High-Speed Induction Motor Drives: Starting Operation Control

Copyri ght © IFAC Control in Power Electronics a nd Elect rica l Drives. Lausanne. Sw itzerla nd . 1983

HIGH-SPEED INDUCTION MOTOR DRIVES: STARTING OPERATION CONTROL P . Ferraris, M. Lazzari and F. Villata Dlpartz'mento di Elettroteeniea, Fa eolt a dz'Ingegn eria, Politeenieo dl' Torino, Italy

Abstract. The problem here analysed concerns the drives providing the use of high speed induction motors. There is an increasing tendency to use this type of motors (having rated frequencies comprised between some hundreds and some thousands hertz) for those applications requiring the utilization of grinding and milling machines. In such cases the supply can be constituted by a square wave static inverter and closedloop speed regulation is often not necessary. For such reasons and other discussed in the paper, the starting of these motors is normally realized through the adoption of an open-loop ramp in the supply voltage and frequency. During the starting, transient current overloads may occur caused by a too fast increasing of the supply voltage and frequency; that may be even more verified, owing to the wide spread of the parameters concerning the motors that a given inverter is required to drive; of particular importance is often also the friction torque due to the bearings. The starting transient is considered as the more significant among all the other situations for which a variation in the shaft speed is necessary. The current overloads may cause different types of trouble: for instance the intervention of the maximum current protection, which is often present in such a type of drives. In the paper the starting transient are studied through a suitable dynamic model of the drive and a simple method for altering the starting characteristics of the system is proposed and analysed. The method is based on the alteration of the frequency variation rate, when the motor current exceed a given limit. Such adaptive control, for which a motor current feedback has to be provided, allows to increase the starting time only when it is necessary, lea ving the speed transient conform to the present ramp in the other cases. The prediction of the quantitative effect of the technique are very interesting; experimental and simulation results allow to better clarify the terms of the problem and the obtainable results. Keywords. Mach ine tools; electric drives; a.c.motors; acceleration control; adaptive systems.

INTRODUCTION

related to the shaft bearings performances (ball bearings, magnetic bearings, fluid bearings, etc.) and to the thermal problem solutions.

In the industrial processes, mainly for preCIsion mechanical production, the use of high speed motors is becoming a fact in rapid evolution, assuming sometimes the role of important factor for the global technical and economical result; that is especially concerning grinding and milling machines but other application fields are rapidly assuming concretenes (special pumps, textile processes, etc.). The actual utilization increment rate is mainly concerning the use of special induction motor, also if the adoption of other motor types may be easily predicted in the next years.

The thermal problem, very often solved through a devoted refrigeration fluid circuit, is hea vily affected by the extremely high ratio between the handled power and the active electromagnetic materials volume. factor aggravating the difficulties is A constituted by the limitation in the internal temperature, owing to mechanical expansion problems and to effects on the bearings life. That makes not only of great importance the entire thermal problem, but makes also necessary to provide very fast monitoring and protective systems (F erraris, P.; M. Lazzari and F. Villata 1981); it may be remembered that a temperature increasing of about one hundred degrees per second may happen under particular overload conditions, with heavy drawbacks. In such conditions, fast monitoring and protective devices

The actual practical upper limits are of about 3 kHz for the supply frequency and of about 100kW for the handled power; very refined technical solutions have to be adopted for both the electromagnetical parts and the machanical components; at the present aim, it is important to underline the particular realizing difficulties

