User Guide

Version 7.0

 2008 Opal-RT Technologies Inc. All rights reserved for all countries.
Information in this document is subject to change without notice, and does not represent a commitment on the part
of OPAL-RT TECHNOLOGIES. The software and associated files described in this document are furnished under a
license agreement, and can only be used or copied in accordance with the terms of the agreement. No part of this
document may be reproduced or transmitted in any form or by any means, electronic or mechanical, including
photocopying, recording, or information and retrieval systems, for any purpose other than the purchaser's personal
use, without express written permission of OPAL-RT TECHNOLOGIES Incorporated.
Documents and information relating to or associated with OPAL-RT products, business, or activities, including but
not limited to financial information; data or statements; trade secrets; product research and development; existing
and future product designs and performance specifications; marketing plans or techniques, client lists, computer
programs, processes, and know-how that have been clearly identified and properly marked by OPAL-RT as
proprietary information, trade secrets, or company confidential information. The information must have been
developed by OPAL-RT and is not made available to the public without the express consent of OPAL-RT or its legal
counsel.
ARTEMiS, RT-EVENTS, RT-LAB and DINAMO are trademarks of OPAL-RT TECHNOLOGIES, Inc. MATLAB, Simulink,
Real-Time Workshop and SimPowerSystem are trademarks of The Mathworks, Inc. LabVIEW is a trademark of
National Instruments, Inc. QNX is a trademark of QNX Software Systems Ltd. All other brand and product names are
trademarks or service marks of their respective holders and are hereby acknowledged.
We have done our best to ensure that the material found in this publication is both useful and accurate. However,
please be aware that errors may exist in this publication, and that neither the authors nor OPAL-RT TECHNOLOGIES
make any guarantees concerning the accuracy of the information found here or in the use to which it may be put.
Reference Number: RT-LAB Users Guide
Published in Canada

Contact Us
For additional information you may contact the Customer Support team at OPAL-RT at the following coordinates:

Tool-Free (US and Canada)

1-877-935-2323 (08:30-17:30 EST)

Phone

1-514-935-2323

Fax

1-514-935-4994

E-mail

[email protected]
[email protected]
[email protected]

Mail

1751 Richardson Street

Suite 2525
Montreal, Quebec
H3K 1G6

Web

www.opal-rt.com

OPAL-RT Technologies Inc.

TABLE of CONTENTS

CHAPTER 1: INTRODUCTION
What to Expect from this Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
About ARTEMiS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Intended Audience and Required Skills and Knowledge . . . . . . . . . . . . . . . . . . . 6
Organization of Guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
CHAPTER 2: QUICK START
Getting Started (Off-line simulation). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Getting Started (RT-LAB real-time simulation) . . . . . . . . . . . . . . . . . . . . . . . . . 9
CHAPTER 3: USING ARTEMIS
Six Pulse Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
14-Thyristor Frequency Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Medium Power Network. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
ARTEMiS modeling of transformer saturation . . . . . . . . . . . . . . . . . . . . . . . . . 18
CHAPTER 4: STATE-SPACE SPACE (SSN) SOLVER BASICS
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
The ssnSSN_lib.mdl library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Usage of the SSN Nodal Interface Block in a model . . . . . . . . . . . . . . . . . . . . 27
1st Real-life case: 12-pulse HVDC system . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2nd case: 3-level NCP inverter and SSN Real-Time Impulse Events . . . . . . . . . 32
3rd case: Inlined Thyristor Valve Compensation in SSN . . . . . . . . . . . . . . . . . 34
Static Var Compensator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
SSN parallel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
CHAPTER 5: REFERENCE
ARTEMiS Guide 45
ARTEMiS Distributed Parameters Line 53
ARTEMiS-SSN Frequency Dependent Line 63
ARTEMiS-SSN WideBand Line 71
ARTEMiS Stubline 79
ARTEMiS Transformer with Switched Saturable Core 91
ARTEMiS-SSN Nodal interface Blocks 101
ARTEMiS Distributed Parameters Line with Variable Internal Fault distance 107
ARTEMiS MMC 1P Cell 113
ARTEMiS MMC 2P Cell 117
ARTEMiS TSB 2-Level 121
OpReplaceSpsBlocks 125

2008 Opal-RT Technologies Inc.

OPAL-RT Technologies Inc.

TABLE of CONTENTS

CHAPTER 6: KNOWN LIMITATIONS (ARTEMIS V6.0 RELEASE)

PC Memory limitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
ARTEMiS limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
SimPowerSystems 4.6- 5.0 limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

2008 Opal-RT Technologies Inc.

Introduction

1.1

What to Expect from this Guide

This guide explains the ARTEMiS add-on for SimPowerSystems blockset.

1.2

About ARTEMiS
ARTEMiS stands for Avanced Real-Time ElectroMechanical Simulator. It is a plug-in to the
SimPowerSystems blockset for Simulink that enables hard real-time simulation of SimPowerSystems
models. The objective of hard real-time simulation is that all iteration of the model are completed in a
prescribed amount of time at each time step.
The hard real-time simulation objective is different that the typical simulation objective. In a normal
simulation, one wants the smaller total simulation time or said in another way, the smallest average
simulation time step. The hard real-time simulation objective is to have the smaller maximum time
step.
The second main objective of real-time simulation is to maintain the simulation accuracy to a certain
level. This is a potential problem in real-time simulation because it is made with fixed-step solvers.
Compared with typical variable-step solvers, the usage of fixed-step solver can lead to innacuracies
because there are no built-in accuracy check within the solvers. A typical variable step solver will
implicitly compare it results with a higher order algorithm to verify accuracy. With a fixed-step solver,
there is no such verification and larger time step always degrade simulation accuracy in some way.
ARTEMiS help reach real-time simulation objectives in several ways. By its characteristics, ARTEMiS can
extend the range of time step to achieve both speed and precision for a specific real-time application.
In applications where network switching causes numerical oscillations that cannot be damped at time
step above minimum hardware limits, ARTEMiS solvers good damping properties successfully damp the
spurious oscillations. Furthermore, in applications where some underdamped or high frequency
components, relative to the fastest possible sampling time must be taken into account, ARTEMiS
improves the precision of those components compared to the trapezoidal or Tustin methods.
Since version 6.0, ARTEMiS offers a new solver called State-Space Nodal which combines the accuracy
potential of state-space methods with the natural ability of the nodal approach to handle circuit with a
large number of switches. Consequently, ARTEMiS is no longer limited with regards to the
number of coupled switches in a circuit. which provides some edges in the modelisation of microgrids or distribution networks in which artificial decoupling methods like stublines cannot be used but
also in more standard systems like HVDC or SVC.
In switched power system, ARTEMiS now comes with automatically Inlined Interpolation methods for
thyristors and 2-level voltage inverters. These very efficient solver methods detects and compensates
for switching events that occur in the middle of time steps. The option works in conjunction with the RTEVENTS blockset and uses real-time optimized interpolation techniques to improve accuracy.
ARTEMiS provides special model options. It comes with a saturable transformer model based which use
flux as state and that does cause algebraic loops. One can also use SimPowerSystems continuous time
machine model in conjunction with Simulink higher order fixed-step solvers. In some case, the use of
higher order solver can increase notably the precision.

ARTEMIS Users Guide

Q022014-01

Intended Audience and Required Skills and Knowledge

Introduction

Finally, ARTEMiS comes with specialized models for real-time simulation such as ARTEMiS Distributed
Parameter Line and ARTEMiS Stublines that enables distributed simulation of power systems on several
CPUs or cores of standard PCs using RT-LAB. The ARTEMiS plug-in is especially designed to work in the
RT-LAB real-time environment and shall prove very effective in helping the typical user reach its realtime simulation objectives.

1.3

Intended Audience and Required Skills and Knowledge

The ARTEMiS User Guide is intended for ARTEMiS users. It is recommended that you be
familiar with Simulink and SimPowerSystems before getting started.

1.4

Organization of Guide
The followingl guides come with the ARTEMiS documentation:
Installation Guide
User Guide

1.5

Conventions
Opal-RT guides use the following conventions:

Table 1: General and Typographical Conventions

THIS CONVENTION
Bold

INDICATES
User interface elements or text that must be typed
exactly as shown.
Emphasizes or supplements parts of the text. You can

Note:

disregard the information in a note and still complete a

task.

Warning:
Recommendation:

Describes an action that must be avoided or followed to

prevent a loss of data.
A suggestion that you may or may not follow and still
complete a task.

Code

Sample code.

Italics

Reference work titles.

Blue Text

Cross-references (internal or external) or hypertext links.

Q022014-01

ARTEMIS User Guide

Quick Start

This chapter describes how to use the ARTEMiS Add-On to SPS in RT-LAB. For a quick start on how to use the new
State-Space Nodal Solver of ARTEMiS, please refer to the SSN basics chapter of this guide.

2.1

Getting Started (Off-line simulation)

To use the ARTEMiS Add-On to SPS:
1.

Start MATLAB. Open a Simulink model that uses the blocks from the
SimPowerSystems blockset like SPS demo power_monophaseline.

Figure 1:Time Domain and Frequency Domain Testing of Single Phase Line

2.

From the MATLAB command window, open the ARTEMiS library prompt by
typing artemis.

ARTEMIS Users Guide

Q022014-01

Getting Started (Off-line simulation)

Quick Start

Figure 2:MATLAB

The ARTEMiS library window is displayed.

Figure 3:Library ARTEMiS

3.

Click on the ARTEMiS block and drag the ARTEMiS Guide block into your
model.

Q022014-01

ARTEMIS User Guide

Getting Started (RT-LAB real-time simulation)

Figure 4:Time Domain and Frequency Domain Testing of Single Phase Line

4.

Run your model. Once the ARTEMiS Guide block is placed in a model, the
linear part of power system is simulated using the fixed time step algorithms
and options specified in the ARTEMiS Guide dialog box. If both the ARTEMiS
Guide block and the Discrete System block from SimPowerSystems blockset
are present in a model, the ARTEMiS Guide block has precedence.

2.2

Getting Started (RT-LAB real-time simulation)

After the model has run offline successfully, the following step is to modify the model to run it in real-time within
RT-LAB.
The first step to convert this model to RT-LAB and to exploit the parallel simulation capability of RT-LAB is to
convert the SPS Distributed Parameter Line to a ARTEMiS Distributed Parameter Line (DPL). Both DPL models
have the same underlying equations but the latter is design to be used inside RT-LAB. The ARTEMiS DPL model
can be found in the ARTEMiS library under the ARTEMiS group.
When this DPL model is used, the resulting electric model is effectively decoupled into 2 different state-space
systems containing topologically connected elements (RED and BLUE groups of the figure below). RT-LAB will
then compute these state-space systems in different cores/CPUs during real-time simulation.

ARTEMIS Users Guide

Q022014-01

Getting Started (RT-LAB real-time simulation)

Quick Start

So, step by step:

Select all blocks located in the subnetwork 1 in the figure above
and press Ctrl-G to create a new subsystem.
Move the ARTEMiS block inside the subsystem.
Rename this subsystem to SM_Subnetwork_1. The following figure
displays the content of the SS_Subnetwork_1 subsystem.

10

Q022014-01

ARTEMIS User Guide

Getting Started (RT-LAB real-time simulation)

Select all blocks located in the subnetwork 2 and press Ctrl-G to

create a new subsystem.
Add a ARTEMiS Guide block inside the subsystem.
Rename this subsystem to SS_Subnetwork_2. The following figure
illustrates the content of the SS_Subnetwork_2 subsystem.

ARTEMIS Users Guide

Q022014-01

11

Getting Started (RT-LAB real-time simulation)

Quick Start

Select the 3 remaining blocks, normally the two scopes blocks and
the Mux1 block and press Ctrl-G to create a new subsystem.
Rename this subsystem to SC_Console.
Add the RT-LAB opcomm block between the inports blocks and the
content of the subsystem. Dont forget to set the number of inports
of the opcomm blocks to 3. Refer to the RT-LAB user guide for
more help.
The following figure illustrates the content of the SC_Console
subsystem after the modifications described above have been
made.

Modify the solver parameters of the model; select one of the fixedstep solver, like ode3 for example, and change the fixed-step size
to 50e-6.
Organize the top level blocks according to the following figure.
IMPORTANT: the powerGUI block must be at the top level
and each subsystem must contain an ARTEMiS block

12

Q022014-01

ARTEMIS User Guide

Getting Started (RT-LAB real-time simulation)

Save your model.

Your model is now ready to be compiled with RT-LAB. Refer to the
RT-LAB User Guide for more help. If your have set the sample
times of your model with a variable set in the workspace(ex: Ts),
you should set the model initialization function with <Ts=50e-6;>
in File->Model Properties->Callbacks->InitFcn
IMPORTANT NOTE: A single ARTEMiS block can also be put in the
top-level of the RT-LAB ready model. At compilation time, RT-LAB
will make a copy of this block with identical parameters in all
separated subsystems.

ARTEMIS Users Guide

Q022014-01

13

Getting Started (RT-LAB real-time simulation)

Quick Start

14

Q022014-01

ARTEMIS User Guide

Using ARTEMiS

ARTEMiS, the Advanced Real-Time Electro-Mechanical Simulator, is a simulation toolset that includes
the ARTEMiS Plug-in to the SimPowerSystems blockset. The ARTEMiS Plug-in is a performanceenhancing add-on product for the SimPowerSystems blockset. It is easy to use: simply add the
ARTEMiS Plug-in block to any Simulink model containing SimPowerSystems blockset blocks and the
model runs using the ARTEMiS improved algorithms.
The ARTEMiS Plug-in offers the following advantages to the standard SimPowerSystems Blockset:
Real-time computational capability. More than simply providing
faster simulations, ARTEMiS is designed to enable real-time
computation of SimPowerSystems blockset circuits. The following
considerations were taken into account for the design of ARTEMiS:

State-Space Nodal (SSN) solver which provides all the

advantages of nodal methods such as enabling the real-time
simulation of circuit with hundred of switches and node
count approaching 1000.

Improved modeling of some power system elements such

as: Saturable transformer model (which can be simulated
at fixed time step in a non-iterative manner in ARTEMiS)

Distributed multi-processors simulation capability of

complex power systems with ARTEMiS Distributed
Parameter Line and ARTEMiS stubline models .

Compatibility with OPAL-RT's RT-LAB suite of products for

easy integrated parallel simulation design process.

Higher precision for linear circuits with high frequency

components: ARTEMiS improves the SimPowerSystems blockset's
precision of simulation compared with the standard fixed-step
integration methods such as trapezoidal or Tustin, especially for
circuits whose variables have high frequency components.
Less numerical oscillations without the need for artificial
stabilizing snubbers: ARTEMiS uses L-stable integration methods
that are free from the numerical oscillations that often affect the
standard SimPowerSystems blockset fixed-step integration
methods such as trapezoidal or Tustin. The effect of ARTEMiS on
numerical oscillation can be seen even on simple cases such as SPS
demo power_monophaseline.
The following section are examples of what can be acheived with the ARTEMiS plug-in to the
SimPowerSystems blockset using the SSN solver. The Simulink demo section of ARTEMiS contains many
more examples.

ARTEMIS Users Guide

Q022014-01

15

Distribution Grid Example

Using ARTEMiS

3.1

Distribution Grid Example

Figure 5 shows a a typical configuration of distribution grid in France and such a model is used to study
the impact of various load profiles and control strategies. This distribution system (F1a) with an on-load
tap changer (OLTC) transformer at the feeder point, 120 3-phase time-variable loads (TVL), 3-phase
fault and more than 650 equivalent EMTP nodes is simulated in real-time with the SSN solver.
15 SSN nodes were used in the separation of the F1a network (the 5 red dots in image).

Figure 5:F1a distribution grid

3.1.1 Real-Time performance

The F1a model with OLTC and one 3-phase fault can be simulated in real-time at a time-step of 70 s
on an Intel-Xeon Processor-E5-2687W (Xeon V3), using only 4 cores out of the 20 available, without
any delays or stublines. See [1] for more detail on this model. This performance is currently out of
reach for other real-time platform such as Hypersim or RTDS because of there are no long transmission
lines in distribution systems.

3.2

Bipolar HVDC system with switched filter banks

In this section, SSN is used to simulate in real a complete 12-pulse bipolar HVDC link with switched
filter banks. This test model is a 2000 MW (500 kV, 2 kA at each pole) HVDC link used to transmit
power from a 500 kV, 5000 MVA, 60 Hz network to a 345 kV, 10 000 MVA, 50 Hz network. The rectifier
and the inverter are 12-pulse converters and the link is bipolar. The rectifier and the inverter are
interconnected through two 300 km distributed parameter line using 0.5 H smoothing reactors. The
transformer tap changers are not simulated and fixed taps are assumed. Reactive power required by
the converters is provided by a set of capacitor banks plus 11th, 13th and high pass filters for a total of
600 MVAR on each side. These capacitors and filters can be switched; each station has 7 switched
banks.

16

Q022014-01

ARTEMIS User Guide

Simulation Results

Line(300km
)
transfo

6-pulse
thyristor
rectifier

6-pulse
thyristor
rectifier
500kV
60Hz

ACfilters(600M
Vars)

345kV
50Hz

ACfilters(600M
Vars)

Line(300km
)

Figure 6:Bipolar 12-pulse HVDC link with switched filter banks

In Fig. 6, the SSN state-space groups are indicated by the colors. Stations (with 2 poles each) on both sides of
the lines are simulated on two different cores of the RT-LAB simulator. The natural delay of transmission line
enables this decoupling. Each station circuit, comprising the two 12-pulse rectifiers are each simulated with a
single SSN solver instance, without decoupling delays. See [2] for more details.

