What can the power system learn from the internet?
Every now and again I find myself using a bit of technology and realise I have no idea how it actually works. I recently had this realisation about the internet whilst marvelling at Netflix’s ability to stream a whole film seamlessly across the network.
With a bit of reading (I can highly recommend this book), I discovered just how elegant the design of the internet is. It is a truly remarkable invention, made more remarkable by the fact they got so much of it right first time around. Yes, there are some things you would do differently if you started again now. The first design had no encryption and only accounted for 4 billion IP addresses which in retrospect was too few for all the internet connected devices today. We can perhaps chalk this down to the humility of the designers who had no idea that we would want to conduct banking over their network or connect our fridges and lightbulbs to it.
The more I read, the more I realised that there is a lot we could learn about how to adapt our power systems for the energy transition. Whilst the parallels between the internet and power networks aren’t always clear cut, it is worth remembering that the internet has solved many of the problems facing power networks in the next decade. Specifically:
- Rapid growth in devices and connections
- Congestion at the network edge
- Fundamental shifts in the types of load — the internet has moved from delivering small blocks of text to large streams of video. The power system will have to contend with big lumps of domestic load like EV charging and electric heat
Below I have picked out a few of the internet design ideas that jumped out at me and may be interesting when thinking about our power networks.
Operated by protocols not control centres
The amazing thing about the internet is that no one body is controlling it. There isn’t a central control centre dispatching every packet to its final destination or dictating how many data centres we need or where they should be. Instead the internet operates via a set of well defined protocols. These protocols mean that the system maintains robust, optimal operation without the need for a central clearing house or coordinator. Clearly there is some supervision and a mix of public and private bodies help shape these protocols and the infrastructure that delivers them. It is, however, a far cry from the world of central dispatch and clearing we have in the power system today.
Self stabilising
The most elegant thing I discovered about the internet is how it deals with congestion. It is a beautifully simple but immensely effective technique. The amount of data sent in any one go by a server is ramped up and down based on how fast the client computer is acknowledging receipt of that data. This means that as the network gets more congested, the server backs off and thus acts to ease congestion. It’s a self stabilising feedback loop that allows all the individual nodes on a network to cooperate without actually needing to communicate with one another directly. This is a very important concept in networks and one that we desperately need to solve in our low voltage networks. Where there is limited network capacity, you have two choices. Either you ration the available capacity amongst all the end users (e.g. everyone gets a maximum of 2kW to power their home); this is a highly inefficient way of sharing a network. Or, you find a way to get the users to cooperate to share that network capacity efficiently. One way to achieve cooperation is by having a central actor to coordinate (a system operator). The challenge with this approach is that the system operator quickly becomes a bottleneck and barrier to innovation. The other way is to design cooperation through protocol, as the internet has done.
We have already achieved this kind of approach in one part of the electricity system. Frequency response, where devices listen to the system frequency and act to regulate it, is a great example of cooperation through protocol. Once the service has been tendered, it is hands off for the system operator as the devices on the network make the right decisions autonomously. We need to design more of our power system markets and protocols like this. This frees up the system operator to be less of a director and more of an efficient market facilitator.
In the low voltage networks, real time feeder congestion price signals communicated between the local substation and the homes connected to them could achieve efficient cooperation without central control.
Modular by design
In a previous blog post I described how, like any complex system, we need to decompose our power system into smaller modules (or layers), each with their own optimisation goals and with clear interfaces between them. The internet is designed exactly this way. It is broken into layers with each layer focussing on a specific task. The link layer looks after the physical transmission of packets across each hop. The internet layer focusses on routing the data from A to B. The transport layer (in TCP world) makes sure nothing gets lost. The application layer deals with making that data useable for a human.
Whilst the power system may not decompose in exactly the same way, the principle still holds. Systems work best when split into modules which have a clear function and interfaces and don’t have to worry about the functionality of all the other modules. In the power system, this might look more like a hierarchy of optimisation goals starting at the home and working outwards to the whole transmission grid.
Approach to security
When the internet started to be used to transfer sensitive data like bank details, a security layer needed to be added. Rather than having to send every packet to a central clearing house to be authorised, the internet uses the SSL/TLS protocols to allow for the client and server to communicate securely over an open network. Certificate authorities act as the authenticators of identity without having to see or process any of the data that is being transferred securely.
We could learn from this in the energy system. Our energy data systems tend to be centralised organisations who control both the flow of data/service requests as well as authority. The DCC, who operate the data network for smart meters, is a good example of one of these centralised data bodies. We need to find better ways of securely communicating with and dispatching assets at the grid edge without these messages having to go through central clearing. This is certainly not an advert for blockchain, the internet security layer was invented in 1999, long before the advent of decentralised ledgers and is deliverable through mature technologies available today (blockchain being one of many available solutions).
A major difference
One of the major differences between the power networks and data networks is that data clearly has a start and end node whereas we treat power networks as a central pool which you put energy into and take energy out of. The ability to track data across the data networks leads to some of its optimisation advantages. For instance the congestion feedback described above is derived from the direct server/client connection.
The power sector can learn a bit from this point of difference. If we were able to track and model energy flows better, we could dynamically price congestion or charge more effectively for which bits of the network participants are using. The former is the basis for locational marginal pricing. The latter is local balancing where we reward optimising energy portfolios that minimise pressure on the networks. By treating everything as a central pool, we push all the optimisation responsibility onto the network operator and even the best SO in the world can’t beat the innovation, competition and resultant optimisation that price signals and markets deliver.
Conclusion
It’s clear, as an industry, that we need to move our network and system thinking on a paradigm to keep up with the rapid changes we are seeing in how we use electricity. The successes of the internet to rapidly scale, adapt to new use cases, run without central ownership or control and be a platform for unimaginable innovation could teach us a lot in how to deal with our own challenges. If I was calling the shots I’d be pulling in the best minds from the world of data networks to see how we can think a bit differently.