Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 176 (2017) 85 – 95

Dynamics and Vibroacoustics of Machines (DVM2016)

Simulation of Direct Drive Electromechanical Actuator

with Ballscrew
L. Bilyaletdinova*, A. Steblinkin
Central Aerohydrodynamic Institute named after N.E. Zhukovsky (TsAGI), Zhukovskogo 1, Zhukovsky 140180, Russia

Abstract

The paper contains the description of electromechanical actuator (EMA) dynamic mathematical model developed by TsAGI in
the frame of European-Russian collaborative project «RESEARCH» for the direct drive electromechanical actuator which was
under CESA company development. The model was implemented within the MATLAB/Simulink software. The major focus of
the model is the modularization and scalability of actuator components, as well as the focus on actuator mechanical non-
linearities in the region of low input control signals as well as the possibility to use that model for flight mission cavity thermal
analysis. The model has various switchable levels of components detail, which provides a balance between the accuracy and the
value of the integration step in dealing with a variety of modeling tasks. Due to the possibility of taking into account a number of
factors it can be used for tasks of two types. At first, it can be used for tasks requiring small integration step and high precision
simulation of high-frequency processes (detailed analysis of the static and dynamic characteristics of EMA, the optimization of
actuator heat dissipation, controller setting, etc.). Also it can be used for the tasks requiring rapid model calculation (analysis of
the closed-loop stability «aircraft-flight control system-actuator», actuator characteristics evaluation, etc.).
© 2017 The Authors. Published by Elsevier Ltd.
© 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of the organizing committee of the international conference on Dynamics and Vibroacoustics of
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Machines. under responsibility of the organizing committee of the international conference on Dynamics and Vibroacoustics of Machines
Peer-review

Keywords: direct drive EMA, electromechanical actuator, ballscrew, backlash, friction, simulation, Simulink;

1. Introduction

Fuel burn reduction has become the priority for aircraft operators due to environmental and cost implications.
Such interest has encouraged the aerospace industry to look at ways of minimizing fuel consumption, concluding

* Corresponding author. Tel.: +7-926-828-15-58.

e-mail: kedr23k@yandex.ru.

1877-7058 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the international conference on Dynamics and Vibroacoustics of Machines
doi:10.1016/j.proeng.2017.02.276
86 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95

that the approach towards a More Electrical Aircraft (MEA), i.e. an aircraft that uses electricity to operate its
consumers, could provide some benefits. This approach tends to remove hydraulic actuators and the hydraulic
systems, as they require a significant amount of maintenance, are heavier and more energy consuming [1, 2].
Despite the potential benefits such as weight reduction, increased safety and reliability, lower consumption,
reduced maintenance costs, etc. electrical actuators are not fully embraced by the industry due the limitations they
present in their current development stage, which prevent them from being used massively or in critical areas, such
as aircraft primary flight controls actuation [2]. These limitations are high actuator weight, safety concerns due to
the jamming probability of electromechanical actuators, thermal behavior and electromagnetic compatibility issues.
The intention to solve these issues through the development, manufacturing and experimental validation
processes was the basis for the co-founded Russian-European R&D project called RESEARCH [www.research-
project.aero] which objective was the development of the regional jet elevator control system based on electrically
powered actuators: electromehcanical (EMA) and electrohydrostatic actuators. The development of EMA for this
system had being performed by CESA company. The actuator’s overall view and main characteristics are
represented below.

Fig. 1. EMA isometric view (CESA).

Table 1. EMA parameters.

