|Search||Hot Links||What's New!|
Please let me remind all of you--this
material is copyrighted. Though partially funded by NASA, it is still a private
site. Therefore, before using our materials in any form, electronic or otherwise, you need
to ask permission.
There are two ways to browse the site: (1) use the search button above to find specific materials using keywords; or,
(2) go to specific headings like history, principles or careers at specific levels above and click on the button.
Teachers may go directly to the Teachers' Guide from the For Teachers button above or site browse as in (1) and (2).
An engine 12 inches long and 4 inches in diameter, and weighing only
4 pounds, is ready to be
flight-tested in a model airplane.
by Franklin van den Hout and Jo Koullen
Since the introduction of the first operational turbojet
engine some 50 years ago by Frank Whittle and his team in Great
Britain, the engines have become bigger and bigger to meet the demands of high fuel efficiency and specific thrust. Six
years ago, a team of researchers in the Netherlands began to go against this trend by designing and manufacturing a very
small turbojet engine.
Measuring 12 inches long and 4.3 inches in diameter, this 4-pound engine posed a significant engineering challenge.
Following many hours of static runs and tests, the engine will soon be flight-tested in a model of a de Havilland Vampire III
Because of the engine's size, its thermodynamic cycle ischaracterized by relatively high operating temperatures, very
low component pressure ratios and efficiencies, and the high rotational speed of its rotor. The low Reynolds numbers
encountered (1.5x105) indicate the dominance of frictional forces and losses plus their influence on boundary-layer
behavior. The result is a very sensitive engine in terms of operation and off-design behavior.
Goals for the engine's design-point performance at ISO conditions were gross thrust of 9 pounds, specific fuel consumption of 0.225 kilogram per newton-hour, compressor airflow of 0.33 pound per second, compressor pressure ratio of 1.90, turbine-inlet temperature of 1,256°F, rotational speed of 100,000 rpm, and maximum sound-pressure level of 90 decibels at 39 inches. A turbine fuel of JP3/JP4 kerosene and natural gas was used.
The off-the-shelf rotor assembly is a Garrett Airesearch Model T3/T31 60-TRIM. This assembly, with its overall diameter of 2.3 inches and weight of 10.5 ounces, has proven to be an excellent choice. The high-peak-efficiency,backswept-vaned compressor has a wide stable operating range of airflow for a given rotational speed and is already of excellent aerodynamic design. The radial inflow turbine, made of Inconel 713 alloy, is capable of handling operating temperatures up to approximately 1,652°F.
Today's lightweight turbocharger rotors have very short shafts, leaving little space between compressor and turbine rotor (approximately 3.5 inches) to add other components such as the combustor, compressor diffuser, and turbine nozzles. Positioning all of these components within this limited space would be impossible.
Therefore, instead of an annular straight-flow combustor arrangement, a reverse-flow
annular configuration was chosen, to be situated aft of the turbine. Although the length
of the turbojet engine increased, the overall diameter was kept very small. For
applications in model jet aircraft, overall body diameter is more critical than length. A
small diameter also allowed for unlimited axial enlargements of the combustor and
eliminated the complex integration of a straight-flow combustor with the bearing module,
Placing the combustion chamber behind the turbine isolated the much colder bearing module and compressor arrangement from the high temperatures encountered in the combustor and turbine. Measurements have shown that bearing temperatures are kept at a surprisingly low 50°F throughout the stable-speed operating range of the engine.
Special bearings and other complex cooling arrangements are therefore unnecessary.
The designers' experience with pulse jets fueled by liquefied petroleum gas led them to select kerosene fuel for this turbojet engine. Kerosene has good combustion characteristics when properly atomized and is much safer to handle than gasoline or liquefied petroleum gas. Although very flammable when atomized, kerosene has less of a tendency to form explosive mixtures quickly when a leak occurs, such as in pumps or feed lines. Safety in fuel handling and engine operation has been a serious design issue.
The use of kerosene did require the addition of a booster pump for fuel pressurization. The main difficulty with kerosene fuel was proper atomization or evaporation within the limited space of the combustion chamber. Ample margins in temperature (thermal stress) and rotor speed (component loadings) have been provided to ensure long life, particularly for the hot-section components. The maximum operating turbine-inlet temperature of 1,382°F is well below the turbine's capability. The maximum rotational speed of the rotor during takeoff is approximately 105,000 rpm, which is also well below the engine's design limits.
