Professional Documents
Culture Documents
net/publication/282201427
CITATIONS READS
7 2,449
2 authors:
All content following this page was uploaded by Zdenek Cerovsky on 09 January 2019.
Self reactance Xd in direct axis ’d’ and self reactance Xq in quadrature axis ’q’ influence the load characteristics
of the permanent magnet machine. Their calculation methods are known. All calculations should be verified by
measurements. Measurement with AC variable voltage and frequency source applied on stator winding is correct
when eddy currents do not exist. Other methods evaluate the DC current decay after initial direct current was applied
on the stator winding. Accuracy of these methods is lowered because of the numerical evaluation of current decay
curve. Another method performs the measurement on the machine loaded as a motor. The load angle β is measured
by a machine set of four machines. The accuracy is lowered by higher harmonics induced in no load in the helping
synchronous machine. The load angle can be also obtained by geometrical construction of the vector diagram. In
this paper the evaluation of the generator steady state regime of the machine is used. The generator works with
different loads consisting of capacitance, inductance or arbitrary impedance or resistance. When the generator
transfers from no load to load regime the rotor is shifting and the load angle changes. This angle is measured.
The newly proposed measurement methods are simple in special cases of loads. Results of measurements and their
experimental verification are published in the paper.
Key words: Electric machines, Parameter identification, Permanent magnet machinery, Synchronous machines,
Variable speed drive
Ključne riječi: Električni strojevi, identifikacija parametara, stroj s permanentnim magnetima, sinkroni stroj, elek-
tromotorni pogon promjenjive brzine
1 INTRODUCTION field lies also in electric vehicle drives [9] and aircraft [10].
Permanent magnet synchronous machines have influenced
Progress in power electronic converters brought radi- the usage of electronic converters on the other hand too
cal change of synchronous machine usage area. They op- [11–13].
erated with constant frequency usually [1, 2]. Synchronous
machines supplied by electronic converters are first of all Synchronous machine usage area caused construction
widely used as permanent magnet synchronous machine modifications according to different technologies and areas
(PMSM) in variable speed drives [3–8]. New application where machines are used. Other changes in construction
were caused by permanent magnet development [14]. Self can be used also for machines where eddy currents in
reactance Xd and Xq depend not only on machine con- the ’dq’ axes exist and cannot be omitted.
struction but also on the permanent magnet material qual- In this paper the evaluation of steady state of the
ity. Classical salient-pole synchronous machines with rotor machine working as generator is used. The generator is
winding for field current have Xd > Xq and their ratio is loaded with capacitance or inductance when Ld induc-
usually tance is measured. The generator is loaded with arbitrary
Xd
∈ h1.3; 2.5i. impedance or resistance when Lq inductance is measured.
Xq This load causes that the torque angle (load angle) β oc-
This proportion changes in PMSM sharply. Many curs. When the machine transfers from no load to load
PMSM have Xd and Xq nearly the same. But indispens- regime the rotor is shifting and the torque angle changes
able number of PMSM has the ratio from zero to β. This angle can be measured by the posi-
tion sensor or by the stroboscope. The vector diagram in
Xd steady state is depicted in Fig. 1. The angle β is the an-
∈ h0.5; 1.5i.
Xq gle between stator voltage U1 and voltage Ub induced by
permanent magnet flux.
That means that Xd < Xq is also possible [8, 11, 15–
17]. The calculation methods of direct axis reactance Xd To construct the diagram following values must be
and quadrature axis reactance Xq are known. Also the fi- known:
nite element method is possible. All calculations surely
R1 stator resistance
contain uncertainty caused by nonlinear iron features, ge-
ometrical complexity, permanent magnet material features Xd stator self reactance in ’d’ axis
and by accepted presumptions [18]. That means calcula-
tions should be verified by experiments any case. Reac- Xq stator self reactance in ’q’ axis
tances of variable speed PMSM are usually expressed as U1 stator voltage
inductances for practical reasons. There were used differ-
ent measurement methods for direct axis inductance Ld I1 stator current
and quadrature axis inductance Lq in literature. In [19]
two methods are used. The first method is using an AC- ϕ phase angle.