353

354

P. Ferraris, M. Lazza ri a nd F. Vill a t a

must be introduced which have to statically switch-off the system on the basis of the motor current analysis; for the case of the inverter fed motors, which represent the natural solution, the protecti ',le technique based on the input inverter current limitation, has to be avoided. All that brings to a global technical approach for which, for a good system protection, overload current conditions have to be avoided during normal transient phases of the working cycle, like, for example, the starting and braking operations. From such point of view, a ramp variation of the speed has to be considered of normal adoption; but presently, the necessity of reducing the dead time of a working cycle and the increasing of the project activity concerning automatic multi-spindles machines makes necessary to reduce, as much as possible, the ramps duration; it must be also noted that a gi ven inverter is very often used for dri ',ling a set of different motors, in the same working cycle too. From the other hand an inferior limit to the ramp duration is imposed not only for thermal reasons but for mechanical motivs depending on the bearings too. For a given motor, or motors family, a minumum starting or braking duration may be then defined; such duration may result not compatible with the presence of the protective devices, introducing a potential cause of breaking in the working process. It has to be noted that a high speed spindle, realized through a high frequency motor, is more and more frequently treated as a normal tool in machining centres, providing the use of different tools for different working phases; the system government is performed by computers having not only the task of selecting the spindle, but also the one of changing, during the cycle, the preset shaft speed, normally through steps in the setting signal. At such regards it has also to be observed that the mechanical losses, notwithstanding the quality of the special bearings, are particularly high with respect to the one concerning the normal motors. In addition, they may be heavily variable depending on different factors, like temperature, working age, actual speed, running in phase (the bearings are, in most of the applications of interest for the present investigation, subject to periodical substitution), etc •• That obliges to take into account a parameter whose variability influences the way of interfering with the system's protection, and that may result also of the same magnitude order of the rated motor torque. The problem is then of obtaining from the supply converters a behaviour which should result adapti ve with respect to the mentioned variability in the working conditions. The problem of an adaptive optimization of the control strategies, concerning the motor drives under examination, is very large, and it is of particular interest for the efficiency of the future generation of automatic machining centres; in this paper the problem is limited to the aspects concerning the working speed variations, by considering mainly the starting phase which under

such a point of view may represent the worst situation. PROBLEM ST A TEMENTS The more suitable supply system for the considered motors is constituted by six-steps inverters, with the use, in constant expansion, of power transistors; the adoption of a waveform of square wave type represents in practice the only one of general interest, for some different reasons: apart from some particular cases, it is impossible to adopt, with satisfying results, P.W.M. modulation techniques useful for obtaining fundamental frequencies of some kHz; also when a modulation approach is technically possible, other reasons related to the losses, both in the inverter and in the motor, may reduce the amount of the advantages of the solution; the high speed of the rotor, also at the inferior extreme of the working speed interval, makes neglectable the disturbances caused by the alternati ',le torques. The system regulation is then obtained by varying in opportune way the commutation frequency and the input DC voltage of the inverter. The speed regulation is normally obtained without esternal speed feedback loops. The reasons of such a choice are: the speed value precision is not of great importance for the process results; the obtainable speed constancy and precIsion is good enough, because of the low nominal slip value of the motors in exam; the adoption of a tachometer system may be very difficult in the most of the cases the working speed range is normally small; control techniques based on the electrical motor state may allow very satisfying performances. Then the starting phase is performed by increasing in a coordinate way the D.C. link voltage and the inverter frequency; the necessity of a soft start for allowing a good bearings beha viour brings to define a minimum value for the starting ramp duration; the problem is then to make the system adaptive in order to alterate as low as possible the preset ramp for limiting the transient slip value and current value within the protection limits, notwithstanding the modification of the system and of its describing parameters, as it was described in the previsions section. F or a better comprehension of the general problem meaning, one may observe the transient reported in the Figs. 1, 2, 3, 4, 6 and 7; they furnish the starting current evolution for two types of motors and for three different preset starting duration. The motors under test have the following rated characteristics

Hi gh- s pe ed Induc tion Mo tor Driv es

355

Motor A

Pn Vn In fn poles pair number

= 5.5 kW = 350 Volts = 15 A = 600 Hz =1

Motor B

Pn Vn In fn poles pair number

= 7.5 kW = 350 Volts

= 20 A = 1000 Hz

=2

Fig. 3.

Phase current: 10 A/div Time: 100 msec/div.

Motor A starting 2 sec

Fig. 1.

with

a

preset

duration

of

Phase current: 10A/div Time: 100 msec/div.

Motor A starting 10 sec

with

a

preset

duration

of

Fig. 4.

Phase current: 20A/div Time: 100 msec/div.

Motor B

starting 10 sec

with

a

preset

duration

of

The transient current waveform, and the related maximum value is, obviously, depending on the selected preset ramp type. PROPOSED METHOO DESCRIPTION

Fig. 2.

Phase current: 10A/div Time: 100 msec/div.

