Introduction
Power systems planning is the process of proposing new infrastructure for the development of the power system. This is done by attending to technical, social and economic principles, that in the end merge into an economic valuation of the social and technical aspects.
Common practice
The following methodologies are common practices in power systems planning. Each has its advantages and disadvantages.
Electric model based planning: This process consists in evaluating some operational points of the grid with electricity simulation software. Power flows, contingency analysis and dynamic studies are performed in order to determine which changes in the grid are the most beneficial, giving a higher priority to the technical criteria.
Advantages: Highly accurate.
Disadvantages: Very labor intensive. The scenarios and investments are chosen by hand due to the laborious process, hence probably accidentally omitting better choices.
Techno-economic model based planning: These types of modeling techniques, also known as expansion planning, are versions of a classical econometric problem in which the investments in the grid are entered into an mixed-integer optimization software that chooses the combination of investments and when to make them, that produce the most benefit for the system. There are plenty of flavors of these family of problems; Some consider only transmission expansion, while others consider generation and transmission expansion. Some are deterministic while others consider stochastic uncertainty in the inputs.
Advantages: Optimality of the solution. It considers many operational scenarios.
Disadvantages: Somewhat labor intensive. Approximations may make them inapplicable to some problems. Not (easily) scalable to a large number of nodes since the whole problem is formulated inside a Mixed Integer Programming (MIP) solver, which may have a very hard time finding a suitable solution for the complete problem at once.
Mixed approach: Another usual practice is to simulate the market with an optimization software inside a MIP solver and then verify it with electrical simulation software. In this approach, a simplified version of the power system is introduced into an optimization problem that dispatches the generation, sometimes considering some light technical constraints, simulating a competitive market. Some of those market results are then simulated in more detail using electrical simulation software with the complete infrastructure model. This process is repeated for a number of investments.
Advantages: Closer to the real interaction of market and system operator.
Disadvantages: Very labor intensive. The market model and the electrical simulation model are often incompatible. The scenarios and investments are chosen by hand due to the laborious process, hence probably accidentally omitting better choices.
Without aiming at having a comprehensive compilation of the practice, the listed ones constitute most of what is being done at the moment in the industry. Something that all the listed practices have in common is that they are labor intensive. This is reflected in a recent survey aimed at understanding the state of the tools used at leading TSO and ISO companies.
Vision
The advanced Grid Insights vision is not radically different, but it is executed radically differently. Yes, we need optimization, yes, we need electrical calculation, but above all, we need automation. But automation done right.
A common conception of automation is to couple existing pieces of software aiming at joint operation. Unfortunately, experience has proven that trying to couple software vendor product A with software vendor product B is a dead end, because A and B were not designed to interoperate. Sometimes this sort of automation is against the interest of the vendors.
After several initiatives trying to couple vendor products, it was clear that it takes less resources to replicate the required functionality than hacking A and B to work in a consistent and maintainable manner. Furthermore, since those products are moving targets, to be able to keep them dancing together, it is necessary to incur in fruitless adaptation costs every time the vendor software changes their inner workings. A perfect example of this are vendor files being changed artificially from release to release to force an upgrade.
The mixed approach, combining several vendor products is king, but it is labor intensive and that curtails our ability to come up with the best portfolio investments. Regarding power systems planning, the perfect end result would be to have a Pareto front for the decision makers, instead of a handful of scenarios. But why?
Usually, due to the very labor intensive process and resource constraints, it is close to impossible to produce the hundreds of scenarios needed to compose a Pareto front (a cost-benefit curve of equivalent utility). Such a curve would inform the decision makers about which solutions provide the best value at each cost level. This is far better than providing a bunch of points of utility cost that are hard to extrapolate to the solution space (the complete picture of possibilities)
Image 1: This picture shows thousands of investment scenarios evaluated automatically. It also shows the Pareto front formed by those investments that produce the most benefit at each cost point.
So how do we get there?
A better power systems planning methodology
To achieve the above image of a power systems planning, you need:
A very fast electrical solver.
A very fast generation dispatch program.
A very fast black-box solver.
A traceable database.
Do you see a trend here? We need very fast software, and we need it to be interoperable, much like puzzle pieces. In this way, we can build innovative new processes that are not feasible with the business as usual software.
The algorithm works like this:
Prepare the investments. By hand for now, but it would be nice if a machine could come up with candidate investments…
Start the black-box solver. This solver will propose combinations of investments to test:
For each investment combination, run the black-box:
Run generation dispatch for a time period.
Run power flow for the same time period.
Evaluate the economic costs and benefits and evaluate an objective function for the solver.
Once the solver has finished proposing and checking investments, trying to find the better ones, we plot the results, generate reports, etc.
In the end, every investment evaluation is available, or at least, re-simulatable easily for further analysis. This screening process can be as simple or complex as one has time for since the black-box function can be anything you want, and it is all automatic.
Decision makers are usually thrilled to see the Pareto front instead of the usual scenarios.
Paying the price
As the saying goes, there is no free lunch. Coming up with fast and interoperable software is not an easy feat, especially if there is no comparable experience. It is, however, an investment that pays dividends later.
Image 2: Chart showing the effort of in-house development, depending on the approach taken: In gray, using vendor-locked software, in green, using interoperable software.
The image above, shows the experience that has motivated the creation of Advanced Grid Insights; One can definitely buy vendor solutions that are usually locked for extension. These solutions are easy to adopt in the beginning since the early functionality is already present. However, when more advanced functionality is required, but such functionality is out of scope for the vendor, you are forced to hack the vendor tool to serve your innovation.
On the other hand, you could create your own software, that you can extend with time, and adapt to your needs. If this is done correctly, the cost of extending and making the tool interoperable with others is almost trivial. This works as a good investment.
There is a third option not depicted, which is, to start your own software and end up exactly like the vendor-locked solutions. No one wants that, and it is to be avoided by having the right in-house talent.
Conclusion
The common practice in power systems planning is very labor intensive. This makes it hard to be able to produce the best results, and takes a toll on the personnel in the form of burn outs. The clear way out is automation.
Automation is an investment; It may be a good one or a bad one, depending on the approach you make. Experience advocates on building the proper toolbox, rather than relying on vendor-locked solutions.
Decision makers are usually thrilled to see the Pareto front instead of a handful of scenarios. To be able to produce those results, it is required to invest in getting a hold of the simulation process. That means producing software with a well defined architecture that allows for interoperability and scalability.
“We want everything as automated as possible, we want the best results possible, and we want them to be traceable.” - Red Electrica's planning department.
Advanced Grid Insights has been founded around that premise.
Comments