source with variable voltage and frequency applied on sta-
tor winding for synchronous inductances measurement. 2.1 Measurement of stator self inductance in ’d’ axis
The rotor must be fixed in the phase axis on which the Vector diagram in Fig. 1 will change when the PMSM
direct inductance Ld is measured and then shifted 90 ◦ is loaded as generator by capacitance or inductance only.
electrical degrees for Lq measurement. The method is cor- For the case of the capacitance load see Fig. 2.
rect when in the machine eddy currents do not exist. The The stator current vector I1 = IC is situated in the
second method, described also in [20], evaluates the DC direct axis ’d’ as its magnetic flux linkage Ψc = Ld · I1 .
current decay after initial direct current was applied. In Voltage Ub is induced by the permanent magnet flux.
[15] a similar voltage step method was used. Accuracy of
these methods is lowered because of the numerical evalua- The magnetic flux linkage Ψc = Ld · I1 will be added
tion of the current time curve. Current decay can last sev- to the magnetic flux linkage of the permanent magnet
eral minutes in turboalternators as mentioned in [20]. Pa- which induces the voltage Ub . It holds exactly when R1 is
per [3] uses the steady state working when the machine is zero or very small. For the case of inductance load the mag-
loaded as a motor. Under this condition the magnetic fluxes netic flux linkage Ψ will be subtracted from the magnetic
in ’dq’ axes are constant. The torque angle (load angle) β flux linkage of the permanent magnet. The direct axis in-
is measured by a machine set of four machines. The accu- ductance is the function of the magnetic circuit and perma-
racy is lowered by higher harmonics induced in no load in nent magnet quality. We can measure the voltage Ub of the
the helping synchronous machine. Paper [17] uses also the PMSM in no load. The direct axis reactance X1d + X1σ
steady state analysis. The load angle is obtained by geo- can be calculated from the difference of no load voltage
metrical construction which lowers the accuracy too. and load voltage. It holds when we omit the resistance R1 .
The voltage drop on the resistance R1 is small when com-
pared with the stator voltage U1 . The calculation can be
2 MEASUREMENT OF PMSM PARAMETERS IN
corrected with respect to the voltage drop R1 · IC . The cor-
STEADY-STATE CONDITION
rection is usually not needed. The vector diagram is shown
The newly proposed measurement methods are de- in Fig. 2 for the case when the machine was loaded with
scribed in this section. They are simple and accurate. They capacitance. Voltage drop R1 · IC is assumed as very low.
R1 · Ic
sin ε = . (3)
Ub
L1d + L1σ
Td = , (4)
Rd
U1 − Ub · cos ε
Td = . (5)
Rd · ω · Ic
Procedure is the same when we load the machine with in-
ductance.
Z loading impedance
ϕ phase angle
δ = 90 − ϕ − β. (10)
Current components in ’q’ and ’d’ axis I1q and I1d are:
~ and B:
We can write for the angle γ between vectors A ~ (U~1 − R1 · I~1 ) − C
~
X1q + X1σ = − , (15)
I~1q
Ub
cos γ = . (8) X1d + X1σ
A L1q + L1σ = . (16)
ω
We can write for the angle β:
The computation will be very simple if the measure-
β = α ± γ. (9) ment on the PMSM is performed in the regime of re-
sistance load only. In such a case the angle ϕ = 0, the
Sign plus holds for the case Ld < Lq . Compare Fig. 1 current vector I~1 is a real negative number and the vec-
and Fig. 3. tor U~1 − R1 · I~1 = U1 + R1 · I1 . We get from Fig. 3 and
ϕ = 0 for C~ = (U1 + R1 · I1 ) · e β · cos β.