Figure 7:Bipolar DC-link filter bank switching test overall result (left) and ITVC algorithm effect on the DC-current
of the HVDC link (right)

3.2.1 Simulation Results

As a test case, we made the DC-link energization. Then, when the DC-link reached its nominal current, we began
to connect filter banks 1 to 6 at 0.4 sec. intervals, followed by simultaneous disconnection off all banks at 5 sec.
As expected and shows in Fig. 7, each time a filter bank is put on-line, it provides the AC network reactive power,
i.e. raising the AC voltage amplitude and DC-link controllers reacts by raising the firing angle to keep the DC-link
current at nominal level.
The SSN method implements 'Inlined Thyristor Valve Compensation' algorithm (ITVC). The ITVC method is used
to maintain the simulation accuracy even when the thyristor switching occurs between time-steps. Fig. 7 shows
the effect of the ITVC algorithm: it decreases by 50% the jitter caused by the sampling of thyristors gate signal
by the fixed-time step simulation time frame. It must be understood that this simulation imprecision, with ITVC
algorithm is turned off, is caused by fixed step sampling and would be similar in any fixed step software like
EMTP, PLECS or PSCAD. The figure also shows a very regular current pattern with ITVC in action.
ARTEMIS Users Guide

Q022014-01

17

Real-time simulation speed

Using ARTEMiS

3.2.2 Real-time simulation speed

The bipolar HVDC link with switched filters can be simulated at a time step of 49 s on 6 cores of a 3.3 GHz Xeon
V2 (3 cores per station). Higher speed can be expected if Xeon V3 are used.

3.3

ARTEMiS modeling of transformer saturation

3.3.1 SPS modeling of transformer saturation

The native saturable transformer model of SPS has no linear part in its magnetization branch. The
magnetization branch is instead modeled entirely by a current injection. The current injection is
computed from a table of the phi=f(i), with the flux being computed from the integral of voltage across
magnetization branch.

Figure 8:Saturable transformer model in SPS and ARTEMiS

In the SPS model, the non-linear Lsat component of the transformer is completely modeled by a
current injection computed from the phi=f(i) characteristics

Figure 9:piecewise segment flux-current characteristic of the magnetization branch of a saturable transformer

In SPS in particular, one can specify a residual flux only when the segment 1-2 has infinite slope as
mentioned in the SPS documentation.

3.3.2 ARTEMiS transformer model

In ARTEMiS, a slightly different approach is used that modify the current injection curves by including
the linear part of the magnetization curve inside the state space equations describing the system. The
modification are as follow:

18

Q022014-01

ARTEMIS User Guide

Advantages of the approach

1-The first segment of the phi=f(i) characteristic is included in the linear part of the state-space system
described by ABCD matrices.
2-This linear part is extract from the original phi=f(i) characteristic.
3-The flux across the branch is computed from its linear part phi_linear=L_linear*I_linear
4-A current injection in parallel to the linear inductive branch is used to model the saturation.

Figure 10:Modified injection characteristic in ARTEMiS caused by the inclusion of the first segment in the linear
part of the state-space system

The method can be viewed as follow: in normal mode (non-saturated), the magnetization branch is part
of the ABCD state-space system and the branch flux phi is obviously equal to L*i. When saturation
occurs, it is like connecting other inductance in parallel to the first one. The important thing to notice is
that the voltage across these two inductance is the same, so is the total flux that would be obtain by
integration of the voltage across the branch and therefore this flux can be derived from the linear
branch and used for current injection.
The differences with SPS native model are the following:
1-The ARTEMiS saturable transformer model requires a non-infinite 1 segment slope to so a state can
exist in the ABCD matrices. If not, ARTEMiS will add a very large one.
2-Residual flux can be specified even if the first segment do not has an infinite slope. The implication of
this is that the flux will move from the start of the simulation but in a very slow manner because of the
very high inductance. The model is therefore adapted to transformer re-energization tests.

3.3.3 Advantages of the approach

The main advantage of the ARTEMiS model is that is can provided accurate fixed-step simulation results
without algebraic loops. In SPS, this algebraic loop is caused by the usage of a discrete-integrator
(trapezoidal method) in the transformer itself. In ARTEMiS, this flux is computed in the linear part of
the state-space system.
SPS provides ways to break this algebraic loop but this can degrade accuracy. See the demo section for
more details.

ARTEMIS Users Guide

Q022014-01

19

Demos

Using ARTEMiS

3.3.4 Demos
ARTEMiS provides demos linked to the saturable transformer model.
artemis_power_ctsat.mdl: single phase transformer energization test.
The demo shows an increased accuracy of ARTEMiS over SPS at a time step of 50s. The flux response
of SPS is wrong at 50s. The ARTEMiS response at 50 s matches correctly the SPS response at 1 s.

Figure 11:SPS response for artemis_power_ctstat.mdl at 50s

Figure 12:SPS response at 1 s matching the ARTEMiS response at 50 s

20

Q022014-01

ARTEMIS User Guide

Initial flux setting

3.3.5 Initial flux setting

Since ARTEMiS 5.1.4, the initial flux of the transformer can be specified in the transformer mask. In difference
with the SPS method, ARTEMiS magnetisation branch is part of the ABCD state-space equation of the simulated
system and initial states are set by the state variables.
However, some SPS transformer model dont allow the initialisation of magnetization flux. The following table lists
what type of transformer support initial flux setting thought the transformer mask. When not supported, the user
must set manually the magnetization inductance initial current in the POWERGUI panel of SPS.

Table 2: List of SimPowerSystems transformer model (R2008 A-B)

Model

Direct Mask initial flux setting support

Saturable transformer

no

Multi-winding transformer

no

Zigzag Phase-Shifting Transformer

no

Three-Phase Transformer (Three Windings)

yes

Three-Phase Transformer (Two Windings)

yes

The ARTEMiS demo untitled:

Fixed-time step simulation of 3 phase saturable

transformer without algebraic loop explains how to compute and set manually the initial flux of a

transformer through the Initial States panel of the SPS POWERGUI.

3.3.6 Limitations of the approach

When testing case with inrush current caused by residual flux, the approch can cause a small flux decay du to the
non-inifinite linear part of the transformer core inductance.

3.4

References
[1] C. Dufour, S. Alma, S. Cuni, G. Scrosati, G. Valvo, G. Sapienza, "Renewable integration and protection studies
on a 750-node distribution grid using a real-time simulator and a delay-free parallel solver", CIRED-2015, Lyon,
France, June 15-18, 2015
[2] C. Dufour, L.-A. Grgoire, J. Blanger, "Solvers for Real-Time Simulation of Bipolar Thyristor-Based HVDC and
180-cell HVDC Modular Multilevel Converter for System Interconnection and Distributed Energy Integration",
2011, CIGR conference proceedings, Recife, Brazil, April 3-8, 2011.

ARTEMIS Users Guide

Q022014-01

21

References

Using ARTEMiS

22

Q022014-01

ARTEMIS User Guide

State-Space Space (SSN) solver tutorial

This section explains how to use the SSN solver of ARTEMiS in typical applications. Examples of SSN nodes
selection are given and some advanced ussage question are answered at the end.

4.1

Introduction
The State-Space Nodal (SSN) method can be considered as a nodal method. The main difference is how
the nodal branch or groups are made. In SSN, the user selects the way the groups are made. These
groups are computed by a state-space method while the interface between the groups is solved by a
nodal method. By making large groups for example, the number of equivalent nodes to be solved by
the nodal method can be limited. At the same time, by choosing wisely the groups, the number of
switches per groups can be limited and full-precalculation can be made. See [1] for a detailed
explanation on the SSN theory.
Within the Simulink/SimPowerSystems environment, the SSN presents some challenges for the normal
user to achieve real-time simulation. The main challenge is to correctly designed the SSN model using
SSN Nodal Interface Blocks to make groups of reasonable state and number of switches and to also
limit the number of total nodal nodes connecting these groups.
The SSN also includes powerful features like:
1- Inlined interpolation of thyristor firing
2- Inlined Interpolation of voltage inverter in a manner similar to the RTeDRIVE models (TSB).
3- Real-time Impulse event detection.
4- Iterative method for MOV and switches.
These will be explained through a series of example.

4.2

The SSN library

The ARTEMiS library contains the nodal interface blocks along with some other utility blocks used in the
SSN algorithm.

ARTEMIS Users Guide

Q022014-01

25

The SSN library

State-Space Space (SSN) solver tutorial

Figure 12:NIB is the SSN library

The blocks seen in Figure 1 are called 'SSN nodal interface blocks' or simply NIB. They represent the
nodes of the nodal method used in SSN. These SSN nodes connect state-space described groups that
must respect some causality laws. For example, in state-space approach, one cannot connect a current
source in series with an inductance. Similarly, the SSN nodal interface blocks must respects the same
laws. To achieve this, the block port has a type I (for current source) and V (for voltage source) and
this type must be chosen to respect causality laws. Taking the 1-phase SSN nodal interface block as an
example: the block has an I-type port and V-type port as selected on its dialog box below
.

Fig 1a Dialog box of the 3-ph SSN nodal interface block

26

Q022014-01

ARTEMIS User Guide

Usage of the SSN Nodal Interface Block in a model

4.3

Usage of the SSN Nodal Interface Block in a model

Fig. 2 shows the usage of SSN blocks in a SPS models. The model used for this test is named
'ArtemisSSN_simple_switched_case.mdl'.

Figure 13:Example of nodal interface blocks

Some basic rules are to be followed when using the SSN blocks.
1-SSN-nodal interface block (NIB) connected to inductive groups must have V-type port (view it has a
Voltage source connected to an inductive element. V-type NIB contains a virtual voltage source)
2-SSN-nodal interface block (NIB) connected to capacitive groups must have I-type port (view it has a
current source connected to an capacitive element, I-type NIB contains a virtual current source)
3-The Main ARTEMiS Guide must have 'Enabled State-Space Nodal (SSN) method' set. Note that the
solver used for the SSN method is trapezoidal like in EMTP-type software. The ARTEMiS 'Discretization
method' item only apply to part of the network that do not use the SSN solver, like the inverter side of
the HVDC model of the next section.

ARTEMIS Users Guide

Q022014-01

27

Disabling SSN (for validation purposes in pure SPS)

State-Space Space (SSN) solver tutorial

Figure 14:Fig. 3SSN and ARTEMiS block options.

4.3.1 Disabling SSN (for validation purposes in pure SPS)

For comparison purposes, if you disable the 'Enable State-Space Nodal method (SSN)' checkbox, the
model will run using standard ARTEMiS method with the SNN nodal interface blocks still inside the
model. This is because the SSN nodal interface blocks are simply null current/voltage sources that do
not change the simulation when the SSN method is turned off.

4.4

1st Real-life case: 12-pulse HVDC system

The SSN method will now be shown on a 12-pulse HVDC system. The HVDC case is interesting because
it offers many possibilities as how make the groups. The HVDC system is also interesting because it
contains many switches: 2 6-pulse valve groups and possibly 20 or more switched filter banks on ACbus and DC-bus. The SSN method was designed in mind to cope with this type of real-time simulation
challenge.

28

Q022014-01

ARTEMIS User Guide

1st example of SSN groups

Figure 15:AD_GRID04 12-pulse HVDC model.

4.4.1 1st example of SSN groups

The most basic SSN separation we can make to use the SSN method is to use the filter bank connection
point as a SSN node. Consequently, we need to understand the causality of the groups we are going to
define from this (3-phase) node.

Figure 16:Interface type from the chosen SSN node

One can make the following observations: the transformer has an inductive impedance as seen from
the SSN node. The source also has an inductive type impedance. Finally, the filter group has a
capacitive impedance as seen from the SSN node because one of its component is a simple capacitor.
The SSN method was applied here only to the rectifier side of the HVDC system. The inverter side is still
simulated by standard state-space method of SPS/ARTEMiS. The SSN could be applied also to the
inverter side needless to say.

ARTEMIS Users Guide

Q022014-01

29

Adding groups

State-Space Space (SSN) solver tutorial

Using the filter connection point as a node, we end up with 3 SSN groups, described on the figure
below.

Figure 17:Resulting groups with 3 SSN nodes.

4.4.2 Adding groups

In this example, we increase the number of groups by separating the transformer from the thyristors
valves, producing an additional group and 9 nodes in total. From a real-time simulation perspective, the
addition of 6 nodes will slow down the simulation but separating the transformer from the valves will
produce much smaller group equations. Especially, the states of the transformer will no longer have to
be precomputed 2^12 times with the valves.
Also, because the thyristors have RC snubber attached to them, they are better considered as a
capacitive group requiring an I-type SSN interface.

30

Q022014-01

ARTEMIS User Guide

Separating the valves groups

Figure 18:HVDC system with 4 groups and 9 nodes

4.4.3 Separating the valves groups

Making the two 6-pulse valves groups as 2 SSN groups has the advantage that memory requirements
are minimize because there are only 2^6=64 permutations per group instead of 2^12=4096 for one
12-valve group. It may even allow the simulation to run entirely inside the L1 or L2 cache of microprocessors, so it may speed up the simulation even if we now have 11 SSN nodes.

Figure 19:HVDC system with 6 groups and 11 nodes

ARTEMIS Users Guide

Q022014-01

31

Adding switched filter banks

State-Space Space (SSN) solver tutorial

4.4.4 Adding switched filter banks

Figure 20:HVDC system with switched capacitor banks.

This last example is interesting because we added two switched filter banks to the AC-bus of the model
without adding any nodes! This is caused by the facts that all the filter SSN groups (including the
breakers) are connected to existing SSN node in the circuit.

4.5

2nd case: 3-level NCP inverter and SSN Real-Time Impulse Events
The SSN algorithm is able to detect Impulse Events during simulation. By Impulse Event, we mean the
instantaneous opening or closing of a switch (most often a diode) following the open or closing of
another switch in the system. This happens for example in a buck converter in which the free-wheeling
diode turn-on instantaneously when the forced switch (IGBT or MOSFET) opens.
In real-time simulation it happens that this type of event is difficult to simulate accurately. The reason
is that switch natural conduction conditions are usually evaluated at the beginning of a time step, so if
a forced switch change state, its effect is can only be detected on the next time step.
In ARTEMiS and ARTEMiS-SSN algorithm, we use the fact that the state of a system cannot change
instantaneously when a switch changes of conduction state. We can therefore re-evaluate the switch
voltage after any forced switching by simply re-evaluating the outputs of state equations.
In the ARTEMiS-SSN algorithm, some caution is to be taken for the Impulse Event Detection to work
correctly because this Impulse Event verification is made on a group basis.

32

Q022014-01

ARTEMIS User Guide

2nd case: 3-level NCP inverter and SSN Real-Time Impulse Events

Figure 21:Three-level NCP inverter system in SSN

The above figure depicts a 3-level Neutral clamped inverter drive system in SimPowerSystems and
SSN. Each arm is composed of 4 IGBT/Diode pairs plus 2 clamping diodes.
The real-time simulation of this model is really challenging because it is composed of 30 coupled
switches. In the solution above, since SPS has a switch model for the IGBT/Diode pairs, the internal
number of switches reduce to 18, which make real-time simulation still impractical because of the high
number of matrix permutation to compute (2^18). The solution in SSN is to put each arm in a separate
group of 10 switches (6 internal SPS switches, considering the IGBT/Diode pairs as one device).

Figure 22:SSN group separation for real time simulation

ARTEMIS Users Guide

Q022014-01

33

Impulse Events in SSN

State-Space Space (SSN) solver tutorial

The resulting model has 5 groups and 6 nodes. The inverter was separated at the arm level to obtain 6
SPS switches per group, which can be precomputed and run in real-time after. Note that the NIB
blocks ARE the nodes in SSN.
Key aspect of the separation with Impulse Events: the NIB that connects to the output of the inverters
(A connector) are of I-type to provide the group with an image of the continuity variable during
impulse events, that are the transformer input currents. This so-called continuity variable is the model
variable that will not force the instantaneous switching in the inverter. In the above example model,
suppose that all IGBT are turned-off suddenly. The transformer input current will surely drop but
without discontinuities. Its this current that will force some anti-parallel diode in the inverter to turn-on
momentarily. Such methods was actually used by ABB to simulate such 3-level NPC converters [2].

4.5.1 Impulse Events in SSN

This last point is important to understand. It is caused by the fact that the SSN algorithm does not
make multiple iteration of equation to verify Impulse Events like instantaneous diode turn-on effects. It
only re-evaluate the Outputs of a group for natural switch threshold crossing each time a forced switch
is activated. This can be done on the basis that the states of a systems cannot change instantaneously
on a switching action.
In general, a switched device using diodes as free-wheeling diode (for example) will have a branch that
force the continuity of the current at switching time. This element must either be grouped with the
switching elements (best case) or the NIB must have a I-type interface to give an image of this element
in the groups for the SSN Impulse Event Detection to work. In the example of the 3-level NCP inverter,
this element is the transformer primary leakage inductance.

4.6

3rd case: Inlined Thyristor Valve Compensation in SSN

The Inlined Thyristor Valve Compensation (ITVC) method is a real-time method to compensate the
sampling effect of thyristor by the fixed step time frame. Simply explained, each time a thyristor firing
pulse is generated, it must wait the next time step to be taken into account inside the simulation. If
the pulse arrive just before the fixed step frame, the error is minimal but when it occurs just after, then
the error is bigger because the wait last almost a full time step. Because the firing pulse are not
synchronized on the simulation time step, it usually results in a low-frequency jitter on important
system variable, often confused with controller instability.
The ITVC methods is designed to compensate this effect, in off-line and HIL simulation. It is so efficient
that it is always active in the ARTEMiS (State-Space and SSN).
We will explain the ITVC method on the HVDC example with 6 groups, 11 nodes SSN separation.

