FUEL METERING PUMP DEVELOPMENT AND MODELING
Jiří Toman, Vladimír Hubík
UNIS, a.s., Department of Mechatronics, Jundrovska 33, Brno, Czech Republic
Vladislav Singule
Faculty of Mechanical Engineering, Brno Univesity of Technology, Technická 2, Brno, Czech Republic
Keywords: Aircraft electronics, BLDC motor, Control system, Fuel pump, Identification, Motor drive, Simulation.
Abstract: This article would like to inform reader about the research and development steps and practices which were
used during the development of a control system for a fuel metering pump. The aims of the development
were to design a control system for a brushless DC (BLDC) motor in accordance with rigid aviation
standards and to verify new development practices and tools allowing faster and much simpler final
certification. The paper comprised definition of first requirements, preliminary hardware (HW) and software
(SW) design, system modeling and simulation, system optimization, detailed design, verification and
testing. In addition, the first results measured on an evaluation sample are presented.
1 INTRODUCTION
The designed control system shall ensure safe and
reliable control of the BLDC motor that drives the
fuel metering pump (FMP) (Bose, 2009). FMP
operation shall be ensured under various external
conditions, including harsh environments and
unknown initial states such as pressure in the
system, position of the rotor or necessary starting
torque of the BLDC motor.
1.1 Architecture
1.1.1 System Architecture
The fuel system will be composed of electrical,
mechanical and hydraulic equipment necessary to
measure, to provide and to control the required fuel
flow (Cochoy et al., 2007).
The metered flow will be distributed to a fuel
manifold.
Figure 1: Block diagram of the system architecture.
1.1.2 FMP Architecture
The FMP is composed of three main components:
 Fuel pump with a by-pass relief valve,
 28 Vdc brushless DC motor – BLDC,
 Control system (CS) of the FMP with power
electronics.
A by-pass relief valve shall be part of the FMP in
order to protect the equipment from overpressure.
The FMP architecture is shown in Figure 2.
FUEL OUTLET
SYSTEM ARCHITECTURE
FMP
ECU (CONTROL SYSTEM)
FUEL INLET
28 VDC
INTERFACE WITH
ECU
419
Toman J., Hubík V. and Singule V..
FUEL METERING PUMP DEVELOPMENT AND MODELING.
DOI: 10.5220/0003530804190426
In Proceedings of the 8th International Conference on Informatics in Control, Automation and Robotics (ICINCO-2011), pages 419-426
ISBN: 978-989-8425-74-4
Copyright
c
 2011 SCITEPRESS (Science and Technology Publications, Lda.)

Figure 2: Block diagram of the FMP.
1.2 Control System Requirements
Requirements for the control system were defined
during the early stage of the project in consultation
with external aviation specialists. These
requirements include electrical characteristics,
reaction of the control system to defined incidents,
start-up characteristics, control commands etc.
The FMP’s variable flow rate will be controlled
by the Electronic Control Unit (ECU) to maintain
constant speed of the Auxiliary Power Unit (APU).
In parallel with the pump, a by-pass relief valve will
protect it against overpressure. The FMP shall
provide the fuel inlet temperature to the APU ECU.
The FMP shall control the fuel flow by varying the
pump speed for the following APU operations:
 Start the APU (open loop: no closed loop
control on APU speed for the ECU),
 Maintain the APU speed at a constant value
(closed loop on APU speed for ECU) and
transfer to the FMP via the digital data bus
and/or the analog setting input.
The most important requirements of the control
system are listed below.
1.2.1 Interface Characteristics
 The control system of the FMP should be
controlled via a digital data bus or a single
analog signal,
 CS together with the BLDC motor should be
supplied from an AC/DC bus with the option
of supply from a battery,
 CS should receive only start/stop and flow
throttle commands,
 CS should be able to indicate the flow level,
power bridge current, status word, FMP mode,
built-in-test results and start/stop command
acknowledgement,
 The nominal power consumption should be
less than 250W, the peak power consumption
could be up to 500W for a defined period of
time.
1.2.2 Functional Requirements and
Performance Characteristics
 CS shall ensure safe and soft start of the
BLDC motor under any conditions.
 During operation the CS shall maintain the
fuel flow at required values.
 CS shall enable internal parameters setting in
a dedicated maintenance mode.
 CS shall be able to precisely meter fuel flow
during a defined interval.
 The maximum underflow should be less then
2% at Δω > 40% and the fuel flow should
stabilize in 240 ms.
 In case of stop command the FMP shall stop
in less than 100 ms.
 Minimum fuel flow is 3 l/h and maximum 117
l/h.
1.2.3 Physical Requirements
 CS should be designed for minimum
dimensions and weight.
 CS cooling shall be designed with active fuel
cooling to withstand ambient air temperature
over +55°C.
 CS shall ensure normal non-degraded function
under the defined circumstances and lifetime.
 Nominal operational temperatures shall be -
55°C to +85°C, short-term operating
conditions shall be -55°C to +125°C.
 During APU operation, the maximum steady
state flow in worst temperature conditions
(air: +85°C, fuel: +65°C) will be 76 l/h.
 The CS shall restart without problem after
having been soaked 4 min at +125°C.
Beyond the above stated requirements, the
control system and the FMP shall be designed in
accordance with many other requirements regarding
atmospheric pressure, temperature variations and
humidity, shocks and vibrations, lightning and
electrostatic discharges. For certification purposes
the development, testing and verification must be
performed in accordance with the aviation standards
”RTCA/DO-178B – Software Considerations in
Airborne Systems and Equipment Certification”
(RTCA, Inc., 1992), ”RTCA/DO- 254 – Design
Assurance Guidance for Airborne Electronic
Hardware” (FAA Advisory Circulars, 2005) and
”RTCA/DO-160F – Environmental Conditions and
Test Procedures for Airborne Equipment” (RTCA,
Inc., 2007). Failure mode analysis (FMEA) of the
designed hardware and testing of the SW code by
means of dedicated tools are mandatory parts of the
FMP ARCHITECTURE
INTERFACE WITH
ECU
MOTOR
BLDC
FUEL
PUMP
CS with power
electronics
U,V,W
FUEL OUTLET
FUEL INLET
n
FUEL SENSOR
28 VDC
ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics
420

