Documente Academic
Documente Profesional
Documente Cultură
January 1997
Abstract - llhis paper describes an lIllproved model of rotating loads of loads consistmg of numerous mductnon motors can be found m
m complex load areas. studres dealmg wah the behamor of motors supphed by non-
The first section of the paper concentrates on the theoretical aspects smusoidal waveforms (mvextm) Such studies focus mamly on
of load modeling in harmonic penetration studies with a particular motors considered mdivldually and not as an mtegrated part of the
reference to the rotatbg part of load. The second section illustrates network;
the results achieved firom the expenmental surveys conducted both - the choice of the scale factor, to be used m order to extend the
m the laboratory and at a ma30r industrial customer, m order to above mentioned models to the different ratmgs of mdudion motors
validate the model proposed in the field. The third section describes present m the network, is often qulte a complex task.
the cahbration stage of the proposed model based on results h m a In harmmc penetration studies, there are some mica1 srtuations
statistical survey conducted at manufacturers on parameter variation (for example. a specific network topology, a hgh percentage of
VS bequency with a large sample of both LV and MV induction rotatmg load, or a w d e W b u t i o n of power factor comxtlon
motors. Finally, SOme guidelines are also provlded for a correct use capacitors) that ewdence the need to adopt a hlgher accuracy levei m
of the proposed model m system studies concemmg harmoruc eqmvalent cucwts smulatmg the load area
penetration within electric distribution networks. ' h s paper descnbes a model of complex load areas. able to
represent more accurately the rotatmg part of the load.
Introduction The proposed model can be used for smulatmg large complex load
areas wahin harmomc studies for whch the mjection current method
well known, consumers' h e a r passive loads play an mportant has been fouud to be stdl sufficiently accurate
role m harmonic penehsltion studies, smce they pnmanly a f € a
frequency response of the system m near-resonance conditions, L Load Modeling In Harmonic Penetration Studies
bemg the mam damping cmpom"pon. AU the typical h ear passive
loads present m the various collsul~ler segments (resldenhl, Previous research work has clearly shown that the load models most
.
commercial hsdustrial) may be grouped as follows: commonly used in power system harmonics literature indicate quite
- stahr loads (lighting, resistive heatmg, etc.) w t h electrical a marked difference m frequency response when they are used to
charactmacs generally well represented by the total installed active match field measurements on a distribution network supplying large
and reactive power (P. Q)or by the total demands at fimdamental load areas with a high percentage of rotatmg loads [11.
fiequency (with capacitive compensation separately represented); In receat decades, many studies have been paformed by specialists
- rorahng 1 4 (induction motors for the most wdespread m rotating machinery in order to characterize more accurately the
applications) with electrical charactenstics that can no longer be behavior of induction motors with an increase m supply hquency
adequately simulated through their adivdreactive demands at (see e.g. [2,3]).'Ihe relevant models have been shown to accord well
fundamental with induction motor performances in respect of both non-sinusoidal
~ncomputer smulptton of harmoruc distortion m power systems, supply at high frequencies and during transients. Even though the
considerable work has been done. both on calculation methodology above equivalent circuits are suitable for smgle-motor
and on models of network components best reflected 1 ~ expenmental
. representation: they, in fact. do not seem to be directly lltdizable to
tests and field measurements results. However. some di8Bculties still simulate large groups of motors in system studies such as harmonic
exlst m achieving a sufficiently accurate m o d e h g of load areas, penetration analyses in distribution networks.
especially whea conslstmg of complex loads wth a hl& percentage Recent works by International Committees on this topic have
of rotatmg equipment. recogrued the need. as in other power system studies. to model the
m e maln causes are the foUowmg load as an aggregate. though evidencing some uncertainties that call
- there iswcreal
i f f id UI achevmg synthesis. due to the fact that for experimental studies specifically concemed with network
the network load is very seldom perfectly h o w ~1 its distribution levels.