Sign minus holds for the case Ld > Lq . Compare
Fig. 1 and Fig. 3. We get from (15):
For this decision the measurement of the angle β is
needed. It is possible to measure it by the position sensor (U1 + R1 · I1 ) − C · e β
X1q + X1σ = − =
or by the stroboscope as shifting the axes when the ma- I~1q
chine rotor transfers from no load to load regime or on the (U1 + R1 · I1 ) − (U1 + R1 · I1 ) · e β cos β
contrary. = − =
−e β · I1q
• In case if Ld > Lq it holds β < α. (U1 + R1 · I1 ) · (1 − e β cos β)
= − =
−e β · I1q
• In case if Ld = Lq it holds β = α. U1 + R1 · I1
= · (e− β − cos β) =
• In case if Ld < Lq it holds β > α. I1q
U1 + R1 · I1
= · (cos β − sin β − cos β) =
I1q
U 1 + R1 · I1
= · sin β =
I1 · cos β
U 1 + R1 · I1
= · tan β. (17)
I1
U1 + R1 · I1
X1q + X1σ = · tan β, (18) Fig. 5. Measurement arrangement
I1
X1q + X1σ U1 + R1 · I1
L1q + L1σ = = · tan β.
ω I1 · ω 3.1 Influence of saturation in the machine during
(19) measurement
Saturation of the magnetic circuit in PMSM is caused
It is only the measurement in the regime of resistance by two factors. The first factor is the permanent magnet in
load needed. Only voltage U1 , current I1 and angle β be- itself and the second factor is the armature reaction caused
tween U1 and Ub are needed. The angle β can be mea- by load. Permanent magnets have good quality nowadays.
sured by the position sensor or by the stroboscope. Angle β Nevertheless the coercive field strength Hc and remanent
marks the ’dq’ axes shifting when the machine rotor trans- magnetic flux density Br are limiting factors for the ma-
fers from no load to load regime or on the contrary. chine design. Therefore the saturation level of the magnetic
circuit at no load is rather low. The influence of the arma-
3 PERFORMED MEASUREMENTS ture reaction on the magnetic saturation in PMSM is lower
as in machines using excitation winding. Let us compare it
Measurements were performed according to the Chap- using the magnetic circuit in Fig. 6. The reluctance of the
ter 2. Measured machine was: Synchronous machine armature stack is assumed to be much smaller than the re-
1 kW, 140 V, star connected, 3 phases, 4.6 A, 133.33 Hz, luctance of the air gap. Let us consider the case when the
2000 min−1 , pole number 2p = 8, with buried permanent armature reaction is zero. It means that the armature coil in
magnets symmetrically distributed. Photo of the measured Fig. 6 is without current.
machine is in Fig. 4. Measurement arrangement is in Fig. 5.
from which we get: with MMF 100 A is given by (24) ∆Bδ = 0.0628 T. That
means that the influence of the armature reaction is in ma-
Br chine with excitation winding 3.358 times stronger then in
Bδ = Br ·δ
. (21)
1+ g·Hc ·µ0 PMSM.
In all these cases were the air gap magnetic flux densi-
When the armature coil current is present and causes ties Bδ calculated under the assumption that the reluctance
the MMF it will hold: of the armature stack is negligible. Nevertheless the results
Br Bδ show that the saturation effect can appear in PMSM only at
Bδ = Br − · · δ ∓ MMF , (22) very high loads and is much more weaker then in machines
g · Hc µ0
with exciting windings. Approximately similar results hold
from which we get: in ’q’ axis because also here the magnetic flux crosses the
permanent magnet material. Magnetic flux crosses in this
MMF
Br · 1 ± g·Hc case the side extension parts of permanent magnet poles.
Bδ = Br ·δ
. (23) Therefore the measured values can be used in prevailing
1+ g·Hc ·µ0 loads. When we want to calculate load characteristics of
the machine at high loads it is better to measure the self
The sign ’+’ holds in case when the MMF supports
inductances at high currents but with attension not to de-
the permanent magnet. The sign ’-’ holds in case when
magnetize the permanent magnet.
the MMF acts against the permanent magnet.
Let us assume that the magnetic circuit is magnetized 3.2 Measurement of stator self inductance in ’d’ axis.
only with excitation winding (not depicted in Fig. 6). The
permanent magnet is not used and the armature reaction Machine was loaded with capacitive current.
causes additionally the same MMF as in the previous
case. The armature reaction MMF causes additional air gap 3.2.1 Values measured
magnetic flux density Bδ :
Stator effective phase voltage U1 = 58.38 V, phase
MMF voltage induced with permanent magnets Ub = 55.71 V.