development and both of these tasks were
successfully performed at the final stage of the
development. Compliance with the above mentioned
aviation standards during the development was
needed for demonstration of ability to design,
develop and certify complex electronic control
system for the aviation industry, using new
approaches based on utilization of COTS
components and integrated simulation tools.
2 CONTROL SYSTEM
MODELING AND
SIMULATIONS
Based on previous experience, every development of
a control system starts with definition of
requirements and system modeling. A mathematical
model describes all parts of the control system
together with the system under control. This gives
the developers an exact idea about system behavior,
reactions and the ability to verify different control
strategy and algorithms in the early stage of the
development. In addition, the control system can be
tested by means of hardware in the loop simulation,
e.g. using a tool such as dSPACE, without any
previous hardware design. This is a considerable
advantage since the team could predict many
mistakes, dead ends and even damage to the first
evaluation samples. Unquestionably, these are
benefits that considerably decrease the development
time and costs.
This part summarizes development of the
mathematical model of the dynamic system, i.e. the
fuel metering pump, using the mathematical
modeling approach and methods of experimental
identification. Next the development of the optimal
control is discussed, according to the specific
behavior demands. The system for identification is
an experimental evaluation sample of FMP that is
intended for aircraft engine fuel delivery. Increased
demands are put on control electronics, because
application reliability and safety according to actual
aerospace standards are crucial. Identification and
optimization of the controller constants have been
performed in the MATLAB/Simulink environment.
2.1 Identification
It is necessary to set up a mathematical model to
control the whole system according to the response
demands. Because the dynamic system consists of
electrical, mechanical and hydraulic parts, and is
therefore non linear, a mathematical description
based on the physical principles would be very
complicated. For this reason a “Black box” method
is used. This method is based on analysis of the
single step response. The block schematic of the
identification principles is shown in Figure 3.
IDENTIFICATION ARCHITECTURE
MOTOR
BLDC
FUEL
PUMP
CS with power
electronics
U,V,W
FUEL OUTLET
FUEL INLET
n
FUEL SENSOR
IDENTIFICATION
TOOLBOX
SET POINT
TRANSFER FUNCTION
OF SYSTEM
Figure 3: Block schematic of the identification process.
The monitored parameters are input voltage step
change and rotational speed response of the BLDC
motor. These characteristics are then imported into a
MATLAB environment for further post-processing
using the “Identification toolbox”. The result of the
identification process is the description of the
system, which consists of two transmission first-
order functions. The achieved transmission
equations are 1 and 2.





