Model of Elevator Drive With Jerk Control

Bojan Kneževi Branko Blanuša

Faculty of Mechanical Engineering Faculty of Electrical Engineering
University in Banja Luka University of Banja Luka
Banja Luka, BiH Banja Luka, BiH
[email protected] [email protected]

Darko Mareti
Faculty of Technical Sciences
University in Novi Sad
Novi Sad, Serbia
[email protected]

Abstract— A strong jerk which can cause passengers discomfort simulation results are given in fifth section. Obtain results are
occurs when electric elevator starts and stops. If uncontrolled, summarized in conclusions.
the jerk have influence on the electrical and mechanical
subsystem performance and reduce the system life expectance. II. MODELING OF POSITION TRAJECTORY WITH JERK
Modern drives have ability to control speed, so we can indirectly CONTROL
control the jerk. This paper describes the control of real gear-less Jerk is defined as first derivation of acceleration. So, if the
electric elevator drive with vector-controlled induction motor.
acceleration have step form, it causes a big jerks which have a
Position trajectory is determined by the need that jerk is changed
negative effect to passengers and on the mechanical system of
by the predefined function. A mathematical model of the drive is
presented, and it is tested through computer simulations.
elevator. Also, the accuracy of the positioning is reduced and
the overshoot and settling time increase. This is the case with
Keywords-electric elevator; jerk control; induction motor; the frequently used trapezoidal form of speed. For this shape of
speed, acceleration has step form. Significant improvement can
I. INTRODUCTION be achieved if the acceleration changes following the smooth
The most common solution for vertical transport of curve. This provides less jerks and eliminate their harmful
passengers and goods in the residential and commercial effects [4].
buildings is an electrical elevator. Its construction is not A case when the jerk in the process of acceleration and
substantially changed since the time when designed the first deceleration has the sinus function is suggested (Fig. 1):
commercial elevators of this type. Control structures, safety
elements, drive endurance and its durability and economy are
­ 2π
constantly improved. Modern passenger elevators request high
° jm sin T t , t ∈ [ 0, T ]
transport speed, low jerk, precise positioning, simple and °°
efficient control and small number of sensors [1]–[3]. j = ®0, t ∈ [T , t2 ] , (1)
Particularly interesting in modern elevators is the comfort and °
convenience of passengers. Today there are elevators which °− jm sin 2π ( t − t2 ) , t ∈ [t2 , t3 ]
reach the nominal speed of 18m/s in order to increase their °̄ T
availability in high office buildings with a high frequency of
traffic. In these elevators there is a strong jerk, when they where t3 is total duration of movement, T is a time of
starting and stopping, which can cause discomfort of the acceleration/deceleration, t2=t3-T is a moment when the object
passengers. Modern control drives can realize movement of the starts deceleration and jm the maximum amplitude of jerk.
elevator with controlled jerk, so the idea of this work is very
Based on (1) the acceleration in particular intervals can be
relevant and justified, especially when the effect of jerk in the
projects of elevators are not usually analyzed. expressed by the following relation a = ³ j dt + Ca , velocity as
Organization of paper is as follows: v = ³adt + Cv and the displacement as d = ³vdt + Cd .
Mathematical model of position trajectory which include Constants Ca, Cv and Cd are determined from initial
jerk control is presented in second section. Description of conditions. Acceleration is calculated from the expression for
system configuration, elevator with drives and drive sheave is jerk (1):
presented in third section. Position control of drive is described
in fourth section. Qualitative analyses of described model with

978-1-4577-0746-9/11/$26.00 ©2011 IEEE

 § 2π · j T § 2π ·
a = ³ jm sin ¨ t ¸ dt = − m cos ¨ t ¸ + Ca . (2) 20

© T ¹ 2π © T ¹ position

acceleration [m/s ], jerk [m/s ]

3
position [m], velocity [m/s],
15

Constant Ca is determined from the condition that the

10
acceleration is equal zero in t=0:

2
5

j T § 2π · j T velocity

a = − m cos ¨ ⋅ 0 ¸ + Ca = 0 Ÿ Ca = m . (3)
2π 2π
0
© T ¹ jerk acceleration

0 2 4 6 8 10 12
time [s]
Expression for acceleration in the period t
[0,T] is:
Figure 1. Reference of jerk, acceleration, velocity and position.

