Coordination signals and automatic demand-responseKirjoittaja Pedro H. J. Nardelli BCDC Digital
Thanks Florian Kühnlenz for valuable discussions and feedback.
Another day I was waiting my bus in front of a traffic light and I started wondering about the reasons why it is necessary to have these lights. In some busy streets, they can coordinate who can go and who cannot. Likewise, in some health centers, you can select what is your problem and then you get a number assigned.
These examples are situations that involve some kind of coordination based on signaling. It appears obvious that this rule-based coordination is needed. Imagine the same situation without this coordination signal: messy traffic, crazy pushing and pulling in lines, discussions about priorities in treatments etc. In such cases coordination is needed.
Although similar, the different phenomena have their peculiarities, which serve as a classification tool. Take the traffic light as an example. The cars stop in the red light, not only because they know that this is what is expected, but also because everyone knows that the others are aware of this and are expecting this behavior. So, the traffic light as a coordinator implies that everyone knows how to behave, and that everyone knows that the others know how to behave. Their goal is somehow shared.
In the health center, the person knows his/her number, but doesn’t know how many people are in his/her line, so it is impossible to estimate how long it is going to take and if the situation is fair based on his/her own health state compared to the others. In this case, the way the service works gets unclear.
These scenarios are based on information broadcast for coordinating agents. Every case is different, but this kind of differentiation from the perspective of ”who knows what” introduces a natural characterization.
Appliances and automatic demand-response
The so-called Internet of Things (IoT) and their constituent smart appliances are expected to play a big role in the modernization process of the electricity grid. With the inclusion of more intermittent sources of energy, supply gets less predictable and smart appliances are expected to react so demand responds by adjusting their usage according to the available power.
In electricity grids, the frequency of the alternate current shall be constant at 50 Hz in Europe and reflects a balanced supply and demand. If the supply is higher than demand, frequency rises. If supply is falling short, frequency drops. The grid frequency is the same for regions connected in the same power grid (synchronous region). So everywhere in Finland, Sweden, Norway and parts of Denmark, the frequency seen by appliances is the same. Hence, frequency can be seen as an indicator of real-time supply and demand state, available anywhere connected inside a given synchronous grid.
Then, if smart appliances know the state of the grid, they can react accordingly. Fridges, for instance, can adjust their cooling cycles as a reaction to the frequency. So, if frequency is too low, they postpone their cycles a little bit, reducing the load on the system. It looks like a great solution. But, it comes with a big problem.
Let’s come back to your classification of who knows what. Every smart appliance knows the frequency and also knows the others are aware of it. They are subject to the same broadcast signal (the grid frequency), but they do not know how their peers will react. If they act based on the signal, a kind of coordination will happen: every smart fridge will do the same at the same time. If this would actually happen, the idea of demand-response goes in the opposite direction as it has been designed. The unintended synchronized behavior will create instabilities, which may even lead to a blackout. In this case, the smart appliance acts against the system, what ends up to be not so smart.
Is there any solution?
Synchronizing behavior like the one described before is possible to be solved based on randomization or central coordination. Randomization implies that each device is programmed to react to a given situation in a random manner. For example, if they see a frequency drop, they only change the cycle with a given probability. The central controller solution coordinates which appliance is acting or not.
This kind of problems have been studied in many different fields (from communications engineering to traffic control) but it still poses huge challenges for every specific system. In the case of electricity grid, there are many nuances that make the implementation harder than in other scenarios. Whoever designs the system needs to consider: multiple time scales from sub-second level up to a whole day, different kinds of appliances, coexisting management strategies, electricity traded as a commodity in the market, weather conditions and many others factors.
The challenges are posed. The first step towards an effective management policy is to understand the nature of the phenomena to be managed. If methodological biased is done from the beginning, smart ideas may lead to undesirable outcomes when applied in practice. When coordination is needed, it is important to understand how information about the physical system is shared, who knows what and how decisions about actions are done. A careful management design shall be done looking at all these aspects as a complex whole. The modern electricity grid cannot be reduced to aspects of physical electricity grid, information networks or decision-making procedures; it is the totality of the system in action that matters.
Traffic light: https://upload.wikimedia.org/wikipedia/commons/9/91/Modern_British_LED_Traffic_Light.jpg
State of the power system: http://www.fingrid.fi/en/electricity-market/power-system/Pages/default.aspx
Pedro H. J. Nardelli
Oulu University, Centre for Wireless Communicatios CWC