= 415 ∙
1
9∙10

 + 1
(1)





=9∙

0.12 + 1
(2)
2.2 FMP Model and Simulation
According to the obtained identification results a
system model in MATLAB/Simulink environment is
created. The model is made of basic Simulink blocks
from common libraries and its structure is depicted
in Figure 4.
Figure 4: The basic diagram of the FMP simulation.
The measured voltage step change is the input value
to the block and the output presents the transient
response of the model. Both parameters are stored
and displayed in the Scope block. The measured
rotational speed transient characteristic of the FMP
Scope
Output signals
identification_out.mat
Input signals
signal_matrix.mat
Identificated system
U[V]
N[RPM]
3
3
3
2
FUEL METERING PUMP DEVELOPMENT AND MODELING
421

is also added to the structure and therefore makes it
possible to compare response of the model with the
real system.
Comparison of the model and the real FMP
system response is shown in Figure 5. The blue
colored line represents the transient characteristic of
the model, the red colored line transient represents
the transient characteristic of the real FMP. This
result is sufficient for controller algorithm
development to the target embedded platform.
Figure 5: The simulation results of the controller respond.
2.3 FMP Model and Simulation
The “Controller” block has been added to the
previous model. A cascade controller structure of
discrete proportional – summation - differentiation
(PSD) controller is implemented inside this block
(Leonhart, 2001). The block schematic of the PSD
controller is depicted in the Figure 6. The rotational
speed controller has priority over the power
controller.
Figure 6: Inside structure of the PSD controller.
The difference between the required and actual
values of rotational speed is one input parameter to
the “Controller” block. Another input is the input
voltage. The actual rotational speed is driven by
feedback from the FMP model. The controller
output is the actuating signal, which has voltage
representation. The block schematic of the controller
loop is depicted in the Figure 7.
Figure 7: The block schematics of the control loop.
The PSD controller constants were iteratively
changed to evaluate quality of control according to
the given requirements. The FMP response was
compared using the least squares method according
to the limit characteristics and the lowest difference
from the calculated required characteristic. The
“Criterion” function (block) was then created to
limit characteristic definition, which determines the
controller requirements. The maximum overshoot
(upper limit) and required dynamic response of the
controlled system (lower limit) are also defined in
the block. The upper limit is evaluated as a
percentage of the required rotational speed. The
lower limit is then determined by the time constant,
which defines the maximum time for system
stabilization for the required speed. The lower limit
is realized in the model as a first-order transfer
function with the required time constant.
The system response with optimal PSD constant
settings is shown in Figure 8. There are also
depicted the limit characteristics and required value
of the rotational speed. From the results it can be
seen that the developed controller fulfills the
controller quality demands and it is possible to use it
in a real system on the target platform. The whole
methodology proves to be useful in development of
the optimal control algorithms. Time and cost
reduction of the whole development cycle is evident.
Values provided by the simulations of the control
system were used in the hardware design which was
the next step of the project.
3 HW DESIGN OF THE
CONTROL SYSTEM
The control system is of a modular design and
consists of three parts that can be interchanged
according to customer needs. The three basic
ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics
422

