Comparison between Thrust and Horse-Power

Because the turbo-jet engine is rated in thrust and the turbo-propeller engine in s.h.p., no direct comparison between the two can be made without a power conversion factor. However, since the turbo-propeller engine receives its thrust mainly from the propeller, a comparison can be made by converting the horse-power developed by the engine to thrust or the thrust developed by the turbo-jet engine to t.h.p.; that is, by converting work to force or force to work. For this purpose, it is necessary to take into account the speed of the aircraft.

The t.h.p. is expressed as FV/(550 ft. per sec.) where

Since 1 horsepower is equal to 550 ft.lb. per sec. and 550 ft. per sec. is equivalent to 375 miles per hour, it can be seen from the above formula that 1 lb. of thrust equals 1 t.h.p. at 375 m.p.h. It is also common to quote the speed in knots (nautical miles per hour); one knot is equal to 1.1515 m.p.h. or 1 pound of thrust is equal to one t.h.p. at 325 knots.

Thus if a turbo-jet engine produces 5000 lb. of net thrust at an aircraft speed of 600 m.p.h. the t.h.p would be

$\left(5000\times 600\right)/375=8000$

However, if the same thrust was being produced by a turbo-propeller engine with a propeller efficiency of 55% at the same flight speed of 600 m.p.h., then the t.h.p. would be

$\left(8000\times 100\right)/55=14,545$

Thus at 600 m.p.h. 1 lb. of thrust is the equivalent of about 3 t.h.p.

Effect of Forward Speed

Since reference will be made to “ram ratio” and Mach number, these terms are defined as follows:

From the thrust equation, it is apparent that if the jet velocity remains constant, independent of aircraft speed, then as the aircraft speed increases the thrust would decrease in direct proportion. However, due to the “ram ratio” effect from the aircraft forward speed, extra air is taken into the engine so that the mass airflow and also the jet velocity increase with aircraft speed. The effect of this tends to offset the extra intake momentum drag due to the forward speed so that the resultant loss of net thrust is partially recovered as the aircraft speed increases. A typical curve illustrating this point is shown in Figure 10–3. Obviously, the “ram ratio” effect, or the return obtained in terms of pressure rise at entry to the compressor in exchange for the unavoidable intake drag, is of considerable importance to the turbo-jet engine, especially at high speeds. Above speeds of Mach 1.0, as a result of the formation of shock waves at the air intake, this rate of pressure rise will rapidly decrease unless a suitably designed air intake is provided; an efficient air intake is necessary to obtain maximum benefit from the ram ratio effect.

As aircraft speeds increase into the supersonic region, the ram air temperature rises rapidly consistent with the basic gas laws. This temperature rise affects the compressor delivery air temperature proportionately and, in consequence, to maintain the required thrust, the engine must be subjected to higher turbine entry temperatures. Since the maximum permissible turbine entry temperature is determined by the temperature limitations of the turbine assembly, the choice of turbine materials and the design of blades and stators to permit cooling are very important.

With an increase in forward speed, the increased mass airflow due to the “ram ratio” effect must be matched by the fuel flow and the result is an increase in fuel consumption. Because the net thrust tends to decrease with forward speed the end result is an increase in specific fuel consumption (s.f.c.), as shown by the curves for a typical turbo-jet engine in Figure 10–4.

At high forward speeds at low altitudes the “ram ratio” effect causes very high stresses on the engine and, to prevent overstressing, the fuel flow is automatically reduced to limit the engine speed and airflow.

The effect of forward speed on a typical turbo-propeller engine is shown by the trend curves in Figure 10–5. Although net jet thrust decreases, s.h.p. increases due to the “ram ratio” effect of increased mass flow and matching fuel flow. Because it is standard practice to express the s.f.c. of a turbo-propeller engine relative to s.h.p., an improved s.f.c. is exhibited. However, this does not provide a true comparison with the curves shown in Figure 10–4, for a typical turbo-jet engine, as s.h.p. is absorbed by the propeller and converted into thrust and, irrespective of an increase in s.h.p., propeller efficiency and therefore net thrust deteriorates at high subsonic forward speeds. In consequence, the turbo-propeller engine s.f.c. relative to net thrust would, in general comparison with the turbo-jet engine, show an improvement at low forward speeds but a rapid deterioration at high speeds.

Effect of Afterburning on Engine Thrust

At takeoff conditions, the momentum drag of the airflow through the engine is negligible, so that the gross thrust can be considered to be equal to the net thrust. If afterburning is selected, an increase in takeoff thrust in the order of 30% is possible with the pure jet engine and considerably more with the bypass engine. This augmentation of basic thrust is of greater advantage for certain specific operating requirements.

Under flight conditions, however, this advantage is even greater, since the momentum drag is the same with or without afterburning and, due to the ram effect, better utilization is made of every pound of air flowing through the engine. The following example, illustrates why afterburning thrust improves under flight conditions.

Assuming an aircraft speed of 600 m.p.h. (880 ft. per sec.), then momentum drag is: 880/32.2 = 27.5 (approximately). This means that every pound of air per second flowing through the engine and accelerated up to the speed of the aircraft causes a drag of about 27.5 lb.

Suppose each pound of air passed through the engine gives a gross thrust of 77.5 lb. Then the net thrust given by the engine per lb. of air per second is 77.5 – 27.5 = 50 lb.

When afterburning is selected, assuming the 30% increase in static thrust given previously, the gross thrust will be 1.3 × 77.5 100.75 lb. Thus, under flight condition of 600 m.p.h., the net thrust per pound of air per second will be 100.75 – 27.5 = 73.25 lb. Therefore, the ratio of net thrust due to afterburning is 73.25/50 = 1.465. In other words a 30% increase in thrust under static conditions becomes a 46.5% increase in thrust at 600 m.p.h.

This larger increase in thrust is invaluable for obtaining higher speeds and higher altitude performances. The total and specific fuel consumptions are high, but not unduly so for such an increase in performance.

The limit to the obtainable thrust is determined by the afterburning temperature and the remaining usable oxygen in the exhaust gas stream. Because no previous combustion heating takes place in the duct of a bypass engine, these engines with their large residual oxygen surplus are particularly suited to afterburning and static thrust increases of up to 70% are obtainable. At high forward speeds several times this amount are achieved.

Effect of Altitude

With increasing altitude the ambient air pressure and temperature are reduced. This affects the engine in two interrelated ways:

Graphs showing the typical effect of altitude on thrust, s.h.p., and fuel consumption are illustrated in Figures 10–6 and 10–7.

Effect of Temperature

On a cold day the density of the air increases so that the mass of air entering the compressor for a given engine speed is greater, hence the thrust or s.h.p. is higher. The denser air does, however, increase the power required to drive the compressor or compressors: thus the engine will require more fuel to maintain the same engine speed or will run at a reduced engine speed if no increase in fuel is available.

On a hot day the density of the air decreases, thus reducing the mass of air entering the compressor and, consequently, the thrust of the engine for a given rpm. Because less power will be required to drive the compressor, the fuel control system reduces the fuel flow to maintain a constant engine rotational speed or turbine entry temperature, as appropriate; however, because of the decrease in air density, the thrust will be lower. At a temperature of 45°C, depending on the type of engine, a thrust loss of up to 20% may be experienced. This means that some sort of thrust augmentation, such as water injection, may be required.

The fuel control system controls the fuel flow so that the maximum fuel supply is held practically constant at low air temperature conditions, whereupon the engine speed falls but, because of the increased mass airflow as a result of the increase in air density, the thrust remains the same. For example, the combined acceleration and speed control fuel system schedules fuel flow to maintain a constant engine rpm, hence thrust increases as air temperature decreases until, at a predetermined compressor delivery pressure, the fuel flow is automatically controlled to maintain a constant compressor delivery pressure and, therefore, thrust. Figure 10–8 illustrates this for a twin-spool engine where the controlled engine rpm is high-pressure compressor speed and the compressor delivery pressure is expressed as P₃. It will also be apparent from this graph that the low-pressure compressor speed is always less than its limiting maximum and that the difference in the two speeds is reduced by a decrease in ambient air temperature. To prevent the L.P. compressor overspeeding, fuel flow is also controlled by an L.P. governor which, in this case, takes a passive role.

The pressure ratio control fuel system schedules fuel flow to maintain a constant engine pressure ratio and, therefore, thrust below a predetermined ambient air temperature. Above this temperature the fuel flow is automatically controlled to prevent turbine entry temperature limitations from being exceeded, thus resulting in reduced thrust and, overall, similar curve characteristics to those shown in Figure 10–8. In the instance of a triple-spool engine the pressure ratio is expressed as P₄/P₁, i.e., H.P. compressor delivery pressure/engine inlet pressure.

