Flat Investment Levels: End to RE Cost Reductions?
A view from the stratosphere:
A few weeks ago, Ed Kelly posted a new blog entry at Stratosolar. He reviewed the latest Bloomberg report on global clean energy financing. As Ed sees it, the figures reported spell trouble for conventional views of RE. Anyone counting on wind and solar alone to cut carbon emissions in time to avoid the worst effects of climate change may be sadly disappointed.
The numbers that Bloomberg reports are quarterly RE investments, broken down by global region. The key bar chart is shown below.
As one would expect, there’s a lot of variability quarter to quarter, year to year, across different regions. While open to interpretation — the uptick starting in 4Q 2016 could be taken as the start of a new growth phase — the moving average looks roughly flat since 2011. As Ed points out, that is not the sign of the booming market we’d expect if RE resources were truly competitive with fossil fuels. If something doesn’t change, RE will not grow fast enough for us to hold a 2℃ line on global warming.
Subsidies still rule
Despite impressive reductions in the specific cost of solar panels and wind turbines, it seems that investment is still heavily dependent on subsidies. The type and size of subsidies vary across regions, but whether they be feed-in tariffs, investment tax credits, accelerated depreciation allowances, renewable portfolio standards, or whatever, the pattern looks the same: cut back on subsidies and investment drops.
The pattern is not so evident if one looks only at quarterly installed capacity. Globally, that’s been rising pretty consistently. But the rise over the last six years seems due almost entirely to cost reductions. Each investment dollar is buying more installed capacity than it did earlier. That’s encouraging, but is it enough? Can reductions in the cost per kWh for RE resources alone get us to where we need to be while monetary investment levels remain flat?
In theory they could, if prices continued an exponential decline and if costs for the components declining in price remained dominant for the systems as a whole. But both predicates are problematic.
Basic problem #1
There are two basic problems here. The first is rooted in the fact that the RE cost reductions, though impressive, still follow the scaling pattern characteristic of industrial learning curves. Each doubling of production volume tends to bring some percentage of cost reduction. For a surprising range of products, it’s roughly a 20% drop in the cost per unit of production. That’s only a rule of thumb, and there are factors beyond production volume that affect costs. But a 20% reduction in unit cost seems to be the typical return on the capital spending associated with a doubling of production volume.
The point to understand is that cost reductions of this sort depend on a growing market — or on a growing market share for the most efficient manufacturers in the case of market supplied by a number of competing vendors. The cost reductions are the result of increased productivity after investment in new equipment, new designs, and new, more efficient processes. Manufacturers have to expect that the investment will pay off in increased sales revenue. Otherwise it’s a losing proposition and won’t happen.
A consequence of that dynamic is that if the market stops growing, these “learning curve” type cost reductions taper off. There can still be “collateral” cost reductions due to spillover from other still-advancing technologies that play a role in production. The cost of factory robots is an example. But unless they affect the cost of inputs, collateral cost reductions still require capital investment in new equipment. Capital investment in new equipment becomes hard to raise in a market that isn’t growing in monetary terms. So it’s a general rule of manufacturing that when the market stops growing, cost of product stops dropping. It may even begin to rise.
For the last six years, the overall global market that manufacturers see — i.e., the monetary revenue they receive from sales of new RE capacity — has been relatively flat. Market share for efficient producers, however, has been growing as vulnerable competitors are squeezed out. But for PV panels, that process seems largely played out. The market has become dominated by a small handful of large super-efficient producers who have little to gain from undercutting each other’s margins. Hence the long term outlook for PV panel prices would appear stable to rising. A similar situation likely applies to wind turbines, though it’s a different set of players.
I should note that this is by no means a consensus view. The overall BNEF report itself, of which the cited clean energy investment report is a chapter, projects a continuation of price declines. But the basis is not clear. If it’s merely a forward extrapolation of trend lines for the last two decade without recognizing the role of market dynamics, it doesn’t inspire confidence.
Basic problem #2
The second problem is that the elements that have been falling in price — wind turbines and PV panels — have fallen so far that they’re no longer the dominant elements in the systems required to deploy them. Not, at least, if one honestly considers the full extent of those systems. In fact it’s not clear that the rate of deployment would be much accelerated if the cost of PV panels and wind turbines per se fell to zero. Other factors have become limiting.
