24-09-2014, 04:41 PM
Validation and integration of a rubber engine model into
an MDO environment
[attachment=68482]
Abstract
Multidisciplinary design optimization (MDO) is a technique that has found use in the
field of aerospace engineering for aircraft design. It uses optimization to simultaneously
solve design problems with several disciplines involved. In order to predict aircraft
performance an engine performance simulation model, also called “rubber engine”, is
vital. The goal of this project is to validate and integrate a rubber engine model into an
MDO environment.
A method for computer simulation of gas turbine aero engine performance was created.
GasTurb v11, a commercial gas turbine performance simulation software, was selected
for doing the simulation models. The method was validated by applying it to five
different jet engines of different size, different type and different age. It was shown that
the simulation engine model results are close to the engine manufacturer data in terms of
SFC and net thrust during cruise, maximum climb (MCL) and take off (MTO) thrust
ratings. The cruise, take off and climb SFC was in general predicted within 2% error
when compared to engine manufacturer performance data. The take off and climb net
thrust was in general predicted with less than 5% error. The integration of the rubber
engine model with the MDO framework was started and it was demonstrated that the
model can run within the MDO software. Four different jet engine models have been
prepared for use within the optimization software.
The main conclusion is that GasTurb v11 can be used to make accurate jet engine
performance simulation models and that it is possible to incorporate these models into an
MDO environment.
Introduction
Advanced simulation techniques have become increasingly important in the field of
aerospace engineering for aircraft design. This thesis work deals with incorporating an
existing commercial jet engine performance modeling software into a multidisciplinary
design optimization software framework under development at Bombardier Aerospace.
Background
Multidisciplinary design optimization is used more and more for predicting aircraft
performance in conceptual design of aircraft. In an MDO software environment, multiple
disciplines, e.g. weight estimation, aerodynamics, engine performance, stability and
control, are incorporated and optimized simultaneously. The advantage of this is that it
captures the interactions between the disciplines and has the possibility of finding an
optimum that is superior to the optimum found if each discipline is optimized separately.
Engine performance for a big range of engine operating conditions is vital for predicting
aircraft performance. Traditionally, tables of engine performance called thrust tables were
provided by engine manufacturers. The thrust tables contain thrust, fuel flow and other
engine parameters for different flight conditions and engine thrust settings.
When an aircraft concept with a new engine is studied, the engine manufacturers need to
be contacted in order to get engine performance data. In order to avoid this, there is a
need for a generic engine model, called “rubber engine”, that can produce the
performance parameters normally found in thrust tables.
Engine control system
“Control systems must be designed to prevent aircraft engines from destroying
themselves” [3]. The main functions of an engine control system are to produce the
commanded amount of thrust in a repeatable way for both transient and steady operating
conditions while maintaining stable engine operation within safe mechanical operating
limits over the whole flight envelope. A modern jet engine has an electronic engine
control system that is controlling engine parameters according to PLA, ambient
conditions and thrust management tables. Because of the complexity of an electronic
engine control system, it is difficult to model the thrust rating structure of a modern
engine.
Going back to Equation 2.2, the net thrust consists of a difference in momentum and a
difference in pressure. These properties are difficult to measure in flight and it is
therefore not practical to use the engine net thrust directly in the control system. Instead,
the thrust has to be related to another engine parameter that is measurable [3]. The most
practical measure of the engine thrust is to relate it to the operating line in a fan map.
Both rotational speed and pressure ratios are readily measured in the fan which has
benign environmental conditions compared to the hot parts further back in the engine.
The corrected fan speed correlates very well to engine mass flow and is therefore a useful
parameter for controlling the thrust. A typical control system for a commercial engine
attains the thrust ratings by controlling the corrected fan speed as function of ambient
temperature, altitude and Mach number, with corrections applied to take the bleed air and
power offtakes into account.
Proprietary information from Bombardier and engine
manufacturers
In the role as an intern at Advanced Design at Bombardier, there was proprietary
information from engine manufacturers available. The information comprised of thrust
tables, technical reports, presentations, engine manuals, engine deck software, engine
deck software manuals and the knowledge of the experienced engineers at Bombardier.
An engine deck is an analytic steady state engine performance prediction software that
Bombardier can obtain from engine manufacturers when a new or improved aircraft
design is proposed. The engine deck provides the engine performance in the engine
operating envelope.
The engine deck output variables vary from engine to engine, but in general, the engine
manufacturers tend to hide the design parameters of their engines, especially burner exit
temperature. Some engine decks output details like cooling flows, component efficiencies
and nozzle coefficients but most engine decks hide those parameters for the user.