Stroke ±75 mm
Supply voltage 270 VDC
Maximum rod velocity 80 mm/s
Rated velocity at max op. load 70 mm/s
Maximum operating load 30 kN
Stall load 45 kN

Being a highly critical application with lots of cross-related physical domains to be taken into account, the
development of such system should rely on wide and intensive simulation activities through the whole development
cycle. As the result the development of the detailed EMA dynamic mathematical model was started by the
consortium in order to obtain static and dynamic actuator characteristics assessment and detailed evaluation, to tune
the electronic control unit (controller) parameters, to check the elevator control system thermal behavior through the
complete flight mission profile and to confirm the compliance of the closed control loop «aircraft – flight control
system – actuation system» parameters with the requirements (stability margins and step responses parameters), as
it’s well known that the actuators characteristics (especially in the region of small control signals) can drastically
influence the stability of an aircraft [3]. In addition, the possibility to provide to the industry the highly detailed and
validated model which can be used within further EMA development programs on the early design stage is obvious.

2. Model Description

The EMA model was developed within the MATLAB/Simulink software with the use of the following principals:
 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95 87

 The model decomposition into sub-models is dictated by the actuator physical structure (see Fig. 2);
 The model should cover different detailing levels of sub-components to fulfill the different simulation
objectives while balancing between the model preciseness and the required integration time-step;
 The principal of maximum abstraction was used while grouping parts of the model into different blocks;
 The model was done completely parameterized with the use of initialization m-file.

Fig. 2. EMA general overall scheme.

Within the project the general scheme shown above was simplified as no reduction stage is used in the actuator
and no mechanical output chain is not simulated, therefore EMA simulated consists of four main parts:
 Controller which gather all the feedback signals from actuator and implements the logical functions
(algorithms including PWM, voltage normalization etc.) to control the actuator power electronics;
 Power electronics based on IGBT-switches which transmit input DC voltage applying necessary voltage
levels on the three phases of the electrical motor;
 Electrical motor based on permanent magnet synchronous motor (PMSM) to transmit to electrical power to
the power of the rotational movement by means of electromagnetic torque;
 Ballscrew to transmit the input rotation of the electrical motor shaft to the linear displacement of the rod.
As the results the upper level of the model is divided into four corresponding components (blocks).

Fig. 3. EMA model structure (upper level).

Table 2. EMA model upper level interfaces.

Input signals
Xin mm Commanded rod position
External linear force acting on the rod («+» corresponds to the force direction pushing the rod inside
Fext N
the actuator, «-» corresponds to the force direction pulling the rod outside the actuator)
Ballscrew efficiency which will change with the temperature calculated in the AMESim thermal model
n -
(with the co-simulation technique)
Output signals
Xout mm Actual rod position
Losses W Power losses for every power part of the EMA: power electronics, PMSM, ballscrew
88 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95

The «Gain = 1» block was inserted into the feedback line for closing/opening the position loop for further
automated analysis of the open control loop frequency response to determine EMA stability margins.
The EMA model has sub-models with different levels of details for every part of the EMA. All of these models
can be used for different purposes taking into account its specific problem-oriented features. In addition, the models
will require absolutely different integration time-steps which will also influence on the design fields for which these
models can be used. The switching between model types is done by changing the corresponding logical switch.

2.1. Electronic control unit (ECU)

The ECU provides the control logic to form the motor control voltages basing on the feedback signals obtained
from EMA motor sensors and rod position sensor. The control is quite typical for such applications and is realized
by three PID regulators inside position, velocity and current control loops providing space-vector field oriented
PMSM-control. The development of ECU algorithm and hardware implementation was done by Tecnalia company.

Fig. 4. Controller overall scheme.

The control algorithm realized inside the ECU model depends on the motor type (1-phase or 3-phase) used for
simulation: the simplified control algorithm with reduced PID regulators is used for 1-phase BLDC-motor while
Enhanced controller is used for 3-phase motor.

2.2. Power electronics

Power electronics block with two different sub-models provides one of the following functions:
 The direct transmission of control voltage, calculated by ECU, to electrical motor windings without
simulation of power electronics elements;
 PWM of control voltage and integration of blocks, from that full invertor model was composed in according
to electrical scheme.
The first sub-model can be used while simulation 1-phase motor or 3-phase motor when no power electronics
integration is needed. The second one can be used only with 3-phase motor.