Types of Engine Test Bed

This section describes the configuration of engine test beds, and provides design to guidelines to ensure that:

Many other parameters are measured besides those mentioned above. For these, accuracy depends on good measurement practice rather than overall test bed design.

Outdoor Sea Level Thrust Test Bed

This is illustrated in Figure 10–10, and consists basically of an open air stand supporting an engine and providing thrust measurements. The effects of cross wind on entry conditions are negated by a large mesh screen fitted around the engine inlet. The immediate test bed area is free of obstructions to the airflow, to ensure the validity of the thrust and airflow readings. This is the most definitive thrust test bed, as for indoor test beds the thrust and airflow measurements are corrupted by the flow field generated by the side walls. Outdoor test beds are sited in remote areas, to minimize the environmental disturbance of the noise produced. Because of the resultant logistic difficulties and the impact of adverse weather conditions, indoor testing is preferred in most countries, with measurements calibrated versus outdoor facilities.

Indoor Sea Level Thrust Test Bed

Here a similar engine arrangement to that of Figure 10–10 is mounted indoors, as shown in Figure 10–11. The air flowpath to the engine is crucial, as flow disturbance must be minimized. The engine nozzle efflux enters a detuner, which exhausts hot gases and provides sound attenuation.

For a given engine the measured thrust may be up to 10% less than the value that would be recorded on an outdoor test bed. This is due to unrepresentative static pressure forces acting on the engine and cradle, caused by the velocity of air within the cell passing around the engine. This air is entrained into the detuner by the ejector effect of the engine jet; it prevents hot gas reingestion and also cools the detuner. Furthermore, if the engine final nozzle is unchoked the test bed configuration can cause a rematch due to the local static pressure distribution at the nozzle exit. An indoor test bed gives useful all-weather availability, but unless the test bed is purely functional it should be calibrated against an outdoor test bed, as discussed later, to determine the effects of the static pressure field and rematch.

Key guidelines for designing an indoor test bed are presented below. These minimize the measured thrust deficiency described above, and also prevent extreme flow fields such as vortices from entering the airmeter. Owing to the complex flows within a test bed, some adjustability and subsequent experimentation may be required.

Indoor Sea Level Jet Bed or Turboshaft Gas Generator Tests

This test bed is employed when a turboshaft engine gas generator is tested with a final nozzle and exhaust detuner instead of its power turbine. The theoretical exhaust gas power is calculated based on gas generator exit flow, pressure, and temperature, and is the power that would be available from a power turbine of 100% efficiency with no duct pressure losses. Such testing is necessary when the power turbine either remains with the installation, or is supplied by another company. As this applies mainly to land-based engines, an altitude chamber would not normally be required. Key points are as follows:

Indoor Sea Level Shaft Power Bed

This is used for turboprop or turboshaft engines, with output power measured directly. Figure 10–12 shows the key features of such test beds. The main differences from an indoor thrust bed are as follows.

On shaft power test beds some device must absorb the engine output power, providing suitable characteristics of load versus speed. There are several possibilities:

Altitude Test Facility (ATF)

Thrust or shaft power test beds may be housed within an altitude test facility (ATF), which reproduces the inlet conditions resulting from altitude and flight Mach number. Figure 10–13 shows the key features of an ATF. Unlike a sea level test bed the plant must provide a continuous airflow even without the engine operating, to maintain reduced pressure and temperature.

Figure 10–14 illustrates two possible layouts of ATF plant. To simulate both ambient conditions and flight Mach number engine inlet total pressure and temperature must be controlled to the ram (free stream total) values for the altitude and Mach number. Also, the static pressure at the nozzle exit plane must be set to that of the test altitude. These parameters are mostly sub-ambient at altitude, hence common features of the various types of ATF are substantial pressure reduction, chilling and drying capabilities, and recompression of discharge air back to ambient. Accurate measurement of thrust is complex.

The other possibility for testing at flight conditions is a flying test bed as described below, where the engine is mounted on an aircraft. The main advantages of the ATF are:

Flying Test Bed

A flying test bed is also often used for major aeroengine programs. Typically a four engined aircraft is modified to mount a single, new development engine at one berth. Compared with an ATF the advantages are:

However, as mentioned there are no direct measurements of thrust and mass flow. These must be calculated as follows.

Measurements and Instrumentation

Engine tests use differing amounts and sophistication of instrumentation, depending on their purpose. Many development tests require detailed performance investigation, hence pressures and temperatures are measured at virtually every station, as well as power or thrust, shaft speeds, fuel, and airflow, etc. At the other extreme, for production pass-off or endurance testing only a minimum of measurements are taken beyond those of the production control system, such as ambient conditions, power or thrust level, and fuel flow.

This section provides background to the instrumentation used for each possible measured parameter. Indicative accuracies and coverage requirements are provided, along with a summary of how the instruments work.

Engine testing is very expensive; hence to ensure good quality data are obtained the importance of the following cannot be overemphasized.

For all of the above an understanding of likely accuracy levels is required.

Pressures

Pressures are measured for a number of reasons:

Figure 10–15 illustrates the main elements of a typical pressure measuring system. Local small holes called tappings allow the engine gas stream pressure to reach a measuring device outside the engine, via fine capillary tubes of 1–2 mm diameter. Generally at least three circumferential locations are used at a station, to increase coverage and to allow error detection by comparison of readings. A leaking line usually reads low, though thought should be given as to whether lines pass through higher pressure regions. An alternative is to gang the lines from several tappings together into a manifold and then read the manifold pressure as a pneumatic average. Though this method is relatively inexpensive a leak in any one pressure line is difficult to detect. The measuring devices used are described here.

Manometers

For pressures below around 2 bar, older test beds have used water or mercury manometers, where the height of a column of liquid in a glass tube is read visually. Small corrections are applied for the temperature of the liquid column. For a well-designed system accuracy is around 0.25%. As automatic data recording has become more prevalent manometers have virtually disappeared.

Transducers

Modern test beds use a transducer, where a pressure difference causes movement of a diaphragm, which is converted to an electrical signal. The other side of the diaphragm may be at ambient pressure or a vacuum, giving gauge and absolute readings respectively. The diaphragm movement is converted to a voltage, which is read by the data logging system. For many transducers the conversion uses an energizing voltage and a resistive straingauge on the diaphragm. Alternatives are piezo-electric, which generate their own voltage, or inductive. Calibration curves relate the electrical signal to pressure levels, and are obtained by either a dead weight tester that applies a known force and hence air pressure, or comparison with other calibrated transducers. Transducer designs are optimized for various pressure ranges; they should ensure operation is in the most linear part of the range, typically 10–90% of full scale. Transducer temperature should be controlled as this affects the straingauge resistance; even compensating circuitry does not fully eliminate the effect. Accuracies quoted herein are for a controlled transducer temperature.

Steady state, typical accuracies are around 0.1% of full scale for the basic transducer; however, a good overall accuracy is 0.5% for engine pressures. This figure allows for calibration drift, hysteresis, engine stability and pressure profiles, transducer nonlinearity, and drift in the voltage supply. Cell static pressure is less prone to engine effects, hence an accuracy of 0.25% is obtainable.

For steady state testing, many pressure tappings are read in turn via scanivalves, where rotating valvery connects a transducer to each tapping in turn. For best accuracy the transducer is kept in a temperature controlled environment, and known reference pressures are also read one or more times per scan giving continual, automatic update of the transducer calibration. With best practice overall accuracy can be around 0.25% as most transducer error effects are eliminated, leaving those due to engine pressure profiles and stability during the time taken to complete a scan. Such a system is unsuitable for transient use due to intermittent reading and volume packing of the pressure lines. The latter effect means around 4 minutes' stabilization is required before an accurate pressure reading can be obtained.

Comments on pressure measurements at the various engine stations are presented below.

Ambient Pressure—Barometers

Barometers are used to measure ambient pressure, with an accuracy of around 0.1%. They fall into two main categories:

Test Cell Static Pressure

This is required for all engines where the intake is inside a test bed, and is measured in at least two places of low cell velocity. Usual locations are on the side walls in the plane of the nozzle exit, at the same height as engine centerline. The instrumentation comprises the open end of a 1–2 mm diameter capillary tube surrounded by a perforated “pepper pot,” which removes the effects of incident velocity. The tube is then connected to either a transducer or water manometer.