Although the rotor is extremely lightweight, its rotational speed is high and the risk of noncontainment should not be ignored. During tests of the preliminary design, an unexpected axial movement of the compressor occurred at 40,000 rpm. The aluminum rotor rubbing against its casing caused the rotor to decelerate quickly to rest. There was no severe damage to the other components, and any broken pieces remained within the engine.
A modular design concept has been maintained throughout the entire engine. All components have been designed to be easily manufactured and put into series production. The turbojet engine has been divided into several modules that are combined by means of bolts, press-fit connections, and special clamps.
A typical bell-mouth-shaped intake was used in this engine for good inlet flow characteristics. Aircraft speeds of approximately 175 miles per hour have been anticipated. A new divergent intake is therefore being considered to reduce the inlet air velocity to acceptable levels and thereby prevent negative-incidence stall at the inducer vanes of the impeller. The intake and compressor casing are two separate items made of an aluminum alloy. The compressor casing also includes a messing nozzle that feeds compressed air to the compressor rotor to accelerate the rotor to ignition speed.
The machining of the rotor casing had to be highly accurate.Large tip clearances are detrimental to overall engine performance and compressor efficiency, especially on an inlet diameter of only 1.8 inches. Tip clearances have been kept between 0.008 and 0.012 inch. A tip clearance of less than 0.008 inch required too much effort for radial and axial rotor displacement control and alignment, and was not pursued.
The proposed engine layout and the selected radial compressor required compressed air
to be redirected to an axial flow. The rotor diameter of 2.3 inches was small enough to
include a radial first-stage vaned diffuser without exceeding the target overall diameter
of 4.7 inches. A 90-degree axial redirector channel and a second-stage flow straightener
are integral parts of the diffuser and were designed to provide an axial flow of air
without too much swirl. The entire diffuser has been manufactured from a single piece of
Impeller exit velocities were estimated to be around Mach 0.55 for the turbojet engine. At Mach 0.5 and above,compressibility effects have to be taken into account when designing a diffuser; above Mach 0.6, these effects are considerable. The severe pressure gradients at the inlet to the diffuser are caused by large changes in air density, and could easily result in early boundary-layer separation and thus poor diffuser performance. The divergence angle or area ratio at the inlet of the diffuser should therefore be small (i.e., conservative) but may increase progressively toward its exit.Such trumpet-shaped diffuser designs can be difficult to manufacture.
For this design, however, easy-to-manufacture trumpet-shaped diffuser channels were made possible by so-called symmetrical half-moon-shaped diffuser vanes. Many small engines use similar designs.
Calculations for this small diffuser can be only approximate because of factors that
are difficult to quantify, such as boundary-layer behavior and the nonuniform inlet
conditions of the air entering each diffuser passage. Creating the diffuser
therefore required a mixture of art (a good eye for shape) and science.
Splitting a flow of air such that each diffuser passage has the same air-mass flow and
pressure is difficult. The imbalance between adjacent diffuser passages increases the
likelihood that the compressor will undergo a surge, as does an increase in the number of
vanes. Increased surface area, friction, and diffuser blockage are side effects that must
be taken into account when a large number of vanes is selected.
Diffusers having 15, 16, and 17 radial vanes were tried. The final configuration chosen was 17 vanes because this number
had the best performance within the part-speed operating range of the engine. An important aspect with respect to the
number of vanes is aerodynamically induced blade vibration caused by the blade-passing frequencies of rotor and stator
vanes. An unbalanced ratio of the number of diffuser vanes to the number of compressor rotor blades is preferred. To
damp aerodynamically induced blade vibration and to even out any imbalances in airflow and pressure to each passage, a
vaneless space of 0.2 inch was introduced between the rotor exit and diffuser inlet.
During the tests of early designs, low-frequency rumbles could readily be heard at certain operating points of the engine. Pressure fluctuations also indicated such flow instabilities and the sensitive nature of overall engine performance.
