SoFT Analysis

1. What is SoFT?

Is SoFT a methodology, product, service or what?
In fact, SoFT is all of the above. First, SoFT is a methodology to obtain the world’s most accurate electrical system simulations. Second, SoFT is a product to implement this methodology. Third, the SoFT product is packaged into a service offer.

What is the SoFT methodology?
The SoFT methodology consists of 3 steps:

1. On-site SoFT measurements of the electrical network components (e.g. motors, transformers, cables, etc).

2. True and fully frequency-dependent modeling of all network components.

3. Simulation and analysis of the electrical system.

What is the SoFT product?
The SoFT product consists of an advanced measurement system, and of sophisticated algorithms to transform the measurement data into accurate simulation models. These models can than be imported into standard simulation programs such as EMTP, MATLAB, etc.

What is the SoFT service?
The SoFT service is ABB’s offer to carry out the 3-step SoFT methodology for utilities and industrial customers. The service offer includes an additional 4th step:

4. Recommendation of safe, cost-effective solutions to mitigate the effects of potential problems.

Is SoFT a software?
No, SoFT is not a software and it is not a simulation program.

2. For whom & for what is SoFT?

What type of customers do you have?
The SoFT customers is a large and diverse group, which cuts across industries, equipment manufacturer and service providers. It can be roughly divided into 5 categories:

1. Power plants & Utilities
2. Industry (oil & gas, steel, tunnels, pulp & paper, etc)
3. Power system consultants
4. Manufacturers of power transformers & MV/HV equipment
5. Insurances & indirect customers

Is the SoFT offer tailored to each customer category?
Yes.

• Customers of categories 1 & 2 above typically request the complete SoFT service offer.
• Customers of category 3 may run their own simulations, but request SoFT measurements and modeling for improving their analysis.
• Customers of category 4 sometime request more detailed measurement data to follow up and monitor how their product quality varies over time and between production series.
• Customers of category 5 usually request SoFT in the name of their customers to provide them the best possibly analysis tools and services.

Why are customers interested in SoFT?
The reasons for using SoFT are

• to perform fast and accurate reliability studies,
• to speed up commissioning,
• to quickly resume plant operation in case of a fault,
• to design system protection equipment,
• to analyze and resolve overvoltage and insulation breakdown issues,
• to investigate any other kind of electrical system problem by means of simulation.

Is SoFT limited to electrical transient studies as the name SoFT -- Simulation of Fast Transients -- suggests?
No, SoFT is not limited to electrical transient studies, but can accurately examine and resolve a large range of electrical questions and issues. Transient studies belong to the most difficult ones and hence the name SoFT.

I want to perform simulations myself. Can you help me construct accurate simulation models for the electrical equipment in my simulation?
Yes, we can perform measurements and construct SoFT simulation models for you to use in your own simulations.

I want to perform measurements and construct SoFT simulation models myself. Can I buy the SoFT product instead of the SoFT service offer?
Yes. However, a more cost-effective solution might be to lease the SoFT product for the required time. Along with a SoFT product purchase or lease we offer 1 week of training per operator.

3. What makes SoFT so unique?

What does SoFT do better than other simulation tools?
In contrast to other simulation tools, which rely on theoretical assumptions and equivalent circuits, SoFT measures the true and fully frequency-dependent behavior of the electrical equipment. These measurements reveal the interplay between the three phases, the equipment interaction, and system resonances to achieve the most authentic and accurate power system studies in the world.

Many companies perform simulations. Is ABB’s SoFT really better than all the others?
Yes. In simple cases you will not notice the difference. However, as the complexity and the accuracy requirement increases, highly accurate models in the full dynamic range (10 Hz - 10 MHz) become essential. SoFT is the only tool in the world providing such simulation models. In fact, SoFT measurements reveal much more information than even the manufacturers themselves know about the equipment they produce.

What do outsiders think about ABB’s SoFT?
Numerous specialists inside and outside ABB endorse the SoFT technology. SoFT has received the prestigious Swiss Technology Award 2006 for its high degree of innovation and far-reaching functionality. Our four development SINTEF in Norway, Hydro-Quebec in Canada, Zürcher Hochschule Winterthur and PI Electronics in Switzerland testify to the uniqueness and genuine quality of SoFT, as do the developers of EMTP, the most widespread simulation software in the industry. However, the most important supporters of SoFT are our customers, for which we solve critical and challenging problems.

4. The SoFT service offer

What do you need to make a SoFT analysis quotation?
To make a quotation we need:

• A detailed schematic of the electrical system (a single-line diagram)
• A description of issue to be investigated including:
  • What is the issue?
  • When is the SoFT analysis needed?
  • What are the consequences of equipment failure?
  • Plant name, type & location
  • Voltage/power levels
  • Prior actions taken (repairs, investigations)

What does a SoFT study cost?
Normally, a SoFT study cost €20-70k.

How long time does a SoFT study take?
On-site measurements are usually performed in 1-3 days. Back-office modeling, simulation and analysis are usually completed within 2-3 weeks after the on-site measurements.

Can our own personnel (instead of ABB) perform the necessary measurements for discount purposes?
No. The measurements are so complicated and specialized that they can only be performed with a dedicated SoFT measurement unit, which has taken ABB many years to develop. Of course, if you have bought or leased such a unit, then you can carry out the measurement yourself.

Shall the measured electrical network components (motors, transformers, etc) be disconnected from line voltage and short-circuited to ground for safety reasons? For how long time?
Yes, the components must be disconnected for about 1 hour when the SoFT measurements are performed, but all components must not be disconnected at the same time. No, they shall not be connected to ground because it invalidates the measurement results.

How many components must be measured in a SoFT study?
All components (motors, transformers, filters, busbars, etc) in the studied network must be measured in order to generate accurate SoFT models with a few exceptions:

• If there are two or more identical components, just one need to be measured.
• If a component cannot be measured, it can instead be modeled theoretically based on product sheet data. Naturally, such a model will be less accurate and will not capture the full dynamics.
• It is possible to combine many components into one measurement. In one case, a feed transformer, two interconnected busbars with a 2 km cable, and 4 LV transformers were bundled into one measurement to see how these components interacted with a motor, which had experienced overvoltages.

Our transformer blew up. Can you perform measurements on it when it is damaged?
No, SoFT measurements on damaged components will not yield the correct result for the further analysis. However, often the industrial facility experiencing the damage or the manufacturer have an undamaged identical component as spare part. Then this undamaged component can be measured instead.

We have a problem with switching surges. Are real switching operations required for the SoFT measurements in this case?
No, real switching operations are not required. Instead, SoFT brings the electrical network into simulation software, such that the effect of switching can be studied there without stressing the network. However, on-line measurements of real switching operations are usually offered to the customers as an extra option, such that they can compare these with the simulations.

Can SoFT monitor my electrical network on-line, 24 hours a day and notify me of potential faults?
No, SoFT does no online monitoring.

5. Technical questions

What measurements does SoFT perform?
SoFT performs Sweeping FRA (Frequency Response Analysis) in up to 7x7 dimensions in the frequency range 10 Hz – 10 MHz on electrical network components one at the time. For instance, a 3-phase motor with disconnected star-point has 6 terminals, and the resistive, inductive and capacitive coupling between all of them is measured to construct a frequency-dependent 6x6 admittance matrix.

Are SoFT measurements similar to other FRA measurements?
No. No FRA equipment on the market has any of the SoFT features in the following list:

• Standard FRA equipment requires 36 reconnections to measure a 6x6 admittance matrix, whereas SoFT performs the complete measurement procedure automatically without any measurement cable reconnection. Thus, SoFT is much faster and avoids the risk of connection errors.
• SoFT applies a combination of inputs on several terminals of the measured component in order to excite its eigenmodes, in order to boost the sensitivity of the measurements.
• SoFT dynamically adopt the frequency spacing and measurement procedure to pinpoint critical resonances and frequencies.
• The termination of the terminals not being measured is altered between GND, 50 Ohm and 1 MOhm depending on the measurement object.
• SoFT completely compensates for the measurement cables to achieve true measurement results independent of the measurement tool.
• SoFT automatic detects incorrect grounding/connection of measurement object.

How does the SoFT modeling work?
The modeling algorithms of SoFT takes the measured admittance matrix as input and output a state space model encoded in ABCDE matrices, which is later fed to a simulation program. The modeling consists of two steps:

1. fitting to ensure that the model accurately captures the full dynamics in the complete frequency range (10 Hz - 10 MHz)
2. passivity enforcement to ensure that the model is robust and provide stable simulation. Technically, it entails ensuring that all the poles are in the left-hand s-plane.

The base of the SoFT algorithms is described in the references below and on the Vector Fitting home page. However, much of the later development work has not been published because of ongoing patent processes.

B. Gustavsen and A. Semlyen, “Rational approximation of frequency domain responses by Vector Fitting”, IEEE Trans. Power Delivery, vol. 14, no. 3, pp. 1052-1061, July 1999.

B. Gustavsen and A. Semlyen, “Enforcing passivity for admittance matrices approximated by rational functions”, IEEE Trans. Power Syst., vol. 16, no. 1, pp. 97-104, Feb. 2001.

C. Heitz, A. Steiner: Algorithm for an automated modal model generation from measured data for electrical N-ports. Unpublished report, Dec 2002.

Can SoFT model all kind of electrical equipment? Are there any limitations?
Yes, there are limitations. Measurements are restricted to 1. passive, 2. close-to-linear objects 3. with 7 or fewer terminals 4. with off-line availability. Let’s address these points in order.

1. Active objects such as drives systems in operation can not be measured. Generators can be measured off-line to accurately study their behavior above a few hundred hertz, where the online electro-mechanical influence is negligible.
2. The transition between different states, such as a breaker in operation can not be measured.
3. Non-linear objects such as surge-arrestors cannot be measured and modeled accurately.
4. Three and four winding transformers with more than 7 terminals cannot be straight-forwardly measured. A few tricks are needed to create accurate multi-terminal SoFT models.
5. The SoFT measurements are performed off-line, i.e. the measured objects must be de-energized and disconnected from the rest of the network.

Electrical components falling into the above categories can not be measured. Instead they must be modeled the traditional theoretical way everyone else use in the simulation business. Thus, they will not possess the superior SoFT accuracy and dynamical behavior.