It is now obvious to industry observers and players that the electrical energy landscape is changing rapidly. It is no exaggeration to say that we don’t yet know how our power will be generated, delivered, stored, controlled and used even as short a period of time as a decade from now.
Any technologist who thinks they do know probably either lacks imagination, or has an over confident view of their ability to predict technical advances. This uncertainty in the industry has spilled over to the public sphere - now the energy industry takes up several column inches in any newspaper on any given day, and has been used by our politicians and journalists for contests on public policy.
Whilst we cannot predict when and how change will unfold, this should not prevent us from taking steps to manage the change that is underway. Power system analysts use a variety of tools to assess the best way to design a new system, augment an old one and manage an existing network. These tools are perfectly adequate for the standard systems we have now which were first developed in the late 19th and early 20th centuries – but they are already showing the strain of operation in the 21st century, with the introduction of new generating technologies.
The deceptively simplest and most-used power system analysis tool is the power system load flow. As the name suggests, load flow analysis is a technique employed by power systems engineers to calculate the power flows through an electrical network. The current state of the art, which typically makes use of commercially available power system software, makes this calculation virtually routine.
So why would it need to be changed in response to the new generation technologies?
There are a couple of issues with traditional load flow analysis that become painfully clear to electrical engineers (or often prospective electrical engineers) the first time they are introduced to the technique.
Its modus operandi is to ‘guess the answer’, and then improve the guess by mathematical iteration. It is an inherent weakness of this methodology that it is not guaranteed to produce the right answer. When the answer is wrong, experienced engineers usually pick this up straightaway but it can (and still does) often mislead the novice.
However, the fact that sometimes the answer is wrong is not why the load flow calculation needs re-invention.
It is simply not up to the task that we are now asking of it.
As more and more renewable generation is added to the grid the intermittent nature of the generation becomes more of a challenge for planning and managing the power system.
Solar PV output obviously follows daily cycles, but as the graph above and below indicates, the output varies with the seasons and with cloud cover over the installation. All of the power generation graphs in this article present one month of data with each day arranged in sequence from day one to day 31 of the month.
Whilst solar installations produce output only in day light hours, and (excluding the effect of clouds) produce the much the same output from day-to-day, wind farm generation output is much more variable.
Wind farm output appears to be far more random as the model above and below indicate, but it is not as unpredictable as the daily graphs imply. In reality, the output of wind farms displays significant regularities which are not readily discerned without detailed analysis.
Gas turbines are sometimes dispatched on an opportunistic basis, taking advantage of high market prices and switched off at other times.
Even traditional base load units (in Australia these are usually coal fired) are not strictly speaking constant in output; they often vary their output because of maintenance cycles, unplanned and planned outages, and market bidding behaviour.