An Introduction to Blockchain and Retail Electricity Markets
The conversation surrounding blockchain in energy seems to recycle the same tired arguments for or against without much nuance. We know that current blockchains are costlier than the Visa network. This is not a “serious flaw”. It is a design trade-of. And yet somehow there are other papers still thinking blockchains are all about “cheaper, faster transactions” compared to legacy centralized systems. Both of these views miss the point. Blockchains exchange the speed and ease of centralized databases for the benefits of a decentralized system: one without a need for a centrally enabling actor, or trusted third-party. Centralization comes with its own downsides, and the grid is extremely centralized as is. But do the wide-ranging efforts at decentralization call for blockchain?
At this point, transaction costs and scalability should be thought of in terms of network security, governance, and user agency. There is a cost to building any network, and the “inefficiencies” associated with blockchain are for a reason. On the other hand, we all understand the hype surrounding blockchain, and how many on the other side of the debate throw around terms like “disinter-mediation” without any real concern for what it means.
In practice, a blockchain is a slow, expensive, immutable, and distributed database. The benefits can mean greater network security and the protection of individual users from censorship. For, as Nick Szabo puts it, “trusted-third parties are security holes” and can also lead to censorship. It’s time to move beyond looking at blockchains — especially on the electricity grid — from a purely cost based or data throughput perspective, and start recognizing the very real trade-offs they entail.
The electric grid has emerged as a popular example of a sector screaming for blockchain, so it seems time to discuss what that actually means. Thus, this article is meant as an overview of the problems to be faced by retail electricity markets, and whether or not certain kinds of blockchains are the proper solution for those problems. A second article will explore the issue in more depth with different projects’ actual solutions, like Grid+, Exergy, EnergyWebX, and PowerLedger, and provide some bare-bones technical overview.
Currently, the centralized-ISO model works at the wholesale level from a reliability standpoint, so why shouldn’t we expect it to work at the distribution level? Prosumers make the DSO entirely different than the ISO: large generating plants have no reason to be connected to other nodes, but prosumers do. So, a centralized architecture works best at the wholesale level, but may not at the distribution level. The way the bid stack is handled by the ISO will not function the same way that it does with the DSO. We’ve all heard plenty about transactive energy, so it is important to note that a DSO, by definition, must look very different from the ISO indeed.
Transactive energy means building a decentralized, autonomous, real-time, two-way distribution system with high degrees of variability and uncertainty in the assets being traded (renewables). This requires a radically different approach to current data-handling mechanisms used on the electric grid. Blockchain is one feasible solution. A centralized DSO is too. What are the trade-offs of each?
Some Data Problems on the Electric Grid
The problem facing grid operations modernization and liberalization is threefold: censorship, cyber-security, and real-time capabilities. While grid modernization efforts vary greatly across the country, truly integrating renewables requires far more than any current initiative. While in the past centralized architecture may have made sense, any failure to solve these problems henceforth can be laid at the feet of incumbent institutions. Whether that is due to negligence, protection of their own interests, or the failings of their chosen methods is unimportant.
For now that the technological capabilities exist to integrate, control, and manage distributed assets, any prosumer not able to sell as much power into the grid as they would like is being censored. Furthermore, the grid is already vulnerable to cyberattack — especially at the distribution level — and given our newfound knowledge of the failings of centralized databases (e.g. Equifax, Facebook), the prospect of millions of newly interconnected assets of users in their home points to the need for a renewed seriousness in how data will be handled on the modern grid. Lastly, the fact that solar, wind, storage, and other assets can be unpredictable, respond far faster than current grid assets, are non-inertial, and distributed, calls for a new paradigm in grid operations.
These concerns lead to the following questions: in discussing the emergence of retail markets and a DSO, can traditional mechanisms developed at the wholesale level with ISOs handle the needs of a distributed system? Can these mechanisms be updated, or is something else entirely called for? Can we accomplish our aims with a traditional, centralized grid operator and database, or is a distributed blockchain-based system required?