34

Q022014-01

ARTEMIS User Guide

3rd case: Inlined Thyristor Valve Compensation in SSN

Figure 23:HVDC system with ITVC (RT-LAB top level separation in 3 CPUs)

In the model, a Firing Pulse Unit was designed with RT-EVENTS, a replica of the Simulink FPU with RTE
blocks.

ARTEMIS Users Guide

Q022014-01

35

3rd case: Inlined Thyristor Valve Compensation in SSN

State-Space Space (SSN) solver tutorial

Figure 24:RT-EVENTS based Firing Pulse Unit

The RT-EVENTS blockset enables to keep in memory the in-step events of this type of firing pulse unit with
multiple comparators.
Since release 6 of ARTEMiS, the way RT-EVENTS connects to ARTEMiS solvers has been simplified. The ARTEMiS
solver now requires only a double value between zero and 1 to activate and compensate thyristors switches. If
the value equal only exactly 1 and 0 (as in regular SPS), the simulation is not compensated. But if the value is
between 0 and 1, the value is taken as the time ratio of the gate event within the time step. Ex: a value of 0.6
would mean that the event occurred at 60% after the beginning of the time step.
Now, in common model a simple RTE converter block will do this job as in the following figure. If and only if the
RT-EVENTS compensation item of the block is set to Enabled, the HVDC simulation will also be compensated.

36

Q022014-01

ARTEMIS User Guide

3rd case: Inlined Thyristor Valve Compensation in SSN

Figure 25:Interface of RT-EVENTS and ARTEMiS (V6 and later)

The ITVC algorithm action is very impressive considering it overall negligible computational cost. The following
figure shows the DC current of the HVDC during energization.

ARTEMIS Users Guide

Q022014-01

37

Static Var Compensator

State-Space Space (SSN) solver tutorial

Figure 26:HVDC energization

If we now look closer at the Idc current and rectifier firing angle, the effect of the compensation is quite obvious.

Figure 27:Zoom on DC-link current and firing angles at of the rectifier side

On the above figure, we observe a very characteristic low-frequency jitter on both DC-link current and firing
angle, quantities linked by the HVDC control. When the ITVC is OFF, there is a approximate 10 Hz jitter on both
values that is not present with ITVC in function. This jitter is typical of fixed-step solvers and would be present in
all fixed-step based simulation algorithms (EMTP, PLECS, SPS, PSIM, etc...).

4.7

Static Var Compensator

A 300-Mvar Static Var Compensator (SVC) regulates voltage on a 6000-MVA 735-kV system. The SVC consists of
a 735kV/16-kV 333-MVA coupling transformer, one 109-Mvar thyristor-controlled reactor bank (TCR) and three
94-Mvar thyristor-switched capacitor banks (TSC1 TSC2 TSC3) connected on the secondary side of the
transformer. Switching the TSCs in and out allows a discrete variation of the secondary reactive power from zero
to 282 Mvar capacitive (at 16 kV) by steps of 94 Mvar, whereas phase control of the TCR allows a continuous
variation from zero to 109 Mvar inductive. Taking into account the leakage reactance of the transformer (15%),
the SVC equivalent susceptance seen from the primary side can be varied continuously from -1.04 pu/100 MVA
(fully inductive) to +3.23 pu/100 Mvar (fully capacitive). The SVC controller monitors the primary voltage and
sends appropriate pulses to the 24 thyristors (6 thyristors per three-phase bank) in order to obtain the
susceptance required by the voltage regulator [1].
Each three-phase bank is connected in delta so that, during normal balanced operation, the zero-sequence tripple
harmonics (3rd, 9th...) remain trapped inside the delta, thus reducing harmonic injection into the power system.
The power system is represented by an inductive equivalent (6000 MVA short circuit level) and a 200-MW load.
The internal voltage of the equivalent can be varied by means of programmable source in order to observe the
SVC dynamic response to system voltage sags.

38

Q022014-01

ARTEMIS User Guide

Static Var Compensator

Figure 28:SVC compensated electric network

With the SSN solver, the natural way to decouple the system is to use the common connection point of the TCR
and the 3 TSCs, resulting in 4 groups of 6 switches each and nodal matrix of size 3 only, thus very efficient in
computational terms. The TCS groups are interfaced with I-type SSN Nodal Interface Blocks while the TCR and
network group is interfaced with a V-type block (hint: it is clearly inductive so it must be driven by a Voltage
source for causality reasons ->V-type).
With the ITVC compensation of thyristor firing, very accurate simulation ca be achieved. The above figure shows
the simulation results for a slow scan of the TCR bank firing angle. The figure below shows a typical effect of
thyristor-based system in fixed step simulation. In that case, a kind of quantization effect occurs on the system
output reactive power, as it shows some discrete step effects. With the ITVC compensation of ARTEMiS and SSN,
the reactive output of the systems is smooth with regards to the firing angle.

Figure 29:Effect of firing compensation in ARTEMiS

ARTEMIS Users Guide

Q022014-01

39

Inlined Voltage Inverter Compensation (IVIC)

State-Space Space (SSN) solver tutorial

4.8

Inlined Voltage Inverter Compensation (IVIC)

In ARTEMiS v7, 2-level inverter (2LI) can be compensated inline in SSN, which results in compensated
simulation similar to the well-known TSB of OPAL-RT but without the internal delays associated with TSB.
See the online demo ssn_itsb_2levelvsc_hvdclink.mdl for more details.
Note that IVIC does not work for 3-level NPC inverters (3LNPCI). However, there is a demo called
ThreeLevelInverter_IVIC_RTE2.mdl that use IVIC method in a surrogate circuit of the 3LNPCI.

4.9

SSN parallel (without delays)

Starting with version 6.3 of ARTEMiS, the SSN solver is now capable of being executed on multiple processors. In
most cases, this will significantly improve the performance of the solver.
Note that this mode requires RT-Lab 10.3 or later.
The parallel mode is activated only if multiple cores are allocated for a specific subsystem. To allocate additional
cores, use the advanced view of the assignation page:

Figure 30:Subsystem settings page

Select the ... in the Cores column to activate the core selection dialog. There, select the number of cores and
optionally, select specific cores. Unspecified cores are allocated automatically by the system using the assignation
rules described in the RT-Lab User Guide.

40

Q022014-01

ARTEMIS User Guide

Iterative methods in SSN

Figure 31:Subsystem Core Selection Dialog

When multiple cores are allocated for a given subsystem, SSN groups are automatically divided among available
cores so that the total computation time is minimized. Thus, the number of groups and their relative size will
have a direct impact on the possible performance improvements for the parallel mode. In some rare cases, the
situation could be worse or without significative gains. In any cases, some experimentation is often needed to
find the optimal configuration.

4.10 Iterative methods in SSN

This section explains the new iteration feature of ARTEMiS (available with version 7 and later). With this option, it
is now possible to run real-time simulation of Surge Arresters (often called Metal Oxide Varistor or MOV) and
common switches with iterations to ensure the global accuracy of the simulation. These iterations are different
than the ones that SSN makes inside each group. The new iteration method are global, meaning that a change of
segment or operating point in one SSN group will force the global check for iteration among all iterative groups.
The use of this new iteration feature is determined, for MOV, by the use of a specific MOV block in the ARTEMiS
Library and, for switches, of a specific NIB connection. We call these simulation methods iMOV and iSWITCH
respectively.

4.10.1iMOV
To use iMOV in SSN, the user must use the special iMOV block provided in the ARTEMiS->SSN->Iterative Models
library. The on-line demo SSN_iMOV_power_surgnetwork.mdl is used next to illustrate this in more details.

ARTEMIS Users Guide

Q022014-01

41

iSWITCH

State-Space Space (SSN) solver tutorial

Figure 32:iMOV online demo SSN_iMOV_power_surgnetwork

Basically, the iMOV model must be connected directly to X-type NIB anywhere in the model. Provided that the
Maximum number to iteration (iMOV and iSWITCH) parameter of the ARTEMiS GUIde is set high enough
(typical value of 5 is ok for mots cases), the iterative solution of SSN will be used to solve this network. The
MATLAB prompt should also highlight that iMOV are used in the simulation.
Third-Party Rule block detected: SSN_iMOV_power_surgnetwork/ARTEMiS Guide1
ARTEMiS-SSN: approx. memory required: 0.017752 Mb (including nodal matrix)
SSN group info
Group 1 : 12 states, 9 inputs, 9 outputs, 2 switches.
Group 2 : 0 states, 4 inputs, 2 outputs, 0 switches.(SSN Iterative MOV)
Group 3 : 0 states, 3 inputs, 2 outputs, 0 switches.(SSN Iterative MOV)

4.10.2iSWITCH
In the case of iSWITCH, the standard SPS switches are used. The connection method determines if the iSWITCH
method is used by SSN. This is illustrated in the next figure with the online demo
SSN_iSWITCH_6pulse_rectifier.mdl.

Figure 33:iSWITCH online demo SSN_iSWITCH_6pulse_rectifier

For an iSWITCH model and method to be detected by SSN, the switch must be:
1- alone in its SSN group
2- interfaced with V-type NIB
3- have no internal snubber capacitor, nor inductance, i.e. the switch must not contain any L-C state.
42

Q022014-01

ARTEMIS User Guide

Miscellaneous solver options

This is the case for the SSN_iSWITCH_6pulse_rectifier.mdl demo as shown in the figure below.

Figure 34:iSWITCH connection and parameters requirements

When an iSWITCH is detected by SSN, the MATLAB prompt should highlight it correspondingly like the following:
SSN group info
Group 1 : 2 states, 2 inputs, 4 outputs, 0 switches.
Group 2 : 9 states, 6 inputs, 9 outputs, 0 switches.
Group 3 : 0 states, 4 inputs, 3 outputs, 1 switches.(SSN Iterative switch group)
Group 4 : 0 states, 4 inputs, 3 outputs, 1 switches.(SSN Iterative switch group)
Group 5 : 0 states, 4 inputs, 3 outputs, 1 switches.(SSN Iterative switch group)
Group 6 : 0 states, 4 inputs, 3 outputs, 1 switches.(SSN Iterative switch group)
Group 7 : 0 states, 4 inputs, 3 outputs, 1 switches.(SSN Iterative switch group)
Group 8 : 0 states, 4 inputs, 3 outputs, 1 switches.(SSN Iterative switch group)

4.11 Miscellaneous solver options

The SSN solver has been improved since ARTEMiS version 7.x with the advent of iterative models and methods.
The modifications are related to the node order within the nodal matrix. Nodes connected to iterative models are
ordered last, more to the bottom/right of the admittance matrix. This enables an efficient partial refactorization
of the admittance matrix during iterations. Nodes connected to non-iterative switched SSN groups are ordered
just before these. At the top/left of the admittance matrix are then the nodes of non-switched SSN groups. Since
these groups have constant discrete admittance, some form of pre-factorization can be made on these nodes of
the admittance matrix.

ARTEMIS Users Guide

Q022014-01

43

Explanation of the reason of improved speed with SSN

State-Space Space (SSN) solver tutorial

The previous SSN solver came with an optional routine to re-order the nodes in a tri-diagonal way. This solver is
normally disabled in ARTEMiS v7+. In case that this old solver is needed, the following variable must be declared
in the MATLAB workspace during simulation and compilation.

op_ssn_use_lu_tridiagonal_method=1;
The value 1 is not important, only the existence of the variable is checked. Note that the variable is now in
lowercase letter.

4.12 Explanation of the reason of improved speed with SSN

This section objective is to explain the fundamental reasons behind SSN speed improvement over regular statespace methods, in both real-time and non-real-time applications.
The reason is simple: SSN make less multiply-addition operation for the same network.

B
rk0

xn+1=

A
k

xn +

S
S
Nm
eth
o
dw
ith2 g
ro
u
p
so
f x/2stateseach
B
rk0

B
rk1

B
k

m
atrixsets
un+1
k=
1...64

A
1m

xn+1=

A
2n

U_no

S
tatesp
acem
eth
o
dw
ithxstates

B
1m
xn +

I_non+1= Y
3,3

B
rk1

B
2n

m
atrixsets
un+1 m
=1...8
m
=1...8

u_non+1

Figure 35:State-Space vs. SSN comparison

In Fig. 35, we compare SSN and state-space (SS) for a system with x states and y inputs. In SS there is
therefore:
x*x+x*y mult-add operations (operations in SS method)
to be done at each time step. In the SSN method, two SSN groups separated by 3 nodal connection points
(separating the complete network in two equal part let say) are defined. In this case, the algorithm would imply
2 state-space system iterations with x/2 state variables each, plus a 3x3 nodal matrix inversion. This results in:
(x*x+x*y)/2 mult-add operations (operations in SSN method)
plus 3x3 matrix factorization.
If the number of SSN nodes is small compared to x, then SSN cut approximately by half the total number of
operations required by the state-space iterations.
More generally, if the Q SSN groups are balanced in size, the state-space part of SSN will be Q times faster than
pure state-space. Then the challenge is to keep the admittance matrix small. Full LU factorization of an
admittance matrix is O(r 3) operation , where r is the rank or row size of the admittance matrix.
Since ARTEMiS v7.0, the SSN LU factorization routine is made according to the 1969 Dommel paper, in which only
modified row/column of the admittance matrix are refactorized. Therefore, the real criteria in SSN is to keep the
switched nodes of the admittance matrix small. For example, applications in voltage control with few switches
and many controlled PQ injections can have large admittance matrix without noticeable impact of the real-time
computational speed. This is because unswitched row/column of the admittance matrix are pre-factorized before
going into the real-time loop.

44

Q022014-01

ARTEMIS User Guide

SSN Switch management in Real-Time applications

4.12.1 SSN Switch management in Real-Time applications

Fig. 35 shows that the SSN method makes the complete state-space equations block-diagonal by virtue of the
decoupling induced by the approach. The decoupling is partial, however, as the nodal solution links all parts of the
network equations. Using a single CPU approach, the computational gain can approach a factor of 2 if the statespace iteration becomes more important, in comparison to the same 3x3 matrix inversion.
Furthermore, the problem of memory storage of switch permutations is solved here: each group contains only the
pre-calculated set of matrices for the switch contained within the groups. Taking again the specific example of
Fig. 35, which is composed of two three-phase breakers, full pre-calculation of circuit modes in the standard
state-space approach would require the storage of
2^6=64 permutations of states-space equations of size x,u
In the SSN approach, two sets of 2^3=8 system matrices need to be stored (one for each group),that is:
2*(2^3)=16 permutations of states-space equations of size x/2,u/2
plus 3x3 admittance matrix.
thus drastically reducing memory requirements. As the SSN groups and switches are determined by the user, he
therefore has full control on the memory requirements.
Separation of switches is always possible because they can be modeled as a separate group in the proposed
method. In that case, only the D matrix subsystem is non-empty, and the group admittance is included directly in
the global admittance matrix in way similar to standard nodal method.

4.13 Useful tricks and practical issues with SSN

SSN is build on top of SimPowerSystems, which is a closed code. Therefore, there exists some limitations in using
SSN that are explained here. Other limitations come from the fact that SSN is a nodal method.

4.13.1

Which SSN solver should I use?

On the SSN solver pane, many solvers are available. It may happen that one specific solver works better than
others without clear reasons. (The same is true about regular Simulink solvers by the way).This is explained
briefly in [4].
Trapezoidal is the same than EMTP but without Backward Euler steps at switch change. Beware of numerical
oscillation!
Backward Euler: Backward Euler is probably the most stable solver of all, actually an L-stable one. The Backward
Euler method is known to damp too much some components of simulation however.
art5 is the order 5 L-stable Pad approximant to the matrix exponential. In this solver, the input term is
considered to vary linearly between time steps.
art5 with Backward Euler nodal interface: this is the same as art5 for the discrete A matrix. But the discrete B
matrix is different because the input term is considered fully implicit (u[n+1] only, no u[n] term) during
integration. This solution is often more stable numerically than art5.
art5 with balanced zero hold nodal interface: his is the same as art5 for the discrete A matrix. But in this case the
input term is considered as constant and taking 2 values during integration (half the step at u[n] and the next
half v[n+1]).
The art5 variations are provided in case the art5 method does not work well.

ARTEMIS Users Guide

Q022014-01

45

Avoid extreme values of parameters

State-Space Space (SSN) solver tutorial

4.13.2

Avoid extreme values of parameters

Use realistic Ron breaker resistance of 1e-4 Ohms, not 1e-12. Extreme values may make the nodal admittance
matrix badly conditioned.
Also avoid unrealistically high transformer core values.

4.13.3 Current sensors direction error

When using current measurements, it can happen that SPS tries to takes this measurement in a branch that is
located in a different SSN group. The current implementation of SPS current measurements blocks looks for
branches on the + side of the sensor.
When this happens, a clear error message will appear and will explain to you that you must invert the current
measurement ports and add a Simulink gain of -1 to the output of the block, giving the exact same value of
current reading.

4.13.4 How to choose V-type, I-type and X-type of NIB?

X-type is simple: its only used for custom SSN groups and frequency dependant line models [3].
V-type and I-type must be chosen to avoid illegal connection of current source and voltage source in the
determination of state-space equations. This is because of the underlying state-space formulation used to
determine SSN groups equations.
In many case, but V-type and I-type are valid (they are if they do not trigger SPS connectivity errors). But
typically, V-type NIB, which contains a virtual voltage source, is better when connected to inductive inputs of SSN
groups and I-Type NIB, which contains a virtual current source, is better when connected to a capacitive input of
groups, like a shunt capacitor.