∆Bδ = µ0 · . (24)
δ Capacitive current IC = 1.117 A. Frequency dur-
ing measurement f = 99.16 Hz, angular velocity
Let us calculate the increasing of the air gap mag- ω = 623.04 s−1 , revolutions 1487.4 min−1 , phase re-
netic flux density ∆Bδ caused by armature winding sistance R1 = Rd = 0.963 Ω.
MMF 100 A.
Parameters of the magnetic circuit are listed in Table 1.
3.2.2 Results obtained
The magnetic flux density in the air gap when the cir-
cuit is magnetized only by permanent magnet is given by Using (1) we obtain the reactance Xd at frequency
(21) Bδ = 0.7020 T. f = 99.16 Hz:
The magnetic flux density in the air gap when the cir- U1 − Ub 58.38 − 55.71
cuit is magnetized by permanent magnet and by armature Xd = X1d +X1σ = = = 2.39 Ω.
Ic 1.117
reaction with MMF 100 A is given by (23) Bδ = 0.7207 T. (25)
The increasing of the magnetic flux density in the air Using (2) we get stator self inductance:
gap is ∆Bδ = 0.0187 T.
X1d + X1σ 2.39
The increasing of the magnetic flux density in the Ld = L1d +L1σ = = = 0.003836 H.
air gap when the magnetic circuit is magnetized by ex- ω 623.04
(26)
citation winding and additionally by armature reaction
Using (4) we get the time constant:
0.003836
Table 1. Parameters of the magnetic circuit Td = = 0.003983 s. (27)
Length of the permanent magnet g = 0.005 m 0.963
Length of the air gap δ = 0.002 m The angle can be omitted because:
Coercive field strength
Hc = 750 000 Am−1 Rd ·Ic 0.963·1.117
(28)
of the permanent magnet sin ε = Ub = 55.71 = 0.0193,
Permanent magnet ε = 1.106 , ◦
Br = 1 T
remanent magnetic flux density cos ε = 0.9998.
3.3 Measurement of stator self inductance in ’q’ axis. Measured machine has (L1q + L1σ ) > (L1d + L1σ )
with the ratio:
Machine was loaded with resistance only. L1q + L1σ
= 1.47.
L1d + L1σ
3.3.1 Values measured
4 EXPERIMENTAL CONFIRMATION OF MEA-
Stator effective phase voltage U1 = 25.92 V. Cur- SURED PMSM PARAMETERS
rent IR = 2.265 A. Frequency during measurement Measured parameters can be used for PMSM charac-
f = 52.5 Hz. Angular velocity ω = 329.87 s−1 , revolu- teristics prediction. Let us use the experimentally obtained
tions 787.5 min−1 . Phase resistance R1 = Rd = 0.963 Ω. inductances Ld , Lq and calculate the voltage characteris-
Angle β = 8.510 ◦ was measured with position sensor. tics of the permanent magnet generator. Let us load the
Graph of voltage Ub is measured in no-load and is depicted generator with a circuit formed with different impedances
in Fig. 7. ZZ = RZ + ω · LZ or ZZ = RZ − ω·C 1
where the resis-
Z
tance RZ will be several times changed.
Calculation results are shown in Fig. 8 and Fig. 9.
ACKNOWLEDGMENT
Paper was supported by Josef Božek Research Center
of Engine and Automotive Technology at Czech Technical
University in Prague.
REFERENCES
[1] T. Lipo, Analysis of Synchronous Machines. Boca Raton,
USA: CRC Press, Taylor & Francis Group, 2012.
[2] H.-O. Seinsch, Principles of Electric Machinary and
Drives. Stuttgart, Federal Republic of Germany:
B.G.Teubner, 1993.
