See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/3274456

Dynamic Harmonic Evolution Using the Extended Harmonic Domain

Article  in  IEEE Transactions on Power Delivery · May 2003

DOI: 10.1109/TPWRD.2003.809731 · Source: IEEE Xplore

CITATIONS READS
115 100

3 authors:

J.J. Rico M. Madrigal

Universidad Michoacana de San Nicolás de Hidalgo Instituto Tecnológico de Morelia (ITM)
45 PUBLICATIONS   594 CITATIONS    57 PUBLICATIONS   888 CITATIONS   

SEE PROFILE SEE PROFILE

Enrique Acha
Tampere University
151 PUBLICATIONS   5,470 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Multi-Terminal VSC-HVDC Systems View project

PhD research project at University of Glasgow View project

All content following this page was uploaded by J.J. Rico on 09 March 2022.

The user has requested enhancement of the downloaded file.

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003 587

Dynamic Harmonic Evolution Using the Extended

Harmonic Domain
J. Jesus Rico, Member, IEEE, Manuel Madrigal, Member, IEEE, and Enrique Acha, Senior Member, IEEE

Abstract—A novel methodology is presented in this paper for
conducting transient and steady-state analysis of electric networks
containing nonlinearities and switching plants components. The
method is termed the extended harmonic domain (EHD) method as
may be seen as an extension of the harmonic domain (HD) method
used in steady-state analysis. It is shown in the paper that EHD is a
natural approach for conducting dynamic and steady-state studies
of the evolution of harmonics in power circuits containing non-
linear components and flexible ac transmission systems (FACTS)
devices. It is also shown in the paper that EHD provides a suitable
basis for extending the well-known steady-state power quality in-
dices to the transient range. To illustrate the use of the theory, a
three-phase linear circuit and a three-phase PWM-STATCOM are
modeled using EHD. The results are compared against those pro-
vided by industry standard tools as PSCAD/EMTDC and Matlab.
Index Terms—Extended harmonic domain, Fourier series, har-
monic domain, linear time periodic systems, switching functions,
transient and harmonic analysis.
a convenient way to extend the power quality indices used in
steady-state applications to encompass transient responses. A
simple three-phase circuit, with tuned filters, is used to derive
“physical” harmonic currents, as opposed to those derived
using Fourier analysis. The “physical” harmonics are compared
against those derived using the EHD method. A three-phase
circuit containing a STATCOM is also solved in EHD, and this
example is used to calculate power quality dynamic indices.
II. APPROXIMATION OF OPERATORS
The main idea behind the approximation of operators is that
a function , can be approximated in the interval ,
to arbitrary accuracy, by the Fourier series [1]–[3]
(1)

I. INTRODUCTION where , , where is the guess period

of in steady-state (e.g., the period of the system source).
I N power systems dynamics, simulation software plays a
very important role. Several digital programs are used on a
regular basis throughout the world. Programs such as EMTP,
The coefficients in are a function of time since the in-
terval under consideration slides as a function of time . Each
PSCAD/EMTDC, Pspice, SABER, AND NETOMAC are coefficient is calculated by
typical examples of simulation software. In these packages,
relevant differential equations with suitable switching and (2)
initial conditions are solved simultaneously in the time domain.
The paper presents a direct approach for calculating the tran- Equation (2) computes the time-evolution of the harmonics as
sient and steady-state solution of electric power networks by the window of length slides over the actual waveform .
using a formulation based on orthogonal basis and operational If the vector of basic functions is used, then the function
matrices. The new approach lead to an alternative state space in (1) can be approximated as
here referred to as the EHD method. In this simulation envi-
ronment, the coefficients of the orthogonal basis are the state (3)
variables. The solution process allows for the computation of
the time evolution of the harmonic coefficients, in a manner where
that is akin to window of incremental length sliding over the
actual waveform. One salient characteristic of the EHD method
is that linear time periodic (LTP) systems are easily reduced (4)
to linear time invariant (LTI) systems, opening the door for
the use of a wide range of tools and methods to be readily ..
.
available for the solution of LTP systems. The method provides
(5)
Manuscript received August 29, 2000. This work was supported in part by
the National Council of Science and Technology (CONACYT) México and in ..
part by the Instituto Tecnológico de Morelia. .
J. J. Rico is with the Facultad de Ingeniería Eléctrica, University of
Michoacán, Morelia, CP 58000, México (e-mail: jerico@zeus.ccu.umich.mx).
M. Madrigal and E. Acha are with the Department of Electronics and Elec-
trical Engineering, University of Glasgow, Scotland, G12 8LT, U.K. (e-mail: The derivative of with respect to is given by
e.acha@elec.gla.ac.uk).
Digital Object Identifier 10.1109/TPWRD.2003.809731