Motor A starting 5 sec

with

a

preset

duration

of

The method under exam, as it has been already mentioned, is based on the assumption of altering the preset ramp which controls the input speed signal, by using in an opportune way the motor current measurement. Being the ramp obtained through the integration of signal Vi, up to the preset level corresponding to the preset speed, a signal obtained by measuring the motor current is added with the opportune sign, in order to obtain a ramp slop decrement when the current value

P. Ferraris, M. Lazzari and F. Villata

356

exceed a preset value.


The global system regulation philosophy is described in the Fig. 5; the current transducer is obtained through a system of amperometer transformers supplying a rectifying bridge; the feed-back signal Vf is obtained by loading the rectifier with a shunt resistor and by filtering it through a suitable low-pass circuit.


= Lrird

+

Misd


=Lri rq

+

Misq

(3)

The introduced quantities have the following meaning:

Fig. S. In a summing block the reference value Vr and Vf are opportunely summed up and the result is assumed for realizing the signal to be integrated in order to have at disposal the reference frequency and voltage signal. The obtained "current" value Vf may be approximatively retained representing the maximum value of the motor first harmonic current. Let us consider, for simplicity, the case of speed value increments; if the gain G is introduced conforming to the equation:

with the condition: 0

~

Vi

~

Vi

s4>sd

+

s
Vsq = VfVSN - VfVTN the phase voltages VRN, VSN, VTN are calculated in the following way, with reference to the described feed-back method:

Vi with

=MAX (I iR I, I is I, I iT I )

(S)

Vf imax =---+-RfCf Cf

(6)

= Vr - G(V f - Vr)

O~Vi~Vr

sW

F or testing the obtainable results, the simulation of the proposed system has been carried out; for the motor model the Eqs. (Z) and (3) were adopted, concerning a variables transformation obtained by using two reference axes d and q fixed with respect to the stator frame. +

(4)

imax

MA THEMA TlCAL MODEL

= Rsisd Vsq = Rsisq

Vsd = VTVRN - VfVSN - VfVTN

o

Vr has to be choosen in order to correspond to the upper limit for the maximum value of the first harmonic motor current. Normally such an upper limit may be considered, for the motors under exam not exceeding 1.S times the rated value. The described procedure may be, of course, adopted in order to allow to control both increasing and decreasing reference speed signals.

Vsd

The equivalent transformed supply voltages Vsd and Vsq are obtained by manipulating the real motor phase voltages VRN, VSN, VTN through the following transformation:

(8)

s6 Where wand e are respectively the electric pulsation and the electric angle of the rotating field. The potentials VR, VS, VT of the inverter outputs are expressed as it follows: = Edc SQW (6)

o

= Rrird + S
W r
o

= Rri rq + s
W r
(Z)

= EdcSQW(6 - Zn /3)

= EdcSQW(6 - 4n /3) = (1/3)(VR + Vs + VT)

(9)

High-spe ed Induction Motor Drives Where Edc is the DC link inverter voltage, variable with wand the SQw(e) operator is defined to be: SQW SQW

= 1 for

= 0 for

0 < e ~ 1T 1T < e s. 21T

with periodicity 21T and then: VRN

= VR - VN

VSN

= Vs - VN

VTN

= VT

(10)

- VN

357

Also in this case, it is well evident the waveforms modification with respect to the ramps variation. It is very important, also, to observe that many factors of practical character affect the transients, such as the friction evolution, the V-f law at low speed, the magnetic saturation, etc •. The magnetic saturation has been here neglected, because of its small importance on the qualitative results of the proposed method. For what it concerns the real starting evolution of voltage and frequency at very low speed value, it is important to remember that different slopes at Iow frequency are often adopted in order to contemporary obtain: soft electrical transient;

METHOD VERIFICA nON FINAL RESULTS For ver ifying the proposed method a mixed experimental and simulation activity has been developed. The tested motor was the motor B. In the Figs. 6 and 7 two different starting current registrations are reported.

Fig. 6.

Phase current: 20 A/div Time: 100 msec/div.

Motor B

starting with a preset duration of 5 sec

higher voltage value at a gi ven frequency for compensating the stator voltage drop. For what it concerns the frictions, as it has already been remembered, they cannot absolutely be neglected. In the Figs. 8 and 9 the starting current evolution for the considered motor are reported, as obtained by simulating, as well as possible, the strong static friction torque and the particular voltage frequency law. The results, if the difficulties of well identifying the real system working way at very low speed are considered, may be retained to be satisfying.