The engine decks are used for creating thrust tables. A thrust table contain the most
important performance parameters i.e. fuel flow, net thrust, ram drag and T44 for different
altitudes, Mach numbers, ISA values and engine thrust ratings.
The preferred source of information is an engine deck because the user can specify the
flight conditions and installation losses. For some engines, there were no engine decks
available and the information in the thrust tables was used. This can cause problems since
the thrust tables do not always specify the assumptions regarding humidity, FHV and
installation losses.
Design point
The engine design point is where all of the engine components are matched at their
design condition and perform at their design pressure ratio, efficiency and flow. For
subsonic transport type aircraft the design point is typically at TOC or TO [12]. All other
conditions are called off-design. At off-design the component pressure ratio, efficiency
and flow are different from the design point values.
The real engine design point is not known in general, therefore an assumed design point
have to be guessed in order to make a model of a particular engine. The choice of
assumed design point is dictated by the availability and quality of data about engine
parameters.
To make an engine model, the model parameters should be matched to the known engine
parameters at the assumed design point. The main parameters that describe a turbofan are
OPR, burner exit temperature, component performance, FPR, engine total mass flow and
BPR.
The OPR, the burner exit temperature and component efficiencies define the
thermodynamic cycle and the thermodynamic efficiency of the engine core. The total
engine mass flow gives the physical size of the engine together with the BPR which
defines the relative size of the engine core to the overall engine. The BPR and FPR are
important for the engine propulsive efficiency. In addition to the component efficiencies,
there are other component performance parameters such as pressure losses, nozzle
coefficients and shaft mechanical efficiencies. Also cooling flows and installation losses
have to be defined in order to get a complete engine model.
The internal engine cooling is very important for the SFC. Unfortunately, it is very hard
to find reliable data about the internal cooling flows. If better values not are available, a
cooling flow model derived by Shakariyants [13] based on statistical data was used. This
model gives the cooling flow as a function of the SOT and employs a technology level
factor t and a model constant k. The model is seen in Equation 2.9 and Equation 2.10 and
Implementation in GasTurb v11
This section describes and discusses the implementation in GasTurb v11 of the method of
creating an engine model presented in Chapter 2. It can be thought of as a user manual on
how to create an engine model with the purpose of predicting SFC at cruise and net thrust
for the MTO and MCL ratings. The resulting engine model file can be used in GasTurb
with GUI or with the GasTurb dynamic link libraries (DLL’s) without GUI for the MDO
framework.
Description of GasTurb v11
GasTurb is a commercial gas turbine and aero engine modeling software developed by
Kurzke [1]. It lets the user design both the thermodynamic cycle at design and off-design
conditions as well as the geometry and disk stress calculations for turbojet, turbofan and
turboprop engines. It runs under Windows and has a graphical user interface which
makes it user friendly and easy to use. Along with the main program, there are programs
to create and modify turbine and compressor maps and a big collection of non-proprietary
compressor and turbine maps.
Description of DLL’s
At the time of writing, Bombardier had access to one DLL for the mixed flow two spool
geared turbofan with booster type of engine as illustrated by Figure 0.4. This engine type
can also be used to model two spool turbofans without gearbox by setting the gear ratio to
unity or a boosterless engine of either the geared or standard engine type. To model a
boosterless engine, the inner FPR in GasTurb is set to unity and the real inner FPR is
assigned to the booster PR. A DLL for the unmixed flow two spool geared turbofan
(Figure 0.3) is planned to be bought by Bombardier but was not available during the
project. With DLL’s for these two engine types, it is possible to model all jet engine types
considered for Bombardier aircraft. Engine model files for use with GasTurb DLL’s were
created for four different jet engines during this project.
The DLL contains a set of functions for calculating steady state performance in a similar
way as an engine deck. In this implementation a C++ program was written to call the
functions within the DLL. This C++ program reads an input file which specifies the
ambient conditions, power setting and installation losses. Then it calls the DLL with a
GasTurb v11 engine model file and calculates the point defined in the input file. Finally
an output file is created with calculated properties such as net thrust, SFC, pressures and
temperatures at different stations within the engine. Example input and output files are included in Appendix A. For a detailed description of the variables in these files, the
reader is referred to the DLL user manual [15].
Choose engine cycle
The first choice the user must make is the engine cycle. There is a big selection of engine
cycles in GasTurb v11 but only the geared mixed flow turbofan and the geared unmixed
flow turbofan are considered because of the possibility to use these types of engine model
files with the DLL’s.