0885-8977/03$17.00 © 2003 IEEE

588 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003

where the operational matrix of differentiation is defined as Using the terminology described in Section II, (7) is amenable
to an alternative representation, which is more useful for as-
sessing system properties

where

..
. Dropping the coefficients yields a more compact state
space equation

..
. (8)

These equations are the core equations of the EHD method-

The product of two functions is given as
ology whose solution gives full information of the harmonic dy-
namic behavior of variables and .
(6) Equation (8) is solved using a standard integration method
where initial and final simulation times and , integration
where the matrix has a Toeplitz structure, and it is formed time step , and initial conditions for the state variable are
with the harmonic content of always required. Matrices , , , , and vector are
constant within a time step , provided the functions , ,
, , and do not change in magnitude and frequency
.. .. .. .. in the time step. These matrices take complex numbers, and the
. . . .
.. .. .. .. solution vectors and are complex vectors which evolve
. . . . with time, changing at every time step .
.. ..
. .
.. .. .. .. A. Harmonic Domain (HD)
. . . .
.. .. .. .. A particular case of (8) is the set of equations which govern
. . . . the system’s steady state solution (i.e., when )

III. EXTENDED HARMONIC DOMAIN (9)

In principle, the time periodic excitation and response asso-
ciated with physical circuits can be expressed by the sum of
a number of sinusoid terms of different harmonic frequencies.
This is a property that facilitates the understanding and calcula-
tion of such circuits operating under a wide range of operating
conditions. From an engineering standpoint, it makes sense to
develop numerical procedures that exploit this property.
In power electronic devices, a periodic driving force causes
the switching sequences to also be periodic at each operating
point. This opens the possibility of switching devices being
thought as being linear between two consecutive switching
times and piece-wise linear within a period, giving rise to a
system that is indeed linear and time periodic.
For simplicity, consider a first-order state space model
(7)
where , , , and have a period [e.g.,
].
The period will be assumed to be throughout; hence, the
term -periodic will be used. Therefore, (7) may be referred to
as a LTP system or a -periodic system.
Equation (9) is the well-known representation used in HD as
applied in power systems harmonic analysis. This theory has
been used in the past to represent transformer nonlinearities, arc
furnaces, synchronous generators, and power electronic-based
devices [9]–[13].
In steady state applications, the matrices , , , ,
and vector are constant, and the solution of and is car-
ried out directly using (9). Moreover, a very useful character-
istic of (9) is that it is amenable to a direct method for EHD
initialization.
This is a very useful, inherent characteristic in EHD since
initialization methods are not straightforward to implement in
time domain methodologies. For example, a recent paper has
reported that no full initialization method has yet been imple-
mented in any of the EMTP programs for solving systems with
switching devices and nonlinear components [4]. Instead, sim-
pler approaches are used. A simple method is to perform an ap-
proximate steady-state solution with nonlinear branches discon-
nected or represented by linearized models. Alternatively, some
EMTPs have either a “snapshot” or a “start again” feature. Using
a standard EMTP solution, a snapshot is taken and saved once
the system reaches the steady-state, later runs are started using
this information.
RICO et al.: DYNAMIC HARMONIC EVOLUTION USING THE EXTENDED HARMONIC DOMAIN
Fig. 1. Circuit implemented in PSCAD/EMTDC.
IV. DYNAMIC HARMONIC EVOLUTION USING EHD
The three-phase balanced circuit of Fig. 1 is used to show how
harmonics evolve with time. The voltage source in phase has
the following representation V with
period , and parameters and H, and
initial currents 1, 2 and 0 A in the inductors, respectively.
A tuned filter in series with the neutral is used to separate
the dc component from the third harmonic when time domain
simulations are carried out using PSCAD/EMTDC.
589
The state space equation that describes the circuit of Fig. 1,
with no filter, is given by
(10)
where the variable of interest is .
Equation (10) is expressed in EHD by
(11)
where is the identity matrix and is the operational matrix
of differentiation. Also in this case, the variable of interest is
.
From (11), the vectors , , and have values different
from zero for the fundamental and third harmonic (e.g., for