Some of those other factors are obvious and non-controversial. For grid-scale PV, there’s land acquisition and the permitting process, plus site preparation, installation, and grid connection. Similar considerations apply for wind farms as well. But the biggest cost factors beyond solar panels and wind turbines are indirect. They relate to intermittency, and are more controversial.
The issue with intermittency — and the reason, IMO, that RE deployment remains tied to subsidies — is that there is no established market for energy “as and when it happens to be available”. The market that exists is for energy on demand. Wind and solar don’t deliver that. On their own, they can’t. They can only be elements within a larger system that is able to address the market that actually exists.
Grid-connected wind and solar systems are currently parasitic, in the sense that they exist on top of and depend on a host system from which they siphon resources. Specifically, the revenue they generate is diverted from revenues that would otherwise go to the owners and operators of the host system. So long as the host system’s generation capacity is still required to meet demand when the RE resources are not delivering, the total system cost is raised. The result is that the cost of electricity rises — a giant indirect subsidy to renewables at the expense of ratepayers.
Toward market-based RE deployment
For success in the electricity market, it’s the overall cost of full solutions that matter, not the cost of as-available kilowatt-hours. A “full solution” is one able to reconcile available supply with demand. There are three avenues of approach: energy storage, long distance transmission, and demand side management. They aren’t mutually exclusive; competitive solutions will involve mixes of all three. But however the reconciliation is achieved, the cost of doing so must be considered as part of the cost of intermittent RE. When that’s done, wind and solar are still not competitive with untaxed fossil fuels.
Advocates for intermittent RE would prefer to dismiss or play down those costs. Many contend that the required pieces are already in place, needing only modest upgrades to accommodate high levels of RE penetration. The data in Bloomberg’s global investment report suggest otherwise. So too does the way regional investment levels drop when subsidies are cut back.
Of the three approaches for reconciling electricity supply and demand, energy storage gets the most attention. And most of that attention is focused on batteries. The electric vehicle market has led to sharp gains in cost-performance of battery systems — a trend expected to continue. That should be of major benefit for RE systems. However, there’s a lot of confusion about how much storage capacity is actually needed, and what cost targets the storage must meet.
The confusion is understandable when one considers that there is no single market for energy storage. There are different application domains within the overall storage market. There’s a degree of overlap, but overall the capacity and cost requirements span orders of magnitude. Different technologies are likely needed. Let’s take a quick look.
Domains for energy storage
There are four application domains for energy storage that I find useful to distinguish. They’re based on cycling times and capacity:
- At the low end is peak shaving and supply firming. Cycle time is minutes to at most about an hour. This level is enough to keep the output from wind or solar farms smooth and predictable. It avoids bumps in the curve of demand that other generators must supply. That in turn reduces forced cycling of those other generators on short notice. It doesn’t, however, eliminate the need for those generators to be available.
- An order of magnitude above that is the storage capacity needed to accommodate the regular diurnal cycle of solar power. The cycle time is one day. If storage at this level is available, it can reduce the need for peaking generators. It also establishes a floor on hourly wholesale prices of otherwise surplus power. When there’s no other demand for it, the as-available energy can be stored.
- Above that is the storage capacity needed to bridge extended periods of adverse weather — dunkelflaute as Germans now refer to periods of dark overcast skies with little to no wind. The storage required for this is many times the requirement for diurnal cycling. The fact that it may only be tapped a few times per year makes the economics particularly challenging.
- The highest and most demanding level of storage would be for addressing seasonal variation. Here the issue isn’t a few days of near-zero output during adverse weather, it’s a whole season of substantially reduced RE output that has to be covered.
Electric vehicles have pushed down the cost of battery storage a long way, but batteries are still only cheap enough to address the first of these application domains at a cost that’s competitive with dispatched generation from untaxed fossil fuels.
Diurnal cycle buffering requires several kWh per kW of RE capacity. It’s the minimum level of storage needed to break intermittent RE free from its “as available” trap. Consequently there’s a lot of interest in it. However the battery technologies currently available are still too expensive. There’s hope that continued growth in the EV market will change that, but it remains to be seen. It may be that other technologies will prove more suitable.
The next level, bridging for extended periods of adverse weather, requires yet another order of magnitude capacity increase and cost reduction. That’s if it’s to be accomplished from storage. There’s no prospect that conventional storage batteries will ever become cheap enough to address this segment of the storage market. New types of flow batteries might possibly manage it, but presently, generation from stored fuel is the only economically viable option.