If the engine that is to be modeled has a centrifugal compressor stage a simplification has
to be made and the centrifugal compressor is modeled as an axial compressor. This
makes no difference at the design point calculation, but at off-design this could be a
source of error because the maps used reflect an axial compressor. In the map collection,
there are maps for axial-centrifugal HPC’s. If such a map is used in the modeling, the
behavior of an axial-centrifugal HPC would be captured. This was not done for the
modeling of the engines with centrifugal compressor stages.
Pick an engine cycle in the drop down list as illustrated in Figure 3.1, chose
“Performance” as the “Scope”, “Design” as the “Calculation Mode”, then press “Run”.
The user is prompted to choose an engine model file, chose the “Demo_xx.XXX” in
the “GasTurb11” folder, where “xx.XXX” is dependent on the type of engine selected.
Starting with the demo file gives a good starting point and reasonable values for all
engine design parameters.
Geometry and weight prediction
GasTurb v11 has the possibility to predict the geometry and weight of a jet engine. It is
based on a model where each component is modeled as a separate part with a material
density, thickness and surface area. Not every component of a real jet engine is included
in the model and therefore a “Net Mass Factor” is used to take this into account. More
than a 100 additional input parameters are needed to accurately model the geometry and
weight of an engine. The additional input includes the following parameters:
Lengths, angles and entry/exit radii for inlet, inlet cone, burner, ducts, bypass,
exhaust.
Strut, guide vane and blade aspect ratio and pitch.
Material properties, material thickness.
Stress margins for disk stress calculations.
GasTurb has default values for all of these parameters and an attempt was made to model
the investigated engines in the geometry and weight tool. The number of compressor and
turbine stages was given as input, remaining parameters were left to their default values.
The investigated engines that use a centrifugal compressor stage were modeled as axial
flow compressors with approximately the same pressure ratio. This is because only axial
flow turbomachinery is considered in the weight and geometry tool. This assumption
could increase the predicted engine length and decrease the predicted engine diameter
due to geometrical differences between an axial and a centrifugal compressor stage.
The results of the investigation are illustrated in Figure 3.4. The max diameter is
underestimated for all engines but not close to the real engine value. Estimated length and
weight is sometimes too high and sometimes too low. The poor result was expected given
the lack of detailed input data.
The inputs that are needed are not readily available at the Advanced Design department.
Furthermore, this tool is more suited for conceptual design of an engine than for engine
performance and weight prediction within an MDO software framework. Therefore it was
decided that other methods should be used to predict the engine weight and size. A
reasonable method would be a statistical model based on existing engines where the
weight is related to the thrust (thrust/weight ratio).
Case studies
The engine modeling method was applied to 5 different engines. The results are
considered sensitive information in terms of actual numbers and no values or scales are
given in text, figures and tables. For all results, the data labeled “GasTurb” is the data
created by the GasTurb engine model. The “Thrust table” or “Engine deck” label
represents engine data obtained from the engine manufacturer’s steady state performance
program. Whenever an error is described as a percentage, it is calculated as seen in
Equation 4.1. A negative error means that the value predicted by the GasTurb model was
lower than the thrust table or engine deck value. A positive error means that the value
predicted by the GasTurb model was higher than the thrust table or engine deck value.
GasTurb engine model setup
An engine deck was available for Engine 1. Therefore the uncertainties regarding certain
input parameters, e.g. intake pressure loss, bleed air and power offtake were removed
because the user is in control of them when running the engine deck. Many engine
parameters such as efficiencies and nozzle coefficients were hidden from the user. E1
does not have a booster stage but was modeled as a geared mixed flow two spool
turbofan with booster in order to use the engine model file with the DLL. Therefore the
inner FPR was set to 1 and the real inner FPR set as the IPC PR.
E1 engine model estimated input parameters regarding the assumed design point:
The assumed design point was selected as uninstalled 43 000 ft, M0.82 at 94% of
maximum cruise rating.
Fan isentropic efficiency was iterated to match T16.
HPC isentropic efficiency was iterated to match T3. This is a bit low, comparing
to Table 3.2.
Iteration for correct inlet mass flow.
Outer FPR calculated from engine deck output.
Inner FPR 1 because engine without booster.
IPC PR – is the inner FPR, value comes from Engine 1 documentation.
HPC PR calculated from engine deck output.
LPT efficiency and exhaust pressure losses adjusted to obtain correct P6 pressure.
Power offtake according to documentation.
No HPC bleed.
Assumed the relative enthalpy for the port 2.8 bleed to get the correct bleed
pressure compared to the assumed design point engine deck output. Turbine cooling flow calculated from W2 – W3. Uncooled LPT NGV and LPT
rotor was assumed, in line with the cooling flow model presented in Figure 2.4.
Thrust reverser leakage flow calculated as total mass flow in – total mass flow
out, Wleakage = W12 + W2 + WF – W8.