vector these values are and , respectively).
The initial conditions given to the dc currents in phase , , and
(i.e., 1, 2 and 0, become entries in the 0-th harmonic term of
the vectors , , and , respectively. A total simulation time of
25 s with ms and 50 harmonics were used. Fig. 2(a)
shows the harmonic dynamic behavior of the magnitude of ,
where only the dc term and third harmonic appear. Fig. 2(b)
shows the angles of these harmonics and Fig. 2(c) shows the
time representation.
Fig. 3 shows simulation results obtained with
PSCAD/EMTDC. It should be noted that the time
representation of the harmonics obtained with EHD matches
the currents flowing through the tuned filter.
A. EHD and WFFT Comparison
The use of special filters is not the normal way to derive
harmonic information from transient waveforms. Instead, win-
dowed fast Fourier transforms (FFTs) are used. Hence, results
from EHD and a windowed FFT (WFFT) technique are com-
pared. In order to illustrate these two techniques, a disturbance
was applied after the ground current reached the steady state, a
zero initial condition and were selected for this ex-
ample. A solid phase to ground short circuit was applied at 15
s. Fig. 4(a) shows the dynamic behavior of the current . For
the WFFT technique, a sliding window with a period
was used. The sliding window in Fig. 4(a) is shifted to the right
every time step. The magnitude of the harmonics using the EHD
and those calculated using the WFFT are shown in Fig. 4(b).
Also, Fig. 4(c) shows the fundamental frequency and third har-
monic only.
It should be noted that before the disturbance (steady state),
the harmonics obtained with the EHD and with WFFT matches
at all in magnitude and phase angle, in this case only the third
harmonic exists in the current . After the disturbance, using
the EHD only, the fundamental and third harmonic appears, and
using the WFFT appears a wide range of harmonics. It is only
after the sliding window pass the disturbance the fundamental
590
(a)
(b)
(c)
Fig. 2. Ground current using EHD.
and third harmonic matches again between the two methods.
Due to the shape of the waveform, it is rather clear that only
the fundamental and another harmonic (third) seem to appear.
The EHD confirms this thought but this is not achieved by the
WFFT. It is likely that additional refinements in the use of the
WFFT method may lead to more accurate results than the ones
presented in Fig. 4(b), but the great difficulties of doing so are
well documented in a recent paper [16].
IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003
Fig. 3. Ground currents using PSCAD/EMTDC.
Fig. 4(c) shows the angles of the fundamental and third har-
monic current obtained with the EHD and WFFT methods. Key
observations can be derived from these results: i) the use of the
WFFT transform gives rise to a misleading harmonic spectrum;
ii) the fundamental and third harmonic currents derived with
both techniques show quite a different dynamic evolution; iii)
the angle of the fundamental and third harmonic current using
WFFT shows a very substantial mismatch when compared to
the result using the EHD method.
The EHD and the FFT methods take quite a different ap-
proach to solving the dynamic evolution of harmonics. The dif-
ference lies mainly in the fact that the EHD is a direct method
to solve dynamic systems and the FFT is not. Additional advan-
tages of the EHD over the FFT technique is that does not have
problems with window size selection, edge phenomenon, and
aliasing.
V. STATCOM MODEL USING EHD
The STATCOM comprises a voltage-source inverter, a dc
capacitor, and a coupling transformer. Fig. 5 represents a
STATCOM connected to an ac system, where the exchange of
the reactive power between the STATCOM inverter and the ac
power system can be controlled by changing the voltage mag-
nitude between the STATCOM converter output voltage and
the bus voltage at the common connection point. Provided the
STATCOM has provisions for generating and absorbing active
power, its exchange can be controlled by changing the phase
angle between the STATCOM inverter output voltage and the
bus voltage at the common connection point. The STATCOM
converter uses IGBTs which can be turned on and off at a rate
considerably higher than the power frequency. The IGBT’s
switching sequence is governed by a pulse-width modulation
(PWM) technique to control the output harmonics. In this
situation, the STATCOM acts as an independent voltage-source
with harmonic control.
A. PWM Converter and Switching Function
The switching function is calculated using a technique for
harmonic elimination [6]. It can be shown that in order to elimi-
RICO et al.: DYNAMIC HARMONIC EVOLUTION USING THE EXTENDED HARMONIC DOMAIN 591

Fig. 5. STATCOM equivalent.