It is easy to conflate the problems of centralized data handling mechanisms with an unwillingness to change on the part of ISOs, utilities, and regulators. Wholesale markets settle on 15-minute intervals, power producers must often have larger capacity than 1 MW to participate, and much of the country still lacks smart metering capabilities. This is true largely because grid operations haven’t required anything more granular than that in the past. However, the problems facing the grid today — namely, integrating renewables and distributed generators — do require greater granularity. This is to say that the current state of the grid does not necessarily point at the failings of centralized databases.
Yet, in the purview of the demands of building a modernized grid and the interests of the current grid’s users, it is easy to see what the grid needs for data handling at a top level. Whether or not this system is a central database or distributed one is what is up for debate: The modern grid demands a system that handles real-time, secure, trust-minimized transactions with embedded regulatory oversight and governance, without censoring users.
In order to make the case for blockchain, one must first make the case for an open system. Blockchain will do nothing if utilities maintain their monopoly status on distribution grids. Thus, the following discussion assumes the creation of a third-party DSO, a segmented market, and also a changing role of the utility. The removal of franchise rights would be nice, too.
Cyber and Network Security
In going with a centralized DSO model, the first issue at hand is that the DSO becomes a data honeypot of epic proportions: a single centralized database controlling millions and potentially billions of connected assets. These aren’t the commercial and industrial connected assets of the wholesale markets. These are peoples homes. One need look no further than the Facebook or Equifax headlines of this year to understand the dangers of centralized databases with sensitive information. Distribution grids are almost entirely unprepared for the amount of data they are about to be handling from a cyber-security standpoint. What’s worse, energy management systems may end up connected to non-energy related user data through Google Home, Alexa, and the like.
Power grid resilience and security are front-and-center in national policy debates. However, the focus on saving…kleinmanenergy.upenn.edu
If one node goes down in a distributed network, the network keeps running; in a centralized DSO model, the whole network crashes. The shear weight of data on the electric grid is about to increase astronomically. The burden of such on current ISO’s is unremarkable. However, as we think of a more distributed system, we must recognize that the failure of a centralized database would mean the grid going down. Even if more expensive from a transaction cost basis, a blockchain-based system may avoid externalities stemming from grid outages and centrally insuring costumer transactions.
It is quite simple to see that centralized databases, which have proven quite vulnerable in the past, are an issue on the modern grid. Single points of failure must be avoided. What could a malicious party get away with if the DSO were compromised?
Blockchain offers one promising solution to this problem. A distributed ledger has no single point of failure and as such could potentially be more secure in managing user data than a centralized DSO.
Currently, utilities are financially incentivized to censor the prosumer, and they are doing just that. The most obvious example of this is that even with NEM, prosumers are not allowed to produce more than they consume. What this means is that oftentimes, roofspace is left uncovered by solar panels so that homes don’t overproduce. While this is a tariff structure issue and not a data issue, what it amounts to is the user being censored. They should be able to inject as much power into the grid as they please, and be compensated for it by proper market signals.
This censorship is because the two main business models of utilities currently are owning the poles and wires, and a markup on kwh procured through wholesale markets. A retail market hurts their payment settlement business because they sell fewer kwh to customers, but not their ability to act as a toll person for the poles and wires. We’ve seen previously that the modern utility should not be able to do both of these things simultaneously, for this exact reason.
However, even in an idealized, segmented DSO where utilities no longer own poles and wires, the utility being able to procure wholesale power and not own behind-the-meter assets incentivizes them to censor users. What is needed is a grid architecture that allows users direct access to markets, without an intermediary if they so choose, in order to force the hand of utilities to support both buying and selling of assets.
In a bidirectional, nodal network such as the electric grid, the ability to have direct access to markets without an intermediary is what pressures any chosen intermediary (the utility) to act properly. While “disinter-mediation” is used frequently in conjunction with blockchain, the truth is that a well-regulated, non-participating, but still centralized DSO could also do this. We need a grid architecture that allows users direct access to markets to avoid being censored.
Thus, this is admittedly more of a regulatory issue than a technological one, but tradeable smart contract enabled PPA’s offer an underlying technological process that is politically immutable. For example, in a centralized DSO system, regulators or utilities would still potentially have the political clout to manipulate markets through the PSC. For example, in Ontario the newly elected government canceled hundreds of renewables contracts, and in PJM, there is a growing issue with MOPR’s. When integrating fixed-cost 25 year assets, consumers and developers need to know that the rules of a market aren’t going to change with the political winds. A defining feature of blockchain is that it’s very hard to change the governing rules.
