scavenging and end of compression. This flow was Journal of the JIME Vol.37, No.2

5 Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates 129 analysed regarding its velocity components(rotational and upwards), near wall profile, temperature and pressure. The results were assessed by all partners using their specific experience. lubricating oil injection was fixed. The time window for Determination of spray parameters and nozzle geometry In the second part of the simulation-based predevelopment phase the interaction between the droplets of a spray and the general flow inside the cylinder of an MAN B&W K90MC engine was investigated. The target was to determine the influence of different spray parameters on the spray's capability to produce a wall film. The solution approach chosen for this investigation was based on a CFDsimulation strategy. The optimum spray parameters were established and the main nozzle parameters necessary to generate such a spray were derived. These nozzle parameters were: the hole diameter, ratio of hole diameter/hole length, spray direction, spray cone angle, operation pressure and pump stroke length. Unfortunately, some of these parameters cannot be directly determined by means of calculations or simulations. Consequently, the fine tuning had to be done using experiments. Development of wind tunnel design The subject investigated in the final part of the simulations was the layout of an appropriate channel. test This channel was to be used for experimental investigations of different nozzles with respect to sprays and wall films produced by these nozzles. Therefore, the test channel not only had to represent the general geometrical dimensions of the cylinder but also the typical flow field inside the cylinder as it was determined during the first part of the investigations. Thus, the target was to fix the layout of the test channel in such a way, that the shape of the channel produces an internal flow field similar to the flow conditions inside the cylinder. A second target was to minimise the dimensions necessary for this test channel in order to simplify the handling of the channel and to minimise the size of the 'laboratory' needed. As a result of the simulations the geometry of the channel was fixed, the segment to be used for the experiments was described and advise was given of how to establish the inflow conditions into the wind tunnel. FULL SCALE LABORATORY TESTS The purpose of the laboratory tests was to demonstrate the results from the computer under conditions simulations as close to reality as possible before mounting the equipment in cylinders in operation. In this way the largest possible safety for eliminating mistakes before mounting in a vessel was obtained. A test rig was designed and constructed for the laboratory tests. The central part of the test rig is designed as a twisted "wind tunnel" made from transparent plastic (acrylic). The diameter of the inner side of the tunnel's outer wall is 900 mm, corresponding to the cylinder diameter of the MAN B&W K90MC engine on which the equipment was later to be mounted. A blower supplies the tunnel with air flow in the twisted part of the tunnel corresponding to the ultimate velocity of the scavenging air in the engine at the time of dosing. The transparent twisted part of the tunnel allows visual observation of the dosing process and the resulting distribution the tunnel surface. of the oil on The valve injecting the oil is mounted in a cylindrical cover in the curved outer surface of the tunnel. The valve is placed somewhat after the change between the straight and the twisted part of the tunnel, but in the lower half of the tunnel. At this place the effect of the flow shifting between the straight and the twisted part has decreased so much that it has no practical significance. Fig. 5 shows a photo of the actual test rig where the valve position can be seen. Air velocities in the tunnel are measured by means of a Pitot tube. A modified traditional timed Hans Jensen lubricator was applied for dosing the oil

6 130 Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates with the deviations pressure was minimal in the oil supply tube siderations regarding tion of oil which the prevailing among other things was estimated evapora- to be minimal with the carrying scale tests it was verified at have no negative with the applied valves effect on the timing up to 15 meters even in supply long. The tests showed at each lubricator pump frequencies out of the full that the oil compressibili- ty and the tube compliances oil is dosed on con- temperatures. In connection tubes due to the high and based pump stroke that at the and the dosing rates valid for the actual engine. Fig.5 Wind Fig. 6 shows a representatively Tunnel sure recording. to the valves. delay for the pressure Cross-sectional dimensions, climbing angle and positioning of the dosing determined based on the prior mentioned valve in the tunnel liseconds imately atmospheric in the wind pressure approximately deviate from the values which viscosity as the cylinder prevailing in an engine influence oil viscosity It was estimated in tolerable, at liner oil used the timing is deter- the engine 60 mm. piston Even a consid- as far as the delay is concerned as the demand for the timing is is far less where the oil is dosed during of the piston rings. The results 85 degrees when approximately the passage temperature 1,200 m/s, i.e. approx- the same period strict than for systems, is the same to be approximately During erable uncertainty possible 5 mil- corresponding the sonic speed in the oil. This time delay has moved an oil was applied, at room temperature which is presumed Celsius. Both parameters best mined. was Celsius. to compensate for the temperature was the and the air temperature 20 degrees In order tunnel pres- that the time wave is approximately for a 6.2 m tube length, can simply be included The static pressure simultaneous this appears to a velocity of approximately were computer simulations. operation. From shown for the wind in fig. 6 are obtained with the correct tunnel tests. Similar tests results. To save time, the tests have been concentrated on that the risk in connection oil have confirmed with the an MAN B&W K90MC engine. assume these It is reasonable that as test and computer form so well with each other, simulations to con- as is the case, com- puter simulations should be sufficient basis for implementation in other engine types and sizes. Video has been recorded pling equipment with high speed with a picture frequency sam- of 1,000 pictures per second. The video sequences have confirmed the assumption that the oil acts as expected as compared to the presentation and the com- puter simulations. Fig.6 Pressure the Recordings, Lubricator and immediately immediately after By running before and manually drawing a transparent piece the Valve Journal of the JIME Vol.37, No.2 46 a few pump strokes with the lubricator the distribution of paper 日本 マ リンエ ンジニ ア リング学 会 誌 of the oil on on the transparent 第37巻 第2号(2002-2)

7 wirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates 131 Fig.7 Oil Distribution on Wind Tunnel Walltypical Example twisted part of the wind tunnel, a series of drawings of the distribution contours have been produced. These show good conformity with the predictions of the computer simulations. Fig. 7 is a reproduction of such a distribution graph. The darkest areas represent the thickest layer of oil. The cone angle of the oil droplets has in the tests shown to be somewhat smaller than assumed in the simulations but the oil distribution on the gcylinder wall h showed to be in reasonably good conformity with the simulations. It has been demonstrated that a modified traditional HJ lubricator is capable of delivering the pressure necessary for the formation of oil droplets. Tests have been carried out at 80 bar without problems, however, based on the test results, the pressure level for the operation tests were set at bar. TESTS IN SERVICE The practical test has been carried out onboard M/V gsovereign Maersk h, a container vessel equipped with a 12-cylinder MAN B&W K90MC, mark VI engine having the following particulars of special interest for this test, see table 1. This engine type has, since the very first went into service, seen quite a development in fuel equipment in order to combat heavy coke deposits in the combustion chamber and the exhaust system. The coke formation on the piston top land reached such dimensions that it wiped off the oil from the cylinder sliding surface. As a consequence of the above the standard fuel valves were exchanged with so-called slide valves as described by Pedersen[3], and furthermore a piston cleaning ring,(pcr) positioned in the top of the liner was introduced to prevent coke from touching the liner, as described by Mikkelsen &Bryndum[4]. The PCR was installed in all cylinders prior to starting the test and the slide valves were introduced during the tests at 13,200 engine running hours. The average wear rate of the liners until the slide valves were introduced were mm/1000h (scuffed liners excluded) at a cylinder oil dosage between 0.65 and 0.85 g/bhph(mcr) after running-in of the liners and rings. It was decided to install the new lubrication system on two cylinders, with cylinder 7 as the first and cylinder 12 next - after thorough test of system components in cylinder 7. This decision was proved right, as some components actually needed modification before functioning as expected. At the time of writing the system has been in operation for about 14,000 hours and apart from measuring the liners in the traditional manner and monitoring the cylinder oil dosage, a special test of 15 days duration with an activated top piston ring has been performed. Traditional wear measurements In the cylinder lubrication oil dosage and the cylinder liner wear is shown covering the entire test period(that is to be continued for some time). It is clearly seen that some difficulties were

8 Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates SIP LUBRICATION, gsovereign MJERSK h Lube Oil Dosage&Liner Wear Fig.8 SIP Lubrication, gsovereign Maersk h encountered on cylinder 7 during the first nearly 4,000 running hours, hence the dotted line in that period for the liner wear and similarly for cylinder 12 for the first 1,000 hours. What went wrong was unforeseen lubricator adjustment problems leading to an uncontrolled reduction of cylinder oil dosage. Furthermore the timing needed adjustment. It is, however, worth to mention that no scuffing occurred on liner and rings during that period. After the rectification of the system the development in liner wear has been rather moderate. In the latest period the wear rate has been less than 0.02 mm/ 1000 h. At the same time the lube oil dosage has gradually been reduced and then kept at the fairly low level of 0.43 g/bhph(mcr). It is of course difficult to measure such a low wear development with a high accuracy using the traditional cylinder calibration gauge. However, the wear pattern on the liner surface has since the start of the test been carefully observed and the slow disappearance of the wave cut has thus confirmed that the measurements were reasonably accurate. Distributing the oil evenly high up in the cylinder liner and above the standard lube oil quills could raise fear that it would lead to oil starvation in the lower part of the liner. The wear measurements taken from top to bottom and in four diagonal directions show, however, that the specific wear all over the liner is very low and actually below some of the best results obtained in a similar, comparable engine with the traditional lubrication system. Fig. 9 shows a comparison of wear distribution in the liner between the traditional and the spray lubrication system. The lube oil dosage for the traditional system is 0.65 to 0.7g/BHPh to be compared with the dosage shown for cylinder no. 12 in fig. 8. In order to compare the absolute obtained wear result in the test cylinders and to the highest possible degree eliminate any uncertainty, reference cylinders on the same engine have been selected. Actually the first 6 cylinders of the engine is now equipped just as the two test cylinders in respect of liner type, piston ring configuration and fuel valves. For comparison the wear rates of 4 control cylinders during the test period are 0.172, 0.138, and 0.070mm/ 1000h respectively. A parameter of great influence on absolute cylinder wear measured over a period of time is the engine load. This has been discussed by Dragsted151, who suggested a so-called load profile number, LPN to be used in the evaluation of the severity of the engine operation(lpn=the average load in percent of MCR +the number of running hours at a load higher than 80% MCR in percent of all running hours). The LPN for the MAN B&W 12K90MC engine during the discussed test period has been 132, which 48 Journal of the JIME Vol.37, No.2

9 Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates Fig.9 Cylinder Liner Wear Pattern for a Traditional and a SIP Lubrication System according to previous experience is considered rather high even for container vessel engines. The influence of engine load on piston ring wear will be demonstrated in the following section. Radio-Nuclide measurements To determine the influence of engine load on the wear at the piston ring/cylinder as liner system under real operation conditions a Radio-Nuclide wear measurement was carried out at one of the cylinders equipped with SIP lubrication.[6] Online measurements were carried out throughout the whole voyage, off-line measurements were made during port times. The results of both measurements were in good agreement. Only little running in effects were observed. -The measured wear rates of the top piston ring varied between 0.01mm/1000h and 0.065mm /1000h. Tendencies between oil injection timing and wear, oil injection dosage and wear and load and wear could be measured. -Measurements with similar engine operation conditions led to similar wear rates confirming reproducible functionality of the system. -The cylinder oil timing chosen for the entire service test (standard timing) showed a generally low wear rate, which increased with increasing engine load. -The measurements with changed injection timing confirmed the fact that the SIP lubrication system is not sensitive to such changes, considering the big steps of plus/minus 15 degrees crank angle. a Fig.10 Measured Wear Rates dependent on Engine Load, Lubrication Oil Injection Timing and Oil Dosage Generally, the variations in the injection timing led only to small changes in the wear rates which, however, all are at a very low level. The earlier timing led to increased wear at higher loads whereas the later injection timing did not show any load dependence. -The oil dosage was reduced in several steps. The measured wear rates did not increase even for the lowest dosage of 0.43 g/bhph (MCR). The potential of the system to reduce lubricating oil consumption without trade-off in liner wear was confirmed. During all measurements no abnormal wear phenomenon were recorded. This was confirmed by the final inspection of the liner at the end of the measurements. The wear rates remained at a very low level during all the tests. CONCLUSION The experiments and tests carried out with the SIP system, of injecting the cylinder oil into the air swirl, has proven that it is possible to reduce the cylinder oil consumption and at the same time attain acceptably low cylinder liner and piston ring wear. The scope for cylinder oil savings will for most users be close to 50%. Journal of the JIME Vol.37, No.2

10 Swirl Injection Lubrication Low Cylinder Oil Consumption without Sacrificing Wear Rates From May to November 2001 installation has been done on additionally 47 cylinders in service. Results are not yet available, but indications are positive. The target with the project has been achieved. Considerable reduction in cylinder oil consumption is possible without harming neither the liner, the piston ring nor the piston condition. It is the intention to keep monitoring the test cylinders, and attain service experience from additional applications to make guidelines for optimizing the use of the system. What are the functional limits of the system, and within these limits, what are the parameters to be used for fixing the operating costs in the trade off between liner wear and cylinder oil consumption. REFERENCES [1] Lauritsen, S., Dragsted, J., Buchholz, B., gswirl Injection Principle h, CIMAC Congress 2001 [2] Patent application PCT/DK99/00599 glubricating system for large diesel engines h [3] Pedersen, P. S., gthe Intelligent Engine-One Solution to NOX Reduction h, ISME 2000-PD-22. [4] Mikkelsen, U., & Bryndum, L., glarge Bore Engines Problems and Countermeasures h, The Motor Ship Marine Propulsion Conference [5] Dragsted, J. gengine Cylinder Condition, from delivery and some 10 years ahead h. CIMAC Congress 1993, paper D15. [6] Bludszuweit, S., Schwarte, J., Prescher, K. & Richter, B., ga Super-Sensitive Method for Wear Measurements Inside Ship Engines h, CIMAC Congress 1998