2.3. Electrical motor

The electrical motor model is represented by two sub-models: 1-phase and 3-phase motors. The widely known
electrical equations [4, 5] describing motor performance depends on the motor type chosen by the user. The practice
shows that 1-phase motor with equivalent electrical parameters can represent the main dynamic mechanical
properties of 3-phase motor in many cases when studying respectively low frequency band (up to 30 Hz) – the
bandwidth of aircraft flight surface control.

2.4. Ballscrew

The ballscrew model provides the following functions:

 Transformation of the rotational movement of the motor shaft into linear displacement of the actuator rod.
This is done by two sub-models implementation: simplified and complete reductor;
 Calculation of the friction forces in the ballscrew, such as seal friction force and thrust bearings dynamic
friction (thrust bearings viscous friction, radial bearings friction torque are summed in «Motor» model);
 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95 89

 For the complete reductor model two frictions models are available (simplified/complete).

Fig. 5. The general view of the ballscrew reducer.

2.4.1. Simplified reductor (One-mass system)

The simplified ballscrew model performs the described above functions taking into account the following
assumptions:
 The ballscrew have infinite linear and rotational stiffness. So the rotation of the motor shaft directly
transformed into linear displacement of the rod:

p
X  (1)
2

 The ballscrew backlash influences on the system behavior only as hysteresis in the displacement of the
actuator rod and can be even excluded from the simulation in some cases;
 all active and passive forces inside the ballscrew are referred to the motor shaft when using the simplified
reductor model.

Fig. 6. The schematic representation of the simplified ballscrew.

J em - motor shaft inertia, kg·m2;

J stiff - inertia of the ballscrew rotating parts stiffly attached to the motor shaft, kg·m2;
BL - backlash, mm;
M ext - mass of linear moving parts, kg.

Fig. 7. The schematic representation of the simplified ballscrew.

As shown the backlash is inserted only into the signal line of rod displacement calculation but at the same time
the forces/moments calculation (ballscrew input torque Tin – the reaction of the ballscrew to the motor motion) is
not influenced by the EMA being in the backlash zone thus introducing errors in actuator behavior in comparison to
the real one. Nevertheless the value of the backlash is quite small (less 0.02mm) thus it can be taken into account in
such a way for the simplified EMA model.
90 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95

2.4.2. Complete reductor (Two-mass system)

The complete ballscrew model performs the described in 2.3 functions taking into account:
 The limited value of ballscrew linear stiffness;
 The backlash of the ballscrew;
 Correct calculation of friction influence (viscous friction and radial bearings friction torque are referred to the
motor shaft and are calculated in «electrical motor» model. Seal friction force and thrust bearings friction
torque are calculated after the backlash in the «ballscrew» model);
 Correct inertia calculation (The inertia of the motor shaft and rotating parts are referred to the motor shaft
and are calculated in «electrical motor» model. Mass of linear movement parts is used in calculation after
backlash in «ballscrew» model).

Fig. 8. The schematic representation of the complete ballscrew model.

As it seen from the scheme above the only difference of the complete ballscrew model is that it’s assumed as
two-mass system due to the presence of the backlash in the mechanical motion (and forces) transmission line which
leads to the necessity to take the linear stiffness C [N/m] into account. The cross-relation of the «complete reductor»
model’s sub-elements can be observed in the figure 9.

Fig. 9. «Complete ballscrew» model overall structure (Upper level).

Backlash/Elasticity model
The model implements the equations that connect linear movement of motor shaft and linear movement of output
shaft with force, caused by movement of output shaft, taking into account backlash:
 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95 91

 Fin  sign( xin  xout )( xin  xout  B / 2 1000)  C, xin  xout  B / 2  1000
 (2)
 Fin  0, xin  xout  B / 2 1000

xin- the linear movement of ballscrew part before the backlash from the motor shaft side), m;
xout -linear movement of output shaft, m;
B - backlash, mm;
C - linear stiffness, N/m;
Fin - ballscrew inner force, caused by compressibility and the difference between input and output movement, N.