4.13.5 Why cant I use SPS Multimeter blocks in SSN?

The reason is that in the current implementations of SPS, the presence of SPS Multimeter block force the merging
of all the circuits in the diagram even if they are independent. SSN cannot currently deal with this situation.

4.13.6 Illegal current source in series error message when using I-type NIB.
The I-Type NIB block contains a virtual current source, which is there for SSN to determine SSN groups. This
source is visible by SPS and it sometimes create connectivity error by which this current source is put in series
with another current source or an inductor. (Physically, current sources are not supposed to be connected in
series with inductor, its like trying to force a system state to a known value)
In ARTEMiS v7, I-type NIB comes with a shunt resistor of high value (default 1e9 Ohms), that do not interfere wit
h the simulation accuracy and avoid this connectivity error.

4.13.7 PC Memory limitation in SSN

When compiling an ARTEMiS model, lack of memory may occur, especially when the size of the SSN
group matrices ABDC are big and the number of switches is high.
ARTEMiS tried to calculate all matrix permutation in all SSN groups of the electrical network. For
example, for a very small network (only 10 states, 12 inputs and 11 outputs), the state-space matrices
would be like:
x' = Ax+Bu

(A is [10x10], B is [10x12], x is [10x1], x' is [10x1], u is [12,1])

Y = Cx+Du

(C is [11x10], D is [11x12])

And the number of switches is 15, the size of memory required is:
46

Q022014-01

ARTEMIS User Guide

Optimizing SSN groups for best real-time performance

( (10*10 + 10 + 10*12 + 12 + 10) + (11*10 + 10 + 11*12 + 12 + 11) ) * 2^15 * 8 = 138MB

Now the same calculation for a network with 40 states, 40 inputs and 40 outputs is 1.76GB.
In SSN, the number of switch per SSN group is limited to 15 but should ideally be less than 10. It is ALWAYS
POSSIBLE in SSN to limit the number of switch per group to a certain number, just by putting switches in
different groups. The MATLAB prompt gives an indication of the memory used by the SSN matrices, including the
nodal admittance matrix. As a general guideline, this memory usage should be kept low whenever possible. Value
under 100 Mb are generally ok on Xeon V2 or V3 computers.
.....
Y has 318 zeros out of 676 elements

ARTEMiS-SSN: approx. memory required: 71.4192 Mb (including nodal matrix)

SSN group info
Group 1 : 10 states, 24 inputs, 24 outputs, 12 switches.
Group 2 : 14 states, 21 inputs, 24 outputs, 9 switches.
Group 3 : 3 states, 9 inputs, 10 outputs, 6 switches.
Group 4 : 0 states, 18 inputs, 20 outputs, 10 switches.
Group 5 : 0 states, 15 inputs, 15 outputs, 6 switches.
Group 6 : 12 states, 28 inputs, 28 outputs, 10 switches.
SSN nodal matrix is of rank 26 (47.0414 % of zeros)
ARTEMiS network information: the network contains 127 general EMTP equivalent nodes (RLC branches, sources, etc..)

.....

4.13.8 Optimizing SSN groups for best real-time performance

When looking at an electric network to be simulated with SSN, two main criteria should be followed:
1- Try using nodes that will provide the higher SSN group number/number of SSN nodes ratio.
For example, choosing an SSN node at the feeder point of star-connected distribution grid (Figure 36) will create
5 SSN groups while using a single (3-phase)SSN node. This is an excellent choice. After this, the grey point are
the next best location of SSN nodes because they split each existing SSN group into 2.
The switched filter bank node of the bipolar HVDC example is another good example of excellent SSN node
location because it creates many SSN groups with a single SSN node.
2- Try to correctly balance the size of the SSN group.
This is better when using the parallel SSN option in real-time. To help localize a specific SSN group, one can
attached an SSN Group tag to different groups in the model, like shown in Fig.37. The MATLAB prompt listing
will then refer to the different tag used:
ARTEMiS-SSN: approx. memory required: 0.59939 Mb (including nodal matrix)
SSN group info
Group 1 : 3 states, 7 inputs, 8 outputs, 2 switches.
Group 2 : 21 states, 9 inputs, 6 outputs, 3 switches. TAG: capa
Group 3 : 12 states, 16 inputs, 18 outputs, 6 switches. TAG: xfo
Group 4 : 6 states, 7 inputs, 4 outputs, 1 switches. TAG: rect
SSN nodal matrix is of rank 4 (0 % of zeros)

ARTEMiS network information: the network contains 43 general EMTP equivalent nodes (RLC branches, sources....

ARTEMIS Users Guide

Q022014-01

47

Optimizing SSN groups for best real-time performance

State-Space Space (SSN) solver tutorial

Figure 36:Star-shaped distribution grid: best SSN node location

Figure 37:Online demo model: ssn_power_hvdc_itvc.mdl with SSN Group tag blocks
48

Q022014-01

ARTEMIS User Guide

References

4.14 References
[1] C. Dufour, J. Mahseredjian , J. Blanger, "A Combined State-Space Nodal Method for the Simulation of Power
System Transients", IEEE Transactions on Power Delivery, Vol. 26, no. 2, April 2011 (ISSN 0885-8977), pp. 928935
[2] W. Wang, J. Blanger, C. Dufour, A. Douzdouzani, "A Novel and Flexible Test Stand for Medium Voltage Drives
Using a Hardware-in-loop (HIL) Simulator", 2011 PCIM Europe Proceeding, Nuremburg, Germany, May 8-10,
2011
[3] Ramos-Leanos, O.; Naredo, J. L.; Mahseredjian, J.; Dufour, C.; Gutierrez-Robles, J. A.; Kocar, I.; "A
Wideband Line/Cable Model for Real-Time Simulations of Power System Transients," Power Delivery, IEEE
Transactions on , vol.27, no.4, pp.2211-2218, Oct. 2012
[4] C. Dufour, J. Mahseredjian, J. Blanger, J. L. Naredo, "An Advanced Real-Time Electro-Magnetic Simulator for
Power Systems with a Simultaneous State-Space Nodal Solver", IEEE/PES T&D 2010 - Latin America, So Paulo,
Brazil, Nov. 8-10, 2010

ARTEMIS Users Guide

Q022014-01

49

References

State-Space Space (SSN) solver tutorial

50

Q022014-01

ARTEMIS User Guide

Reference

This section describes the various blocks and functions provided with ARTEMiS.

ARTEMIS Users Guide

Q022014-01

43

Reference

44

Q022014-01

ARTEMIS User Guide

ARTEMiS Guide
Library
ARTEMiS (Advanced Real-Time ElectroMagnetic Simulator)

Block
The ARTEMiS Guide block is the main discrete simulation parameter control block of
ARTEMiS from which the different ARTEMiS solvers can be selected.

Figure 32:ARTEMiS Guide Block

Mask

ARTEMIS Users Guide

Q022014-01

45

ARTEMiS Guide

46

Q022014-01

ARTEMIS User Guide

Figure 33:Mask of the ARTEMiS Guide Block

Description
The ARTEMiS Guide block is used to discretize the linear part of the state-space system
generated by the SimPowerSystem blockset (SPS). It implements strictly fixed time step
simulation of SPS schematics and offers alternative to the Tustin discretization method of
the SPS to increase numerical stability and precision. In contrast to the simulation technique
of the SPS, the 'ARTEMiS Guide' block precomputes and discretizes all state-space matrices
for all combinations of the switch topologies thus permitting hard real-time simulation.
Since v6, ARTEMiS offers a new simulation algorithm called State-Space Nodal (SSN), which
combines the accuracy advantages of state-space methods together with the advantages of
nodal methods with regards to switch management. Basically, there is no switch count limit
with the SSN solver.
Since v6 also, the interpolation method have been changed to Inlined methods which are
more efficient in terms of calculation and more easy to use. The interpolation methods are
now active by default because of their simplicity.

Parameters
General tab
Sample Time (s): Sets the sample time for the fixed time step simulation of the electrical
part of the SPS model.This sample time should be the same as the one entered in the SPS
PowerGUI block.
Enable State-Space Nodal method: When checked, activates the use of SSN methods in
SPS subsystems where nodal nodes have been defined using SSN Nodal Interface Blocks.

ARTEMIS Users Guide

Q022014-01

47

ARTEMiS Guide

NOTE: the option Dynamic calculation of switch pattern matrix permutations is now obsolete
ARTEMiS. This option was provided in case of large switch count within a state-space method. With
SSN, this limitation does not exist anymore.
NOTE: the option RLC load substitution by Dynamic Load model is now obsolete. The option was
provided to deal with some flaws in the SPS load flow routines. Since R2011B, SPS comes with a new
load flow routine corrected these issues.

State-Space Solver Options tab

This option apply only to the parts of the model that dont use the SSN solver.
State-Space discretization method: Sets the discretization method used by the ARTEMiS
algorithm for the normal state-space system, not the one using SSN method. Four different
methods are available:art5 (default), art3, art3hd and trapezoidal. The art5 and art3
discretization methods are highly stable and very-accurate integration methods. Both are
immune to numerical oscillations caused by switch operations in power networks. The art5
method is theoretically more accurate than art3, as it approximates the matrix exponential
Taylor expansion to the 5th term, while art3 and trapezoidal approximate to the 3rd and
2nd terms, respectively. The art3hd discretization methods a highly-stable method with
good precision, especially in highly non-linear networks like the demo example provides with
SPS called power_surgnetwork. mdl. The art3hd method is the only integration method
capable of simulating the power_surgnetwork.mdl model with a time step greater than 90us.

SSN tab
These options apply only when the SSN solver is being used in a model.
SSN solver: type of solver used for the SSN method. The SSN algorithm solve a model as
two parts: state-space groups connected in a nodal method. The state-space groups can be
solved by state-space discretisation similar to standard ARTEMiS (art5 solver, an order 5
matrix exponential approximant) while the nodal part can be solved by Trapezoidal,
Backward Euler or Balanced-zero-hold, a mix of Backward and Forward Euler. The default
method is Trapezoidal. Other methods are provided for help only. In case of numerical
oscillations at nodal connection points, the Art5 or Backward Euler method can provide a
solution.
Disable SSN group iterations: SSN iterates the solution within each group normally. For
example, if 2 IGBTs are in the same SSN groups, iterations are made to find the correct and
coherent switch pattern at all time steps. This does not apply if the switch are in different
SSN groups. This option disable the internal group iteration process. This can accelerate
simulations in rare cases.
SSN Groups maximum number of iterations: This set a limit on the number of iteration
within each SSN groups. See Disable SSN group iterations option.
Maximum number of iterations (iMOV and iSWITCH): this option controls the number
of global SSN iteration when iMOV and iSWITCH models are detected inside an SSN model.
This option only apply to iMOV and iSWITCH models. See Iterative Model in SSN section
for more details.

48

Q022014-01

ARTEMIS User Guide

Advanced tab
Extinction angle computation (individual thyristor only): this option makes ARTEMiS
compute the extinction angle of the thyristors of the model. The computed extinction angle
are outputted at the 1 output of individual SPS thyristors (see figure below).

This output angle is a number between 0 and 1 representing the time elaspsed between the
previous time step hit and the actual zero-crossing of current that cuased the thyristor to
open. The option is also available with the SSN algorithm. See demos
artemis_ITVC_ExtinctionZeroCrossing.mdl and
artemis_ITVC_ExtinctionZeroCrossingSSN.mdl for more explanations.
Use continuous time machine models: this option will force SPS to use continuous-time
machine models inside the fixed-step simulation scheme. The machine are modeled with
Laplace integrators and the main Simulink fixed-step solver (ode1 to ode5) will be used to
iterate the machine models.
Distributed Parameter Line model type: this option allow to swap between ARTEMiS DPL
model or SPS DPL model. Only the latter one can be used for load flow calculations. The
ARTEMiS Distributed Parameter Line models are required to enable the parallel simulation of
subnetworks separated by them in the RT-LAB framework.

Inputs
None

Outputs
None

Other notes
Number of switches in ARTEMiS state-space solvers and SSN
The SSN method main purpose is to uplift the limitation on the number of switches that a
model can contain in state-space approach. There is always a SSN group separation method
that will allow full pre-calculation of all matrices and real-time simulation. Switches can even
be alone in a group by themselves like in EMTP.
With the ARTEMiS-SSN solver, the switch limitation of old ARTEMiS state-space solvers is
waived but the user must create groups with a limited number of switches to limit memory
usage by the stored matrix permutation inside the groups. Consequently, switch number
should be limited to reasonable number in each group(12 and lower for example per SSN
group).

ARTEMIS Users Guide

Q022014-01

49

ARTEMiS Guide

At the limit of SSN, one can mimic classic EMTP approach and use only one switch and
nothing else in a SSN group. This is perfectly valid in SSN but would produce huge nodal
admittance matrix and may be not efficient in real-time when there are many switches.

Interpolation methods
ARTEMiS v6 and later automatically incorporates many interpolation methods that were
previously manually enabled. There are 3 types of interpolation implemented in ARTEMiS:
Impulse Event Detection: This type of interpolation occurs when a forced switch action
instantaneously induce a limit condition on another natural switch like a diode. A good
example of this is in buck converter where the opening of a IGBT instantaneously put the
free wheeling diode in conduction. This type of event is now supported by default in
ARTEMiS v6 and later.
Inlined Thyristor Valves Compensation (ITVC, ITVC-SSN): this algorithm corrects the
firing jitter of thyristor valves caused by fixed step sampling of the gate signals. It
automatically activates if the gate signal is a double number ranging continuously from 0 to
1. The number (ex: 0.458) indicates the in-step delay since the last sample time. The
method deactivates if the number is the usual binary number used to control switches. The
method is implemented in both state-space and SSN algorithms.
Inlined Voltage Inverter Compensation (IVIC-SSN): available in SSN only, the IVICSSN method will compensate the simulation of voltage inverter modeled with SPS Universal
Bridge blocks in a matter equivalent to RTeDrive TSB blocks. It automatically activates if the
gate signal is a double number ranging continuously from 0 to 1. The number indicates the
in-step delay since the last sample time. The method naturally account for all working
modes of the inverter, including high impedance case. It must be used in conjunction with a
SSN-defined load (motor, filter, etc....) to work correctly.
Other types of interpolation: custom interpolation methods are proposed in some SSN
demos. See the demo ThreeLevelInverter_IVIC_RTE2.mdl fr example.

Direct Feedthrough

N/A

Discrete sample time

Yes

RT-LAB XHP support

Yes

Work offline

Yes

MATLAB version compatibility

Although RT-LAB is compatible with many MATLAB versions, the ARTEMiS plug-in for
SimPowerSystems is more limited due to fine variations in the SPS code.

ARTEMiS version

MATLAB Compatibility

7.0.x

R2011B

Related Items
ARTEMIS Distributed Parameters Line, ARTEMiS Stubline, ARTEMiS-SSN Nodal
interface Blocks.

50

Q022014-01

ARTEMIS User Guide

ARTEMiS Distributed Parameters Line

Library
ARTEMiS

Block
The ARTEMiS distributed parameters line block implements an N-phases distributed
parameters transmission line model optimized for real-time simulation.