Figure 8: The system response with the optimal PSD
settings.
modules are Control and Communication Unit
(CCU), Power Electronics Unit (PEU) and I/O Unit
(IOU).
3.1 Control and Communication Unit
The main microcontroller is mounted on the Control
and Communication Unit (CCU). The CCU is
replaceable according to application performance
requirements and the architecture of the control
system. The core of the CCU is a multipurpose
microcontroller (MCU).
The CCU has a unified interface for all the
analog and discrete signals that are used for control
and communication with other control system
modules. Using a unified interface enables
replacement of the CCU in case of system
enhancement or maintenance.
The electronic control system for the FMP can
provide selected information about its internal states
and measured values to the higher level control
system via an internal communication network. The
higher level control system can be a Flight Control
Computer (FCC) or a multifunction avionic display
placed in the pilot’s cockpit.
3.2 Power Electronics Unit
The Power Electronics Unit (PEU) consists of a full
H-bridge comprising six power switching
transistors. Motion of the BLDC motor is controlled
by a switching power supply to the particular coils
of the BLDC motor.
An integral part of the PEU is the Protection
module that measures temperature, current and
voltage on the BLDC motor. If any parameter
exceeds a limit value the protection module sends a
fault signal to the MCU. Detection of the fault signal
causes disconnection of the load from the power
supply source.
3.3 I/O Unit
BLDC motor signals/sensors - depending on actual
configuration, the electronic control system could
operate either in sensor or sensor-less mode.
In sensor mode, signals from Hall sensors are used
as feedback. These sensors are usually mounted
inside the BLDC motor by its producer. These
signals are triggered and used as inputs to the control
MCU.
Sensor-less control mode operates on the
principle of sensing induced voltage caused by Back
Electro Motive Force (BEMF) on one of three
BLDC motor phases. Feedback is extracted by
means of BEMF and zero-cross detection.
Functionality and safe operation of the actuator
is ensured by monitoring of selected parameters and
restricting the actuator’s fault operation. If one or
more signals exceed their limit value, the fault is
detected and appropriate action is taken.
3.4 Evaluation Sample
The control system was designed to fit the
requirement of mounting into the FMP to create a
monolithic box with an explosion-proof design. The
development runs in accordance with the aviation
standards RTCA/DO-254 (FAA Advisory Circulars,
2005) and RTCA/DO-160F (RTCA, Inc., 2007).
The electronics of the control system are
designed in a custom tailored shape with logical
partitioning according to the functions performed.
Logical partitioning of control, power electronics
and sensing board provides the added advantage of
custom configuration and much simpler service.
4 CONTROL SYSTEM SW
DESIGN
The main aim taken into consideration was that the
control system had to be portable and easy to
implement on a common 16-bit MCU. The final
control system is written in the C programming
language and is implemented in a DSP MCU.
Algorithms are designed with regard to the high
criticality of the application; therefore no artificial
methods or fuzzy control algorithms could be used.
The requirement for high reliability also limits the
code complexity, thus simple but efficient software
algorithms are used wherever possible.
FUEL METERING PUMP DEVELOPMENT AND MODELING
423

Software design was preceded by detailed
decomposition of system requirements, interface
definitions, data flow and control flow. These
requirements and definitions determined the final
form of the source code. And so their thorough
evaluation was extremely important for design of
control algorithms. Proper definition and evaluation
simplified the software development cycle and
eliminated errors caused by further implementation
of additional functions.
The algorithm consists of initialization part,
motor start-up, closed loop control, interrupt service
routines and communication routines.
4.1 Initialization
Initialization part serves for initial hardware set-up,
parameter setting and power-on self test. During this
stage all the parameters and values are checked
against the standard values. In case of abnormal
values or malfunction, the control system issues a
warning and tries to re-initialize hardware.
4.2 Motor Start-up
In critical applications, it is necessary to ensure
correct startup of the motor. Thus, many simulations
were done before implementing the control
algorithm into the controller. The motor start-up
algorithm ensures reliable start-up of different types
of BLDC motors. Sinusoidal and conventional
trapezoidal methods of control are used.
4.3 Closed Loop Control
After initialization and motor start-up sequence, the
control algorithm switches to closed loop control.
Closed loop control algorithms consist of the two
nested PSD controllers - the speed controller and the
power controller (Leonhart, 2001).
The power regulator sets the desired value by
means of PWM. It compares the desired value from
the superior speed regulator and the measured
current through the BLDC and sets the output value
according to their variance.
The speed regulator sets the desired value for the
current controller. Actual rpm speed could by
measured by Hall sensors or using a Back Electro-
Motive Force (BEMF) (Holtz, 2007).
4.4 Interrupt Service Routines
Interrupt service routines serve for performing
repeated tasks that evaluate critical values such as
power electronics temperature, current flowing into
the BLDC motor, DC bus voltage, etc.
Separate interrupt service routines also serve for
input/output processing and measurement.
5 FIRST EVALUATION SAMPLE
TEST RESULTS
To evaluate the performance of the control system
and the designed electronics, two types of evaluation
test benches were used. Firstly, the in-house
developed evaluation test bench. First measurements
and simulated dynamic testing were performed on
this evaluator and different control algorithms were
tested and controller variables were set.
The control system electronics were then
mounted on the FMP and performance tests were
carried out on an evaluation mock-up platform that
simulates a real fuel circuit. These tests were
performed with the help of an external company
with particular emphasis on start and stop sequences
of the FMP.
The start sequence of the fuel pump, shown in
Figure 9, was verified for a step change request from
50 percent of the fuel flow. This means that the
starting flow level was 43 l/h (3250 rpm of the
BLDC motor) at 2 MPa of back pressure. The
required flow after step change should be 92 l/h
(7300 rpm of the BLDC motor) at 3.9 MPa of back
pressure.
Figure 9: The FMP start sequence measurement.
Figure 9 compares the optimized pump
characteristic with the original values. With a
precisely tuned motor controller it is possible to
achieve a start time of less than 200 ms and to fulfill
the customer requirements.
The next important feature of the fuel pump is
the stop time performance. This behavior is crucial
for the turbine control system and dynamic of the
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
40
47
54
61
68
75
82
89
96
T: 0.506
t [s]
Q [l/h]
T: 0.151
optimized fuel flow
fuel flow
Q = 90%
Q = 100%
delta = 529 [ms]
delta = 174 [ms]
ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics
424