Engine Static Pressures

If the axis of a tapping is perpendicular to the flow direction then it will read static pressure, as no dynamic head will be recovered. Wall tappings give a more accurate reading than side or rearward facing tappings on immersed probes, as the presence of the probe disturbs the flow. Static pressure tappings may be used in place of total pressure readings if a calibration has already been obtained versus the total pressure reading. Such calibrations become tenuous, however, if there is swirl angle variation, as this changes the local Mach number.

Engine Total Pressures

To achieve full recovery of the stream dynamic head, and hence measure total pressure, the pressure tapping is mounted in a probe that points its axis towards the direction of gas flow. If the flow angle varies by more than ±5°, a Kiel head should be employed, as illustrated in Figure 10–15. This uses a chamfered entry to recover effectively the stream dynamic head for incidences of up to ±25°. Above gas temperatures of around 1300 K total pressure probes are not normally viable as they will require cooling and hence become so large that associated pressure drops are prohibitive.

Coverage requirements depend on how well understood the pressure uniformity is at a station, and should be agreed with the relevant component designer. Many radial and circumferential locations may be addressed via either multiple heads on vane leading edges, or several multihead rakes inserted into the gas stream. The heads are often placed on centers of equal flow area to assist in data averaging. Calibration of rig versus engine instrumentation standards may also be undertaken. Coverage in the cold end (compressors) should be at least three off rakes with one to five heads, depending on engine size. If large radial or circumferential non-uniformities are likely then more coverage may be employed, such as downstream of an aeroengine fan where up to 10 heads are used. For the hot end there is often significant swirl hence greater coverage may be needed. Around six rakes, or preferably instrumented vane leading edges, are suggested with two to five heads depending on engine size.

Total pressure rakes and wall static tappings may be used in combination, set in the same plane. Static tappings complete definition of the total pressure profile, as static and total pressures are equal at the walls. In addition, if the static pressure is reasonably uniform, which requires low swirl angle, the difference between total and static pressure indicates flow velocity.

Placement of total pressure rakes should consider obvious sources of error such as wakes downstream of struts. Good practice is to derive calibration factors between fitted instrumentation and that giving fuller coverage as listed above.

Differential Pressures

Normally pressure is measured as the difference between the gas stream and ambient, and an absolute pressure level obtained by addition of ambient pressure to the gauge reading. To read a pressure difference between two points, both sides of a transducer may be connected to tappings at the engine stations in question. This allows use of a more precise, lower range transducer, and avoids large inaccuracies due to the subtraction of similar numbers. One disadvantage is that recalibrating such a transducer is not possible without disconnecting the instrumentation, unlike for scanivalve systems.

Transient Pressures

For transient testing, dedicated pressure transducers, designed to optimize transient response, are required for each tapping. This provides a continuous reading, unlike a scanivalve system. The transducers are mounted local to the engine to minimize line volumes, and hence allow fast response to pressure changes; they may be water jacketed to enhance thermal stability. Line length limits are around 5 m for ordinary handling and 1 m for faster transients such as fuel spiking. In the former case a divertor valve may be employed to allow the same tapping to be read by the steady state scanivalve; for the shorter line length space does not permit this. Typical scan rates range from 10 to 500 scans per second. Absolute accuracies are lower for dedicated transient transducers than for a scanivalve system, around 1.5% of full range, and the transducers are more subject to drift.

A calibration curve should be run at the start of each working day to provide a comparison with the steady state instrumentation. In addition a transient maneuver should be performed to check for lag due to divertor valve faults.

Dynamic Measurements of Pressures

Dynamic measurements of pressure address high frequency pressure perturbations, rather than “dynamic pressure” as in the “velocity head.” These measurements are employed to detect flow instabilities such as rotating stall or rumble that can occur in the compression and combustion systems. The accuracy in determining amplitude is low, around 10% of range, as the design is optimized for response and the thermal environment is normally uncontrolled. Probes such as the kistler (resistive) and kulite (piezo) varieties are utilized, which incorporate pressure transducers—the low volume allows response to the very high frequencies involved. The kistler probe is more vulnerable to vibration but can tolerate temperatures up to 350°C, which is 80°C higher than the kulite. The signal may be available in the control room on an oscilloscope and is normally recorded in analog form on tape. Later examination of amplitudes and frequencies helps pinpoint the cause or at least onset of instability. High frequency phenomena may also be shown qualitatively by noise on ordinary transient pressure signals.

Temperatures

Measurement of temperatures provides the following information on engine and component performance.

Temperature measurement is complex. It is vital to follow the good design and working practices outlined herein, otherwise significant inaccuracies may result.

Temperature readings more closely reflect total rather than static conditions, as a rake in the gas stream brings the gas to rest on its surface. In fact it is impossible to measure purely static temperature. The fraction of the dynamic temperature recovered is termed the recovery factor. As for pressure measurement, temperature rakes employ Kiel heads where swirl angle variation is greater than 5°, to ensure the dynamic temperature is recovered over a wide range of incident flow angles. Little error is normally incurred if rake recovery factors are taken as a constant value of 0.94, assuming well-designed probes.

Figure 10–16 illustrates the main elements of a typical temperature measuring system. The following sections describe the instruments used.

Resistance Bulb Thermometers (RBT)

Here temperature is measured via changes in the resistance of a heated material. Platinum is frequently used, hence the common alternative expression PRT (platinum resistance thermometer). In theory resistance thermometers are suitable for temperatures up to around 1000 K, and may give a high accuracy of potentially around 0.1 K if carefully calibrated. They are, however, comparatively delicate, and rarely used actually within an engine; the most common uses are for air inlet temperature and to measure reference temperature in thermocouple systems described later in this section.

Snakes

Snakes are resistance thermometers many meters long, sometimes used to measure average inlet temperature, and may for example be strung out over the inlet debris guard or splitter. One disadvantage is that the large physical size makes accurate calibration impossible, hence a preferred alternative is multiple resistance bulbs. Indicative overall accuracy for a snake is 1–2 K.

Thermocouples

If two dissimilar metal wires are connected at a junction, and the loose ends maintained at some reference temperature, a voltage is generated dependent on the temperature difference between the junction and the reference. Typically the junction is a welded head of up to 1.1 times the wire diameter. Thermocouples are less accurate than RBTs, but more robust. The loose ends’ temperature is maintained by either a UTR (uniform temperature reference) box, whose own temperature is measured by RBTs, or an Icell (ice cell). Single pieces of wire should be used between the hot and cold ends, otherwise measurement uncertainties increase by around 2 K per extra junction. With single pieces batch wire calibration is applicable, where a calibration is obtained of a number of thermocouples made from a particular batch of cable. Providing the results agree, this calibration is applicable to other thermocouples made from that same batch.

Thermocouples that employ different wire materials produce different voltage curves. Standard curves are defined for different thermocouple types, and different classes of thermocouple lie within different tolerances of these curves. Class 1 thermocouples will lie within the greater of 1.5 K or 0.4% of the standard curve, and class 2 within 2.5 K or 0.75%. At higher temperatures drift and hysteresis together contribute a further 3 K inaccuracy, despite heat treatment to improve thermoelectric stability. In addition typically 1 K error will be contributed by both circuitry and conduction effects around the hot junction. Combining all these effects by root sum squaring gives overall system errors at 1000 K of 5.6 K and 8.2 K for class 1 and 2 respectively. Type K thermocouples use chromel-alumel (Ni–Cr/Ni–A1) wire. Type N thermocouples (Nicrosil/Nisil, Ni–Cr–Si/Ni–Si) were introduced around 1990 to provide longer life and improved stability over type K. Overall system errors at 1000 K are reduced to 4.2 K and 7.6 K.

In sitting thermocouples radiation from adjacent surfaces must be avoided, otherwise the temperature measured is not that of the gas stream. For locations where radiation may be severe, shielded thermocouples are employed, which use up to four concentric thin tubes surrounding the thermocouple bead.

The application of the above devices to measuring temperatures at key engine stations is described below.

Air Inlet Temperature

Recommended coverage is at least three RBTs mounted on the intake debris guard, and more if non-uniform inlet temperature profiles are suspected. The test bed layout should be adjusted to ensure that the difference between the readings is less than l K, otherwise it is difficult to be sure that the true average temperature is being measured, and the temperature profile may fundamentally affect engine performance. Snakes or even thermocouples may also be used; however, this results in lower accuracy as described earlier.