Figure 34:ARTEMiS distributed parameters line block

Mask

Figure 35:Mask of the ARTEMiS distributed parameters line block

Description
The ARTEMiS Distributed Parameters Line block implements an N-phases distributed
parameters line model with lumped losses. The model is based on the Bergeron's travelling
wave method used by the Electromagnetic Transient Program (EMTP) [1]. This block is
similar to the SPS distributed parameters line block but is optimized for discrete real-time

ARTEMIS Users Guide

Q022014-01

53

ARTEMiS Distributed Parameters Line

simulation and allows network decoupling. It also allows multi-CPU simulation on an RT-LAB
simulator.
Refer to the SPS Distributed Parameter Line block Reference page for more details on the
mathematical model of the distributed parameters line.
ARTEMiS provides an m-script that converts the SPS distributed parameters line block to an
ARTEMiS distributed parameters line block. See the ARTEMIS Distributed Parameters
Line reference page for more details on this script.
Network decoupling
One of the main advantage of the ARTEMiS line blocks (Distributed parameters lines and
Stublines), by opposition to the SPS lines, is the decoupling of the electric circuit into smaller
subnetworks. This important property allows ARTEMiS to simulate, in real-time, circuit with
more switching elements.
SPS and ARTEMiS solve electric circuits using the common state-space method. One of the
main limitation of this method is related to the switch elements. When an event occurs that
changes the topology of the circuit (or change the state of a switch), SPS and ARTEMiS need
to compute a new state-space matrix. This calculation causes a non acceptable overhead
when simulating a circuit in real-time.
To solve this problem, ARTEMiS stores the state-space matrices of a given set of topologies,
normally the steady-state topologies, in cached memory and uses them when necessary
without having to recalcule the matrices. However, the number of matrices required to cover
all topologies of the system depends on the number of switch elements. When a circuit
contains a lot of switch elements, the number of required topologies is high and it is not
possible to store all matrices in cached memory because of the size of the matrices.
The decoupling property of the line allows ARTEMiS to divide the state-space system in two
different state-space systems and reduce the total size of the state-space matrices in
memory. It also reduces the maximum number of topologies by an important factor.
RT-LAB simulation using a cluster of PCs
The distributed configuration of RT-LAB allows for complex models to be distributed over a
cluster of PCs running in parallel. The target nodes in the cluster communicate between each
other with low latency protocols such as shared memory, FireWire, SignalWire or InfiniBand,
fast enough to provide reliable communication for real-time applications.
However, electrical circuit cannot be easily distributed over a cluster of PCs without changing
the dynamic behaviors of the system. The communication delays degrade the computation.
ARTEMiS lines (Distributed Parameters Lines and Stublines) can be used to distribute a
circuit over a cluster of PCs. ARTEMiS used the intrinsic delay of the line to split the circuit
without affecting the dynamic property of the system. Moreover, SPS and ARTEMiS use
physical modelling lines and connectors to model the circuit. This type of signals cannot be
used by RT-LAB to communicate signals between subsystems, because the RT-LAB opcomm
block only supports basic Simulink signals. The only exception to this rule are the ARTEMiS
Distributed Parameters Line block and the ARTEMiS Stubline block. RT-LAB allows the
insertion of a line block at the root level of the block diagram and the connection of the
physical modelling ports of the block to the real-time subsystems. Also note that the
physical modelling signals and ports do not have to pass through the opcomm block. The
Example in the Characteristics and Limitations section illustrates how to use the block with
RT-LAB.
54

Q022014-01

ARTEMIS User Guide

Parameters
Simulation mode: Defines the mathematical models of the distributed parameters line
used by ARTEMiS and SPS. Here are the available options:
SimPowerSystems: When this option is selected the block uses the
SPS mathematical model that is not optimized for real-time
simulation.
ARTEMiS model: When this option is selected the block uses the
ARTEMiS mathematical model that allows fast real-time simulation
and that allows network decoupling.
Number of phases N: Specifies the number of phases, N, of the model. The block
dynamically changes according to the number of phases that you specify. When you apply
the parameters or close the dialog box, the number of inputs and outputs is updated.
Frequency used for RLC specifications: Specifies the frequency used to compute the
resistance R, inductance L, and capacitance C matrices of the line model.
Resistance per unit length: The resistance R per unit length, as an N-by-N matrix in
ohms/km. For a symmetrical line, you can either specify the N-by-N matrix or the sequence
parameters. For a two-phase or three-phase continuously transposed line, you can enter the
positive and zero-sequence resistances [R1 R0]. For a symmetrical six-phase line you can
set the sequence parameters plus the zero-sequence mutual resistance [R1 R0 R0m]. For
asymmetrical lines, you must specify the complete N-by-N resistance matrix.
Inductance per unit length: The inductance L per unit length, as an N-by-N matrix in
henries/km (H/km). For a symmetrical line, you can either specify the N-by-N matrix or the
sequence parameters. For a two-phase or three-phase continuously transposed line, you can
enter the positive and zero-sequence inductances [L1 L0]. For a symmetrical six-phase line,
you can enter the sequence parameters plus the zero-sequence mutual inductance [L1 L0
L0m]. For asymmetrical lines, you must specify the complete N-by-N inductance matrix.
Capacitance per unit length: The capacitance C per unit length, as an N-by-N matrix in
farads/km (F/km). For a symmetrical line, you can either specify the N-by-N matrix or the
sequence parameters. For a two-phase or three-phase continuously transposed line, you can
enter the positive and zero-sequence capacitances [C1 C0]. For a symmetrical six-phase line
you can enter the sequence parameters plus the zero-sequence mutual capacitance [C1 C0
C0m]. For asymmetrical lines, you must specify the complete N-by-N capacitance matrix.
Line length: The line length, in km.
Measurements: Line current and voltage measurement are not working.

Inputs
N-Phases voltage-current signals

Outputs
N-Phases delayed voltage-current signals.

ARTEMIS Users Guide

Q022014-01

55

ARTEMiS Distributed Parameters Line

Characteristics and Limitations

The ARTEMiS distributed parameters line block does not initialize in steady-state so
unexpected transients at the beginning of the simulation may occur.
The use of the ARTEMiS Distributed Parameter Line disable the Measurements option of the
regular Distributed Parameter Line. Usage of regular voltage measurement blocks is a good
alternative.

Direct Feedthrough
Discrete sample time

No
Yes, defined in the ARTEMiS guide
block.

XHP support

Yes

Work offline

Yes

Example
The example shows how to use the ARTEMiS distributed parameters line to decouple an
electrical network into two distinct subnetworks and consenquently to optimize the time to
simulate the system in real-time. This property also allows ARTEMiS to simulate systems
that contains more switching elements and consequently more complex systems.
Note that the procedure shown below can also be apply to ARTEMiS Stubline block to
decouple subnetworks and optimize real-time simulation.
Open the SPS demo power_monophaseline model by typing the
following command in the command prompt of Matlab:
power_monophaseline;
To become familiar with the example, consult the help and perform
simulation and check the results. The next steps will modify the
demo to use the ARTEMiS solver instead of the normal SPS solver.
Drag an ARTEMiS Guide block from the ARTEMiS library into the
model and set it sample time to 50e-6 seconds.
Set the SPS PowerGUI block to <Discrete> mode with a sample
time equal to ARTEMiS
Change the Distributed Parameter Line line block of SPS to the
ARTEMiS block and copy the original line parameters in the
ARTEMiS Line model. Optionally, one can use the
opReplaceSpsBlocks function. At the MATLAB prompt type:
opReplaceSpsBlocks('power_monophaseline', 'ReplaceSpsBlocks');
The model must be similar to the following figure. Save the model
under the following name : power_monophaseline_artemis.mdl.

56

Q022014-01

ARTEMIS User Guide

 Simulate the model and analyse the results. You will see that the
results are similar to the original model.

ARTEMIS Users Guide

Q022014-01

57

ARTEMiS Distributed Parameters Line

The next steps will show you how to run the model on a cluster of
PCs running RT-LAB. The general idea is to benefit from the
intrinsic delay of the transmission line to split the model into
subnetworks. The mathematical model of the distributed
parameters line of ARTEMiS, contrary to the SPS model, allows
distribution of the line onto two different CPUs. This property also
allows ARTEMiS to simulate systems that contains more switching
elements and consequently more complex systems.
Select all blocks located in the subnetwork 1 in the figure above
and press Ctrl-G to create a new subsystem.
Move the ARTEMiS block inside the subsystem.
Rename this subsystem to SM_Subnetwork_1. The following figure
displays the content of the SS_Subnetwork_1 subsystem.

58

Q022014-01

ARTEMIS User Guide

 Select all blocks located in the subnetwork 2 and press Ctrl-G to

create a new subsystem.
Add a ARTEMiS Guide block inside the subsystem.
Rename this subsystem to SS_Subnetwork_2. The following figure
illustrates the content of the SS_Subnetwork_2 subsystem.

ARTEMIS Users Guide

Q022014-01

59

ARTEMiS Distributed Parameters Line

Select the 3 remaining blocks, normally the two scopes blocks and
the Mux1 block and press Ctrl-G to create a new subsystem.
Rename this subsystem to SC_Console.
Add the RT-LAB opcomm block between the inports blocks and the
content of the subsystem. Dont forget to set the number of inports
of the opcomm blocks to 3. Refer to the RT-LAB user guide for
more help.
The following figure illustrates the content of the SC_Console
subsystem after the modifications described above have been
made.

Modify the solver parameters of the model; select one of the fixedstep solver, like ode3 for example, and change the fixed-step size
to 50e-6.
Organize the top level blocks according to the following figure.
IMPORTANT: the powerGUI block must be at the top level.

60

Q022014-01

ARTEMIS User Guide

 Save your model.

Your model is now ready to be compiled with RT-LAB. Refer to the
RT-LAB User Guide for more help. If your have set the sample
times of your model with a variable set in the workspace(ex: Ts),
you should set the model initialization function with <Ts=50e-6;>
in File->Model Properties->Callbacks->InitFcn

Limitations
Usage in RT-LAB as task decoupling elements
When used in RT-LAB to decouple and separate computational tasks on different cores/CPUs,
the following connection restriction are applicable to the ARTEMiS distributed parameters
line model:
1- The ARTEMiS distributed parameters line must be located on the top-level of the RT-LAB
compatible Simulink model
2- Each ARTEMiS distributed parameters line outports can be connected only to
SimPowerSystems component located inside RT-LAB top-level subsystem (names beginning
with SS or SM prefixes)
3- No connection between ARTEMiS distributed parameters lines is allowed on the top-level.
If such a connection is required, the ARTEMiS distributed parameters block connection lines
must be first routed inside the subsystems individually and the connection between the
ARTEMiS distributed parameters line ports can be made inside the subsystem.

Related Items
OpReplaceSpsBlocks, ARTEMiS Guide, ARTEMiS Stubline

ARTEMIS Users Guide

Q022014-01

61

ARTEMiS Distributed Parameters Line

References
[1] Dommel, H., Digital Computer Solution of Electromagnetic Transients in Single and Multiple
Networks. IEEE Transactions on Power Apparatus and Systems, Vol. PAS-88, No. 4, April,
1969.

62

Q022014-01

ARTEMIS User Guide

ARTEMiS-SSN Frequency Dependent Line

Library
ARTEMiS

Block
The ARTEMiS-SSN Frequency Dependent Line block implements an N-phases distributed
parameters transmission line model with frequency dependence of line parameters.

Figure 36:ARTEMiS-SSN Frequency-dependent Line block

Mask

Figure 37:Mask of the ARTEMiS-SSN Frequency-dependent Line block

ARTEMIS Users Guide

Q022014-01

63

ARTEMiS-SSN Frequency Dependent Line

Figure 38:Mask of the ARTEMiS-SSN Frequency-dependent Line Block: Measurements Parameters

Description
The ARTEMiS-SSN Frequency Dependent Line block implements an N-phases distributed
parameters line model with frequency dependence of line parameters. The model is based
on the Martis model used by the Electromagnetic Transient Program (EMTP-RV) [1][2].
This model is optimized for discrete real-time simulation and allows network decoupling. It
also allows multi-CPU simulation on an RT-LAB simulator.

Parameters
Number of phases: the number of phase of the model (1-2-3-6)
Line data variable: the name of a MATLAB workspace variable containing the FD_line
parameter. The variable is a structure containing the various parameter of the model.
>>fdfit =

Nph: [1x1 struct] ......................................................number of phase

NpolY: [1x1 struct] ........................ number of poles for Yc(s) (Yc=1/Zc)
Ypol: [1x1 struct] .......................................................... poles of Yc(s)
Yres: [1x1 struct] ...................................................... residues of Yc(s)
YDmat: [1x1 struct] ...................................... constant residues of Yc(s)
NpolH: [1x1 struct] .......................................... number of poles of H(s)
Hpol: [1x1 struct]...................................poles of H(s) (propagation function)
Hres: [1x1 struct] ..................................................................residue of H(s)
HDmat: [1x1 struct] ....................................... constant residues of H(s)
taumin: [1x1 struct] .................................. minimum propagation delays
Ti: [1x1 struct] ........................................ current transformation matrix
Tv: [1x1 struct] ....................................... voltage transformation matrix

64

Q022014-01

ARTEMIS User Guide

And each component being itsefl a structure with Data and Name parts. For example:
>> fdfit.NpolY
Data: [3x1 double]
Name: 'Number of poles for each mode in Ycm'
The document untitled 'Obtaining FD-line model parameters from EMTP-RV'
explains how to get these parameters from the fitting routines of EMTP-RV. It can be
found in the installation repository of ARTEMiS:
pathtorepository/ARTEMiS/artemis_version/art_r201XY/auxiliary_routines/marti
_fd_line/line_param/Obtaining_FDline_model_parameters_from_EMTP_RV.pdf

Unique Tag Identifier: a user set string that must be unique for each instance of
this block inside a Simulink model. (Note: in future releases, this parameters will be
set automatically and will not be visible from the user)
Voltage measurement and Current Measurement choice : User decide if we want to
measure others parameters of the line like phase voltage or phase current. (Note : the
voltage is a phase-to-ground measurement)
Voltage and Current Label: User can specify a label for its current or voltage values. In
case the previous choices have been setted up to yes, current and voltage labels provided by
the user are used to retrieve current. This can be performed using label_s for the sending
side of the line ( the one connected to the source) and label_r for the receiving side of the
line.

Inputs
N-Phases voltage-current signals

Outputs
N-Phases delayed voltage-current signals.

Example
Offline usage example
The FD-line model interface with and only with the SSN method. The reason for this is that the FD-line
model is internal coded with the nodal approach.
To make this interface, the FD-line model must be used in conjunction with SSN Nodal Interface Blocks
(NIB) with the X-type interface chosen in the direction of the FD-line. The NIB can connect to other SSN
groups of either V- I- or X-type.
The curve below shows the source energization current while phase C is connected to the 1-1mH
single phase load.

ARTEMIS Users Guide

Q022014-01

65

ARTEMiS-SSN Frequency Dependent Line

Real-time example
The distributed configuration of RT-LAB allows for complex models to be distributed over a cluster of
PCs running in parallel. However, electrical circuit cannot be easily distributed over a multiple cores
and/or cluster of PCs without changing the dynamic behaviors of the system.
ARTEMiS lines (FD-line, Distributed Parameters Lines and Stublines) can be used to make the parallel
66

Q022014-01

ARTEMIS User Guide

simulation of an electric circuit. ARTEMiS used the intrinsic delay of the line to split the circuit without
affecting the dynamic property of the system. See the ARTEMiS Distributed Parameter Line
documentation for a complete example of the usage of ARTEMiS line models in the RT-LAB framework.
For real-time simulation the model had to be prepare according to RT-LAB conventions (SM_ SS_ toplevel Simulink groups for example).
The model below contains 2 FD-line models connecting some source and loads.

The top-level separated model for RT-LAB will have the ARTEMiS-SSN Frequency Dependent Line model
stay at the top-level of the diagram as shown below

ARTEMIS Users Guide

Q022014-01

67

ARTEMiS-SSN Frequency Dependent Line

And with the NIB block inside the SM_Master and SS_Slave subsystems like depicted below:

68

Q022014-01

ARTEMIS User Guide

Compilation of this model in RT-LAB will results in two independent tasks (SM_Master and SS_Slave)
interconnected by the 2 FD-line which will transmit their propagation voltage and currents between the
two subsystems.

Characteristics and Limitations

Usage of the FD-line model in RT-LAB as task decoupling elements
When used in RT-LAB to decouple and separate computational tasks on different cores/CPUs,
the following connection restriction are applicable to the ARTEMiS distributed parameters
line model:
1- The ARTEMiS-SSN Frequency Dependent Line must be located on the top-level of the RTLAB compatible Simulink model
2- Each ARTEMiS-SSN Frequency Dependent Line outports can be connected only to
SimPowerSystems component located inside RT-LAB top-level subsystem (names beginning
with SS or SM prefixes)
3- No connection between ARTEMiS-SSN Frequency Dependent Lines is allowed on the toplevel. If such a connection is required, the ARTEMiS-SSN Frequency Dependent Line block
connection lines must be first routed inside the subsystems individually and the connection
between the ARTEMiS-SSN Frequency Dependent Line ports can be made inside the
subsystem.

ARTEMIS Users Guide

Q022014-01

69

ARTEMiS-SSN Frequency Dependent Line

SSN solver in the ARTEMiS GUIde block

The SSN solver of the ARTEMiS GUIde block must be Trapezoidal when using a ARTEMiSSSN Frequency Dependent Line block. This is because the Trapezoidal solver is used
internally by the ARTEMiS-SSN Frequency Dependent Line block.
Initialisation
The ARTEMiS-SSN Frequency Dependent Line block does not initialize in steady-state so
unexpected transients at the beginning of the simulation may occur.

Direct Feedthrough
Discrete sample time

No
Yes, defined in the ARTEMiS guide
block.

XHP support

Yes

Work offline

Yes

Related Items
OpReplaceSpsBlocks, ARTEMiS Guide, ARTEMiS Stubline, ARTEMIS Distributed
Parameters Line, ARTEMiS-SSN Nodal interface Blocks.

References
[2]

J.R. Marti, Accurate Modelling of Frequency-Dependent Transmission Lines in

Electromagnetic Transient Simulations, IEEE Trans. on Power App. and Systems, Vol. PAS101, No. 1,January 1982, pp. 147-155.

[3] C. Dufour, H. Le-Huy, J.-C. Soumagne, A. El Hakimi, Real-Time Simulation of Power

Transmission Lines using Marti Model with Optimal Fitting on Dual-DSP Card, IEEE Trans. on
Power Delivery, Vol.11, No.1, Jan. 1996, pp. 412-419.

70

Q022014-01

ARTEMIS User Guide

ARTEMiS-SSN WideBand Line

Library
ARTEMiS

Block
The ARTEMiS-SSN WideBand Line block implements an N-phases distributed parameters
transmission line model with frequency dependence of line parameters using full-phase
domain modelisation.

Figure 39:ARTEMiS-SSN WideBand Line block

Mask

Figure 40:Mask of the ARTEMiS-SSN WideBand Line block

ARTEMIS Users Guide

Q022014-01

71

ARTEMiS-SSN WideBand Line

Figure 41:Mask of ARTEMiS-SSN Wideband Line block, Measurements tab

Description
The ARTEMiS-SSN WideBand Line block implements an N-phases distributed parameters line
model with frequency dependence of line parameters. In difference to the ARTEMiS-SSN FD
line model which uses a modal transformation to compute the frequency dependant
parameters, the WideBand model directly models all its frequency dependant parameter in
the phase domain, avoiding the error caused by the use of a constant modal transformation
matrix in FD line model. Cables in particular typically exhibit a strong frequency dependant
of the modal transformation matrix, making the FD line model inaccurate in this case. The
WideBand model is much more accurate with cables in this regards [1]. This model is
optimized for discrete real-time simulation and allows the decoupling of network connected
on each side of it. It also allows multi-CPU simulation on an RT-LAB simulator.

Parameters
Number of phases: the number of phase of the model (1-2-3-4-6-8-10-12)
Line data variable: the name of a MATLAB workspace variable containing the WB line
parameter. The variable is a structure containing the various parameter of the model.
>>wbfit =

Nph: ..................................................................................number of phase

Ng: .............................................................. Number of propagation groups
tau: ................................................................................ Propagation delays
NYc: ..................................................................................... Yc fitting order
YcNpR: ................................................................. Number of real Yc poles
72

Q022014-01

ARTEMIS User Guide

YcNpC: ..........................................................Number of complex Yc poles

YcR: ...................................................................................... Real Yc poles
YcCR: .......................................................... Real part of complex Yc poles
YcCI: ................................................... Imaginary part of complex Yc poles
YcstD: ........................................................................ Yc constant residues
YcRres: ............................................................... Residues of real Yc poles
YcCRres: ................................ Real Part of Residues of complex Yc poles
YcCIres: .......................... Imaginary part of Residues of complex Yc poles
NH: ........................................................................................ H fitting order
HNpR: .....................................................................Number of real H poles
HNpC: ............................................................. Number of complex H poles
HR: ......................................................................................... Real H poles
HCR: ............................................................. Real part of complex H poles
HCI: ...................................................... Imaginary part of complex H poles
HRres: ................................................................. Residues of real H poles
HCRres: .................................... Real part of Residues of complex H poles
HCIres: ............................. Imaginary part of Residues of complex H poles

The document untitled 'Obtaining WideBand line model parameters from EMTPRV' explains how to get these parameters from the fitting routines of EMTP-RV. It can
be found in the installation repository of ARTEMiS:
pathtorepository/ARTEMiS/artemis_version/art_r201XY/auxiliary_routines/wideb
and/line_param/Obtaining_WideBand_ine_model_parameters_from_EMTP_RV.pdf

Unique Tag Identifier: a user set string that must be unique for each instance of
this block inside a Simulink model. (Note: in future releases, this parameters will be
set automatically and will not be visible from the user)
Voltage measurement and Current Measurement choice : User decide if we want to
measure others parameters of the line like phase voltage or phase current. (Note : the
voltage is a phase-to-ground measurement)
Voltage and Current Label: User can specify a label for its current or voltage values. In
case the previous choices have been setted up to yes, current and voltage labels provided by
the user are used to retrieve current. This can be performed using label_s for the sending
side of the line ( the one connected to the source) and label_r for the receiving side of the
line.

Inputs
N-Phases voltage-current signals

Outputs
N-Phases delayed voltage-current signals.

Example
3-phase cable energization
In the demo, a 15 km 3-ph cable (equivalent to 6 phase line considering the sheath) is energized with
an open end. The sheath is grounded with 10 Ohms resistors at both ends. See the Demo Validation
against EMTP-RV section at the end of this document for more info on this test case.

ARTEMIS Users Guide

Q022014-01

73

ARTEMiS-SSN WideBand Line

The WD-line model was fitted with a total of 14 poles for the characteristic admittance Yc(w) and 36 for
the 6 propagation functions H1(w) to H6(w).

To make this interface, the FD-line model must be used in conjunction with SSN Nodal Interface Blocks
(NIB) with the X-type interface chosen in the direction of the FD-line. The NIB can connect to other SSN
groups of either V- I- or X-type.

74

Q022014-01

ARTEMIS User Guide

Energization from balanced source results in receiving end cable core voltage in the figure below.

Going to real-time
The distributed configuration of RT-LAB allows for complex models to be distributed over a cluster of
PCs running in parallel.
ARTEMiS WideBand lines (and also FD-line, Distributed Parameters Lines and Stublines) can be used to
make the parallel simulation of an electric circuit. ARTEMiS used the intrinsic delay of the line to split

ARTEMIS Users Guide

Q022014-01

75

ARTEMiS-SSN WideBand Line

the circuit without affecting the dynamic property of the system. See the ARTEMiS Distributed
Parameter Line documentation for a complete example of the usage of ARTEMiS line models in the
RT-LAB framework.
For real-time simulation the model had to be prepare according to RT-LAB conventions (SM_ SS_ prefix
for top-level Simulink groups for example). This demo model is already like this, the user can optionally
add a RT-LAB console with SC prefix.
Compilation of this model in RT-LAB will results in two independent tasks (sm_master and ss_slave)
interconnected by the WideBand-line which will transmit their propagation voltage and currents
between the two subsystems.

Characteristics and Limitations

Usage of the WideBand line model in RT-LAB as task decoupling elements
When used in RT-LAB to decouple and separate computational tasks on different cores/CPUs,
the following connection restriction are applicable to the ARTEMiS distributed parameters
line model:
1- The ARTEMiS-SSN WideBand Line must be located on the top-level of the RT-LAB

compatible Simulink model

2- Each ARTEMiS-SSN WideBand Line outports can be connected only to SimPowerSystems
component located inside RT-LAB top-level subsystem (names beginning with SS or SM
prefixes)
3- No connection between ARTEMiS-SSN WideBand Lines is allowed on the top-level. If such
a connection is required, the ARTEMiS-SSN WideBand Line block connection lines must be
first routed inside the subsystems individually and the connection between the ARTEMiSSSN WideBand Line ports can be made inside the subsystem.

SSN solver in the ARTEMiS GUIde block

The SSN solver of the ARTEMiS GUIde block should be Trapezoidal when using a ARTEMiSSSN WideBand Line block. This is because the Trapezoidal solver is used internally by the
ARTEMiS-SSN WideBand Line block.
76

Q022014-01

ARTEMIS User Guide

Initialisation
The ARTEMiS-SSN WideBand Line block does not initialize in steady-state so unexpected
transients at the beginning of the simulation may occur.

Direct Feedthrough
Discrete sample time

No
Yes, defined in the ARTEMiS guide
block.

XHP support

Yes

Work offline

Yes

Related Items
OpReplaceSpsBlocks, ARTEMiS Guide, ARTEMiS Stubline, ARTEMIS Distributed
Parameters Line, ARTEMiS-SSN Nodal interface Blocks.

References
[1] Atef Morched, Bjrn Gustavsen,, Manoocher. Tartibi, A universal model for accurate calculation of electromagnetic transients on overhead lines and underground cables, IEEE Trans. on Power Delivery, Vol. 14, No.
3, pp. 1032-1038, July 1999.
[2] C. Dufour, H. Le-Huy, J.-C. Soumagne, A. El Hakimi, Real-Time Simulation of Power Transmission Lines using Marti Model with Optimal Fitting on Dual-DSP Card, IEEE Trans. on Power Delivery, Vol.11, No.1, Jan.
1996, pp. 412-419.

ARTEMIS Users Guide

Q022014-01

77

ARTEMiS-SSN WideBand Line

78

Q022014-01

ARTEMIS User Guide

ARTEMiS Stubline
Library
ARTEMiS

Block
The ARTEMiS Stubline block implements an N-phase distributed parameters transmission
line model with exactly one time step propagation delay. It is optimized for real-time
simulation.
The ARTEMiS Stubline block permits the decoupling of state-space system equations of
networks on both sides of the stubline.

Figure 42:ARTEMiS Stubline block

Mask

ARTEMIS Users Guide

Q022014-01

79

ARTEMiS Stubline

Figure 43:Mask of the ARTEMiS Stubline block

Description
The ARTEMiS Stubline block implements an N-phase distributed parameters transmission
line model with exactly one time step propagation delay. The model is based on the
Bergeron's travelling wave method used by the Electromagnetic Transient Program (EMTP)
[1]. This block is similar to the SPS distributed parameters line block but is optimized for
discrete real-time simulation and allows network decoupling. It also allows multi-CPU
simulation on an RT-LAB simulator.
Refer to the SPS Distributed Parameter Line block Reference page for more details on the
mathematical model of the distributed parameters line.
Network decoupling
One of the main advantage of the ARTEMiS line blocks (Distributed parameters lines and
Stublines), by opposition to the SPS lines, is the decoupling of the electric circuit into smaller
subnetworks. This important property allows ARTEMiS to simulate, in real-time, circuit with
more switching elements.
SPS and ARTEMiS solve electric circuits using the common state-space method. One of the
main limitation of this method is related to the switch elements. When an event occurs that
changes the topology of the circuit (or change the state of a switch), SPS and ARTEMiS need
to compute a new state-space matrix. This calculation causes a non acceptable overhead
when simulating a circuit in real-time.
To solve this problem, ARTEMiS stores the state-space matrices of a given set of topologies,
normally the steady-state topologies, in cached memory and uses them when necessary
without having to recalcule the matrices. However, the number of matrices required to cover
all topologies of the system depends on the number of switch elements. When a circuit
contains a lot of switch elements, the number of required topologies is high and it is not
possible to store all matrices in cached memory because of the size of the matrices.
The decoupling property of the line allows ARTEMiS to divide the state-space system in two
different state-space systems and reduce the total size of the state-space matrices in
memory. It also reduces the maximum number of topologies by an important factor.
RT-LAB simulation using a cluster of PCs
The distributed configuration of RT-LAB allows for complex models to be distributed over a
cluster of PCs running in parallel. The target nodes in the cluster communicate between each
other with low latency protocols such as shared memory, FireWire, SignalWire or InfiniBand,
fast enough to provide reliable communication for real-time applications.
However, electrical circuit cannot be easily distributed over a cluster of PCs without changing
the dynamic behaviors of the system. The communication delays degrade the computation.
ARTEMiS lines (Distributed Parameters Lines and Stublines) can be used to distribute a
circuit over a cluster of PCs. ARTEMiS used the intrinsic delay of the line to split the circuit
without affecting the dynamic property of the system. Moreover, SPS and ARTEMiS use
physical modelling lines and connectors to model the circuit. This type of signals cannot be
used by RT-LAB to communicate signals between subsystems, because the RT-LAB opcomm
block only supports basic Simulink signals. The only exception to this rule are the ARTEMiS
Distributed Parameters Line block and the ARTEMiS Stubline block. RT-LAB allows the
insertion of a line block at the root level of the block diagram and the connection of the

80

Q022014-01

ARTEMIS User Guide

physical modelling ports of the block to the real-time subsystems. Also note that the
physical modelling signals and ports do not have to pass through the opcomm block.

Parameters
Number of phases N: Specifies the number of phases, N, of the model. The block
dynamically changes according to the number of phases that you specify. When you apply
the parameters or close the dialog box, the number of inputs and outputs is updated.
Available number are 1 to 6 and 2 (differential input). This last option is useful when using
ARTEMiS Stubline in case where it do not have to be refered to ground like in stubline
transformer applications.
Per-Unit value specification: Specify if the resistance and inductance value are specified
in per-unit or not.
Resistance per unit length: The resistance R per unit length, in ohms or pu..
Inductance per unit length: The inductance L per unit length, in henries (H) or pu.
Nominal power (VA): Nominal power base (for per-unit values only).
Nominal voltage(V): Nominal voltage base (for per-unit values only).
Nominal frequency (Hz): Nominal frequency base (for per-unit values only).
Sample Time: The block sample time, in second (s).

Inputs
N-Phases voltage-current physical domain connection.

Outputs
N-Phases delayed voltage-current physical domain connection.

Characteristics and Limitations

The ARTEMiS Stubline block does not initialize in steady-state so unexpected transients at
the beginning of the simulation may occur.

Direct Feedthrough
Discrete sample time

No
Yes, defined in the ARTEMiS guide
block.

XHP support

Yes

Work offline

Yes

Example
This section provides an example on how to build a 3-phase stubline transformer. The
stubline transformer will exhibit a decoupling delay between the primary and secondary
sides suitable for distributed simulation real-time simulation of large systems. Such a
transformer could be used to decouple HVDC system equations at the rectifier/inverter

ARTEMIS Users Guide

Q022014-01

81

ARTEMiS Stubline

station transformers and compute each equations in parallel on different CPUs or cores. The
model is part of the ARTEMiS demos and is named artemis_Transfo_Stubline (.mdl).
In the example, we construct a stubline-based 3-phase transformer from an original
SimPowerSystems transformer and compare the no-load and short-circuit responses. The
principle used to build the stubline transformer is to move the secondary windings leakage
inductance and resistance in stublines put in series with the windings themself. This is done
using single-phase transformers first, then adjusting the per-unit stubline parameters and
finally to make the Y ou Delta connections after the stubline.

Figure 3: Model of a stubline transformer

The example uses a SPS 3-phase transformer with the following parameters:

82

Q022014-01

ARTEMIS User Guide

Figure 4: Three-phase transformer parameters

We will build the stubline 3-phase transformer using single-phase transformer using pu
units. Since we will also use pu-based differential stubline (a stubline with no built-in ground
referentials), appropriate single-phase per-unit bases have to be found. First, the total 3phase nominal power has to be divided by 3 when configuring single-phase transformer
inside. Secondly, the 3-phase winding voltage takes into account the connection type (Y or
Delta) in the voltage specification while single-phase transformer has no such thing. Third,
the R-L pu specification of a 3-phase transformer are specified as Y-connection equivalent
values.
In the final, the resulting single phase transformer therefore has the following parameters:

ARTEMIS Users Guide

Q022014-01

83

ARTEMiS Stubline

Figure 5: Single phase transformer parameters (with null secondary R-L parameters)
Note that the single-phase transformer winding that are Y connected have their voltage ratio
cut by a sqrt(3) factor. Also note the nominal power that is cut by a factor 3. The ARTEMiS
Stubline put in the Y connection has the following parameters:

84

Q022014-01

ARTEMIS User Guide

Figure 6: ARTEMiS Stubline parameters (Y-connected windings)

while the ARTEMiS Stubline put in the Delta connection has the following parameters:

ARTEMIS Users Guide

Q022014-01

85

ARTEMiS Stubline

Figure 7: ARTEMiS Stubline parameters (Delta-connected windings)

Note that the bases used are consequent with the parameters of the single-pase
transformer. The R-L per-unit values are the same than in the 3-phase transformer. Only the
base voltage values differ depending on the connection type.
The design of such transformers is often tricky because of the possible errors in the base
conversion. It is always advisable to compare the stubline model with a rererence for noload and short-circuit cases to verify the correctness of the design. This is what is done in
the example where we superpose the voltages and currents of the stubline transformer with
a standard SPS model.

86

Q022014-01

ARTEMIS User Guide

Figure 8: Comparison of stubline- and SPS- transformers values for no-load (before 0.25 sec.) and
short-circuit (after 0.25 sec.)
Finally, this model can be simulated in several CPU if the model is separated in accordance to
RT-LAB rules with the stublines used as inter-CPU decoupling elements placed on the toplevel of the Simulink model.

ARTEMIS Users Guide

Q022014-01

87

ARTEMiS Stubline

Figure 9: Stublines usage in RT-LAB to decouple compuational task on several cores/CPUs

See the artemis_Transfo_StublineRT.mdl demo for details on how to use the stublines to
decouple and simulate such a model on several cores/CPUs in RT-LAB.

Limitations
Usage in RT-LAB as task decoupling elements
When used in RT-LAB to decouple and separate computational tasks on different cores/CPUs,
the following connection restriction are applicable to the ARTEMiS Stubline model:
1- The ARTEMiS Stubline must be located on the top-level of the RT-LAB compatible Simulink
model (as in Figure 9 for example)
2- Each ARTEMiS Stubline outports can be connected only to SimPowerSystems component
located inside RT-LAB top-level subsystem (names beginning with SS or SM prefixes)
3- No connection between stublines is allowed on the top-level. If such a connection is
required (ex: star-connection neutral point), the ARTEMiS Stubline lines must be first routed

88

Q022014-01

ARTEMIS User Guide

inside the subsystems individually and the connection between the ARTEMiS Stubline ports
can be made inside the subsystem.

Related Items
OpReplaceSpsBlocks, ARTEMiS Guide, ARTEMIS Distributed Parameters Line

ARTEMIS Users Guide

Q022014-01

89

ARTEMiS Stubline

90

Q022014-01

ARTEMIS User Guide

ARTEMiS Transformer with Switched Saturable Core

Library
ARTEMiS (Advanced Real-Time ElectroMagnetic Simulator)

Block
The ARTEMiS-Transformer with Switched Saturable Core implements a 3-phase saturable
transformer in SimPowerSystems model using a switched saturable core method instead of
the current injection with delay of the native SPS transformer models. This type of model is
use to solve instability problems of the current injection methods with delay.
Available models are zigzag-Y , Y-Y, Y-D (30 deg.), in PU and SI versions.

Figure 44:ARTEMiS Transformer with Switched saturable Core

ARTEMIS Users Guide

Q022014-01

91

ARTEMiS Transformer with Switched Saturable Core

Mask

Figure 45:Mask of the ARTEMiS Transformer with Switched saturable Core (zigzag-Y)

Description
The ARTEMiS-Transformer with Switched Saturable Core implements a 3-phase saturable
transformer in SimPowerSystems model using a switched saturable core method instead of
the current injection with delay of the native SPS transformer models. The model is to be
used in conjunction with the ARTEMiS GUIde block.
The model is based on the SimPowerSystems transformer model for the linear part. The
non-linear part, i.e. the saturation is modeled has a switched core inductance. In the linear

92

Q022014-01

ARTEMIS User Guide

region of operation, the first segment of the i=f(flux) characteristic is included in the ABCD
state-space matrix and flux is monitored from there. Whenever the flux reach the 2nd (and
last) segment of the i=f(flux) characteristic, a inductance is switched in parallel with the
linear one and simulation continues with the new configuration of the circuit caused by this
switching action.

Parameters (zig-zag)
Units: Specify the units used (SI or PU) for Zigzag Phase-Shifting Transformer block. Two
different blocks must be used for SI or PU units.
Nominal power and frequency: The nominal power rating, in volt-amperes (VA), and
nominal frequency, in hertz (Hz), of the transformer.
Primary (zigzag) nominal voltage Vp: The phase-to-phase nominal voltage in volts RMS,
for the primary winding of the transformer.
Secondary nom. voltage phase shift: The phase-to-phase nominal voltage, in volts RMS,
and the phase shift, in degrees, for the secondary winding of the transformer.
Winding 1 zig-zag [R1 L1]: The resistance and leakage inductance of the windings 1 of
the single-phase transformers used to implement the primary winding of the Zigzag PhaseShifting Transformer.
Winding 2 zig-zag [R2 L2]:The resistance and leakage inductance of the windings 2 of the
single-phase transformers used to implement the primary winding of the Zigzag PhaseShifting Transformer.
Winding 3 secondary [R3 L3]: The resistance and leakage inductance of the windings 3 of
the single-phase transformers used to implement the secondary winding of the Zigzag
Phase-Shifting Transformer.
Magnetization resistance Rm: This parameter is accessible only if the Saturable core
parameter on the Configuration tab is selected.
Saturation characteristic: The saturation characteristic for the saturable core. Specify a
series of current/ flux pairs (in pu) starting with the pair (0,0).
NOTE: the ARTEMiS-Transformer with Switched Saturable Core only allow a two-segment
saturation characteristic so only 3 pairs of points can be entered (including the (0,0) point).

Parameters (Y-D)
Units: Specify the units used (SI or PU) for Zigzag Phase-Shifting Transformer block. Two
different blocks must be used for SI or PU units.
Nominal power and frequency: The nominal power rating, in volt-amperes (VA), and
nominal frequency, in hertz (Hz), of the transformer.
Primary (Y) nominal voltage Vp: The phase-to-phase nominal voltage in volts RMS, for
the primary winding of the transformer. This winding is always connected in Y.
Secondary nom. voltage: The phase-to-phase nominal voltage, in volts RMS, for the
secondary winding of the transformer.

ARTEMIS Users Guide

Q022014-01

93

ARTEMiS Transformer with Switched Saturable Core

Secondary winding (abc) connection: the type of connection for the secondary windings.
Available connection are: Y-Y, and Y-D (30 deg.). Note that the winding neutral point
connection is always available at both primary and secondary winding.
Winding 1 impedance [R1 L1]: The resistance and leakage inductance of the windings 1
of the single-phase transformers used to implement the primary winding of the Zigzag
Phase-Shifting Transformer.
Winding 2 impedance [R2 L2]:The resistance and leakage inductance of the windings 2 of
the single-phase transformers used to implement the primary winding of the Zigzag PhaseShifting Transformer.
Magnetization resistance Rm: This parameter is accessible only if the Saturable core
parameter on the Configuration tab is selected.
Saturation characteristic: The saturation characteristic for the saturable core. Specify a
series of current/ flux pairs (in pu) starting with the pair (0,0).
NOTE: the ARTEMiS-Transformer with Switched Saturable Core only allow a two-segment
saturation characteristic so only 3 pairs of points can be entered (including the (0,0) point).

Advanced Parameters
Use SPS injection method: Disable the Switching Core Saturation and use standard SPS
injection method to simulate saturation
Disable saturation: Disable the Switched Core saturation (only if Use SPS injection
method is not selected)
Unique Global Tag: Unique tag within the COMPLETE model to route some internal flux
signalsinside the transformer model. If two ARTEMiS Switched Core transformer model with
the same Unique Global Tag are in the same simulation model, an error will occur.

Examples
Example 1: Energization of a zig-zag transformer with an floating source
This example case makes the energization of a 3-phase zigzag-Y transformer on an inductive
load. The load is has about 0.8 p.u. of active power 0.6 pu of reactive power. The

94

Q022014-01

ARTEMIS User Guide

transformer has an total impedance of 0.26 pu and is energized from rest with a balanced
source of 1 pu of voltage.

Figure 46:Test model for the zigzag-Y transformer

The saturation curve is depicted on the next figure.

ARTEMIS Users Guide

Q022014-01

95

ARTEMiS Transformer with Switched Saturable Core

Figure 47:Saturation curve of the zigzag transformer of the test model

The particularity of this model is that it simply cannot be simulated in real-time with
SimPowerSystems only. If one try to simulate this model with the current injection with
delay method of SPS, the model is unstable, event at 0.1 s!
With the ARTEMiS-Transformer with Switched Saturable Core, the model is stable and very
accurate at time step of 30s and more.
The following curves compare the simulation results of the model with ARTEMiS-Transformer
with Switched Saturable Core at 30s with one made with native SimPowerSystems at 1s,
with an algebraic loop. This means that the solver becomes iterative in this case and not
suitable for real-time simulation anyway. It can however be taken for off-line simulation
reference.
The simulations are conducted with positive and negative angle phase shifts to verify the
internal connection of the ARTEMiS models.

Figure 48:Zigzag transformer input current comparison (positive 15 phase shift)

96

Q022014-01

ARTEMIS User Guide

Figure 49:Zigzag transformer output voltage comparison (positive 15 phase shift)

Figure 50:Zigzag transformer input current comparison (negative 15 phase shift)

ARTEMIS Users Guide

Q022014-01

97

ARTEMiS Transformer with Switched Saturable Core

Figure 51:Zigzag transformer output voltage comparison (negative 15 phase shift)

Input-Outputs
A+, B+, C+ (PM-type connectors): 1st winding of zigzag. Positive polarity zigzag winding
connection.
A-, B-, C-: (PM-type connectors). 2nd winding of zigzag. Negative polarity zigzag winding
connection.
a3, b3, c3: (PM-type connectors). 3rd (or secondary) winding connected in Y.
flux: core flux signals (size 3)

Characteristics and Limitations

1- The ARTEMiS-Transformer with Switched Saturable Core can only work with the ARTEMiS
GUIde block present in the model. The first reason is that the ARTEMiS saturable
transformer models are used to provide the core flux readings required by the model. The
2nd reason is that the damping properties of the ARTEMiS art5 solver are required to obtain

98

Direct Feedthrough

N/A

Discrete sample time

Yes

RT-LAB XHP support

Yes

Work offline

Yes
Q022014-01

ARTEMIS User Guide

Related Items
ARTEMIS Distributed Parameters Line, ARTEMiS Stubline, ARTEMiS-SSN Nodal
interface Blocks, ARTEMiS-SSN Frequency Dependent Line.

ARTEMIS Users Guide

Q022014-01

99

ARTEMiS Transformer with Switched Saturable Core

100

Q022014-01

ARTEMIS User Guide

ARTEMiS-SSN Nodal interface Blocks

Library
ARTEMiS (Advanced Real-Time ElectroMagnetic Simulator)

Block
The ARTEMiS-SSN Nodal interface Blocks are used to define nodes and groups of the
ARTEMiS-SSN solver. The SSN (State-Space Nodal) solver is a simulation solver that use
nodal method to couple together, without delays, groups defined by their discretized SPS
state-space equation or any model that has a discrete resistive companion model compatible
with the nodal method of EMTP.

Figure 52:ARTEMiS-SSN Nodal interface Blocks

ARTEMIS Users Guide

Q022014-01

101

ARTEMiS-SSN Nodal interface Blocks

Mask

Figure 53:Mask of the ARTEMiS-SSN Nodal Interface Blocks (3-phase case)

Description
The ARTEMiS-SSN Nodal Interface Blocks (NIB)is used to define nodal point and state-space
groups in a SimPowerSystems schematic within the ARTEMiS-SSN solver. Each block
instance defines a node by itself. The NIB also defines the perimeter of the SSN groups.

Parameters

Number of phase: Set the number of phase for the NIB.

Number of Ports: Set the number of Ports of the block. All phase of a single port connects
to a single SSN group.
Port x type: The Number of Ports parameter sets the number of Port x type (where x=
1 to 16) accesible by the user. For each Port x type parameter, 6 different options are
possible.
V-type(Left) : Voltage type interface to the state-space groups, with ports on the left side
V-type (Right): Voltage type interface to the state-space groups, with ports on the right side
I-type(Left): Current type interface to the state-space groups, with ports on the left side

102

Q022014-01

ARTEMIS User Guide

I-type(Right): Current type interface to the state-space groups, with ports on the right side
X-type(Left): External SSN group type interface, with ports on the left side
X-type(Right): External SSN group type interface, with ports on the right side
These various options are used to connect different types of SSN groups:

Inductive type SSN groups require a V-type interface

Capacitive type SSN groups require a I-type interface

External SSN model, such as the FD-line model, require a X-type interface

Some example will be given to explain this

Examples
Example 1: NIB with I-type and V-type interface
Take the following model, ArtemisSSN_simple_switched_case.mdl, which contains a
switched inductive source connected to a filter bank.

The model has been separated into 2 SSN groups, with the intersection being defined by the NIB. The
NIB interface is I-type in the direction of the capacitor of the filter bank while it is V-type in the
direction of the inductive source. The type of interface is displayed on the block. The NIB also defines
the 3 node that will used internally in the nodal part of the SSN solution.

Example 2: NIB with X-type interface, for SSN external models

The model below simulates a Frequency Dependent Parameter Transmission Line (FD-line) based on
the model originally developped by Marti. This FD-line model is internally coded using the nodal

ARTEMIS Users Guide

Q022014-01

103

ARTEMiS-SSN Nodal interface Blocks

approach and can only produce discrete resistive companion model data like the model discrete
admittance and history current sources. The direct inclusion of the line characteristic impedance Zc()
into a state-space method would have produce huge ABCD matrices becaue of the many states that
compose Zc( ).

For this reason, the SSN approach is prefered when the interface of this type of model to the statespace method of SimPowerSystems. To make the interface, the NIB block must have the type-X chosen
and connected toward the external SSN model, an FD-line model in this case. AS previously, the NIB
also defines the nodal point of the SSN solution. In this case, 6 nodes will be used in the nodal solution
part of SSN.

Input-Outputs
PM-type connectors

Characteristics and Limitations

V- and I-type NIB blocks are used to compute state-space equation of the SSN groups, and
are internally composed of current or voltage sources. State-space equation causality
restrictions apply to these blocks. This is why V-type (internal voltage source) connects to
inductive groups and I-type (internal current source) to capacitive type groups.

104

Direct Feedthrough

N/A

Discrete sample time

Yes

RT-LAB XHP support

Yes

Work offline

Yes

Q022014-01

ARTEMIS User Guide

Related Items
ARTEMIS Distributed Parameters Line, ARTEMiS Stubline, ARTEMiS-SSN Frequency
Dependent Line.

ARTEMIS Users Guide

Q022014-01

105

ARTEMiS-SSN Nodal interface Blocks

106

Q022014-01

ARTEMIS User Guide

ARTEMiS Distributed Parameters Line with Variable Internal Fault distance

Library
ARTEMiS

Block
The ARTEMiS Distributed Parameters Line with Variable Internal Fault Distance block
implements an 3-phases distributed parameters transmission line model with an on-line
modifiable internal fault capability

Figure 54:ARTEMiS Distributed Parameters Line with Variable Internal Fault Distance block

Mask

Figure 55:Mask of the ARTEMiS Distributed Parameters Line with Variable Internal Fault Distance block

Description
The ARTEMiS Distributed Parameters Line with Variable Internal Fault Distance (ADPLF)
block is based on the Bergeron's travelling wave method used by the Electromagnetic
Transient Program (EMTP) [4]. The model implement two DPL lines in series with an internal
mid-point to connect faults. The fault type is specified inside the model (see the example at

ARTEMIS Users Guide

Q022014-01

107

ARTEMiS Distributed Parameters Line with Variable

Internal Fault distance

the end). The fault location is entered as a signal input and can be change during the
simulation without recompiling the models. An error signal is set if the fault location is to
short. Otherwise, the fault distance can be arbitrarily set to any value
For the model to have a fault distance that is variable during real-time simulation, an
approximation is made in the distribution of the line losses. This is explained next:
A standard Bergeron-type line is a lossless line to which lumped resistances are added to
represent the line losses. This is the case of the EMTP and SimPowerSystems DPL models.
Normally, when using two lines in series, the losses should be distributed in proportion of the
respective line lengths. However, if we do this, the total surge impedance of the line would
vary with the line length (i.e. the fault location). This in return would force the recalculation
of state-space matrices and it is not acceptable during real-time simulation.
The ADPLF model therefore fixes the losses distribution without regards to the fault location.
By default, the losses are split in half between the two lines.
Refer to the SPS Distributed Parameter Line block Reference page for more details on the
mathematical model of the distributed parameters line. This may induce some error when
the fault distance is very short. The Maximum fault distance from ABC terminal (%)
parameter can help to minimize this error if the maximum fault distance is known. For
example, if the fault location is located in the first half of the complete line, the losses would
be distributed in a {25%, 75%} way, so to obtain the exact losses reparation of the average
distance of the fault.

Parameters
Number of phases N: Currently, only 3-phase line is supported
Frequency used for RLC specifications: Specifies the frequency used to compute the
resistance R, inductance L, and capacitance C matrices of the line model.
Resistance per unit length: The resistance R per unit length, as an N-by-N matrix in
ohms/km. For a symmetrical line, you can either specify the N-by-N matrix or the sequence
parameters. For a two-phase or three-phase continuously transposed line, you can enter the
positive and zero-sequence resistances [R1 R0]. For a symmetrical six-phase line you can
set the sequence parameters plus the zero-sequence mutual resistance [R1 R0 R0m]. For
asymmetrical lines, you must specify the complete N-by-N resistance matrix.
Inductance per unit length: The inductance L per unit length, as an N-by-N matrix in
henries/km (H/km). For a symmetrical line, you can either specify the N-by-N matrix or the
sequence parameters. For a two-phase or three-phase continuously transposed line, you can
enter the positive and zero-sequence inductances [L1 L0]. For a symmetrical six-phase line,
you can enter the sequence parameters plus the zero-sequence mutual inductance [L1 L0
L0m]. For asymmetrical lines, you must specify the complete N-by-N inductance matrix.
Capacitance per unit length: The capacitance C per unit length, as an N-by-N matrix in
farads/km (F/km). For a symmetrical line, you can either specify the N-by-N matrix or the
sequence parameters. For a two-phase or three-phase continuously transposed line, you can
enter the positive and zero-sequence capacitances [C1 C0]. For a symmetrical six-phase line
you can enter the sequence parameters plus the zero-sequence mutual capacitance [C1 C0
C0m]. For asymmetrical lines, you must specify the complete N-by-N capacitance matrix.
Line length: The line length, in km. This length is the total length of the line, not the
individual length of the 2 line sections used by the model.
108

Q022014-01

ARTEMIS User Guide

Maximum fault distance from ABC terminal (%): This parameter is used to indicate the
maximum fault distance from the ABC side of the line (the side with the fault distance
inport). A 100% is the default value for which the losses are distributed evenly between the
two line section (independently of each section line length). If the maximum fault distance is
known, the losses are then distributed differently to better approximate the average fault
distance.

Inputs
Fault distance in pu: this signal value is the location of the fault in per unit of total line
length with regards to the side of the input connector on the block.
N-Phases voltage-current signals (Physical Connection)

Outputs
Too_short: when equal to 1, this signal output indicates that the fault distance is too short
for the selected simulation sample time. The model requires that the line transmission delay
be at least one sample time of the model. In that case, the user has the option of either
lowering the simulation sample time or increasing the line length or fault distance.
N-Phases delayed voltage-current signals (Physical connection

Characteristics and Limitations

The ARTEMiS Distributed Parameters Line with Variable Internal Fault Distance block does
not initialize in steady-state so unexpected transients at the beginning of the simulation may
occur.
The use of the ARTEMiS Distributed Parameters Line with Variable Internal Fault Distance
disable the Measurements option of the regular Distributed Parameter Line. Usage of
regular voltage measurement blocks is a good alternative.

Direct Feedthrough
Discrete sample time

No
Yes, defined in the ARTEMiS guide
block.

XHP support

Yes

Work offline

Yes

Example
The following example compare the ARTEMiS Distributed Parameters Line with Variable
Internal Fault Distance with a line fault modeled with two distinct line section. The example
helps to put in context the error introduced by the model with regards to the normal
ARTEMiS line model, that implement the standard Bergeron line model with lumped loss.

ARTEMIS Users Guide

Q022014-01

109

ARTEMiS Distributed Parameters Line with Variable

Internal Fault distance

Figure 56:Demonstration model of the ARTEMiS Distributed Parameters Line with Variable Internal Fault Distance

Inside the ADPLF, the user can implement its own fault scheme as seen in the following
figure. In our case, a single-phase fault to ground is implemented.

Figure 57:User fault implementation inside the ADPLF

The main error will arise for faults near the line terminal because a lumped loss of R/8
instead of R/4*fault_length/line_length. Remember that a normal Bergeron line with loss
has R/4 loss at each end and R/2 in the middle with losses proportional to line length
section. In the case of the ADPLF, this loss is fixed and no more proportional to section
length.
110

Q022014-01

ARTEMIS User Guide

The line used for the test is 100 km in length and has a 0.01273 (direct) and 0.3864
(homopolar) Ohms/km series losses. The line has a minimum transmission delay of
approximately 333 s and the minimum fault distance is approximately 15 km for a
simulation time step of 50 s (50/3.33 , see Limitations). The user must use pi-line to
simulate shorter faults. The test consists on a 4-cycle single-phase to ground fault on the
line from steady-state. The line is completely opened at 0.11 seconds. Because the line is
not loaded, the per-fault steady-state current is quite small.
The next two figure shows the results for a very short and a mid-line fault. On the short
fault, one can observe that the input current during the fault is smaller than the reference.
This is caused by the lumped losses of the line end which is bigger than normal.