Inertia model
The block is used to calculate the resulting linear movement of output shaft (and the speed) under the summary
force calculated in «Friction» sub-model described further. The 2nd Newton’s law is used for the linear movement.

Fig. 10. (a) «Backlash/Elasticity» model; (b) «Inertia» model.

Simplified friction model

Both friction models (simplified and enhanced) take into account the following physical phenomena with the
respect to the EMA data available prior to the simulation activity and its objectives:
 Stribeck effect [6];
 Viscous friction;
 Coulomb friction.
But the programmatic implementation of the numerical calculation of these effects are different. The simplified
model calculates the summary force Fsum, which acts on the actuator output part from the ballscrew with the use of
the following equation taking into account force Fin, caused by the movement of input shaft, external force Fext,
seal friction force Fseal and thrust bearings dynamic friction Fbear:

  2  
Fsum  Fin   Fext  Fbear  Fseal   Fin   Fext  sign V   Fext  Cbear  Fseal   (3)
  p 

The equation takes into account the direction of the friction forces, which always counteracts the motion being
reactive forces.

Fig. 11. Simplified friction model.

92 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95

The simplified friction model correctly represents the mechanical behavior of the EMA only in the «moving»
state – when motor shaft speed is not equal to zero and the summary forces have already overcome the striction
force. In addition the use of «sign» function leads to unstable numerical solution in the region of very small speeds.
In opposite the «enhanced friction» model should be used for correct calculation of the «starting» state (the
transition from zero to non-zero speed) and allows to correctly simulate EMA at speeds close to zero-region.
Enhanced friction model [7]
The model calculates the summary force Fsum, which acts on the actuator output part from the ballscrew with the
use of the following equation:

Fsum  Fin  Fst _ res (4)

 F  Fr  sign(V ) if V  0
Fst _ res   ext (5)
 f ( Fin , Fstart ) if V  0

 F , Fin  Fext  Fstart

f ( Fin , Fstart )   in (6)
 Fext  Fstart  sign( Fem  Fext ), Fin  Fext  Fstart

Fst _ res -the resultant force of the static resistance.

Fr - the modulus of the static resistant reactive forces which act during the moving of the mechanism:

2 
Fr  Fseal  Fbear  Fseal  Fext  Cbear (7)
p

Fstart  1,3Fr (8)

Fstart - the modulus of the static resistant reactive forces which act while mechanism starting.
The calculation algorithm is shown in the figure 13. Its right path determines the value and the direction of the
reactive (counteracts the moving) component of the static resistant force while the rod moving. Its left path
determines the value and the direction of the reactive component of the static resistant force in case the rod is not
moving. In this case this component compensates the action of active electromagnetic and external forces acting on
the rod. The middle path of the algorithm describes the motor shaft behavior when the mechanism starting. This
algorithm allows taking into account the difference of the friction force in different states. The strict moving
condition (V≠0) and rest condition (V=0) are replaced by the non-strict conditions (|V|>ԑ) to provide the solution
stability (ε – the small positive value, e.g. ε=0,0001).
The use of the algorithm gives the ability to correctly take into account non-linearity of friction and non-elastic
deformation in both states: moving and resting including the condition of starting and stopping the mechanism. As
the result the «enhanced friction» model provides more accurate simulation results of the cyclic movements with
low control amplitude. Also it gives the ability to use bigger integration time-step (10-4 vs. 10-5 s) with the same
level of «hunting» oscillations [8] (due to the use of integrative position control) in comparison to scenario when
sign-function approximation with tanh-function or Fermi-function is used with «simplified» friction model.
 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95 93

Fig. 12. The algorithm of forces calculation.

Fig. 13. «Enhanced friction» model.

Table 3. «Enhanced friction» model interfaces.