jmT jmT § 2π · III. SYSTEM CONFIGURATION

a= − cos ¨ t¸ . (4)
2π 2π © T ¹ The analyzed elevator considers electrical drive and drive
sheave that is mounted directly to the motor shaft without gear.
Constants for all other values are calculated in a similar Block diagram of the system is shown in Fig. 2. Elevator motor
way and finally we obtain expression for the displacement d is three-phase asynchronous motor with squirrel cage induction
(Fig. 1): motor (AM). Drive converter is current regulated PWM
voltage source inverter (CRPWM VSI) direct current power
supply. Indirect vector control algorithm (or Indirect Field
­ T 2 T3 § 2π · Oriented Control, IFOC) where d, q rotational system is
° jm t − jm 3 ¨1 − cos t , t ∈ [ 0, T ]
° 4π 8 π © T ¸¹ oriented toward rotor flux Ψr* is used for a control of drive.
° T2 Rotational movement of the motor shaft turns into linear
°° jm ( 2t − T ) , t ∈ [T , t2 ] movement over drive sheave directly mounted to motor shaft
d = ® 4π .(5) without gear box. The drive torque is transferred from the drive
° T2 ª ( t − t2 ) T ª
2
2π º
( t − t2 )»º » ,
sheave by friction to the hoisting ropes that are connected to
° jm «( 2t − T ) − + 2 «
1 − cos
the car as counterweight. Weight of counterweight is mcw [kg]
° 4𠫬 T 2π ¬ T ¼ »¼
° (Fig. 3).
°̄ t ∈ [ t2 , t3 ]
Drive sheave rotational speed is the same as rotational
speed of motor shaft. Rotational and linear speed of car are
The changes of j, a, v and d are presented graphically in Fig.
related over radius of drive sheave. Load torque in the function
1. The total displacement can be determined form (5) when of weight in the car can be calculated following the next
t = t3 as: expression:

T2 du § m ·
d ( t3 ) = d m = jm t2 . (6) TL = g¨m− m ¸, (9)
2π 2 © 2 ¹

Following the described procedure, we get velocity in where du [m] is diameter of sheave, m [kg] is a weight in
steady state which is known as maximum value of velocity vm. car and mm [kg] is elevator capacity. The total inertia is
From the expression for vm interval T can be calculated as calculated as:
follows:

T2 2π ⋅ vm du
vm = jm ŸT = . (7)
2π jm

Maximum value of displacements dm is known, so from (6)

time t2 is:
mcw
2π ⋅ d m
t2 = . (8)
T 2 jm m
+c
m
If the condition t3 > 2T (or t2 > T ) is not satisfied velocity
vm has to be reduced. Figure 3. Drive sheave, hoisting ropes, car and counterweight.
 Figure 2. Block diagram of the elevator system with vector controlled induction motor and drive sheave that is mounted directly to the motor shaft
without gear.

2
ignored. Kinetic and static (Coulomb's) friction is modeled as a
§ m ·§ d · disturbance. Kinetic friction is modeled with step function and
J m = J r + ¨ 2mc + m + m ¸ ¨ u ¸ , (10) the static friction is modeled with pulse whose duration is short
© 2 ¹© 2 ¹
at startup.
where Jr is inertia of rotating masses (rotor, brake, sheave, Usually, input disturbances are small, PID controller
etc.) and mc [kg] is a weight of car. operates in linear mode and system limits ωmax (maximum
angular speed) and Tmax (maximum torque) are not reached. In
Position sensor is mounted on the motor and serves to
achieve the position feedback. a case when ΔθΔθmax or ω ωmax torque and/or velocity
reaches the limits and the system operates in nonlinear mode.
IV. POSITION CONTROLLER DESIGN So, it is necessary to define limits Δθmax and ωmax and obtain
linear working mode and aperiodic response without overshot.
Elevator is the positioning system and position controller is
Following procedure is described in [7] and values for
used for the task to provide position reference tracking and
zero error in steady state. To solve the problem of trajectory *
tracking application in elevators different types of position q + +
Tem
controllers are implemented, neural networks PID [5], variable
KP
+ -
structure control (VSC) [6], nonlinear control, etc. Constant
load is usual for an elevator. So, a position controller with
proportional, derivative and integral action (PID) is can be used z
(Fig. 4). It also provides the system with the capability of KI
z-1
tracking the constant ramp profile without an error.
Positioning controller synthesis is carried out for the load
with modeled inertia as in (10). In the case of gearless elevator q KD z-1
drive, which is the subject of interest in this paper, the drive z
motor is running at low revolutions. Therefore, a member in a
transfer function of load, who depends of the speed, can be Figure 4. Position controller.
maximum angular speed and maximum position error are given
respectively (T is a sample time): 20

K I Δθ 15

ωmax ≈ , (11)

position [m]
K PT 10

2
2T § K T · 5

Δθmax = max ¨ P ¸ . (12)

Jm © KI ¹ 0
reference
actually
0 2 4 6 8 10 12
time [s]
In this case very simple position controller can be used. The Figure 5. Actually position and reference position of the car.
proportional and integral action is located in the direct path,
and it processes the tracking error. The derivative action is 2.5

placed in the feedback path (Fig. 4). With the derivative action 2

located in the direct path, the closed-loop system transfer 1.5

function has one real zero.. Relocation of the derivative action 1

position error [mm]

0.5
of the position controller into the feedback path eliminates the
0
finite zeros end overshoot. The parameter setting of the
−0.5
position controller is discussed and determined in [7], with the −1
aim of achieving the fastest strictly aperiodic response which −1.5
arriving the target position without an overshoot. Jerk, torque, −2
speed, and position have smooth transient response under given −2.5
0 2 4 6 8 10 12
limits (sec. V), so the undesirable speed and torque oscillation time [s]

are avoided. Figure 6. Difference between reference and actually position (mm).