Fig. 9. Calculated voltage characteristic compared with [3] J. F. Gieras and M. Wing, Permanent Magnet Motor Tech-
measurement. Three phase symmetrical load is capacitive nology, Design and Applications. New York, USA: Marcel
with constant CZ = 0.032 mF and changing resistance RZ Dekker, Inc, 2002.
from 6.00 Ω to 30.0 Ω, f = 106.6 Hz. Points measured are [4] L. Novák, J. Novák, and M. Novák, “Electrically driven
marked with symbols ’+’, full line holds for calculated val- compressors on turbocharged engines with high speed syn-
ues, dash line holds for no load regime RZ → ∞. chronous motors,” in Proceedings of the 8th Electromotion
Conference, EPE, (Lille, France), pp. 600–605, July 2009.
[5] Z. Čeřovský, Commutatorless Variable Speed Synchronous
symmetrical impedance ZZ = RZ − ω·C 1
Z
. Capacitance Motor. Prague, Czech Republic: Academia, 1981.
CZ was hold constant CZ = 0.032 mF while the resis- [6] Z. Čeřovský, J. Novák, and M. Novák, “High speed syn-
tance RZ was changed from RZ from 6.00 Ω to 30.0 Ω. chronous motor control for electrically driven compressors
Then the machine was loaded in turn with impedance on overcharged gasoline or diesel engines,” in IECON Pro-
1
ZZ = RZ − ω·C Z
under the same conditions. ceedings (Industrial Electronics Conference), (Porto, Por-
Capacitance CZ was hold constant CZ = 0.032 mF tugal), pp. 1162–1167, November 2009.
while the value of resistance RZ was three times changed [7] Z. Peroutka, “Selected problems of modern ac motor trac-
RZ = 5.67 Ω; 13.14 Ω; 19.89 Ω. The calculation result is in tion drives,” in 17th International Conference Electrical
Fig. 9 compared with measurements. Points measured are Drives and Power Electronics, EDPE11, (High Tatras, Slo-
marked with symbols ’+’. vakia), pp. 23–36, September 2011.
It can be seen from Fig. 8 and Fig. 9 that the generator [8] F. De Belie, P. Sergeant, and J. Melkebeek, “A sensorless
inductive current demagnetizes permanent magnets of the drive by applying test pulses without affecting the average-
current samples,” IEEE Transactions on Power Electronics,
generator. The voltage drop is relatively high. The capaci-
vol. 25, no. 4, pp. 875–888, 2010.
tive generator current acts positively with permanent mag-
net orientation. The outgoing generator voltage is higher [9] D. T. Sepsi, P. Stumpf, Z. Varga, R. K. Jardan, and I. Nagy,
then is the voltage in no load. “Comparative analysis of ICEVs, and different types of
EVs,” in ELECTRIMACS’2011, (Paris, France), p. 6, June
2011.
5 CONCLUSION
[10] M. Jurković and D. Žarko, “Optimized design of a brushless
Modern electric drives very often use permanent mag- dc permanent magnet motor for propulsion of an ultra light
net synchronous machines (PMSM). Parameters of these aircraft,” Automatika, vol. 53, no. 3, pp. 244–254, 2012.
machines influence the drive features. The saliency of [11] I. Boldea, M. Paicu, G. Andreescu, and F. Blaabjerg, “Ac-
PMSM has often high importance for the drive and its con- tive flux DTFC-SVM sensorless control of IPMSM,” IEEE
trol. The saliency is influenced by machine construction Transactions on Energy Conversion, vol. 24, no. 2, pp. 314–
and by permanent magnet material quality. Measurement 322, 2009.
methods for PMSM parameters identification are impor- [12] J. Bauer, S. Flígl, and A. Steimel, “Design and dimansion-
tant. New methods are described in the paper. They are ing of essential passive components for the matrix conveter
suitable for stator self inductances in direct and quadrature prototype,” Automatika, vol. 53, no. 3, pp. 225–235, 2012.
axes identification. They can be used for all construction [13] J. Lettl, “Control design of matrix converter system,” in
variation of PMSM and for all permanent magnet mate- Joint International IEEE Aegean Conference on Electrical
rials. Results of measurement methods and their experi- Machines and Power Electronics & Electromotion Proceed-
mental verification are published in the paper. Performed ings, (Bodrum, Turkey), pp. 480–484, September 2007.
experiments acknowledge the measurement methods.
[14] D. Žarko, D. Ban, and T. A. Lipo, “Analytical calculation Zdeněk Čeřovský was born in Czech Republic
Feb. 1, 1930. He graduated from Czech Tech-
of magnetic field distribution in the slotted air gap of a sur-
nical University in Prague. He started his car-
face permanent-magnet motor using complex relative air- rier with CKD factory as DC/AC machine de-
gap permeance,” IEEE Transactions on Magnetics, vol. 42, signer. He projected hoisting engines, mill stand
no. 7, pp. 1828–1837, 2006. engines, ship engines, drives for street cars, locos
and electric vehicles. He changed for Czechoslo-
[15] B. Cassimere, S. Sudhoff, and D. Sudhoff, “Analytical de-
vak Academy of Sciences, worked as head of
sign model for surface mounted permanent magnet syn- Power Electronics Department in the Institute for
chronous machines,” IEEE Transactions on Energy Conver- Electrical Engineering. Projected the first street
sion, vol. 24, no. 2, pp. 347–357, 2009. car with induction motors in ČSR together with
CKD. Received Dr.Sc. degree. Later on was director of the Institute
[16] S.-C. Yang and R. Lorenz, “Comparison of resistance- for Electrical Engineering. Changed for Czech Technical University in
based and inductivity-based self-sensing controls for sur- Prague as Professor. Was visiting professor at the UNI in Hannover Ger-
face permanent-magnet machines using high-frequency sig- many. Is member of the Research Center JB of Automotive Technology.
nal injection,” IEEE Transaction on Industry Applications, He was member of Executive Council of the European Power Electronic
vol. 48, no. 3, pp. 977–986, 2012. Association EPE Brussels and member of EPE-PEMC Council Budapest.
He worked in many grants and projects like: Project NATO High Technol-
[17] Z. Čeřovský and P. Mindl, “Double rotor synchronous gen- ogy No 950918, Grant MPO IMPULS. SATREMA, Grant FI-IM5/171.
erator used as power splitting device in hybrid vehicles,” He works for Italy grant agency. He is emeritus professor at Czech Tech-
in Fisita Transactions 2006 and Fisita World Automotive nical University in Prague.
Congress, (Yokohama, Japan), p. F2006P038T, October
2006.
[18] R. Helmer, E. Garbe, J. Steinbrink, and B. Ponick, “Com- Miroslav Lev was born in Czech Republic Sept.
bined analytical-numerical calculation of electrical ma- 20, 1958. He graduated from Czech Technical
University in Prague. He graduated from two fac-
chines,” Acta Technica, vol. 54, no. 1, pp. 1–18, 2009. ulties of the UNI in Prague. He finished at the
[19] D. Klumpner, I. Serban, M. Risticevic, and I. Boldea, Faculty of Electrical Engineering the specializa-
“Comprehensive experimental analysis of the IPMSM for tion Technical Cybernetics and at the Faculty of
Mechanical Engineering the specialization Pro-
automotive application,” in 12th International Power Elec-
duction Machines and Equipment. He was great
tronics and Motion Control Conference EPE-PEMC 2006, deal of his technical activity employed by the Re-
(Portorož, Slovenia), pp. 1776–1783, September 2006. search Institute of Machine Tools and Machin-
[20] K. Kovácz and I. Rácz, Transient phenomena in AC Ma- ing VUOSO, by the factory KONE Lifts and
ITI TÜV Group Süddeutschland. At the present time he is employed by
chines I and II. Budapest, Hungary: Academy of Science,
Czech Technical University in Prague, Faculty of Electrical Engineering,
1959. Department of Electric Drives and Traction. His current interests include
mathematical simulation and programming of systems working in real
time.
AUTHORS’ ADDRESSES
Prof. Zdeněk Čeřovský, Dr.Sc.
Miroslav Lev, M.Sc.
Department of Electric Drives and Traction,
Faculty of Electrical Engineering,
Czech Technical University in Prague,
Technička 2, CZ-16627, Prague, Czech Republic
email: cerovsky@fel.cvut.cz, lev@fel.cvut.cz
Received: 2014-03-18
Accepted: 2014-07-04