Input signals
V m/s The speed of the output shaft
Fin N Force, caused by movement of output shaft
Fext N The linear external force acting on the actuator rod (output part of the ballscrew)
η The efficiency of the screw-nut joint of the ballscrew
Output signals
Fsum N The force acting on the motor shaft from the ballsrew as the result of active (external) and reactive (internal) forces

3. Simulation results

One of the most critical characteristic of the actuator for the primary flight control system of any aircraft is its
bode diagrams or amplitude and phase frequency response. The comparison of EMA frequency response when using
two different types of reductor models is shown in the figure 15.
94 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95

Fig. 14. Frequency response with (a) control amplitude variation (b) external load variation.

The bode diagrams for the different input amplitudes at no-load case are shown in the Fig. 16 (left), there’re are
diagrams for the different values of external load at fixed input amplitude 0.5%Ain=0.375 mm (see Fig. 16, right).

Fig. 15. Frequency response with (a) control amplitude variation (b) external load variation.
 L. Bilyaletdinova and A. Steblinkin / Procedia Engineering 176 (2017) 85 – 95 95

It can be seen that bode magnitude diagrams remain at the same level in the bandwidth 0.1…6 Hz while the
amplitude of control signals varies from 0.5% to 5% Amax and the actuator reaches the speed limit after 2.5Hz
under 10% Amax control signal. Also the actuator’s great stiffness to the external load can be observed – the phase
drop under external load is less than 10deg in the control bandwidth (up to 5Hz). Within this frequency range the
simplified friction model can be used as it’s seen from the Fig. 15.

4. Further works

It should be noted that the current version of the model was developed with the focus on the actuator mechanical
part while keeping power electronics out of precise look, so the next step of its development will be the complete
realization of power electronics based IGBT-switches bridge controlled by PWM as well as the power supply line
with corresponding voltage drops. Such model configuration has been already tested giving the comparable
simulation results (in the bandwidth <30 Hz) but not integrated into the project workflow model. In addition, the
adaptation of the EMA Simulink model for co-simulation with SABER software is planned.
And of cause the EMA model validation based on the specially prepared by project partners validation plan will
be the logical conclusion of the simulation activities. The process will be done through technology validation
campaign in the frame of RESEARCH project to be held on CESA and TsAGI facilities.

Acknowledgements

The authors wish to thank all the RESEARCH project partners for their participation and support. The authors
would also like to express their gratitude to the European Commission and Russian Ministry of Trade and Industry
for their support to the RESEARCH project.

References

[1] J.-C. Derrien, Electromechanical Actuator (EMA) Advanced Technologies for Flight Controls, ICAS 2012 conference proceedings.
[2] A. Alexeenkov [et. al.], New Developments of Electrically Powered Electrohydraulic and Electromechanical Actuators for the More Electric
Aircraft, ICAS 2014 conference proceedings.
[3] S.V. Konstantinov [et. al.], Elecrtrogydravlicheskie rulevye privody s adaptivnym upraqvleniem manevrennykh samoletov (Electrohydraulic
control actuators with adaptive control for maneuverable aircrafts), St.-Peterburg, LETI, 2011, 513 p (in Russian).
[4] Y. Fu [et. al.], Nonlinear modelling and System Analysis of the Linear Electromechanical Actuators, Journal of Computational Information
Systems 11:6 (2015) p. 1983-1995.
[5] D. Woodburn [et. al.], Dynamic Heat Generation Modelling of High Perfomance Electromechanical Actuator, 48th AIAA Aerospace
Sciences Meeting Including the New Horizons Forum and Aerospace Exposition, 4-7 January 2010, Orlando, Florida, AIAA 2010-290.
[6] Andersson S., Soderberg A., Bjorklund S. Friction models for sliding dry, boundary and mixed lubricated contacts // Tribology International
№40. – 2007. – p.580-587.
[7] R. Shreiner, Modelirovanie momentov nagruzki electrodvigateley v MATLAB (The simulation of load torques of electrical machines using
MATLAB), Moscow, Molodoy ucheniy, 2010, p.6-12. (in Russian).
[8] Canudas C., Olsson H., Astrom K., Lischinsky P. A new model for Control of Systems with Friction, IEEE Transactions on Automatic
Control, vol. 40, №3. March 1995. p.419-425.