composition and it is generally unstable. In order to represent the total rotating load with a single equivalent
- the majonty of studies available on equivalents of mduction motor motor, circuit parameters that are needed can no longer directly
groups m a d y focus on the dynamc behavior m transients relate to the structural characteristics and electrical ratmgs of
foilovmg sudden vanations m the voltage supply or else m shght. individual motors, and are therefore difficult to evaluate. Due to the
semi-static frequency and voltage vanations around the rated value. aforesaid difficulties. the model most used for the sunulation of the
- m the majonty of cases. the models available for the representation rotating part of a complex load in harmonic penetration studies [CS]
are directly denved from the classic equivalent cucuit of the
induction motor. and generally share the constancy of the motor
parameters with the frequency. All the models use the induction
motor locked rotor equivalent circuit by assummg the followulg
approximations:
96 WM 125-5 PWRD A paper recommended and approved by
a) the slip is considered equal to 1. whereas it actually differs from 1
the IEEE Transmission and Distribution Committee of the IEEE precisely in the first higher harmonics order (5th harmonic: s = 1.2.
Power Engineering Sociecy for presentation at the 1996 IEEEPES 7th harmonic: s = 0.86. etc.), that is, in those that are often the
Winter Meeting, January 21-25, 1996, Baltimore, MD. relevant component of the total harmonic distortion;
Manuscript submitted July 3 1, 1995. b) the stator + rotor resistance. which IS usually less than 10% of
the locked rotor reactance at fundamental frequency, but that by the various non-hear sources in normal operating conditions
increases w t h the frequency. is generally omitted. [ I1.12]
c) the deep bar effect m the rotor wndmgs w t h the increase m The analysis assumes the rotatmg load to be passive. causmg,
frequency is ignored. whereas it is known that m mdustnal mduction therefore, only voltage drops at each harmonic frequency. already
motors (hundreds of kW) the current density distnbution influence is present m the supply.
remarkable and determmes a (rotonc) resistance mcrease with Both the above-mentioned techniques were used for the
frequency and a decrease of the leakage mductance expenmental test stage m order to measure the impedance offered by
In order to take into account the influence of the phenomenon in c), a tnduction motors. both mdividually and m large goups. to each
f\llther model w a s previously proposed by the authors [l], that harmonic present in the supply wave shape
adopts the followng mam changes.
- the stator + rotor resistance has been considered added m series to
the locked rotor reactance;
- the vanation law of the parameters (R, L) of the rotating part of the
load as a b c t i o n of the frequency, o w g to skm effect m the rotor
bars. is approximated to by a polynomial representation
The representation adopted follows the approach gven m [9], which
has proved the most suitable m order to express serm-empirical Fig. 2 - The expenmental arrangement used in the case of non
models also able to represent non-homogeneous motors, and invasive technique.
therefore complex loads such as those present UI distribution
networks. The expenmental arrangement used m the case of non mvasive
techniques. is reproduced by the sunple diagram m Fig 2. where the
rotatmg load passive equivalent (mdividual mduction motors or
group). t h e harmonic source determmmg the backgound distortion
at the PCC (Pomt of Common Couplmg)and the passive equivalent
unpedance of the network upstream of the PCC are represented
Special care must be taken for a correct mterpretation of the results
obtamed wth the measurements at the PCC In fact. a prelmmary
I
_----- I
------' x, = - -1.2- . l - , , . h . [ l - ( h . f @ ? ]
I f
(4) accurate analysis has to be camed out m order to detennme the
background distortion at the PCC at no-load conditions both for the
static parr rotating parr P .P vanous network configurations and in the vanous loadmg conditions.
The invasive technique has been used only for t a t s performed m the
Fig. 1. The cor&er load equrvalent circuit laboratory. where a low harmonic content w a s present m the supply
voltage The non-mvasive technique w a s used for a11 the tests
Fig. 1 shows the complex load equivalent circuit and the analytical performed by the major industrial customer.
expressions of the parameters, where:
kharmonic orda. ILZ Laboratory tests
fo- funduncatpl fkequency,
R,,,-equivPlent d e s resistance at fundamental frequency, The tests conducted in the laboratory made it possible to mvestigatc
,ljr=locked rotor mean equivalent reactance, only groups of motors with LV supply and low ratmgs (hction of
p-fiaction of the rotating load kW to 11 kW). The harmonic content necessary for tmpedance
a. p. kp parameten of the proposed model. versus frequency characterization was been obtamed through the two
follomng configurations:
IL Model Validation through Experimental Tests - a sut-pulse addc bridge converter injecting distorted cumnt waves
into the network supplying the laboratory premises.