Leakage core to bypass is calculated as the difference between core in and out
flow, W2 – (W6 – WF).
The mixer efficiency was unknown and set to 1.
The mixer area was assumed to be equal to the fan frontal area.
The nozzle petal angle was left at its default value of 10º.
Iterate nozzle thrust coefficient to match the net thrust.
Iterate burner exit temperature for correct fuel flow. No data is given about T41,
but the resulting ITT is 24K too high. This could be due to an incorrect cooling
flow model or because of the wrong temperature drop over the HPT. The HPT
turbine temperature drop is decided by the HPT extracted work,
GasTurb engine model setup
The E2 thrust tables contain a big uncertainty in the form of a “fuel flow factor”. The
purpose of the fuel flow factor is to adapt the engine performance program SFC to
measured data from flight tests. Despite efforts to find the fuel flow factor for Engine 2,
no one at Bombardier could provide this information. Therefore, the engine deck data
was assumed to represent the E2 engine and engine deck output data was used to create
the GasTurb model. This engine model was not converted to GasTurb v11 and only exists
as a v10 cycle data file. Therefore all presented results were created by GasTurb v10.
In order to model the E2 unmixed flow engine in GasTurb some assumptions have to be
made because of the differences in the two models. In the engine deck, the scrubbing
drag as illustrated in Figure 2.1 is modeled and given as a separate term. The scrubbing
drag is small but it cannot be modeled in GasTurb. There are terms generating thrust in
addition to the core and bypass nozzle, including exhaust of core and core plug
ventilation air. This thrust is similar in size to the scrubbing drag, so the net effect is
small. These effects are taken into account by calibrating the GasTurb model net thrust to
the engine deck data by iterating the turbine exit duct pressure ratio.
An intake pressure recovery map was made from data found in an engine manual. The
map was adapted to the GasTurb file format. The map was also verified to give correct
values at cruise by comparing to the available engine deck outputs. There were no engine
deck data available for the TO condition so the validity of the map could not be verified
for this flight condition. The map is seen in Figure 4.7. Note that the intake pressure
recovery is essentially constant with regards to Mach number in the interval M0.4 to
Difficulties and future work
From the first day at Bombardier, it was said that GasTurb v11 would soon arrive to the
company. In the meantime, GasTurb v10 was used. Finally, v11 arrived only a few weeks
before the project was supposed to finish. This meant that the planning did not hold and
at the end of the project a lot of time was spent on upgrading engine model files to v11
and doing the validation again with the new software version.
A solution to the problem with saving component map files in GasTurb v11 (see section
3.6.2) was not found. This did not influence the results in terms of SFC or net thrust. It is
recommended that a solution is found if future work involves changes to the component
map files.
One problem with modeling of jet engine performance is the vast number of parameters
needed to make a detailed model. Many parameters, e.g. component efficiencies, duct
pressure losses and internal engine cooling, were estimated. This was overcome by
calibrating the engine model to match the available engine performance data at the
assumed design point. The calibration was done by manually tuning the unknown
parameters. In the future, the calibration could be done automatically within the
optimization software framework under development at Bombardier.
An accurate performance model of a jet engine requires component maps that describe
the compressor and turbine characteristics. Component maps are engine manufacturer
proprietary information and were not available during the project. The solution was to use
generic component maps and calibrate them so that the predicted performance matched
the engine manufacturer performance data. It is suggested that the calibration could be
automated within the optimization software framework. There is also a possibility of
using publicly available component maps that describe the component performance in a
better way than the default GasTurb component maps.
A difficulty with the MTO and MCL rating net thrust prediction is to find reasonable
limits to the control system. The method presented here relies heavily on the availability
of detailed, perhaps proprietary, engine performance data. It would be desirable to
develop a more generic method to find control system limits. This reduces the
dependence on manufacturer data and makes it less of a data matching exercise. One
conceivable example could be to relate the maximum allowed P3 to the maximum OPR of
the engine.
The rubber engine model could not be used in a full scale multidisciplinary design
optimization run because the MDO framework was still under development during this
project. The engine model was however run separately within the MDO framework. The
setup of optimization loops and the details in how the engine model interacts with the
other disciplines of the MDO software is left for future projects.
Conclusions
It is concluded that the goals set at the beginning of the project were fulfilled. The
presented results prove that GasTurb v11 can be used to make accurate jet engine
performance models. In addition, a method on how to create a model was developed and
is presented in this work. Therefore the validation goal is fulfilled.