Cold End (Compressor) Temperatures

Usually thermocouples are employed, and accuracy is as described earlier. Coverage should be at least three points circumferentially, with rakes having one to five heads radially depending on engine size and the expected radial temperature profile. The heads are usually placed on centers of equal area to provide uniform coverage and assist in data averaging. An aeroengine fan is a special case; due to the temperature profiles and relatively low temperature levels rakes with up to 10 heads are employed.

Hot End (Turbine) Temperatures

Temperature measurement is significantly more difficult for turbines, for two main reasons:

For both reasons the temperature at combustor exit cannot be measured, and measurements are rarely possible at exit from any first HP turbine stage. For measurement stations further downstream, the minimum coverage required is at least eight locations circumferentially, and three to five thermocouple heads radially, depending on engine size. Patternation introduces a further error beyond the thermocouple inaccuracies described above. Rather than using centers of equal area, head placement is often biased towards the walls, where the temperature gradient is steepest.

Transient Temperatures

One further important thermocouple property is response time. Physically large thermocouples take time to respond to temperature changes, due to thermal inertia, and are unsuitable for transient development testing. Response is governed mainly by the time constant of the junction itself. For development testing physically small junctions are employed, mounted to minimize conduction and radiation.

For control system instruments, large robust “production” devices are required. Here the time constant of the thermocouple can be allowed for in the control algorithms, or heat transfer increased by increasing the flow past the thermocouple junction. Where the pressure ratio to ambient exceeds around 1.2, a sonic probe is used as shown in Figure 10–16. A small flow is extracted from the gas path through a venturi surrounding the thermocouple bead and then diffuses before being dumped overboard. Otherwise aspiration is employed, where higher pressure air is injected to draw flow past the bead by an ejector effect.

Liquid Fuel Energy Flow

Measuring fuel flow fulfills two essential purposes:

In all cases the method involves the measurement of volumetric fuel flow, conversion to fuel mass flow based on the actual fuel density, and finally deriving energy flow using the fuel heating value (FHV). To obtain the density and FHV periodic laboratory analysis of fuel samples is required. This must be done at least once per fuel batch delivery, and even each day during key performance testing. Fuel energy flow is calculated from the volumetric flow, FHV and specific gravity.

Volumetric Flow

For liquid fuels three main instruments may be used:

Because fuel flow is so vitally important at least two of these devices should be used in series. Figure 10–17 shows the main elements of a fuel flow measuring system, including typical installation requirements.

Density

To measure density of a fuel sample a hydrometer is used, where a graduated scale on a float is read visually. The “units” are usually relative density or specific gravity, which is simply the ratio of the actual density (mass per unit volume) to a standard value of 1000 kg/m³. The accuracy of measurement is around 0.1%.

Actual density varies with temperature, hence temperature readings are required for both the laboratory sample test and the fuel supply to the engine. For the latter, RTBs are placed in the fuel pipes, ideally ten pipe diameters away from the flow meter to prevent possible flow disturbances corrupting the flow measurement.

Fuel Heating Value

FHV is also obtained via laboratory analysis of samples. The main methods are:

Accuracy of the above methods is around 0.1%. If no laboratory analysis is available, an FHV for kerosene only may be calculated from the density.

Transient Fuel Flow Measurement

For measuring volumetric fuel flow transiently only a turbine flow meter is suitable as its speed gives an instantaneous flow rate. The usable range is around 10–100% of full scale, so different instruments are required for starting and fuel spiking, with ranges of around 20% and 300% of that for normal running. Fuel density and lower heating value are simply the same numbers as for steady state. Both a bulkmeter and flow bottle are unsuitable as they rely on fuel flow being steady.

One backup measurement that may be particularly useful transiently is via the combustion fuel injector flow number. Given the upstream fuel pressure measurement, flow is simply the flow number times the square root of the pressure drop. This is useful during fuel spiking where fuel flow changes so quickly that even a turbine flow meter reading may lag due to inertia. The method may be checked against the main instruments based on the previous steady state position.

Gas Fuel Energy Flow

The measurement process is similar to that for liquid fuels, with fuel energy flow evaluated from volumetric flow, density, and lower heating value.

Volumetric Flow

A bulkmeter cannot be used to measure gas fuel flow as it is compressible, i.e., its density changes with pressure, and the positive displacement principle upon which a bulkmeter relies is only applicable to incompressible liquid fuels. One of two techniques is employed as outlined below.

As for liquid fuel, a turbine flow meter may be utilized, with similar installation rules and measurement accuracies. It is not necessary to calibrate for viscosity.

An orifice plate may also be utilized, and an accuracy of ±1% can be achieved. The installation criteria include a flow straightener at the measuring section inlet, and typically upstream and downstream straight lengths of pipe of 10 and 5 diameters respectively. However, orifice plates have a number of significant disadvantages relative to a turbine flow meter:

Calibration may utilize a series of sonic nozzles in parallel. This is an extremely sophisticated system and would normally be used in a laboratory rather than an engine test bed.

Density

The fact that gas fuel is compressible makes evaluation of its density complicated. The most accurate method of determining gas fuel density is from its pressure, temperature, gas constant, and compressibility. Static pressure should be the average of values measured around five diameters upstream and downstream of the volumetric flow measurement device; this upstream location is purposely after the flow straightener pressure losses. Temperature should be the average of measurements taken upstream of the flow straightener and approximately five diameters downstream of the volumetric measuring device using RTBs. For the gas constant and compressibility term (z) it is essential to analyze a sample using a gas chromatograph and then utilize formulae. The last measure is necessary as in fact very few gas supplies do not have some variability in the gas composition, and relatively small changes can significantly affect the gas constant and compressibility. Failure to do this can lead to inaccuracies of up to ±5%.

Taking all of the above measures will lead to accuracies for gas fuel mass flow of around ±1%. This is the best accuracy achievable, and is not vastly changed whether combining these effects is via root sum square (for 95% confidence) or arithmetic addition (for 100% confidence). The main inaccuracy is in volumetric flow, at best around ±0.75%, while pressure is only accurate to around ±0.25–±0.5%. Temperature has little effect, being only ±0.1 K if an RBT is used, while the compressibility and gas constant should be accurate if the formulae are used correctly.

Fuel Heating Value

The FHV can be evaluated to within 0.1% from the gas chromatograph output.

Air Mass Flow

Measurement of engine inlet mass flow is vital for various reasons:

Total and static pressures are read in a duct of known area, usually where contraction occurs and increases the dynamic head, along with total temperature. Measurements may also be taken at other stations, and where direct measurement is not possible calculations are performed based on flow continuity using design air system assumptions.

Airmeters

As shown in Figure 10–18, two similar instruments exist for measuring engine inlet flow:

The term airmeter is applied to either device. Both require the use of a discharge coefficient (CD), which is the ratio of effective flow area to geometric area. This is less than unity because first, velocity is less than the ideal value, due to skin friction, i.e., total pressure is lost between the upstream measured value and the throat; and second, the flow does not fill the whole of the throat due to boundary layer growth. If the engine face is very close, such as with a flare, bow wave effects may add to this. The discharge coefficient may he determined by several possible methods:

At least 6–12 throat static tappings should be used and read individually to allow error detection; the number depends on size. For a flare, total pressure should be that in the cell. For a venturi there may be some pressure loss upstream of the throat, hence total pressure must also be measured or the loss allowed for by calibration. Total temperature measurement ideally uses RBTs. Overall, a typical value for accuracy is around 0.5% of airflow, providing the tappings remain clean.

Turbofan Core Flow

For a turbofan the mass flow actually entering the core is of greatest interest, and may be determined in various ways.

Air System

Once a value has been established for first compressor inlet mass flow, the air system design data for cooling and leakage flows are initially assumed to obtain values for mass flows at other stations. The methods discussed under “Test Bed Analysis,” or direct measurements of individual air system flows, are used to check this assumption.

Other Measurements

For all engine stations, including entry, measurement of total and static pressures, and total temperature may provide an indicative mass flow measurement. This may be calibrated against some true measurement, either directly or using air system data depending on the location.

Transient Air Mass Flow

For transient testing a flow probe is often fitted, especially at HP compressor entry. This reads a difference between total and static pressures via a dedicated differential transducer, with short length balanced pipework to remove aerodynamic noise. The mass flow is calculated using an effective area and calibration versus steady state data based on inlet flow readings from the airmeter. If the airmeter is a close coupled flare then it may be read transiently, using dedicated transducers with 1 m or 5 m line lengths.