The proposed model was subjected to a comprehensive experimental - a six-step inverter generating a dstorted (square) voltage wave
test, carried old both in the laboratory and by a major industrial supplymg directly the rotatmg loads.
customer. 'Ihe test arrangements were chosen so as to measure The harmonic unpact of the convener on the voltage waveshape was
induction motors harmonic impedance vs. frequency during normal first adequately enhanced by choosing quite a weak pomt of ~ m "
operating conditions, thus taking into due account the motors' rated coupling as well as by adding in senes with the network unpedance
saturation. some variable prevalently mductive unpedances of some mH
All the tests conducted in the two above configurations gave quite
U.1 Methodology for m e a s m g harmonic unpedance similar results [17]. The measurements were performed wth a
spectrum analyzer (maxunum voltage: 600 V. m a x m m frequency:
As it is known. the vanous tests for investigatmg the harmonic 2550 Hz) as well as with a portable PC-based acquisition system
behavior of power systems and of theu components can be equipped with an 8-channel, 25 smultaneous k-samples per channel,
performed using two altemative techniques. invasive and non- 12-bit acquisition board. The current measurements were conducted
invasive [IO]. w t h transducers based on the Hall effect.
The invasive technique requires the injection of harmonic currents at Figure 3 shows an instance of the results obtained from tests
the point of interest for m e a s m g the consequent voltage distortion. conducted on induction motor p u p s The figure illustrates the
permitting the calculation of harmonic impedance as a ratio of imagmary part of impedance versus frequency of a group of motors
complex quantities V and 1 at each different frequency The non- consisting of two 3-kW and an I I-kW three-phase motor operatmg
invasive technique utilizes the distortion level present at the pomt of simultaneously (base impedance is referred to the group rated power
interest for deducmg the harmonic impedance of the network through and 400 V)
the watching of a well-known load (often a capacitor) as well as Similar tests conducted on a smgle induction motor. as illustrated in
through analysis of the frequent harmonic current fluctuations caused Fig 4 a) and b), show that the behavior of the impedance magnitude
420
1s manly due to the contnbution of the Imagmary part which at the mstallations For the measurements camed om at the L\
is more hoticeable as the kequency mcreases The analysis of the busbars, suitable wde-band pass-clamp C T ’ s were used
The presence of large n o n - h e a r sowces charactenzed bv a nch
PCC-16 IO Mv.
0
OS i L L
d0 250 3SO 550 650 ’7k
qHz)
a) L
1
I
I
I
0.5 1
Fig. 5 - The W d i s t n b u r i o n network of the rndusmai cusromer
I!
0.5 t : /
!
250 350 550 650 /O0
b
O.4kl’
fcHz) I-
Fig. 4 - Zl/)diagram for
a 3kR’ rated induction motor in test
conjguranon offig.2;a)Z real part; b) 2 imaginay part.
0
_ _ _ _ --- *-
of one of the large induction motors momtored which were always m
operation (1.25 MW). Fig. 9 also show the h a n n m c unpedance
values calculated usmg the proposed model.
Fig. 7-Reconied (0) and calculated (9values of harmonic impedance In order to extend the validation stage of the model to a more
of a 130 W L V induction motor. representative sample, a comprehensive survey wns conducted at two
large differat European factories.
The purpose was to determine, on a statistical basis, average values
1.5 I ! (analytically calculated) to be assigned to the parameters of the
1 rotating part of circuit m Fig. 1.
The above parameters are e x p d as functions of those of the
simplified induction motor equivalent circuit as follows [13,14]:
where:
R,, = stator resistance,
k=rotorresistrmct,
Fig.8-Reconied (0) and calculated (9values of harmonic impednnce Data, collected on 59 induction LV motors (380 V) with difFemt
of LV lnducrion motors group. ratings (from 180 W to 400 kW), were then subjected tb statistical
processing [IS]m order to evaluate the per unit mean value of 4
I I
and L h as a function of lkqwcy, as well as to find the
- corresponding standard deviation. As concems the stator reslstence
(Rst), the statistical p.u average values found are 0.009 for MV
motors and 0.04 for LV motors.
The analytical fimctions that represent the average values can be
expressed through polynomial forms. such as:
0.1
I11.3 Guidelmes for the use of the model m harmonic studies applied
to complex load areas.