It was also proved that the GasTurb v11 engine models can be run without GUI and
without user interaction in optimization loops within Isight, the software chosen for
Bombardier’s MDO software framework. Therefore the integration goal is fulfilled.
an MDO environment
[attachment=68482]
Abstract
Multidisciplinary design optimization (MDO) is a technique that has found use in the
field of aerospace engineering for aircraft design. It uses optimization to simultaneously
solve design problems with several disciplines involved. In order to predict aircraft
performance an engine performance simulation model, also called “rubber engine”, is
vital. The goal of this project is to validate and integrate a rubber engine model into an
MDO environment.
A method for computer simulation of gas turbine aero engine performance was created.
GasTurb v11, a commercial gas turbine performance simulation software, was selected
for doing the simulation models. The method was validated by applying it to five
different jet engines of different size, different type and different age. It was shown that
the simulation engine model results are close to the engine manufacturer data in terms of
SFC and net thrust during cruise, maximum climb (MCL) and take off (MTO) thrust
ratings. The cruise, take off and climb SFC was in general predicted within 2% error
when compared to engine manufacturer performance data. The take off and climb net
thrust was in general predicted with less than 5% error. The integration of the rubber
engine model with the MDO framework was started and it was demonstrated that the
model can run within the MDO software. Four different jet engine models have been
prepared for use within the optimization software.
The main conclusion is that GasTurb v11 can be used to make accurate jet engine
performance simulation models and that it is possible to incorporate these models into an
MDO environment.
Introduction
Advanced simulation techniques have become increasingly important in the field of
aerospace engineering for aircraft design. This thesis work deals with incorporating an
existing commercial jet engine performance modeling software into a multidisciplinary
design optimization software framework under development at Bombardier Aerospace.
Background
Multidisciplinary design optimization is used more and more for predicting aircraft
performance in conceptual design of aircraft. In an MDO software environment, multiple
disciplines, e.g. weight estimation, aerodynamics, engine performance, stability and
control, are incorporated and optimized simultaneously. The advantage of this is that it
captures the interactions between the disciplines and has the possibility of finding an
optimum that is superior to the optimum found if each discipline is optimized separately.
Engine performance for a big range of engine operating conditions is vital for predicting
aircraft performance. Traditionally, tables of engine performance called thrust tables were
provided by engine manufacturers. The thrust tables contain thrust, fuel flow and other
engine parameters for different flight conditions and engine thrust settings.
When an aircraft concept with a new engine is studied, the engine manufacturers need to
be contacted in order to get engine performance data. In order to avoid this, there is a
need for a generic engine model, called “rubber engine”, that can produce the
performance parameters normally found in thrust tables.
Engine control system
“Control systems must be designed to prevent aircraft engines from destroying
themselves” [3]. The main functions of an engine control system are to produce the
commanded amount of thrust in a repeatable way for both transient and steady operating
conditions while maintaining stable engine operation within safe mechanical operating
limits over the whole flight envelope. A modern jet engine has an electronic engine
control system that is controlling engine parameters according to PLA, ambient
conditions and thrust management tables. Because of the complexity of an electronic
engine control system, it is difficult to model the thrust rating structure of a modern
engine.
Going back to Equation 2.2, the net thrust consists of a difference in momentum and a
difference in pressure. These properties are difficult to measure in flight and it is
therefore not practical to use the engine net thrust directly in the control system. Instead,
the thrust has to be related to another engine parameter that is measurable [3]. The most
practical measure of the engine thrust is to relate it to the operating line in a fan map.
Both rotational speed and pressure ratios are readily measured in the fan which has
benign environmental conditions compared to the hot parts further back in the engine.
The corrected fan speed correlates very well to engine mass flow and is therefore a useful
parameter for controlling the thrust. A typical control system for a commercial engine
attains the thrust ratings by controlling the corrected fan speed as function of ambient
temperature, altitude and Mach number, with corrections applied to take the bleed air and
power offtakes into account.
Proprietary information from Bombardier and engine
manufacturers
In the role as an intern at Advanced Design at Bombardier, there was proprietary
information from engine manufacturers available. The information comprised of thrust
tables, technical reports, presentations, engine manuals, engine deck software, engine
deck software manuals and the knowledge of the experienced engineers at Bombardier.
An engine deck is an analytic steady state engine performance prediction software that
Bombardier can obtain from engine manufacturers when a new or improved aircraft
design is proposed. The engine deck provides the engine performance in the engine
operating envelope.
The engine deck output variables vary from engine to engine, but in general, the engine
manufacturers tend to hide the design parameters of their engines, especially burner exit
temperature. Some engine decks output details like cooling flows, component efficiencies
and nozzle coefficients but most engine decks hide those parameters for the user.