Injected Steam Flow

Steam may be injected into an engine to reduce emissions and increase power. The supplied steam flow must be measured directly, as least for engine development. This is normally accomplished using a vortex shedding flow meter. Here a bluff body within the flow vibrates due to shedding of wakes (just as does a telephone wire in the wind). The frequency of vibration is roughly proportional to the volumetric flow. Such a device is calibrated to determine the exact dependency, versus a choked (critical) nozzle.

Thrust

For aeroengines, measuring thrust is essential for three main reasons:

Figures 10–10, 10–11 and 10–13 showed the basic layout of sea level and altitude thrust test beds. An engine is mounted in a cradle and restrained axially via load cells, which measure the axial force required. These contain springs and are often preloaded and mounted in opposition to maintain stiffness. Standard practice is to calibrate before and after a test run, often by hanging weights, and use the mean as valid during the test. An indicative accuracy level for the load cell is within 0.25% reading; to convert this to thrust requires test bed calibration, which is discussed later. Basically, for indoor test beds the impact of the static pressure field on the engine relative to that in an infinite atmosphere must also be evaluated. This leads to an overall thrust accuracy of around 1%.

In an altitude chamber additional effects contribute to measured thrust as shown in Figure 10–13. The inlet ducting incorporates a slip joint to prevent axial force transmission to the engine; this joint has a seal area larger than the upstream duct. The actual gross thrust produced by the engine is the measured load cell force plus inlet momentum drag and pressure loads acting on the duct and seal areas. Typical accuracy is around 1.5%. For both test beds, measurement of mass flow is also important, to corroborate predictions of momentum drag at flight conditions.

Shaft Speeds

Shaft speed measurements provide valuable component information:

A phonic wheel and pick up are normally used. This method is highly accurate and reliable, and hence available even as a production measurement. The passing of teeth cut in a wheel, integral with the shaft, is sensed by an electromagnetic coil. The pulses generated are converted to a shaft speed, knowing the number of pulses per revolution. Two main methods may be used for this:

Engine Output Shaft Torque and Power

These are highly interrelated as power is simply the product of torque and rotational speed.

A torquemeter forms part of a load carrying shaft, and measures torque by sensing the relative rotation of two ends of a length of shaft. Shaft torque must be measured at an appropriate point along the output shafting, as shown in Figure 10–12, to exclude any losses that are not part of the engine supply. Figure 10–19 illustrates one method where phonic wheels, which also measure speed, are attached to both the shaft and an outer unloaded tube. By this means the teeth are in the same plane, and changes in the waveform picked up indicate angular displacement, and hence torque. Overall torquemeter accuracy is around 0.5 to 1%, depending on engine stability. An alternative method utilizes strain gauges to measure shaft twist; however, such devices are delicate and have not been widely successful.

A hydraulic dynamometer may also measure torque. Input torque is transmitted to the casing by the water and any bearing friction; the casing is freely mounted externally on bearings to allow torque measurement via a load arm and a weighing system.

Humidity

Humidity must be measured so that test data can be referred to either dry or ISO conditions, for evaluation on a defined basis. There are three main methods of humidity measurement:

For all these methods an indicative accuracy level is the greater of 5% relative humidity or 0.2% specific humidity.

Geometric Parameters

Variable stator vanes are common on axial compressors, and recuperated engines often have variable turbine nozzles. To measure vane rotation directly, around three off RVDTs should be used. An RVDT is a “rotary variable displacement transformer,” in which the relative movement of electric coils changes the mutual inductance. Accuracy is good, around 0.5°, assuming all vanes are held at the same angle. If alternatively the linear displacement of the actuator moving the ring is measured, via an LVDT, then other effects are relevant, such as ring tolerances and any slack in the total mechanism. The accuracy of measured vane position is then entirely dependent on the system design.

Turbine throat areas are subject to manufacturing variation and are normally measured geometrically on build, with a repeatability of around 0.1%; method (and hence definition) adjustments may give apparent “step” changes, however. The conversion factor between area and capacity is derived from the turbine aerodynamic design data.

For thrust engines, propelling nozzle areas are required. Defining and measuring certain areas may not be trivial, for example, on a turbofan the bypass nozzle plane may contain struts, or if a mixer is employed its area is required.

Test Bed Calibration

Engine performance measured on indoor test beds will differ from that in an “infinite atmosphere.” Test bed calibration enables data from an indoor test bed to be adjusted to reflect infinite atmosphere conditions. In addition, engines of a given mark may be tested on more than one test bed, each with a different configuration and hence different calibration factors.

For shaft power engines where air is ducted from ambient as per typical installations, and the measurement made at duct entry, the test bed configuration does not normally need to be calibrated for. All that remains is the usual requirement for all instrumentation to be functioning satisfactorily.

Test Bed Definitions

Test bed approval is a formal process, involving regulation by the appropriate airworthiness authority for aircraft engines, and high customer involvement for other engine types. The following basic definitions apply, though other similar terms are also used:

A cross calibration exercise normally involves comparing results obtained running the same engine on two test beds, without any transportation of test bed slave equipment. The benefits of cross calibration are dubious for engine configurations where transportation between test beds would require significant rebuild.

Calibration of Gold Standard Test Bed

For thrust engines, the gold standard bed for the first engine of a type is declared by calibration versus an outdoor test bed, i.e., infinite atmosphere. For shaft power engines direct calibration of the mass flow measurement is required.

Steps in Test Bed Cross Calibration

The object of cross calibrating a test bed is to reproduce the performance measurements obtained on the gold standard test bed, after referral to standard conditions, such as ISA sea level static.

The steps involved are:

Test Bed Audit

Test bed calibration exercises are expensive due to the engine running involved. Once a test bed has been calibrated the process is rarely repeated unless bed configuration is changed, or there is an inexplicable change in production engine performance. However, to maintain test bed approved status it should be audited, typically biannually. Important features to review in an audit are as follows:

Specific Performance Tests

By the end of development the achieved performance standard must be demonstrated. Contractual implications mean both the customer and the engine manufacturer will be keenly interested. During development the standard is achieved via specific performance tests, which have the following aims:

A typical test is a performance curve consisting of around 5 to 10 stabilized points. These cover the range from the minimum power or thrust level important for the application (usually idle) to maximum rating. Stabilization times before taking readings range from around 4–20 minutes, the latter if large heat exchangers are utilized. Shorter times do not allow scanivalve pressure readings to stabilize.

For aircraft engines similar performance tests are conducted in an altitude test facility or flying test bed, to simulate the conditions encountered at cruise accurately. Owing to the time spent, altitude cruise is the most important condition for SFC.

Windmilling

For aircraft engines drag, auxiliary power capability and combustor inlet conditions are measured over a range of Mach numbers and altitudes in either an altitude test facility or flying test bed. Windmill testing is not normally conducted for other engine types.

Support of Engine Component Development

During engine development various tests are necessary where evaluating performance is not the goal but defining its level is vital to meet the primary test objective. For interest, examples of such tests are described next.

Ingestion Tests for Aircraft Engines

Ingestion events cannot be avoided for aircraft thrust engines since adequate filtration is impractical. Testing is a regulatory requirement to ensure satisfactory engine behavior, as outlined below.

Thermal Paint

This records the metal temperatures of key “hot end” components such as combustors and turbine blades. Special paint is applied, which changes to various colors depending on the temperature reached; several paint grades are available to suit different temperature ranges. The engine is first run at low power to heat the carcass to ensure representative seal clearances, hence cooling flows, and then held for between 3 and 10 minutes at the desired high power condition. The paint normally takes 3 minutes to cure; longer times are utilized when engine stabilization times are long, for example with a recuperated engine, though this must balance the risk of paint degradation. The paint is examined on engine strip after the test, or occasionally via borescope inspection. The indicated component surface temperatures are then related to the temperatures of the gas path and cooling air, assessed from the performance measurements and analysis.

Component Temperatures

In addition to thermal paint tests, key component areas such as disc faces may be fitted with thermocouples to measure metal temperatures. This allows the assessment of many operating conditions during a build, and provides safety monitoring during high power running or, in case of cooling flow inadequacy, in early development.

Air System

Pressure and temperature readings may be taken in key air system chambers, to validate design data for cooling flows, cooling temperatures, and pressure margins.

Straingauging

To assess the susceptibility of gas path components to resonant vibration, straingauges are cemented in key locations. These consist of fine wires embedded in a mounting plate; vibration causes the wires to stretch, which changes their electrical resistance. The test normally consists of slow acceleration and deceleration (around 4 minutes each) with various strain gauge selections being recorded. The frequency and amplitude of the measured signal characterizes blade response, identifying potentially harmful resonances. Testing explores the full range of mechanical and referred rotational speeds that the engine type will encounter. Mechanical speed determines the frequency of excitation forces, referred speed produces flow regimes that may cause excitation, and pressure level determines amplitude.