Harder still is sufficient storage to address seasonal variation in supply. Seasonal variation is a non-issue for nuclear power, but for high penetration RE scenarios excluding nuclear, the only 100% RE solutions would involve mixes of seasonal industries, heavy overbuilding of capacity, and curtailment.
One scalable possibility for a seasonal industry is fuel synthesis. It’s a popular idea among RE advocates. The problems are low efficiency and cost of capital. At best only around 40% of energy input to the process can later be recovered. And even if the fuel produced is simply hydrogen from electrolysis, the capital cost of the plant is high enough to make intermittent seasonal operation problematic. Against untaxed fossil fuels, it’s very hard for synthetic fuels to compete.
The unfortunate reality is that it’s very challenging to bridge between “as available” energy resources and the “energy on demand” model on which developed economies have long relied. Not technically challenging; there are any number of workable approaches if cost is no object. The problem is economic viability. Economic viability is signalled by a self-sustaining spiral of rising investment and market size with declining prices. As the Bloomberg report makes clear, that has yet to happen. Dispatched generation from untaxed fossil fuels sets a high bar for clean energy solutions to clear — or one could say a low bar, cost-wise, under which they must limbo.
That, of course, is no news to advocates of nuclear power. That message is central to their advocacy. It may well be that one or more of the next generation nuclear technologies now under development will succeed. If so, it will ultimately leave our present obsession with wind and solar looking quaint.
Or not. With the world ecosphere and future climate at stake, it’s important to hedge our bets. Though support for “100% renewables” and opposition to nuclear seems more ideological than rational, I’m not prepared to dismiss the “100% renewables” vision entirely. It has mainstream momentum, and there are technical approaches to it that could prove economically viable. Leading the pack, in my view, is the approach laid out in the Stratosolar site I mentioned in the opening.
The logic of deploying PV capacity in the stratosphere on tethered platforms is simple enough. The low temperatures and more intense sunlight improve panel efficiency, and the absence of high winds, rain, or hail, and the bone dry atmosphere, could ultimately reduce the cost of the panels deployed. But those are not actually the major benefits. The major benefits relate to taming of intermittency issues.
The Stratosolar approach could, in principle, provide the kind of complete systems solution that would enable a market-based exponential growth in investment levels and capacity. It could support (in this case, literally “support”) sufficient integrated gravity-power storage to cover the diurnal cycle. With conditions in the stratosphere unaffected by cloud cover and weather, PV deployment there would eliminate issues of extended periods of adverse weather. And while it couldn’t eliminate the problem of seasonal variation entirely, it would reduce it.
At the latitude of London, for example, winter sunrise 12 miles above the surface occurs almost 45 minutes earlier, and sunset 45 minutes later, than it does on the surface. More importantly, sunlight reaches the panels with nearly full intensity whenever the sun is above the horizon. As a result, the difference between summer and winter PV production would be much smaller than at ground level. It would be easier to bridge via seasonal industries and generation from fuel.
The Stratosolar approach remains speculative. Tethering of large lighter-than-air platforms floating in the stratosphere has never been demonstrated, and many assume that it is not possible. Calculations based on fluid dynamics and strength of materials say that it should be, but doubts will remain until the concept is physically demonstrated. However, Stratosloar is only one possible approach for taming intermittency. There are many technically feasible ways to do so without resort to dispatched generation from fossil fuels. It’s a matter of finding one or more that can be economically viable against stiff competition from untaxed fossil fuels.
Whatever technology or combination of technologies ultimately get us rolling toward net zero carbon emissions, it’s clear that change is needed. The course that we’re on will not get us where we need to be quickly enough to prevent sea level rise — as one example — from submerging southern Florida.
The current regime of RE subsidies has done what it was intended to do. It has brought the LCOE for as-available RE down to levels that in favorable locations would be quite competitive with electricity from fossil fuels. “Would be”, that is, if LCOE for as-available energy, rather than on-demand power, were the basis for competition. But it isn’t.
The market appears to have limited appetite for as-available energy, even at bargain prices. What’s needed now to advance RE are systems and infrastructure that increase its utility. The approaches, as already noted, are cheap long-distance transmission, cheap energy storage that scales to terawatt-hours, and commercial applications heavily dominated by the cost of energy and able to operate intermittently. Those are now the areas in need of policy support.