Figure 58:Evaluation of the ADPLF model for a fault distance of 15 km.

ARTEMIS Users Guide

Q022014-01

111

ARTEMiS Distributed Parameters Line with Variable

Internal Fault distance

Figure 59:Evaluation of the ADPLF model for a fault distance of 50% (mid-line fault)

If we now make a fault at mid-line point, the two results are exactly the same. This is
normal because the ADPLF assume a fixed losses distribution corresponding to a mid-line
separation. Fault current is lower in this case also as expected for a fault occurring farther
from the power source.

Limitations
Usage in RT-LAB as task decoupling elements
The ADPLF model cannot be used as a separating element in RT-LAB.
Short distance fault limit
The ADPLF model can only implement fault occurring at a distance corresponding to one
time step of propagation of the line (the fastest mode for the 3-phase line). If shorter fault
distance needs to be implemented, a pi-line model is recommended. As a quick rule of the
thumb, considering the speed of light of 300000km/s, a 3.33s/km relation exists between
the minimum time step and minimal fault distance of the model.

Related Items
OpReplaceSpsBlocks, ARTEMiS Guide, ARTEMiS Stubline, ARTEMiS Distributed
Parameter Line

References
[4] Dommel, H., Digital Computer Solution of Electromagnetic Transients in Single and Multiple
Networks. IEEE Transactions on Power Apparatus and Systems, Vol. PAS-88, No. 4, April,
1969.
112

Q022014-01

ARTEMIS User Guide

ARTEMiS MMC 1P Cell

Library
ARTEMiS (Advanced Real-Time ElectroMagnetic Simulator)

Block

Mask

Figure 60:Mask of MMC 1P block

ARTEMIS Users Guide

Q022014-01

113

ARTEMiS MMC 1P Cell

Description
The MMC cell block implement a unipolar bridge with a capacitor. Series RC snubber circuits
are connected in shunt with each switch device. Press Help for suggested snubber values
when the model is discretized. The gates are controlled by Double signals. The following
figure presents the equivalent electrical circuit of the MMC cell block implement a unipolar
bridge.

Figure 61:Equivalent Electrical Circuit of the MMC cell Block

When the upper switch or upper anti-parallel diode conducts, voltage between the Center
and the Common equals Vc (minus internal voltage drops). When the lower switch or diode
of the leg conducts, this voltage is equal to 0 (plus internal voltage drops).
The RC snubber in shunt with the switch are required to solve numerical oscillation. Using
the time step and the equivalent inductance of the circuit the value of the Rsnubber and
Csnubber are given by the following equation.

R Snubber = ------------- Le q
Ts 5

1
C Snubbe r = -------------------------------------22
L --------------- e q Ts 15
Where Leq is the equivalent inductance and Ts the simulation time step.
The resistance in shunt with the capacitor will discharge it. If no discharge resistance is
required, it should be set to inf.
If the upper and lower switches are turn ON at the same time, the error output will
output 1 when the error occur and 0 otherwise. Upon error the capacitor voltage will
become 0 instantaneously and return to normal once the fault is cleared. If the keep
cell short-circuit upon error is checked the capacitor voltage will always remains to 0.
The error output will indicate the total number of short-circuit cell.

Parameters
Snubber resistance: Snubber resistance value, only used in high impedance mode.
Snubber Capacitor: Snubber capacitor value, only used in high impedance mode.
114

Q022014-01

ARTEMIS User Guide

Cell capacitor: Value of the cell's capacitor .

Resistance in shunt with the cell capacitor: Value of the discharge resistance for the
cell's capacitor .
Initial capacitor voltage: Initial voltage value of the cell capacitor. It can be a scalar if all
the capacitors start at the same value. If a vector the same length then the total number of
cell is used, the voltage for each cells capacitor can be set.
Ron: Internal resistance of the selected device, in ohms.
Number of cells: This determine how many cells are connected in series. A maximum of 50
cells can be connected in series. If more then 50 cells are required, a second MMC_1P block
need to be connected in series.
Sample time:Time at which the capacitor voltage will be computed.
Keep cell short-circuit upon error:Check this box if the cell capacitor should
remain to 0 when an error occurs.

Inputs
Upper gate (double): double signals that controlled the upper switch gates. This signal
has to be a vector of same length then the number of cells. A signal value of 1 indicates the
switch is conducting, while a value of zero indicates the switch is OFF.
Lower gate (double): double signals that controlled the lower switch gates. This signal has
to be a vector of same length then the number of cells. A signal value of 1 indicates the
switch is conducting, while a value of zero indicates the switch is OFF.
Center (SPS): Middle point of the cell.
Common (SPS): Common point of the cell.

Outputs
Vc (double): The voltage at the cell's capacitor, vector of same length then the number of
cells.
Error (double): Ouput 0 in normal operation and output 1 if a short-circuit occurs. When
the keep cell short-circuit upon error is checked it will output the total number of shortcircuited cells.

Characteristics

ARTEMIS Users Guide

Direct Feedthrough

No

Sample time

Parameter

Work offline

Yes

Dimensionalized

Yes

Q022014-01

115

ARTEMiS MMC 1P Cell

116

Q022014-01

ARTEMIS User Guide

ARTEMiS MMC 2P Cell

Library
ARTEMiS (Advanced Real-Time ElectroMagnetic Simulator)

Block

Mask

Figure 62:Mask of MMC 2P block

ARTEMIS Users Guide

Q022014-01

117

ARTEMiS MMC 2P Cell

Description
The MMC-2P cell block implement a bipolar bridge with a capacitor. Series RC snubber
circuits are connected in parallel with each switch device. Press Help for suggested snubber
values when the model is discretized. The gates are controlled by Double signals. The
following figure presents the equivalent electrical circuit of the MMC cell block implement a
unipolar bridge.

Figure 63:Equivalent Electrical Circuit of the MMC-2P cell Block

The voltage between A and B is determined by the switching applied to g1 to g4. g1 and g2
must be complementary and so does g3 and g4.
The RC snubber in shunt with the switch are required to solve numerical oscillation. Using
the time step and the equivalent inductance of the circuit the value of the Rsnubber and
Csnubber are given by the following equation

1
R Snubber = ------------- L eq ------------------Ts 5
nbcells

1
C Snubber = ------------------------------------- nbcells
2 2
L --------------- eq Ts 15
Where nbcells is the number of cells in series and Leq is the equivalent inductance

Parameters
Snubber resistance: Snubber resistance value, only used in high impedance mode.
Snubber Capacitor: Snubber capacitor value, only used in high impedance mode.
Cell capacitor: Value of the cell's capacitor .
Ron: Internal resistance of the selected device, in ohms.
Number of cells: This determine how many cells are connected in series. A maximum of 20
cells can be connected in series. If more then 20 cells are required, a second MMC_2P block
need to be connected in series.
Sample time:Time at which the capacitor voltage will be computed.

118

Q022014-01

ARTEMIS User Guide

Inputs
g1 (double): double signals that controlled the upper left switch gates. This signal has to
be a vector of same length then the number of cells. A signal value of 1 indicates the switch
is conducting, while a value of zero indicates the switch is OFF.
g2 (double): double signals that controlled the lower left switch gates. This signal has to be
a vector of same length then the number of cells. A signal value of 1 indicates the switch is
conducting, while a value of zero indicates the switch is OFF.
g3 (double): double signals that controlled the upper right switch gates. This signal has to
be a vector of same length then the number of cells. A signal value of 1 indicates the switch
is conducting, while a value of zero indicates the switch is OFF.
g4 (double): double signals that controlled the lower right switch gates. This signal has to
be a vector of same length then the number of cells. A signal value of 1 indicates the switch
is conducting, while a value of zero indicates the switch is OFF.
A (SPS): Middle left point of the cell.
B (SPS): Middle right point of the cell.

Outputs
Vc (double): The voltage at the cell's capacitor, vector of same length then the number of
cells

Characteristics

ARTEMIS Users Guide

Direct Feedthrough

No

Sample time

Parameter

Work offline

Yes

Dimensionalized

Yes

Q022014-01

119

ARTEMiS MMC 2P Cell

120

Q022014-01

ARTEMIS User Guide

ARTEMiS TSB 2-Level

Library
ARTEMiS (Advanced Real-Time ElectroMagnetic Simulator)

Block

Mask

Figure 64:Mask of TSB 2-Level block

Description
The TSB 2-Level block implements a 2-level IGBT/GTO/MOSFET inverter with 2 active
switches with anti-parallel diodes. The block model uses a switching function approach with

ARTEMIS Users Guide

Q022014-01

121

ARTEMiS TSB 2-Level

interpolation and is also capable of simulating cases where the inverter output is in a highimpedance state (ex: no gating or natural rectification). Deadtime and anti-parallel diode
effects are fully taken into account by this model. The gates are controlled by Double
signals. The following figure presents the equivalent electrical circuit of the TSB 2-Level
block.

Figure 65:Equivalent Electrical Circuit of TSB Block

A numerical RC snubber is required to implement high impedance states when no gating is

present at both switches. Suggested values are given next:

L
R Snubber = ------------Ts 5 e q

1
C Snubbe r = -------------------------------------22
L --------------- e q Ts 15
Where Leq is the equivalent inductance of the load (connected to A port) and Ts the
simulation time step.
The Input DC current is normally equal to the current goiing across the upper IGBT. An
alternative method is also provided based on power balance (Idc*Vdc=Vabc*Iabc). A
compensation factor is provided to adjust this power equation. This factor simply makes the
interpolation across the two last values of Iabc to compute Idc. This can be required in
applications with very low power factors such as an inductive machine with very low
mechanical load.

Parameters
Switch Conduction resistance [Ohms]: Conduction resistance of all switches, including
anti-parallel diodes.
Snubber resistance (in high impedance mode only) [Ohms]: Snubber resistance
value, only used in high impedance mode.
Snubber Capacitor(in high impedance mode only) [Farad]: Snubber capacitor value,
only used in high impedance mode.
Sample time [s]:Time at which the capacitor voltage will be computed.
Anti-voltage reversal equivalent diode resistance [Ohms]: This is the equivalent
resistance of the inverter at voltage reversal on the DC-bus. This resistance is different than

122

Q022014-01

ARTEMIS User Guide

the one of the two anti-parallel diodes (that would start to conduct in case of DC-bus voltage
reversal) because of the switching function method of the model.
Alternate Idc calculation method based on power balance: With this option
selected, the DC bus input current is computed from the power balance equation
(Vdc*Idc=Vabc*Iabc)
Alternate Idc compensation factor (default=0) (0...1): This factor enables some
delay tuning in the power equation Vdc*Idc=Vabc*Iabc by interpolating on the 2 last
time value of Iabc. It can be useful in somelow power factor applications.

Inputs
Gate (double, interpolated): (vector of size 2). Double signals that controlled the upper
and lower switch gates. A signal value of 1 indicates the switch is conducting, while a value
of zero indicates the switch is OFF. A values between 0 and 1 indicates that a switching
action occured during the last time step.
Dc+, Dc-(SPS): The positive and negative side of the DC bus.
A (SPS): The inverter middle point.

Outputs
none.

Characteristics

ARTEMIS Users Guide

Direct Feedthrough

No

Sample time

Parameter

Work offline

Yes

Dimensionalized

Yes

Q022014-01

123

ARTEMiS TSB 2-Level

124

Q022014-01

ARTEMIS User Guide

OpReplaceSpsBlocks

Description
This function helps replacing SimPowerSystems electrical blocks by ARTEMiS electrical
blocks or replacing ARTEMiS electrical blocks by SimPowerSystems electrical blocks. This
function is useful because ARTEMiS provides advanced blocks for real-time simulation; these
blocks contain an optimized implementation of the SPS mathematical model which make
them better suited for real-time simulation.
This function also provides an optional interface to help the user select the blocks to be
replaced. The figure below shows the dialog that appear when the function is called with the
default argument.

Note that ARTEMiS currently only supports the Distributed Parameters line block. Other
decoupling blocks than the Distributed Parameters line will be supported in a future release.

Usage
opReplaceSpsBlocks(modelName, operation, searchDepth);

ARTEMIS Users Guide

Q022014-01

125

OpReplaceSpsBlocks

Inputs
modelName

Name of the model to modify.

operation

Optional argument that specified the type of operation

to perform when replacing the blocks.
1- ReplaceBlocks: open a dialog that will help to switch
between the real-time ARTEMiS blocks and the nonreal-time SPS blocks.
2- ReplaceSpsBlocks: automatically replace all SPS
blocks by their real-time ARTEMiS equivalent,
3- ReplaceArtemisBlocks: automatically replace all
ARTEMiS blocks by their non real-time SPS equivalent.
The default value is ReplaceBlocks.

searchDepth

Optional integer that constrains the model search to a

specific depth.

Outputs
None

Example
To open a dialog that will help to switch between the real-time ARTEMiS blocks and the nonreal-time SPS blocks:
opReplaceSpsBlocks(modelName);
To automatically replace all SPS blocks by their real-time ARTEMiS equivalent:
opReplaceSpsBlocks(modelName, 'ReplaceSpsBlocks');
To automatically replace all ARTEMiS blocks by their non real-time SPS equivalent:
opReplaceSpsBlocks(modelName, 'ReplaceArtemisBlocks')

Related Items
ARTEMIS Distributed Parameters Line, ARTEMiS Guide

126

Q022014-01

ARTEMIS User Guide

Known limitations (ARTEMiS v6.0 release)

The following issues and limitations of ARTEMiS v6.0 are known to OPAL-RT.

PC Memory limitation
When compiling an ARTEMiS model, lack of memory may occur, especially when the size of the matrices
ABDC are big and the number of switches is high. It only happens during the compilation because during offline simulation, the permutation matrices are not all calculates.
Note that the separation and generation process take a lot of memory. Even before the generation
process starts, a big part of the memory is already in used by RT-LAB and by Matlab itself (0.8GB).
ARTEMiS tried to calculate all matrix permutation of the electrical network. For example, for a very
small network (only 10 states, 12 inputs and 11 outputs), the state-space matrices would be like:
x' = Ax+Bu

(A is [10x10], B is [10x12], x is [10x1], x' is [10x1], u is [12,1])

Y = Cx+Du

(C is [11x10], D is [11x12])

And the number of switches is 15, the size of memory required is:
( (10*10 + 10 + 10*12 + 12 + 10) + (11*10 + 10 + 11*12 + 12 + 11) ) * 2^15 * 8 = 138MB
Now the same calculation for a network with 40 states, 40 inputs and 40 outputs is 1.76GB.

ARTEMiS limitations
1- 3-level bridge with ideal switch option not supported in ARTEMiS-DTCSE mode.
2- Maximum number of switch is 28 in a single topologicaly connected network for all mode even
Dynamic calculation
3- Stubline usage causes instability when using ARTEMiS-DTCSE mode.
4- ARTEMiS distributed parameters line have no 'Measurements' options. SPS mesurement blocks are
an alternative for line measurements.
5- ARTEMiS distributed parameters lines are not initialized with stady state currents and voltages. This
can results in some transients at the beginning of the simulation.
6- The trapezoidal solver does not support RT-Events based switch gating signals
7- Circuit containing SPS Multimeter blocks are simulated into a single state-space system.

ARTEMIS Users Guide

Q022014-01

127

SimPowerSystems 4.6- 5.0 limitations

Known limitations (ARTEMiS v6.0 release)

SimPowerSystems 4.6- 5.0 limitations

SPS Neutral blocks: There is a limitation in SimPowerSystems (v4.6 and v5.0) that prevent the effective separation (or decoupling) of independant systems of state-space equation if any SPS Neutral block
is present in the model. This does not affects the simulation accuracy of models but only slows them
down because big matrix systems are formed. In RT applications, this will increased the required memory and probably increase the minimal sample time. The effect is similar to the Ground Connections
problem described next.
Ground Connections: There is currently a bug in SPS 4.6 (R2008a) with regards to separation/decoupling of state-space systems.
If , for example, 3 components are wired together to a single SPS 'Ground', the 3 componants will be
put in the same state-space system.
If the same componants are connected to 3 distinct SPS 'Ground' blocks, then the 3 componants will be
put into 3 differents state-space systems (provided that there is no other connections between the
components)
Electrically speaking, the 2 cases are identical but it affect the capacity of separation. The user is
advised to verify the effective subsystems separation as it appears at the MATLAB prompt at the beginning of the simulation with ARTEMiS. The following prompt output shows that the power_x model is
separated into 2 subsystems:
SimPowerSystems processing circuit #1 of power_x ...
Computing state-space representation of linear electrical circuit ...
(13 states ; 9 inputs ; 8 outputs ; 3 switches)
...
ARTEMiS: approx. memory required for full matrix precomputation: 0.037056 Mb
Ready.
Third-Party Rule block detected: power_transfosat/ARTEMiS Guide
SimPowerSystems processing circuit #2 of power_x ...
Computing state-space representation of linear electrical circuit ...
(13 states ; 9 inputs ; 8 outputs ; 3 switches)
...
ARTEMiS: approx. memory required for full matrix precomputation: 0.037056 Mb
Ready.

128

Q022014-01

ARTEMIS User Guide