Fig. 8.

Phase current: 20 A/div Time: 100 msec/div.

Motor B

starting with a preset duration of 5 sec

Fig. 7.

Phase current: 2oA/div Time: 100 msec/div.

Fig. 9.

Phase current: 20 A/div Time: 100 msec/div.

Motor B

starting with a preset duration of 3 sec

Motor B

starting with a preset duration of 3 sec

P. Ferraris, M. Lazzari and F. Villata

358

For example, in the following, simulated results are reported of the starting transient relative to the same motor, when a proportional voltagefrequency law is used and a static friction torque of about 80% of the motor rated torque is supposed present at the start. Figures 10 and 11 represent the behaviours of the motor phase current, of the supply frequency and of the motor speed when an open-loop starting is realized.

3.0r----r--- --r---- - .

i pu

r

Figures 12 and 13 represent the same quantities calculated when the described feed-back technique is used. It may be observed that, in this case, the frequency increasing rate is automatically adapted in order to limit the motor current during the whole starting time; in particular, in this last case, the motor accelerates more rapidly and, consequently a shorter starting time is obtained without exceeding the upper limit imposed to the current.

'-'''--r-\

·.--- · T . - . - - ,

I

1

I

i

2.0i---'-

1.

1. 0 t----:-f-II

0.

-1. 0

t ' -- _... .

t-f

,

0.0

.2

•1

,

: f

I I J.,____ .I___._.._LS_~ I

I

-3. 0 L - . . - - - - I ______

._----_. f ,

-2. 0 ' - - - ' - '!, --.---1

.3

I

.4

.5

-3.0 ____ •1 0.0

._...L-_ _ _ _ _.....

Fig. 10.

.2

--'--------..... sec. .4 .5 •3

Fig. 12.

'! - -1 , I

.6Ffes---I---r --T i

·6! p.u. v31,es-'1---T--

.5r---!

· 5t--· -:·----~- - -·- ·· · ' -· ·- - -- -.-I' -- - - j ~ i i ! I

ft

::

j.--t-+-i

.4~t-_-

f · I!

;. • 3:.

t

~

.2t

JI

'I

!I

1-1------·[ ·

ii !

I il ;' lI~ ·--·.---- --·:-----·1-

I

•

, I

I ' :I

i

I

•1

i

I

I

r

~

I

~--

0.0~__~__~_c.~~~~~s~e~c.~ 0.0 .1 .2 .3 .4 .5 Fig. 11.

.3

~~

~

i, •

I

4_~ _.L-·_-tl -------L--1!

I

I!

.

I

I

i

'

I

!

1 ,

; i

,

!I I

'

-+---.+.---- . 1"--. :

I

I

,

.

;

I

f

2~ fre~~llc.L.L H ___1t._ ! ; ..

I

i

~

'-i __ l. .. __ $P.

i

I : - ----r-~

, i

I

I

I I

ed .__ -+_~

0.0~__~~~~~~~~~s~ec~.~ 0.0 .1 .2 .3 .4 .5 Fig. 13.

High-speed Induction Motor Drives The reported result6, notwithstanding the practical difficulties of analysing systems which are at the boundary of the actual technological possibilities, let us to deduce: the problem of the current transients at the starting, or when a step in the reference speed value is applied, constitutes a problem which may affect, or the duration of the dead times in a working cycle, or the possibility of well protecting the motor; the proposed method may be considered a good compromis between simplicity and quality of the obtainable adaptive transient current controls; due to the realization of new machine-tool systems with particular working cycle, other overcurrents might happen in some phases (as, for example, at the mechanical load application); that makes probably necessary to

359

improve the activity of the adaptive control methodologies concerning the drives family here considered. ACKNOWLEDGEMENT This work was partially supported by the M.P.I. (Ministry of Education); its development was carried out in the ambit of the research unit of Torino of the G.M.E. of the National Research Council REFERENCES Ferraris,P.,M.Lazzari and F.Villata (1981). Investigation on high frequency motors with reference to overload protection. Symposium Electrical Machines for S ecial on Purposes. Bologna Italy.