The engine decks are used for creating thrust tables. A thrust table contain the most
important performance parameters i.e. fuel flow, net thrust, ram drag and T44 for different
altitudes, Mach numbers, ISA values and engine thrust ratings.
The preferred source of information is an engine deck because the user can specify the
flight conditions and installation losses. For some engines, there were no engine decks
available and the information in the thrust tables was used. This can cause problems since
the thrust tables do not always specify the assumptions regarding humidity, FHV and
installation losses.
Design point
The engine design point is where all of the engine components are matched at their
design condition and perform at their design pressure ratio, efficiency and flow. For
subsonic transport type aircraft the design point is typically at TOC or TO [12]. All other
conditions are called off-design. At off-design the component pressure ratio, efficiency
and flow are different from the design point values.
The real engine design point is not known in general, therefore an assumed design point
have to be guessed in order to make a model of a particular engine. The choice of
assumed design point is dictated by the availability and quality of data about engine
parameters.
To make an engine model, the model parameters should be matched to the known engine
parameters at the assumed design point. The main parameters that describe a turbofan are
OPR, burner exit temperature, component performance, FPR, engine total mass flow and
BPR.
The OPR, the burner exit temperature and component efficiencies define the
thermodynamic cycle and the thermodynamic efficiency of the engine core. The total
engine mass flow gives the physical size of the engine together with the BPR which
defines the relative size of the engine core to the overall engine. The BPR and FPR are
important for the engine propulsive efficiency. In addition to the component efficiencies,
there are other component performance parameters such as pressure losses, nozzle
coefficients and shaft mechanical efficiencies. Also cooling flows and installation losses
have to be defined in order to get a complete engine model.
The internal engine cooling is very important for the SFC. Unfortunately, it is very hard
to find reliable data about the internal cooling flows. If better values not are available, a
cooling flow model derived by Shakariyants [13] based on statistical data was used. This
model gives the cooling flow as a function of the SOT and employs a technology level
factor t and a model constant k. The model is seen in Equation 2.9 and Equation 2.10 and
Implementation in GasTurb v11
This section describes and discusses the implementation in GasTurb v11 of the method of
creating an engine model presented in Chapter 2. It can be thought of as a user manual on
how to create an engine model with the purpose of predicting SFC at cruise and net thrust
for the MTO and MCL ratings. The resulting engine model file can be used in GasTurb
with GUI or with the GasTurb dynamic link libraries (DLL’s) without GUI for the MDO
framework.
Description of GasTurb v11
GasTurb is a commercial gas turbine and aero engine modeling software developed by
Kurzke [1]. It lets the user design both the thermodynamic cycle at design and off-design
conditions as well as the geometry and disk stress calculations for turbojet, turbofan and
turboprop engines. It runs under Windows and has a graphical user interface which
makes it user friendly and easy to use. Along with the main program, there are programs
to create and modify turbine and compressor maps and a big collection of non-proprietary
compressor and turbine maps.
Description of DLL’s
At the time of writing, Bombardier had access to one DLL for the mixed flow two spool
geared turbofan with booster type of engine as illustrated by Figure 0.4. This engine type
can also be used to model two spool turbofans without gearbox by setting the gear ratio to
unity or a boosterless engine of either the geared or standard engine type. To model a
boosterless engine, the inner FPR in GasTurb is set to unity and the real inner FPR is
assigned to the booster PR. A DLL for the unmixed flow two spool geared turbofan
(Figure 0.3) is planned to be bought by Bombardier but was not available during the
project. With DLL’s for these two engine types, it is possible to model all jet engine types
considered for Bombardier aircraft. Engine model files for use with GasTurb DLL’s were
created for four different jet engines during this project.
The DLL contains a set of functions for calculating steady state performance in a similar
way as an engine deck. In this implementation a C++ program was written to call the
functions within the DLL. This C++ program reads an input file which specifies the
ambient conditions, power setting and installation losses. Then it calls the DLL with a
GasTurb v11 engine model file and calculates the point defined in the input file. Finally
an output file is created with calculated properties such as net thrust, SFC, pressures and
temperatures at different stations within the engine. Example input and output files are included in Appendix A. For a detailed description of the variables in these files, the
reader is referred to the DLL user manual [15].
Choose engine cycle
The first choice the user must make is the engine cycle. There is a big selection of engine
cycles in GasTurb v11 but only the geared mixed flow turbofan and the geared unmixed
flow turbofan are considered because of the possibility to use these types of engine model
files with the DLL’s.
If the engine that is to be modeled has a centrifugal compressor stage a simplification has
to be made and the centrifugal compressor is modeled as an axial compressor. This
makes no difference at the design point calculation, but at off-design this could be a
source of error because the maps used reflect an axial compressor. In the map collection,
there are maps for axial-centrifugal HPC’s. If such a map is used in the modeling, the
behavior of an axial-centrifugal HPC would be captured. This was not done for the
modeling of the engines with centrifugal compressor stages.
Pick an engine cycle in the drop down list as illustrated in Figure 3.1, chose
“Performance” as the “Scope”, “Design” as the “Calculation Mode”, then press “Run”.
The user is prompted to choose an engine model file, chose the “Demo_xx.XXX” in
the “GasTurb11” folder, where “xx.XXX” is dependent on the type of engine selected.
Starting with the demo file gives a good starting point and reasonable values for all
engine design parameters.
Geometry and weight prediction
GasTurb v11 has the possibility to predict the geometry and weight of a jet engine. It is
based on a model where each component is modeled as a separate part with a material
density, thickness and surface area. Not every component of a real jet engine is included
in the model and therefore a “Net Mass Factor” is used to take this into account. More
than a 100 additional input parameters are needed to accurately model the geometry and
weight of an engine. The additional input includes the following parameters:
Lengths, angles and entry/exit radii for inlet, inlet cone, burner, ducts, bypass,
exhaust.
Strut, guide vane and blade aspect ratio and pitch.
Material properties, material thickness.
Stress margins for disk stress calculations.
GasTurb has default values for all of these parameters and an attempt was made to model
the investigated engines in the geometry and weight tool. The number of compressor and
turbine stages was given as input, remaining parameters were left to their default values.
The investigated engines that use a centrifugal compressor stage were modeled as axial
flow compressors with approximately the same pressure ratio. This is because only axial
flow turbomachinery is considered in the weight and geometry tool. This assumption
could increase the predicted engine length and decrease the predicted engine diameter
due to geometrical differences between an axial and a centrifugal compressor stage.
The results of the investigation are illustrated in Figure 3.4. The max diameter is
underestimated for all engines but not close to the real engine value. Estimated length and
weight is sometimes too high and sometimes too low. The poor result was expected given
the lack of detailed input data.
The inputs that are needed are not readily available at the Advanced Design department.
Furthermore, this tool is more suited for conceptual design of an engine than for engine
performance and weight prediction within an MDO software framework. Therefore it was
decided that other methods should be used to predict the engine weight and size. A
reasonable method would be a statistical model based on existing engines where the
weight is related to the thrust (thrust/weight ratio).
Case studies
The engine modeling method was applied to 5 different engines. The results are
considered sensitive information in terms of actual numbers and no values or scales are
given in text, figures and tables. For all results, the data labeled “GasTurb” is the data
created by the GasTurb engine model. The “Thrust table” or “Engine deck” label
represents engine data obtained from the engine manufacturer’s steady state performance
program. Whenever an error is described as a percentage, it is calculated as seen in
Equation 4.1. A negative error means that the value predicted by the GasTurb model was
lower than the thrust table or engine deck value. A positive error means that the value
predicted by the GasTurb model was higher than the thrust table or engine deck value.
GasTurb engine model setup
An engine deck was available for Engine 1. Therefore the uncertainties regarding certain
input parameters, e.g. intake pressure loss, bleed air and power offtake were removed
because the user is in control of them when running the engine deck. Many engine
parameters such as efficiencies and nozzle coefficients were hidden from the user. E1
does not have a booster stage but was modeled as a geared mixed flow two spool
turbofan with booster in order to use the engine model file with the DLL. Therefore the
inner FPR was set to 1 and the real inner FPR set as the IPC PR.
E1 engine model estimated input parameters regarding the assumed design point:
The assumed design point was selected as uninstalled 43 000 ft, M0.82 at 94% of
maximum cruise rating.
Fan isentropic efficiency was iterated to match T16.
HPC isentropic efficiency was iterated to match T3. This is a bit low, comparing
to Table 3.2.
Iteration for correct inlet mass flow.
Outer FPR calculated from engine deck output.
Inner FPR 1 because engine without booster.
IPC PR – is the inner FPR, value comes from Engine 1 documentation.
HPC PR calculated from engine deck output.
LPT efficiency and exhaust pressure losses adjusted to obtain correct P6 pressure.
Power offtake according to documentation.
No HPC bleed.
Assumed the relative enthalpy for the port 2.8 bleed to get the correct bleed
pressure compared to the assumed design point engine deck output. Turbine cooling flow calculated from W2 – W3. Uncooled LPT NGV and LPT
rotor was assumed, in line with the cooling flow model presented in Figure 2.4.
Thrust reverser leakage flow calculated as total mass flow in – total mass flow
out, Wleakage = W12 + W2 + WF – W8.