Achieving the highest speeds is often a challenge with limited test hardware, though in-service ambient variability may be addressed via an altitude chamber. Very high cycle temperatures may be reached, often requiring builds with special sizes of turbine and propelling nozzles to provide alleviation.

Bearing Loads

For this test special strain gauged equipment is built in to measure the axial loads on shaft bearings. The engine is run over a representative operating range to allow assessment of bearing lives; stabilization periods may be long (20 minutes and more) as air system pressure levels depend on seal clearances and hence disc growths.

Emissions

Exhaust gases are sampled, usually via a cruciform rake in the exhaust, and analyzed using a spectrometer for levels of pollutants such as unburned hydrocarbons, carbon monoxide, and oxides of nitrogen. Such tests are increasingly important for contractual purposes, as legislation becomes more stringent.

Endurance and Type Testing

All engines complete some form of arduous endurance test before service entry. This may take the following forms:

Handling

The maneuvers tested exceed the most severe to be met in normal service, in order to give high confidence that engines will behave satisfactorily. Examples include:

During these maneuvers key parameters are measured that allow definition of speed versus time, transient working line excursions, weak extinction margins, etc. A suitable scan rate is 10–50 scans per second. For aircraft engines most work is done on sea level test beds; however, some altitude chamber or flying test bed testing is essential to verify effects at the precise conditions to be encountered.

Surge Line Measurement

Surge lines are effected by phenomena such as engine structural loads creating asymmetric tip clearances, inlet distortion, or heat soakage within the HP compressor which lowers the surge line after a decel. It may be necessary to measure actual surge lines in an engine, or at least define a surge-free region, rather than relying on rig or predicted data. Key parameters are measured so that the actual trajectory to surge can be defined. Methods of initiating surge are as follows.

For an HP compressor the most usual method is fuel spiking. Here the working line is raised by momentarily injecting excess fuel, between 100% and 400% extra over around 200 ms. A suitable scan rate is 100–500 scans per second. Either a slave rig is used, or the standard control system reprogrammed.

For LP compressors or fans the most usual method is to build a development engine with a reduced capacity LP turbine or bypass nozzle respectively, to raise the working line. Bleed extraction is used to enable stable operation at high power; the engine is then slowly decelerated until it surges. This process is repeated with various levels of bleed to map out the surge line.

Other less common techniques include in bleeding air downstream of the subject compressor, for example on small engines, or if applicable utilizing heat exchanger bypass capabilities. The first surge on a build may increase compressor tip clearance, which raises operating temperatures and lowers the surge line. These effects should be quantified by further testing, and compressors examined by borescope to check for damage. The deterioration from subsequent surges is much less.

Controller and Engine Operability Tests

Transient performance and control are inseparable. The initial control strategy and algorithms are further developed during extensive sea level and altitude testing. One particular test of interest is surge recovery, where the controller must ensure that should an engine surge for any reason in service it will recover safely. This is demonstrated during development testing via a surge induced by one of the methods described.

Starting Tests

Specific tests addressing the light up region are:

Between light up and idle, tests are required to determine the upper and lower limits of acceptable fueling. Too much fuel results in compressor stall, which is affected by the bleed and whirl levels. Too little fuel results in hang, where at low speed some limitation is reached, such as over temperature, surge or rundown. This may be explored by reducing fuel flow at various starter power levels.

For industrial, marine and automotive engines start testing is usually conducted at the prevailing ambient conditions for the sea level test bed. Problems that may arise in service on cold or hot days are then dealt with as they arise. For aircraft engines cold and hot day start tests must be included in the development program. Methods include taking the engine to a cold or hot climatic region, chilling the engine in a cold room, or using large electrical heaters at engine inlet. Furthermore altitude restart capability compliant with the windmill and starter assist envelopes must be proven by testing in an altitude test facility or flying test bed.

A further test for all engine types is a hot restart, performed shortly after shutdown with the carcass still at high temperature. As well as the effect on seal clearances, heat in the gas path lowers surge lines and raises working lines.

Engine Failure Investigation

Engine mechanical failures during development invariably have a transient dimension. In addition to any transient performance data logging, key mechanical parameters such as vibration, speeds, and pressures must be recorded either analog or at least at 1000 scans per second to cater for such an event. To avoid storing immense amounts of data it may be overwritten, say every 30 minutes.

In the event of a failure this log may be interrogated to distinguish cause and effects. One example is to determine whether a surge occurred before (and hence caused) or after (and hence resulted from) mechanical damage to a compressor. If the surge came first, the amplitude of noise or vibration with a frequency corresponding to the spool speed will increase after the step decrease in compressor delivery pressure and step increase in broad band vibration caused by the surge.

Production Pass Off Test

It is common practice that each engine about to be delivered to a customer, whether new production, overhauled or repair, undergoes a pass off test. The inevitable variation in manufacturing and build dimensions, even within permitted tolerances, results in variation of component and overall engine performance. A typical test sequence is as follows:

For practicality an exception is made for heavyweight gas turbines, however.

Acceptance Criteria

There are various limiting levels that engine performance parameters must meet for pass off. These usually include demonstration of guaranteed power or thrust without exceeding set temperature and speed limits, and achieving guaranteed SFC at this power or thrust. The acceptance criteria are usually stated at some standard conditions, to which measured data is referred. Engines may initially fail their pass off test due to faults with instrumentation or less usually with the build.

In addition to performance parameters certain mechanical integrity issues are also addressed, such as fuel leaks, vibration, oil consumption, measured temperature spread, etc.

Engine Performance Trends

Changes in component manufacture or build practice may cause increasing numbers of engines to fail the pass off test. Identifying these changes early among the general scatter is challenging. Though some special instrumentation may be fitted for the pass off test, more modern practice is to rely on that of the production control system. Instrumentation is therefore often too sparse or inaccurate to suit the conventional analysis methods. Though these may be employed additional methods are necessary, which emphasize comparison with other engines. Figure 10–20 shows an example where pass off data from a production run are plotted for LP turbine inlet temperature:

Steps in Test Data Analysis

Test data requires considerable processing before engineering conclusions may be drawn:

These steps do not always occur in this ideal order. Measurement errors are not necessarily easily detectable, for example they may be systematic and apply to all readings at one station. The analysis process may take a long time to “finish” as new facts alter the understanding. Furthermore, certain methods involving cycle matching combine steps 4 and 5.

Initial Error Detection and Data Averaging

It is essential before using measured data to carry out error detection, for which some techniques are listed below.

Rake data require averaging to form a single thermodynamic mean value at each instrumented station, as represented by the performance model. Such an average is usually either area weighted or mass weighted. Mass weighting is the ideal, but as it requires an accurate knowledge of the static pressure profile, which is difficult when swirl is present, area weighting is more usual. Either way, for the average to be meaningful any faulty heads must be detected and eliminated.

Test Bed Analysis (TBA) Calculations

These compute engine and component performance levels, e.g., thrust, SFC, isentropic efficiencies, directly from the measurements. Calculations proceed sequentially through the engine, evaluating almost all parameters at each station. This requires assumptions for bleeds, pressure losses, and power offtakes. Turbine exit conditions are assessed based on work, measured in the compressors and any power offtakes.

TBA calculations may be performed either at the tested engine condition or at standard conditions such as ISA sea level or ISO. In the latter case the measurements are referred using theta (θ), delta (δ), and the appropriate non-dimensional groups. To account for variation in gas properties due to changing temperature and fuel air ratio, theta is raised to exponents other than the standard 1.0 or 0.5, known as theta exponents. The actual values are obtained by running the engine synthesis model. For illustration, Table 10–2 presents a typical range.

At this stage, instrumentation error detection is again essential, using the derived parameters. Errors may be detected based on considerations such as the following:

One special case regarding TBA calculations is a turbofan, where the core mass flow is not measured directly but must be calculated. Several methods exist, and for convenience the two most commonly used are termed methods “A” and “B” herein. Method “A” uses a core heat balance (see Figure 10–21) and method “B” employs a knowledge of HP turbine capacity. An alternative approach is a transient mass flow probe, which gives indicative rather than absolutely accurate flow measurements.

Uncertainty Analysis

For key parameters produced by the TBA calculations it is essential to evaluate the likely error band resulting from the measurement uncertainty. The first step is to evaluate the potential errors in a calculated parameter resulting from that in each contributory measurement individually; the second step is to combine the effects of these errors. The most common practice is a root-sum-square addition, for 95% confidence. For 99% confidence, simple arithmetic addition is used. The process is also called error band analysis.