(a)

Fig. 6. Switching function.

eliminate the third harmonic, the calculated angle is

and to eliminate the fifth, seventh, 11th, 13th, and 17th har-
monics, the calculated angles are , ,
, , and , respectively. In
a three-phase system, the other two switching functions are cal-
culated by shifting by 120 and 120 , respectively.

B. STATCOM State Equations

With reference to the circuit in Fig. 5, the equations that de-
scribe the STATCOM are

(12)
(b)
where , , and are the switching functions of the PWM
converter.
If the voltage source inverter is assumed to be a power in-
variant circuit, the instantaneous power flowing from the dc to
the ac side may be expressed as
(13)
Use of (12) in (13) yields an expression for
(14)
Also, the current may be written in terms of the dynamic
equation of the capacitor

(15)
The voltage drop across the three-phase impedance of the
STATCOM circuit in Fig. 5 is
(c)
Fig. 4. Comparison of the harmonics using WFFT and EHD.

nate the harmonics, switching angles are required. A typical

switching function is illustrated in Fig. 6. For example, to (16)
592 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003

(a) (b) (a) (b)

Fig. 7. Voltage source disturbances. Fig. 8. STATCOM voltages and currents using EHD.

A suitable combination of (12)–(16) yields a time state space

equation that describes the dynamics of the STATCOM

(17)

Moreover, this equation can be transformed into a LTI system

using the EHD method, as shown in Section III.

C. Case Study
To test the theory presented in this paper, a three-phase,
PWM-controlled STATCOM has been chosen. The PWM
strategy was selected to eliminate the fifth, seventh, 11th, 13th,
and 17th harmonics with a phase shift angle in order
to achieve active and reactive power exchange [5], [8]. The
parameters of the network are , mH,
F. The system voltage source is unbalanced with
the following values , ,
and . The system frequency is
50 Hz.
The HD was used to initialize the EHD solution. In this study,
it is assumed that a disturbance lasting 10 ms takes place at a
time 20 ms. During the disturbance, the voltage reduces to
half of its prefault value, as shown in Fig. 7(a).
Fig. 8(a) shows the EHD solution, where only the voltage ,
current , and voltage in the dc side are shown. Fifty har-
monic terms were used in the simulation. The time representa-
tion of these variables is given in Fig. 8(b). This result shows
that harmonics react very sharply to system disturbances; they
give very accurate information of the instant when the distur-
bance starts and when it finishes. This information is not easily
discerned when expressed in the time domain.
Fig. 9 compares results given by a standard numeric integra-
tion method and the EHD solution, expressed in time domain.
Fig. 9. Comparison of time domain simulations and EHD.
Both results compare very well with each other when 50 har-
monics are used. The accuracy of the EHD solution decreases
when the number of harmonic coefficients reduces (e.g., five
harmonics). Fig. 9 shows the comparison during the disturbance
period. Nevertheless, the solution with a reduced number of har-
monics shows to be a good averaging method. It is important
to remark that if only the fundamental and the dc components
were to be used in the EHD method, it would yield, with suit-
able similarity transformations, a model that is compatible with
those used in small signal stability analysis.
D. Dynamic Electrical Indices
Several electrical parameters are used to enable accurate
power quality assessments of electrical installations and
equipment. The most popular are given in Table I. These
parameters are normally expressed in terms of complex Fourier
coefficients.
RICO et al.: DYNAMIC HARMONIC EVOLUTION USING THE EXTENDED HARMONIC DOMAIN 593

TABLE I
POWER QUALITY PARAMETERS

Fig. 11. Electric quantities in terminals of the STATCOM for the second
disturbance.