Leakage core to bypass is calculated as the difference between core in and out
flow, W2 – (W6 – WF).
The mixer efficiency was unknown and set to 1.
The mixer area was assumed to be equal to the fan frontal area.
The nozzle petal angle was left at its default value of 10º.
Iterate nozzle thrust coefficient to match the net thrust.
Iterate burner exit temperature for correct fuel flow. No data is given about T41,
but the resulting ITT is 24K too high. This could be due to an incorrect cooling
flow model or because of the wrong temperature drop over the HPT. The HPT
turbine temperature drop is decided by the HPT extracted work,
GasTurb engine model setup
The E2 thrust tables contain a big uncertainty in the form of a “fuel flow factor”. The
purpose of the fuel flow factor is to adapt the engine performance program SFC to
measured data from flight tests. Despite efforts to find the fuel flow factor for Engine 2,
no one at Bombardier could provide this information. Therefore, the engine deck data
was assumed to represent the E2 engine and engine deck output data was used to create
the GasTurb model. This engine model was not converted to GasTurb v11 and only exists
as a v10 cycle data file. Therefore all presented results were created by GasTurb v10.
In order to model the E2 unmixed flow engine in GasTurb some assumptions have to be
made because of the differences in the two models. In the engine deck, the scrubbing
drag as illustrated in Figure 2.1 is modeled and given as a separate term. The scrubbing
drag is small but it cannot be modeled in GasTurb. There are terms generating thrust in
addition to the core and bypass nozzle, including exhaust of core and core plug
ventilation air. This thrust is similar in size to the scrubbing drag, so the net effect is
small. These effects are taken into account by calibrating the GasTurb model net thrust to
the engine deck data by iterating the turbine exit duct pressure ratio.
An intake pressure recovery map was made from data found in an engine manual. The
map was adapted to the GasTurb file format. The map was also verified to give correct
values at cruise by comparing to the available engine deck outputs. There were no engine
deck data available for the TO condition so the validity of the map could not be verified
for this flight condition. The map is seen in Figure 4.7. Note that the intake pressure
recovery is essentially constant with regards to Mach number in the interval M0.4 to
Difficulties and future work
From the first day at Bombardier, it was said that GasTurb v11 would soon arrive to the
company. In the meantime, GasTurb v10 was used. Finally, v11 arrived only a few weeks
before the project was supposed to finish. This meant that the planning did not hold and
at the end of the project a lot of time was spent on upgrading engine model files to v11
and doing the validation again with the new software version.
A solution to the problem with saving component map files in GasTurb v11 (see section
3.6.2) was not found. This did not influence the results in terms of SFC or net thrust. It is
recommended that a solution is found if future work involves changes to the component
map files.
One problem with modeling of jet engine performance is the vast number of parameters
needed to make a detailed model. Many parameters, e.g. component efficiencies, duct
pressure losses and internal engine cooling, were estimated. This was overcome by
calibrating the engine model to match the available engine performance data at the
assumed design point. The calibration was done by manually tuning the unknown
parameters. In the future, the calibration could be done automatically within the
optimization software framework under development at Bombardier.
An accurate performance model of a jet engine requires component maps that describe
the compressor and turbine characteristics. Component maps are engine manufacturer
proprietary information and were not available during the project. The solution was to use
generic component maps and calibrate them so that the predicted performance matched
the engine manufacturer performance data. It is suggested that the calibration could be
automated within the optimization software framework. There is also a possibility of
using publicly available component maps that describe the component performance in a
better way than the default GasTurb component maps.
A difficulty with the MTO and MCL rating net thrust prediction is to find reasonable
limits to the control system. The method presented here relies heavily on the availability
of detailed, perhaps proprietary, engine performance data. It would be desirable to
develop a more generic method to find control system limits. This reduces the
dependence on manufacturer data and makes it less of a data matching exercise. One
conceivable example could be to relate the maximum allowed P3 to the maximum OPR of
the engine.
The rubber engine model could not be used in a full scale multidisciplinary design
optimization run because the MDO framework was still under development during this
project. The engine model was however run separately within the MDO framework. The
setup of optimization loops and the details in how the engine model interacts with the
other disciplines of the MDO software is left for future projects.
Conclusions
It is concluded that the goals set at the beginning of the project were fulfilled. The
presented results prove that GasTurb v11 can be used to make accurate jet engine
performance models. In addition, a method on how to create a model was developed and
is presented in this work. Therefore the validation goal is fulfilled.
It was also proved that the GasTurb v11 engine models can be run without GUI and
without user interaction in optimization loops within Isight, the software chosen for
Bombardier’s MDO software framework. Therefore the integration goal is fulfilled.