Radical decentralisation – when does grid defection make sense?
Battery storage is revolutionising the way we think about electricity. Perhaps the most exciting thing about the rapid development of this technology is the greater role for renewables that it is sure to facilitate. For years we have been warned that the challenges of managing intermittent generation will limit the role that it can play. That may no longer be true. But in less developed countries the opportunity is greater still. Many of these countries have no reliable power grid. With battery technologies in play perhaps they will no longer need one – innovators can deliver solutions that bypass the bureaucracy of inefficient state-owned utilities. At a ‘pico’ level this is already happening with the deployment of home solar/battery kits across Africa.
Developers are finding a role for batteries right the way through the value chain: from load shifting, to the provision of ancillary services, to behind-the-meter applications. But it is the opportunity to defect from the grid, or to never build the grid in the first place, that I want to explore here. The Rocky Mountain Institute published a report on the Economics of Grid Defection back in 2014, focused on some of the key US markets. I wanted to look at the economics of the same problem, but for a market without the same legacy infrastructure.
To help, I’ve performed some (basic and simplified) analysis, to compare the levelised cost of putting a microgrid system in place with the cost of putting a transmission cable in place to connect to grid power. I have assumed that the microgrid system combines solar PV, batteries, and diesel back-up. The question that I’ve tried to answer is: For a given battery cost, what is the maximum transmission cable length I would invest in before switching to a microgrid system? There are, of course, lots of assumptions embedded in this analysis; for example, I assume that solar PV costs fall at roughly half the rate of battery costs. The output from some of the analysis is shown below:
The numbers, of course, could be challenged and debated, but the graph shows how the trade-off between off-grid and on-grid solutions becomes much more finely balanced as battery capex costs drop. This is particularly the case where the load to connect is small. The graph suggests that at ~330 $/kWh it could even make sense for loads in countries with an established grid to defect. Battery costs are declining quickly, and this is not far from where many battery technologies are today (see Lazard’s November report on battery costs).
The graph above sows the indifference point for this on-grid/off-grid trade-off at a grid power price of 100 $/MWh. The second graph, below, shows the impact of different power prices, all for a 5 MW system.
This second graph shows the importance of the cost of grid power to the decision to go off-grid. It highlights one of the most pressing energy policy dilemmas in the developing world today: should reform efforts and investment be focused on improving the efficiency of grid power, or on off-grid solutions? The more one focuses on the former, the smaller the potential market for the latter.
I should highlight some of the many simplifications in the analysis:
There is a trade off between battery duration and the need for back-up, which is not explored here
In reality there are lots of different transmission technologies; it is hard to simplify this down to a simple $m/km number as I have here
The amount of investment in lower voltage networks will depend on the density of load
Perhaps more importantly, the analysis ignores the difference between the partly/fully depreciated distribution asset base (for the on-grid case) and investment in new distribution infrastructure for the off-grid case
Every potential application will have its own specific economics.
But in spite of the complexity, battery technologies are poised to shake up the power sector in both the developed and developing world. It’s already happening – see The Guardian’s article on an off-grid town being developed in NSW.