Comparison of Test Results with Expectation

It is usual to produce a pre-test matching model of an engine going to test. For development tests the effects of measured throat areas and any other specific hardware are normally included, whereas for a production engine the expectation—and indeed the requirement—will remain constant. It may be simply expressed via charts relating the measured parameters to thrust or power at standard conditions.

Valid comparison of the test results with expectation requires that both are based on identical conditions of inlet pressure and temperature, installation losses, etc. In addition, for contractual purposes engine performance is normally required at some standard conditions, as stated. To achieve these aims there are two main approaches:

For both approaches the model and test parameters are plotted versus some power setting parameter. For referred data it must be in a non-dimensional form.

Evaluation of Engine Component Performance

Various methods exist for attempting to explain anomalies, which will be due to either unidentified measurement error or engine component performance differences from prediction. Test data analysis is notoriously difficult, as once dealing with “real” data the possibilities for error are vast. The various methods are described below, and should preferably be used in combination.

Traditional Methods

Engineering judgment and logical thought have long been used for test data analysis. Knowledge of what could be suspect is invaluable, and may be based on known potential causes of component deficiency, the state of engine hardware on strip, other knowledge of the test such as what checks were performed for leakage, etc. Comparison of predicted and TBA component efficiencies and capacities gives one view of component performance levels, though being vulnerable to measurement error and rematch effects it is not definitive. This traditional approach together with some trial and error may be used to determine changes to the component performance in the pre-test prediction model which allow it to reproduce all test measurements deemed valid across the power range.

Achieving alignment between a revised model and the measurements is assisted by tables of exchange rates.

Synthesis exchange rates show the change in engine performance parameters if a given component efficiency is in error by l%. Analysis exchange rates show the error in a parameter from the TBA calculations due to a 1% error in each contributory measurement. The advantages of traditional methods are that the result of such an analysis carries high credibility, as the process is easily understood, and no additional setting up is required. The disadvantage is that unlike the more automated methods described below no immediate answer is available.

Cycle Matching

Here an extended performance matching model is run at the tested conditions. It is used both to compare test data with expectation and to explain the results by changes in component performance. For test analysis the matching is extended with measurement values as additional matching constraints, and component scale factors as additional matching guesses. The model thereby changes component performance to replicate the input measurement values.

The main advantage is that analysis is performed automatically, and at the tested condition; this is more accurate than referring test data. In addition engine station data may be produced with however few measurements being taken, parameters at other engine stations being calculated based on the predicted component performance. There are some disadvantages, however. The method can only ascribe differences to component factors that are active in the matching, and the number of these is limited to the number of measurements; if some other component is at fault the answers are wrong. The matching iteration may not always work, or be ill-conditioned such that measurement errors even down to the repeatability level produce large, mutually cancelling changes in several components. Backup calculations are therefore required in online applications.

Matching methods have been successfully used by manufacturers of large turbofans and by some, though not all, airliner manufacturers. Use for industrial engines is not yet widespread.

Methods Based on Probability

Here a computer program “adjusts” both the measurement values and component performance levels so that calculated and measured station parameters match, using tables of exchange rates. By definition there is an infinite number of solutions. That given is the most likely solution, based on input values of the probability of error in components and measurements. Faced with no formal solution this gives a likely answer, and may produce useful suggestions. The solution is not unique, however, and carries little or no weight without engineering substantiation.

Transient Data

Here considerable processing is required to align instrument readings with the equivalent steady state parameters, given the reduced instrument coverage and transducer drift. Once this has been achieved, transient traces are compared with transient performance model predictions, and the model is then adjusted to achieve alignment. The model must already have a good alignment with engine steady state parameters, and adjustments are now made mainly in the modeling of heat soakage for transient operation, where real thermal masses and heat transfer coefficients are not normally known to great accuracy anyway.

Changes in Gas Properties

The presence of water vapor changes the values of these gas properties, which can have a significant effect on the thermodynamic processes throughout the engine. Both CP and R increase significantly, whilst gamma diminishes more slowly; Figure 10–22 shows the variation of these parameters versus water vapor concentration. The molecular weight of water is lower than that of dry air. Formulae 10.1–10.3 at the end of this section give the variation of CP, R, and gamma with water content.

Effects on Component Performance

Engine components behave non-dimensionally and performance is mapped via various forms of parameter groups. Where the full dimensionless groups are used these maps are unique even with varying amounts of water vapor present. Quasi-dimensionless groups are also often used, where the gas properties are omitted and components mapped based on the gas properties of dry air. This form of the groups is useful as it relates the most easily observable engine parameters of flow, speed, pressure and temperature. Changes occur in these quasi-dimensionless maps due to the changes in gas properties described above. The use of such maps when water vapor is present is described in this section.

For a compressor, the presence of water vapor lowers W√T/P at constant N/√T and PR.

This is because the dimensionless group for speed, N/√(γRT), is reduced relative to dry air due to the increased R which outweighs the small reduction in γ. At this lower dimensionless speed the compressor passes a lower dimensionless flow, W√(RT)/P√γ. In addition, at this lower dimensionless flow the quasi-dimensionless flow W√T/P is further reduced, again by the increased R.

For a turbine the presence of water vapor produces little or no change in W√T/P at constant N/√T and ΔH/T. This is because, unlike a compressor, reduced speed normally either raises capacity or has no effect. One further effect of humidity is to increase the specific heat in the turbines, reducing the temperature drop for a given expansion ratio. The impact of these effects on engine performance depends on where through the engine the water is introduced.

Description

Humidity is water vapor naturally present in the atmosphere:

Effects on Engine Performance

Table 10–3 shows qualitatively the effect of humidity on leading engine parameters. Figure 10–23 presents generic exchange rates that may be utilized for first-order accuracy to predict the impact of humidity on key performance parameters. These have been derived by considering all the full non-dimensional parameter groups to be constant at a given operating condition. The impact of humidity is greater than is immediately apparent, as all parameters must be factored. For example, at a specific humidity of 0.04 (100% relative humidity on a 36°C day) shaft power increases by about 1% and LP turbine inlet temperature falls by about 1%. Hence if governing to fixed LP turbine inlet temperature the increase in power would be around three to four times this. The precise impact depends on the engine cycle.

Figure 10–23 may also be used for a first-order correction of engine test data from any humidity level to that for ISO conditions:

Description

Water injection has been employed for both aero- and industrial engines to improve performance, utilizing one of two injection points along the gas path:

Both practices are now largely superseded by other technologies. These are increased temperature capabilities and higher bypass ratios for aeroengines, and dry low emissions combustion technology for industrial engines, where low pollutant levels are achieved without the need for water injection.

Compressor Entry Injection—Aeroengines

The takeoff condition, especially for hot days or high airfields, was particularly important for older engines with zero or low bypass ratio. Water was injected using spray nozzles, and prevented from freezing by adding methanol. The power or thrust gain primarily came from the lower inlet temperature and intercooling within the compressor rather than the additional mass flow of water. If the amount of water injected exceeded the amount that could evaporate within the compressor, further power boost was achieved by similar effects as produced by water injection into the combustor. The weight of tanks, pumps, etc., was a disadvantage, as was the need to supply purified water.

Compressor Entry Injection—Industrial Engines

Industrial engines operating in hot, arid conditions often employ an evaporative cooler upstream of the intake. This consists of a screen with water flowing over and the air flowing through it. The engine experiences a colder, more humid day, and hence can run to higher power and/or lower operating temperatures. Compared with aeroengines water consumption is lower, as only that which immediately evaporates is used. Such equipment would clearly have been impractical for aeroengines.

Combustor Injection—Industrial Engines

Water injection into the combustor has been employed to reduce NO_x emissions, by lowering the temperature peaks in the combustor primary zone and hence the tendency for atmospheric nitrogen to dissociate. Side effects include increased power, but often increased emissions of carbon monoxide (CO) due to lower temperatures, especially for low-pressure ratios and short residence times. In addition combustion becomes more uneven, which may induce mechanical problems due to temperature patternation and noise. Water to fuel ratios (WFR) between 1 and 2 have been employed; as combustor temperatures are very high saturation is not a problem. The large amount of water and associated plant is a substantial disadvantage, the water being purified and pumped to the injection points via pipe work and manifolding. In addition, burners must incorporate water nozzles, and appropriate control capability is essential. As the water supply pipes etc. can be protected from low temperature no methanol is employed. SFC is worse due to the significant amount of heat required to evaporate the water.