The design of the reverse-flow combustor has been largely a process of trial and error. At 4.3 inches long and 3.5 inches
in overall diameter, the combustor has reached an absolute limit with respect to the available volume for reasonably good
combustion characteristics on kerosene fuel. Much effort has been put into testing specific cooling concepts (film and
impingement). After many test runs, which lasted up to 10 minutes each, visual inspections were performed. Color differences on the liner and the appearance of soot, smoke, and carbon deposits provided feedback on the quality of combustion with respect to uniform temperature distribution and local overheating (hot spots). Feedback was also obtained on the proper layout and size of these cooling holes.For these small combustors, a correct balance of several cooling techniques is very important.
Impingement cooling of the hot gases and the combustion-chamber inner walls by air jets has been a predominant technique in the primary and secondary zone of the combustor. Film-cooling techniques have been used extensively in the tertiary zone and around the transition piece near the turbine's first-stage nozzle assembly. Because of the changing direction of the impinging hot gases, this transition piece required extra film cooling. Without this thermal barrier, local overheating and even meltdown could occur very easily.
The two other important aspects of combustor design have been primary-zone recirculation and flame stabilization. The so-called swirl vanes at the entry to the combustor have been eliminated because of difficulties in manufacturing and because kerosene fuel was not intended to be injected by means of individual orifice nozzles. Instead, large-scale primary-zone recirculation using a small number of large air jets has finally led to the successful design of this combustion chamber. When correctly placed, these recirculation holes significantly improved mixing and combustion stability, and later test runs resulted in a spotless combustion chamber.
The use of a radial-inflow turbine required that the hot gases be redirected from an axial reverse flow to a radially inward one. The first-stage nozzle assembly is constructed from a single piece of alloy and includes not only a curved passage (transition piece) to redirect this flow of hot gases but also a set of fixed-geometry radial vanes or nozzles. These vanes are used to accelerate the hot gases and generate the correct velocity triangles at the turbine inlet simultaneously, such that a smooth inward flow through the turbine impeller passages is accomplished at its design point. The fixed-geometry vanes of the assembly have blunt leading edges to allow the hot gases to enter the nozzle passages smoothly from any direction.
Although stability limits for sustained combustion with respect to fuel-to-air ratio
are wide, these limits are narrower for
ignition. Therefore, good ignition characteristics depend greatly on the fuel-injector design and the achievable atomization quality. A well-atomized or evaporated fuel (preferably close to the stoichiometric fuel-to-air ratio) is required in the primary zone. The air temperature and pressure at the inlet to the combustion chamber at low rotational speeds are almost ambient. This is especially detrimental to ignition performance because of the large ignition heat loss and the very poor fuel atomization quality that can actually be achieved.High-quality fuel atomization using plain orifice nozzles was investigated, but small high-performance orifice nozzles require high fuel pressures and heavy onboard boost pumps to achieve fine fuel sprays. These orifices are also difficult to manufacture. If multiple injectors were used, flow division would be the next problem to deal with.
The most important disadvantage is that such orifices tend to create large spray-cone angles, and a finer fuel spray will mean a larger cone angle. The radiation caused by the impingement of burning droplets onto the liner's inner wall is high, especially in a very small combustion chamber. Fuel control, however, is more accurate, and the engine response time caused by changes in fuel flow is very short. Fuel pre-evaporation provided the best solution. The design consists of a fuel pre-evaporator manifold located within the combustion chamber. Because the fuel and the combustion chamber are cold at start-up, the fuel cannot be pre-evaporated unless it is preheated to its high evaporation temperatures just before ignition, which is cumbersome. The solution was to use a natural-gas fuel for start-up and ignition.
Not surprisingly, the use of natural gas has proven to be an excellent choice. At minimum idle speed, the transfer to kerosene fuel is initiated through the same manifold, using synchronized valves. The already hot gases in the combustion chamber then preheat the fuel in the manifold to a high evaporation level before it enters the combustion chamber.Fuel flow is controlled by a variable-speed electric-motor-driven miniature gear pump.
Ignition is accomplished by a 5-kilovolt electronic-condenser discharge spark-ignition unit developed in-house. The position of the spark igniter strongly influenced the start-up and ignition sequence of the engine. When the designers changed the igniter's position relative to the evaporator manifold and the recirculation or cooling holes, the rotational speed at which ignition could actually occur changed from 10,000 rpm to as low as 3,000 rpm.
A trial-and-error process was required to find the right position. Ignition at reasonably high rotational speeds is preferred to prevent flashback caused by low pressures and flows throughout the engine. Ignition speed was finally set to approximately 10,000 rpm.
The rotor-bearing module, the heaviest part of the engine, is the most important module in terms of its construction. Made entirely of stainless steel, it isalong with the compressor diffuser and turbine nozzle assemblybasically the mainframe of the engine. Its alignment with all other components is of prime importance because tip clearances of both compressor and turbine are controlled by this alignment.
The reverse-flow design allows the hot-section components to expand freely to the rear of the engine while the compressor-diffuser/bearing-module combination acts as a fixed pivot. The bearing within this module has a double-overhung arrangement comprising two special ball bearings that can withstand the axial loadings at high rotational speeds. These bearings require preloading in the axial direction. This has been accomplished with a spring.
Both bearings are lubricated and cooled with oil fed from the externally mounted tank by a tube through the compressor casing. This lube-oil system is of the total-loss type. Only a little oil is required during normal operation. A closed-loop system would be too heavy and too difficult to engineer, and would require an overly complex sealing arrangement. The lube-oil droplet flow is controlled by a small orifice of proprietary size, and the oil is fed to both bearings using compressor discharge air. Before start-up, when no pressure is available, a few drops are fed manually. All oil is finally lost in the exhaust duct where it is entrained in the hot gases.
The biggest problems of all were rotor balancing and vibration damping. A two-plane dynamic balancing machine with G1.5 balancing capabilities, available in-house, worked well.
Handling and Operation
To date, the engine has been started up, run normally, and shut down only on a specially designed test bed, on which rotational speed, exhaust temperature, compressor discharge pressure, thrust (static), bearing and oil temperature, and fuel pressure and temperature can be measured.
Oil-droplet flow to the bearing module can be monitored with a sight glass; rotational speed is measured with an optical sensor and static thrust is measured with a load cell. The small lube-oil tank is mounted on the compressor casing of the engine. The test bed also contains the miniature gear-type liquid-fuel pump, the condenser-discharge spark-ignition unit for the igniter, and the synchronized control valves for both natural-gas fuel and kerosene.
The engine is started on natural gas, and the start-up sequence can be done safely by one person. Compressed air from a small electric-motor-driven compressor unit is fed to the start-up nozzle, located on the compressor casing. First the rotor is accelerated with compressed air to 10,000 rpm.
At this point the ignition is turned on, the natural-gas-fuel valve is opened, and light-up occurs, accelerating the engine further to its minimum idle speed of approximately 65,000 rpm. The air starter unit is shut down shortly afterward. The thrust then produced is about 1.1 pounds, while the exhaust temperature is 896°F and the compressor pressure ratio is approximately 1.3.
Changeover from natural-gas fuel to kerosene is accomplished using the same fuel manifold system, by simultaneously closing the gas-fuel valve and opening the liquid-fuel valve. During changeover, the turbojet engine runs on a mixture of gas fuel and kerosene for a few seconds. This method has been very successful. Further acceleration to the engine's maximum continuous speed of 100,000 rpm can then be initiated.
The engine can be shut down simply by closing the liquid-fuel valve after the engine has been decelerated back to its minimum idle speed and maintained at this speed for approximately 1 minute. After rotor runout, the temperature within the bearing module increases slightly from 176°F to 189°F because of radiation effects from the turbine and combustor, as the bearing module is no longer cooled by the relatively cold air passing it during normal engine operation. The team concluded that the only way to increase the specific thrust of the turbojet was to increase the turbine-inlet temperature, which would require costly new materials.
Sensitivity toward component pressure losses and efficiencies increases significantly at very low compressor pressure ratios in combination with high turbine-inlet temperatures. As a result, the off-design performance simulations could be far from reality.
Component interaction on this scale is also extreme. Slight changes in the geometry of the diffuser, for example, may have a tremendous negative effect on the behavior of other downstream components as well as on overall engine behavior. The desirable characteristics of an off-the-shelf item of already excellent design can be lost by the use of incorrectly matched components.
The following information was extracted from the August 1997 issue
of the technical magazine,
Permission was granted to ALLSTAR by the magazine to use the preceding materials.
For more information about this article, please contact John G. Falcioni, the magazine's editor-in-chief
Send all comments to firstname.lastname@example.org
© 1995-20167ALLSTAR Network. All rights reserved worldwide.
|Funded in part by||Used with permission from Mechanical Engineering|
Updated: March 12, 2004