The concern with a centralized DSO is political. Censorship is just the outward effect of centralized decision making, and it is why a segmented market (a distribution grid without utility monopolies) is so important. And while climate change is one of the most polarized issues in politics today, this should not translate directly to markets. The market must be set up to be indiscriminate towards different generating assets, which it is not currently. One can Net Meter solar, but not cogeneration, for example. Ironically, an open market will likely favor renewables, as is happening on wholesale markets. So while this issue is not a purely technical one, a blockchain-based settlement layer may limit the ability to interfere with this process.
Real-Time, Decentralized Operation
As dynamic elements are introduced to the grid, a centrally dispatching entity such as an ISO is far too static of an operator; dynamic assets call for a dynamic market; dynamic markets call for dynamic operators. Humans are generally good at intervening and fixing first order effects, if they can model them properly. In ERCOT, Demand Response programs are operated with call centers, meaning a human being is reacting to dynamic grid events. Human beings with telephones are currently controlling our “grid of the future”. Seriously.
With millions of connected assets, will this be possible? Clouds overhead, wind gusts, or local outages at this level of granularity lead to an inability to properly model (as we do at the ISO level), a grid where we have to regularly respond to local events with 2nd and 3rd order effects (e.g. non-inertial resources means frequency — a second order effect — won’t self-correct in the same way it has with spinning generators).
The current centralized nature of the grid means operators care only for balancing the bulk grid. But on a distributed grid, problems become local; voltage issues and duck curves can happen instantaneously in neighborhoods with high solar penetration levels while the bulk grid runs smoothly. Thus, grid operators will become as effective as the weathermen at predicting grid events, which is to say: not very effective. What this means is that the grid must become increasingly more automated, and controls themselves more distributed.
Put another way, the “omniscient central operator” that those in the DSO camp are calling for cannot, by definition, be omniscient. This is not to say, again, that blockchain works better technologically than a centralized database in this case. Rather, it is to say that, because the issues of variable renewables do not scale linearly, a DSO faces completely different problems than an ISO. We cannot map one-to-one the success of an ISO onto a DSO because 1000 5 kW arrays is not the same as one 5MW array. This isn’t a wholesale market with large, dependable loads and baseload. DG demands grid operation to become more autonomous and sophisticated. Furthermore, we must recognize that this is a trade-off in architecture: a distributed grid will likely mean more frequent local outages, but less frequent grid-wide outages.
Lastly, it’s important to note that real-time markets mean that the payment settlement layer itself will need to incorporate micro transactions, which currently aren’t handled well by legacy transaction mechanisms like credit cards. For example, Visa charges a flat $0.10 fee for a transaction on top of a percentage, making a trade of a few kwh’s far too expensive. It would likely mean potentially complicated and expensive escrow mechanisms with end of the month settlement. In NY, ESCOs are required to put down a $1milllion bond at the wholesale level in order to operate. Who would handle payment imbalances and defaults between retail energy traders?
An actual cryptocurrency offers a solution in the form of instantaneous micro-transactions. While transaction costs in aggregate on current blockchains are more expensive in exchange for the added security, the beauty of them is that, with the absence of flat fees, they can be better at scaling down to small payments. Grid+ is addressing this issue in, in a sense, in the form of pre-paying with cryptocurrency.
It should be apparent now that this discussion goes far deeper than transaction costs and the scalability of blockchains. There are many functional ways that we can build electric grids, but how do we optimize the function with decarbonization, reliability, cost, and prosumer freedom as input variables? Thus, this discussion about a centralized DSO model vs. a distributed blockchain model is not a discussion primarily about data throughput. This is a discussion about security, governance, censorship, reliability, and the trade-offs in efficiencies on the future grid.
Blockchain could provide a real-time, censorship resistant, cyber-secure system. While blockchain doesn’t offer the end-all solution for any single one of the problems detailed above, it does provide a compelling one when weighing them in aggregate. This article is not an attempt to say that that means a blockchain based system would be any better than what a well-run DSO could provide. The answer should become more clear over time, as more distributed generation projects come online and the markets we build encounter real-world issues.