Steam injection, described below, supplanted water injection for installations where steam plant could be used as it overcame the disadvantages associated with the chilling effect of liquid water. As stated, more modern industrial, and even marine, engines are moving instead to DLE combustion.

Effects on Engine Performance

For water injection, pumping power is not an engine performance concern, as the liquid water is incompressible and hence the power requirement is small. Logistic and weight issues of the associated plant are significant, however.

Compressor Entry

Table 10–4 shows the main effects on engine performance of water injection at compressor entry. As outlined, compressor inlet temperature is reduced by water evaporation, producing effectively a colder day and 100% relative humidity. The maximum overall performance boost at fixed SOT on a hot day is:

The SFC improvement is because in non-dimensional terms the engine is suffering less from part load SFC deterioration. By coincidence the methanol prevents any increase in fuel requirement to maintain SOT (though unless liquid water persisted through to the combustor this effect would anyway be small). Saturation of the inlet air determines the feasible lowering of the inlet temperature, hence the initial humidity level has a strong limiting effect.

Figure 10–24 shows the effect on the compressor working lines, which are similar for both LP and HP compressors. At fixed mechanical speed reduced inlet temperature increases referred speed, causing operation to move further up the same compressor working line. At fixed SOT compressor referred speed increases further, again along the same working line.

Combustor

Table 10–5 lists the effects on key engine parameters of water injection into the combustor. Typically at fixed SOT an engine with a combustor design not optimized for emissions would achieve 85% NO_x reduction but with up to a three-fold CO increase. For a 1:1 WFR performance effects are:

The increase in power output is due both to the flow in the turbines exceeding that in the compressors, and the CP of the flow in the turbines being increased by the water. The latent heat of evaporation of water requires additional fuel flow, hence the worse SFC. At fixed SOT the compressor speeds are increased along with engine inlet mass flow.

Figure 10–24 shows that in both cases the HP compressor working line is higher as the HP turbine must pass the injected water vapor flow, raising compressor pressure ratio at a referred speed to maintain the required HP turbine capacity. Because the temperature drop in the turbines is reduced, different levels of power boost are achieved depending on which turbine inlet temperature is considered constant.

Unlike steam injection, described below, water injection into turbines downstream of the H turbine is not beneficial. The large heat absorption in evaporation drastically reduces that turbine’s inlet temperature, and hence also its achievable specific power. The difference for water injection into the combustor is that extra fuel is added simultaneously, which offsets the temperature reduction due to the water. To add extra fuel to offset the temperature reduction for injection into other turbines would require impractically high temperatures in the HP turbine, or the complexity of additional combustors between the turbines.

Compressor Entry

Water injection at entry to the compressor raises few issues as it has no impact on compressor running lines, only resulting in a higher referred speed for a given SOT. The aim is to increase power while keeping within the same operating limits; compressor referred speed is increased only on occasions where it would otherwise be well below its limit. The normal control strategies of temperature and speed limits remain applicable.

Combustor

Steady state the baseline control strategy is invariably constant water fuel ratio at all power levels, and the rated power governed to the same SOT as the dry engine. However on hot and cold days the compressor mechanical or referred speed limits are reached first and the power boost is limited. In practice very cold days often already have referred speed as the limiting parameter even for the dry engine, and no power boost is possible.

Transiently, however, there are two major consequences of water injection that must be addressed by control strategies:

Specific Engine Design for High Injection Rates into the Combustor

Water injection into the combustor beyond around 2:1 WFR would usually require reselection of turbine capacities, to prevent unacceptable surge margin or overspeed as described above. Such modification is rarely if ever undertaken, since most engines must also operate “dry.” However, both “wet” and “dry” cases may be considered in the design phase. Another item to address is increased power turbine entry temperature (PTET) at fixed SOT, due to the reduction in temperature drop in the turbine(s) driving the compressor(s).

The need for turbine capacity changes with high water injection rates arises because for the same compressor operating points compressor work is fixed, but turbine mass flow and specific heat are higher. For example, the requirements for a two shaft gas generator plus power turbine are illustrated below:

By restagger of turbine aerofoils within casting allowances power boosts of up to 30% may be achieved for 4 WFR. As an example, at 6 WFR to retain acceptable compressor working lines and speeds the engine changes would be approximately:

Description

Steam injection is employed on industrial engines to reduce NO_x emissions and boost power. SFC also improves, as steam is usually raised using the engine exhaust heat, in a heat recovery steam generator (HRSG). The chilling effect of steam in the combustor is vastly less and more uniform than that of liquid water, hence there is little or no increase in CO emissions or combustion noise.

Steam injection into other turbines gives around half the SFC improvement and 1/5 of the power boost relative to the combustor, though it costs less in terms of compressor operability. Current systems have various proprietary names, such as STIG (steam injected gas turbine) or the Cheng cycle. Supplementary firing could also be employed, to increase steam production using a boiler with its own dedicated combustion.

Besides the need for steam plant and injection hardware, one significant issue for steam injection, like water injection, is the substantial need for purified water. Higher flow rates are considered than for water injection, as steam injection improves SFC. Owing to the cost and complexity of cooling towers etc. almost all systems are total loss, with the used steam simply passing up the exhaust stack. For all these reasons steam injection is employed almost solely on industrial engines rather than in mobile applications.

Effects on Engine Performance

Table 10–6 summarizes the effects of steam injection on an engine cycle. Typically an aeroderivative industrial engine at constant SOT with 2:1 steam fuel ratio (SFR) would achieve:

Raising steam at the highest possible temperature, dictated by the engine exhaust temperature and HRSG design, gives the best SFC; this reflects a maximum efficiency of energy recovery from the gas turbine exhaust.

In terms of component rematching the effects of steam and water injection are similar, additional power being available from the turbines due to their increased gas mass flow and specific heat. The HP turbine must pass extra flow, as for water injection, resulting in a higher HP compressor working line as per Figure 10–24. At fixed geometry the highest power boost occurs with the highest SFR that can be accepted within surge margin limits, which is normally not higher than 2.

Operability and Control Philosophy

Comments here almost exactly match those for water injection though the tendency to flame out during a decel is less for steam injection as without the need for evaporation it produces less chilling in the combustor.

Specific Engine Design for High Injection Rates into the Combustor

Almost all the same comments apply as for water injection, except that SFC improves rather than deteriorates. For a full turbine redesign, for 6:1 SFR the engine changes would be:

The one difference relative to water injection is that due to the improved SFC going to high SFR is more attractive; however, as with water injection actually doing this is almost unknown in practice due to the compromise to “dry” performance.

In terms of engine cycle selection there are two other notable effects that impact the ability to produce steam:

Description

Condensation is confined to the “cold,” compressor end of the engine and is due only to ambient humidity. The resultant water simply re-evaporates further along the gas path. Condensation may occur once flow conditions cause the local static temperature to be below the saturation value for the local water vapor partial pressure. This is determined by local static pressure and water vapor concentration; saturation corresponds to 100% relative humidity. The actual rate of condensation is determined by residence time, and by minute particles in the air that act as nuclei for water droplets. (See the discussion on the fundamental properties of the phases of water.)

In the cold end there are two regions where water vapor may condense:

Intake

Intake condensation can cause a significant change in engine performance on a test bed. The apparent loss in power or thrust and increase in SFC at fixed fuel flow may be up to 2% for hot ambient conditions, and nearer 0.5% for temperate climates. The reason for this loss in performance may not be immediately apparent, being due to pressure and temperature changes occurring downstream of test bed inlet pressure and temperature measurements. Were it possible to measure accurately the pressure and temperature actually entering the compressor then there would be negligible impact on engine referred performance relative to that without condensation. In the absence of perfect measurements or correction methods, performance testing should not be undertaken during conditions of high humidity or precipitation.

The effects of condensation are an increase in air inlet temperature and a reduction in total pressure, as well as reduced specific humidity in the initial compressor stages. The increased temperature results from the latent heat release while the pressure loss reflects the momentum change in flow acceleration due to reduced density. Once within the engine the condensed water evaporates shortly downstream.

Any engine on the test bed (or even in service) will experience a real difference in performance as referred to the engine intake lip.

Intercooler

For an intercooler system some condensation control algorithm is often employed to prevent condensation causing blade erosion downstream. The coolest air temperature must be maintained above the calculated saturation value by control of the degree of cooling. For example, for a system with an intermediate transport loop between the cold sink and the engine this may involve a partial bypass of the cold sink. The intercooler exit temperature is thereby increased, with two main consequences for engine operation: