Power Simulation Software Packages and Techniques

One of the questions i am occasionally asked by 3rd parties and younger engineers is a variation of 'what simulation package is best?' This post is very much my own perspective as a UK based engineer, so apologies if i overlook a regional favorite then apologies in advance.

This is an interesting question, as on the surface it seems quite innocent and naive, but it actually opens up a whole lot of other questions. The short answer is: It depends! What you are trying to simulate? Who are you simulating it for? What do you know how to use?

Different simulation packages have different strengths, and weakness and are intended for different markets. It’s not a case of which is best, but what do you have, what you know how to use and what is the most suitable for the simulation you are carrying out. A simpler simulation package in the hand of an expert user can be far more powerful than a complex simulation package in the hands of someone who doesn't really understand what they are are doing and what they are looking for.

So first off, the key question is what are you simulating and why? If you are doing high level FEED type studies to get a general flavor of the system response for loadflow and short circuit studies, then most of the packages will give you a similar response. The best package is usually the one you have and know how to use. Some packages have useful abilities to model transformer impedance tolerances easily, and model lumped loads with quick and easy fault contributions. For G99 / Grid Code compliance applications packages such as DIgSILENT Powerfactory and PSS/E are the most popular in the UK, and the preference is largely down to individual experience. For light / medium industrial applications, packages such as ETAP, SKM Powertools and ERACS tend to sit well for industrial sites, as they have good motor and load modelling capability. For medium / heavy applications then DIgSILENT / PSS/E can be more suitable, as they can also show better interfaces with the utility company. In the North American region, i believe that PSLF is a popular alternative to both DIgSILENT and PSS/E.

Next up is who are you simulating it for? This is an obvious question, but usually DNOs, TSOs and Clients will often have specific requirements for software packages. These can sometimes be flexible, but are often fixed, as it is a package that the 3rd party has and is comfortable with. If the Client asks for a simulation in package X and you carry out the study in package Y instead, then trouble could be coming.

Finally what are you trying to simulate? This is the most interesting technically, and despite what the companies may all advertise, every package has its own shortfalls and wrinkles that can make things easier or harder than a competitors package. Predefined models available in the library can make a big difference, in many cases simplifying the model creation time and validation, but can also give woefully misleading results in other cases if shortfalls in library data are not understood. Some packages handle harmonics better than others, and can analyze higher order harmonics well, other packages have good and extensive protection libraries with good user interfaces to make grading easy.

Where things get really interesting is the current push towards EMT type simulations. There has been a general trend in the industry of moving towards EMT type studies for most Grid Connection plants. This is interesting, as it creates a number of issues that may seem counter intuitive. There is an inherent assumption that EMT models are better as they have a higher fidelity of modelling information, as all equipment is represented by full differential equations, instead of by dq frame positive sequence equivalents. This can be especially important in cases of responses for PLLs and fast transients, but is it really more beneficial overall? For short term transients almost definitely but does it help with studies like short circuit, protection, stability and harmonics?

A couple of things i like to think about at this stage is the Garbage In Garbage Out (GIGO) principle and the quote. "All simulations are wrong - but some are useful" (George Box). This is especially the case in the debate between RMS and EMT. For an EMT simulation to be useful i.e. correct, all the input data has to be confirmed and accurate, and for many type of plants, particularly historical ones, this is not easy to achieve and human factors tart to come into play, with engineers making assumptions and simplifications on the input data as they cannot find the information they need for the package. In addition, the additional computational burden of processing of EMT simulations requires very powerful computes and often specialist techniques such as plant model aggregation, which while useful, can vastly over-simplify important network parameters, leading to a false sense of security.

Comparing RMS and EMT is, a false equivalence, they are different tools to do different jobs. EMTs studies are designed to look at specific instances and applications in great detail, they are not really designed to large geographical areas with large amounts of loads, components and study cases / scenarios. Some of the EMT tools are moving towards this, but their under-pinning hierarchy and structure does not allow them to manage large number of cases and options, quickly and easily. I am aware that AEMO in Australia and MHI in Canada are trying to move towards this approach, so i would be most interested in anyone who has direct experience of this approach to advise on how it is working out practically.

To put this into to context, at Aurora we use DIgSILENT Powerfactory and ETAP for most routine power system simulation studies, both have their strengths and weaknesses, but generally we use DIgSILENT for Grid Code studies and ETAP for industrial type plants. For more complex EMT type studies we use DIgSILENT, PSCAD or EMTP-ATP. Lastly for earthing we tend to go for CDEGS or XSGLab. I am not saying any of these simulation packages are better or worse than anything else on the market. I am 100% sure engineers in different countries will have different preferences, but we have found these fit well with the UK market that we operate in.