Fig. 10. Electric quantities in terminals of the STATCOM for the first
disturbance.

In principle, the Fourier transform only applies to voltage and

current waveform that exhibit a periodic behavior. In practice,
windowing FFTs have been used for quite some time to derive
harmonic information under transient conditions. However, the (a) (b)
use of electrical parameters for power quality assessment under
Fig. 12. STATCOM voltages and currents using EHD.
dynamic conditions are used infrequently. One possible expla-
nation is that the use of the windowing FFT does not yield accu-
Fig. 11 shows the electrical indices for a different disturbance.
rate harmonic information and, hence, meaningful electric pa-
It corresponds to a voltage impulse of three times the magnitude
rameters could not be obtained. In this paper, the accurate as-
of phase with a duration of 0.1 ms. As shown in Fig. 7(b),
sessment of the dynamic evolution of harmonics allows for the
the impulse takes place at a time 25 ms. The results show that
extension of such definitions to be applied under transient condi-
the impulse is quite difficult to detect in the voltage and current
tions at every time step. Accordingly, it is possible to talk about
waveforms. However, this small disturbance can be easily de-
dynamic power quantities such as powers , , , and
tected using EHD, as shown in Fig. 12. This extreme sensitivity
; rms values and ; as well as power factor
of harmonics during disturbances could be used to good effect
. For completeness, the distortion factors and
to explain several power quality disturbances.
can also be derived. These transient harmonic indices
may find applications in the design of more robust protection
and control schemes than those based on positive sequence, fun- VI. CONCLUSIONS
damental frequency parameters. The extended harmonic domain methodology has been pre-
Fig. 10 shows the dynamic behavior of these electrical quan- sented in this paper for the first time. It is based on the use of
tities at the terminals of the STATCOM. These results show that dynamic harmonics, orthogonal series expansions, and their op-
as expected, these quantities are constant, for periodic, steady erational properties. The method is suitable for steady state anal-
state operation but register significant changes during transient ysis as well as large signal dynamic analysis of electric networks
conditions due to the dynamic behavior of the harmonics. with embedded FACTS devices and nonlinear loads. The EHD
594
method is an extension of the well-known HD method used in
power systems harmonic analysis. This has the advantage that
a wide range of models developed for HD can be used for tran-
sient analysis with few or no modifications. Moreover, HD can
also be used for EHD initialization. This method yields accurate
information of the time evolution of harmonic coefficients, and
allows for an extension of power quality indices to the dynamic
range. EHD is an alternative linear, time-invariant, state space
equation for which all the powerful tools developed in linear
control design can be applied.
From the simulations presented, it is shown that neglecting
harmonic terms during transient conditions may lead to inaccu-
rate results. To test the theory, a model for a three-phase PWM-
controlled STATCOM has been used. Time responses calculated
with the EHD and time domain simulations were compared and
the many advantages of EHD were highlighted.
REFERENCES
[1] N. M. Wereley and S. R. Hall, “Frequency response of linear time pe-
riodic systems,” in Revised Version of a Paper Presented at the 1990
Conference on Decision and Control, Honolulu, HI, Dec. 5–7, 1990.
[2] S. R. Sanders, J. M. Noworolski, X. Z. Lui, and G. C. Verghense, “Gen-
eralized averaging method for power conversion circuits,” IEEE Trans.
Power Electron., vol. 6, pp. 251–259, Apr. 1991.
[3] V. A. Caliskan, G. C. Verghese, and A. M. Stankovic, “Multifrequency
averaging of DC/DC converters,” IEEE Trans. Power Electron., vol. 14,
pp. 124–133, Jan. 1999.
[4] J. A. Martinez, “Power quality analysis using electromagnetic transients
programs,” in Proc. ICHQP , Athens, Greece, Oct. 1998, pp. 590–597.
[5] J. E. Hill and W. T. Norris, “Exact analysis of a multipulse shunt con-
verter compensator or statcon. Part 1: Performance,” Proc. Inst. Elect.
Eng.-Gener. Transm. Distrib ., vol. 144, no. 2, pp. 213–218, Mar. 1997.
[6] H. S. Patel and R. G. Hoft, “Generalized techniques of harmonic elim-
ination and voltage control in Thyristor inverter: Part I-harmonic elimi-
nation,” IEEE Trans. Ind. Applicat., vol. IA-9, pp. 310–317, May/June
1973.
[7] G. Joos, L. Moran, and P. Ziogas, “Performance analysis of a PWM
inverter VAR compensator,” IEEE Trans. Power Electron., vol. 6, pp.
380–391, July 1991.
[8] D. Shen, W. Liu, and Z. Wang, “Study on the operation performance of
STATCOM under unbalanced and distorted system voltage,” in Proc.
2000 IEEE Power Eng. Soc. Winter Meeting, Singapore, Jan. 2000,
Paper 0-7803-5938-0/00.
[9] E. Acha, “Modeling of Power System Transformers in the Complex
Conjugate Harmonic Domain Space,” Ph.D., University of Canterbury,
Christchurch, New Zealand, 1988.
[10] J. Rico, E. Acha, and T. J. E. Miller, “Harmonic domain modeling of
three phase Thyristor-controlled reactors by means of switching vec-
tors and discrete convolutions,” IEEE Trans. Power Delivery, vol. 9, pp.
1609–1615, July 1994.
View publication stats
IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003
[11] J. Arrillaga, A. Medina, M. L. V. Lisboa, M. A. Cavia, and P. Sánchez,
“The harmonic domain. A frame of reference for power system harmonic
analysis,” IEEE Trans. Power Delivery, vol. 10, pp. 433–440, Feb. 1995.
[12] A. Semlyen, E. Acha, and J. Arrillaga, “Newton-type algorithms for
the harmonic phasor analysis of nonlinear power circuits in periodical
steady-state with special reference to magnetic nonlinearities,” IEEE
Trans. Power Delivery, vol. 3, pp. 1090–1098, July 1988.
[13] E. Acha, A. Semlyen, and N. Rajakovic, “A harmonic domain compu-
tational package for nonlinear problems and its application to electric
arcs,” IEEE Trans. Power Delivery, vol. 5, pp. 1390–1397, July 1990.
[14] B. K. Perkins, “Steady-state solution of the HVDC converter including
AC/DC system interaction by a direct method,” IEEE Trans. Power De-
livery, vol. 14, pp. 1454–1460, Oct. 1999.
[15] A. Semlyen and J. A. Medina, “Computation of the periodic steady state
with nonlinear components using a hybrid time and frequency domain
methodology,” IEEE Trans. Power Syst., vol. 10, pp. 1498–1504, Aug.
1995.
[16] G. T. Heydt, P. S. Fjeld, C. C. Liu, D. Pierce, L. Tu, and G. Hensley, “Ap-
plications of the windowed FFT to electric power quality assessment,”
IEEE Trans. Power Delivery, vol. 14, pp. 1411–1416, Oct. 1999.
J. Jesus Rico (M’00) was born in Purépero, Michoacán, México. He graduated
from the University of Michoacán, Morelia, México, in 1990, and received the
M.Sc. degree from the University of Nuevo León, Monterrey, México, in 1993,
and the Ph.D. degree from the University of Glasgow, Scotland, U.K., in 1997.
He was a postdoctoral fellow at Arizona State University, Tempe, in 1999.
Currently, he is a Researcher at the University of Michoacán, Morelia,
México.
Manuel Madrigal (M’02) was born in Purépero Michoacán, México. He re-
ceived the B.Sc. degree from the Instituto Tecnológico de Morelia, México, in
1993, and received the M.Sc. degree from the University of Nuevo León, Mon-
terrey, México, in 1996, and the Ph.D. degree from the University of Glasgow,
Scotland, U.K., in 2001.
Currently, he is teaching and conducting research in harmonics and power
quality at the Instituto Tecnologico de Morelia.
Enrique Acha (SM’02) was born in México. He graduated from the University
of Michoacán, Morelia, México, in 1979, and received the Ph.D. degree from
the University of Canterbury, Christchurch, New Zealand, in 1988.
Currently, he is Professor of Electrical Power Systems at the University of
Glasgow, Scotland, U.K.
He was a postdoctoral fellow at the University of Toronto, ON, Canada, in
1988.