V. SIMULATION RESULTS
2
Model of the system (Fig. 2) is verified through the
computer simulations using the software package 1.5
MATLAB/Simulink.
velocity [m/s]

The reference position whose mathematical model is 1

described in section II is generated by S-function. Components

of elevator drive and its control are described in sections III 0.5

and IV. The results are shown in figs. 5., 6., 7. and 8. Input reference
parameters for the dynamic of the elevator are: 0 acyually
0 2 4 6 8 10 12
time [s]
--Final position of car: 20 [m],
Figure 7. Actually velocity and reference velocity of the car.
--Nominal speed of car: 2 [m/s],
--Jerk amplitude: 4 [m/s3], 4
reference
--Elevator capacity: 300 [kg], 3 actually

--Weight of the car: 200 [kg]. 2

Fig. 5. shows the reference position (θ*) which leads to the 1

jerk [m/s3]

input of the positioning controller and the actual position (θ). It 0

is obvious that the elevator car reaches the desired position −1

without overshoot following the given reference in the whole −2

trajectory. −3

Difference between the reference and actual position is −4

0 2 4 6 8 10 12
expressed in millimeters and presented in Fig. 6. It can be seen time [s]

that only minor deviations are in the tracking of reference. Figure 8. Actually jerk and reference jerk of the car.
Also, steady-state error is practically zero. VI. CONCLUSION
Fig. 7 shows actually velocity and reference velocity of the The paper describes design of simple elevator drive
car. Reference jerk and actual jerk in the system are shown in controller which provides tracking trajectory, so the jerk is
Fig. 8. Actual jerk has smooth shape, there are no sharp peaks changed at a predefined function. Parameters of real elevator
and its maximum value is under predefined values. Jerk with induction motor drive were included in the model of
deviation from the reference is only for a short interval It has elevator. Simulation of the entire system with designed
small amplitude and smooth shape so its influence to controller is made in the Matlab/Simulink and given results
passenger’s comfort and elevator mechanical system is show that design controller meets the requirements completely,
negligible.
smooth and precise position and speed coontrol and defined
shape and value of jerk are realized (Fig. 5-8)).
Position reference is S-shaped functionn (Fig. 5), so less
torque strokes and less tracking error were obtained.
o Reduced
peak values and smooth shape of electrromagnetic torque
causing lower influence to mechanical subsyystem and suppress
problems like mechanical resonance sustainable oscillations,
etc.
Elevator drive is realized as vector coontrolled induction
motor drive but similar control algorithhm with position
controller can be applied for other types of elevator drive
permanent magnet synchronous motor (PMSM M) drives, etc.
Experimental verification of the results presented in this
paper, are in progress (Fig. 9). The goal is to validate computer
simulations results with measurements on the experimental
prototype.
REFERENCES
[1] Zhang Yajun, Chen Long, Fan Lingyan, “A Design of Elevator
Positioning Control System Model,” IEEE Innt. Conference Neural
Networks & Signal Processing, Zhenjiang, Chinaa, June 8-10. 2008, pp
535-538.
[2] Kukarni, A.B., Nguyen, H., Gaudet, E.W., “A commparative evaluation of
fine regenerative and nonregenerative vector conntrolled drives for AC
gearless elevators,” IEEE Industry Applications Conference, 2002, pp
535-538.
[3] Gulde R., Weeks M., “A position control system design,”
d The 3rd IEEE
International Workshop on System-on-chip for Reeal-Time Applications, Figure 9. Prototype for expeerimental verification of model.
2003, pp 400-405.
[4] H. Z. Li, Z. M. Gong, W. Lin and T. Lippa, “Motiion profile planning for
reduced jerk and vibration residuals,” SIMTech technical reports, Vol. [6] Patxi Alkorta Egiguren, Oscaar Barambones Caramazana, “Robust
8 Number 1 Jan. - Mar. 2007, pp 32-37. Position Control of Induction Motor Drives,” IEEE International
[5] Li Chunwen, Cao Lingzhi, Zhang Aifang, “A Application of Neural Industrial Electronics (ISIE), 20110, pp. 1468-1473.
Network PID Controller to Elevator Control System,”
Sy 29th Chinese [7] Slobodan N. Vukosavi, Digital Control
C of Electrical Drives. Springer,
Control Conference (CCC), Beijing, . 2010, pp 71-74. New York, 2007, pp. 214–217.