Simulation of an INVERSE JET ENGINE MODEL for performance prediction and fault diagnosis in transient operation

1
Simulation of an INVERSE JET ENGINE MODEL for
performance prediction and fault diagnosis in
transient operation
Control Robotics Lab

• Project by Inbal Srebro
• Levin Dmitry

Supervisor Dr. Michael Lichtsinder
2
Motivation

• finding a new and cheaper way to constructing
  engine maps
• Finding a faster algorithm to simulate engine
  as a first step towards real time simulation
• Proving the model is sensitive to error, as a
  first step towards error detection algorithm

3
Objectives

• Development of the Inverse Engine Model (IEM)
• Our goal is to evaluate the engine maps
  when not all the effective (realistic) map
  components are known Compressor-Nozzle IEM,
  Turbine-Nozzle IEM
• Evaluation of Turbine/Compressor Map
• Each turbine/compressor in an engine has
  an operating map. Complete maps are based on
  experimental data or predicted by special
  computer programs.
• Our goal is to write the program which
  will be able to predict operating maps.
• Fast simulation (and control)
• Using IEM algorithm to produce MATLAB
  simulation instrument for map evaluation

4
Single Spool engine station
5
Turbine map
Each turbine/compressor in an engine has an
operating map. Complete maps are based on
experimental data or predicted by special
computer programs The typical turbine map is
shown in the picture below. The x-axis of the
turbine map is pressure ratio(PR). The other axis
is the corrected flow and the z-axis is
efficiency
Evaluated points from data
Lines of
6
Compressor map
7
Inverse Engine Model (IEM)
Command
Program Output
Inverse Engine Model
Evaluation of effective engine component maps
8
CN-Shortened Inverse Engine Model

(without Turbine Map)
Closed Loop Engine
CN-Inverse Model
Maps CompressorNozzle
9
Conditions for construction

(short Inverse Model)
For CN short Inverse Model
10
Advantages of IEM

• May be built when not all engine components can
  be measured
• CPU time decrease compare to conventional model
• May be used for fast calculation of steady-state
  and transient performance of single spool engine,
  and is a step towards real time simulation

11
MATLAB algorithm (flowchart)
12
Simulation results Input
Input to engine (command)
Measure data from transducer
13
Simulation result Turbine map
parameters
14
Simulation result
Turbine map
Data from full conventional model
Evaluated data
Two dimensional plot of turbine map, with
original contour lines (got from the manufactory)
15
3D-graphs Turbine map
The slices
16
Evaluation of Turbine Map
(Using Compressor and
Nozzle maps)
17
Evaluation of Turbine Map
(Using Compressor and
Nozzle maps)
mdot4cor
2500
2750
3000
3250
Evaluated data by Inverse engine model
Pi45
18
Evaluation of Turbine Map
(Using Compressor and
Nozzle maps)
19
Evaluation of Turbine Map
(Using Compressor and
Nozzle maps)
known data from full turbine map
2750
2500
3000
Evaluated data by Inverse engine model
20
Evaluation of Compressor Map
Compressor Map Data Matrix (1500X4)
21
Sensitivity to bias error
In future development, this model will be used
for engine error diagnosis and for that purpose
it has to be sensitive to error a small error in
input should produce larger error in output.
Assuming that single transducer faults could be
present in the engine at any instance, lets
insert the following fault into the model
measurement bias of 1 for the fault transducer
22
Evaluation of Turbine parameters
bias error 1 Error in sensor T05
23
Conclusions

• Shortened Inverse Engine Models can be used for
  engine component map evaluation using data
  acquisition during transient (or steady state)
  operation.
• The investigation shows that relatively small
  measured bias in the inverse model input leads to
  relatively significant output errors -
  Sensitivity.
• Only one strategy of evaluation is described.
  However, alternative shortened inverse engine
  models may be developed for different engine
  component map detection.
• Limitations
• Accuracy of the engine model
• Measurement error (random and bias) problem

24
Conclusions (Evaluation of Turbine
Map)

• The idea of evaluating compressor/turbine
  map which was described above is claimed.
• By evaluating the compressor or/and turbine
  map from inverse engine model(IEM) the data which
  before can be only given by the producer (got
  from long-lasting and very expensive
  experiments) of the engine can now be evaluated
  by just one computer and impressively short time.

25
Future Tasks

• Flexibility
• One flexible program for all Partly map
  simulations
• Noise in the system modeling
• Updating system for being Noise Resistant