whole hydraulic chain. According to the technical
specification, the requirement is a stop time of no
more than 100 ms after receiving the stop command.
The stop time characteristic was much more
difficult to measure. The directly connected speed or
flow sensor to the fuel pump influences the stop time
characteristic. The only way was to measure the
speed from the motor’s internal Hall sensors. There
are three graphs in Figure 10 and Figure 11 that
represent the output signal from the position sensors.
After a simple MATLAB post-process it is possible
to evaluate the actual rotor speed.
Figure 10 shows the stop time characteristic
when the system is not perfectly optimized. The fuel
pump should stop in this case from the nominal fuel
flow of 92 l/h (7300 rpm of the BLDC motor) at 3.9
MPa of back pressure. It is obvious that the technical
specification requirement has not been fulfilled. The
stop time is more than 200 ms.
Figure 10: The stop time characteristic with not optimized
controller.
Figure 11 shows the stop time characteristic of
the optimized controller. The same measurement
method has been used in this case. The stop time has
reduced to 83 ms which is acceptable.
Figure 11: The stop time characteristic with optimized
controller.
6 CONCLUSIONS
Development and certification of any device in the
aerospace industry requires a thorough approach,
including definition of requirements, design of
electronics, software coding and testing. In addition,
very complex documentation must be maintained for
the whole development and lifetime cycle.
Use of COTS components and development tools
enables a faster development cycle with the ability
to carry out modeling and preliminary design in the
early stage of project. Results from this stage can be
used as the basis for hardware and software design
which saves additional costs and greatly accelerates
the process.
The development procedures described in this
article indicate that they can bring about significant
improvements in performance, safety and reliability
of the control system along with reduction of
development time and costs.
ACKNOWLEDGEMENTS
Published results were acquired using the support by
MPOČR project No.FR-TI2/313. Hardware and
software development and physical modules were
realized and project was supported by UNIS, a.s.
This work was also supported by project No.
MSM 0021630518 “Simulation modelling of
mechatronics systems” solved on the Faculty of
Mechanical Engineering, Institute of Production
Machines, Systems and Robotics, Brno Technical
University.
REFERENCES
Bose, B. K. (2009). Power Electronics and Motor Drives
Recent Progress and Perspective, IEEE Trans. On
Industrial Electronics. 56, 581-588.
Cochoy, O. and Carl, U. B. and Thielecke, F. (2007, June).
Integration and control of electromechanical and
electrohydraulic actuators in a hybrid primary flight
control architecture. Recent Advances in Aerospace
Actuation Systems and Components, 1-9.
FAA Advisory Circulars. (2005). RTCA/DO-254 ”Design
Assurane Guidance for Airborne Electronic
Hardware”.
Holtz, J. (2007). Sensorless Control of Induction Machines
– With or Without Signal Injection?, IEEE Trans. On
Industrial Electronics. 53, 7-30.
Leonhart, W. (2001). Control of Electrical Drives (3
rd
ed.). Berlin: Springer.
0 0.05 0.1 0.15 0.2 0.25 0.3
0
5
0
5
0
5
0
5
10
t [s]
U [V]
T: -0.0003
T: 0.258
stop signal
HallA
HallB
HallC
delta = 258 [ms]
0 0.02 0.04 0.06 0.08 0.1 0.12
0
5
0
5
0
5
0
5
10
t [s]
U [V]
X: 0.083
X: -0.00039
stop signal
HallA
HallB
HallC
delta = 83 [ms]
FUEL METERING PUMP DEVELOPMENT AND MODELING
425

RTCA Inc. (1992). RTCA/DO-178B ”Software Conside-
rations in Airborne systems and Equipment
Certification”. USA.
RTCA, Inc. (2007). RTCA/DO-160F ”Environmental
Conditions and Test Procedures for
AirborneEquipment”. USA.
ICINCO 2011 - 8th International Conference on Informatics in Control, Automation and Robotics
426