It's not coal vs wind - it's gas vs batteries

First – with apologies – a rant!

As investors it is concerning that the energy debate continues to create fractures, with lots of misinformation (dust in the air) and diversion reinforcing policy uncertainty.  Australians would expect governments at all levels to provide a policy framework that promotes an affordable, sustainable and secure electricity supply.  Irrespective of assets being publicly or privately owned – the whole supply chain and its end users suffer from the uncertainty. 

This situation risks another investment strike.  An investment strike, particularly combined with the significant portion of Australia’s coal fired generation that is coming up to the end of its useful life, results in shrinking operable capacity.  Shrinking capacity partnered with higher peak demand (both due to weather as well as one-off increases in demand, such as from the commencement of LNG operations in Queensland) is a recipe for higher and more volatile prices as well as potential supply shortfalls.

Anyway, that’s enough, my colleagues are keen to pull me off the soapbox and get me onto the real material …...

Time to rethink baseload?

One of the topics of discussion over the last week has been about the merits of coal fired vs wind generation.  Australia has plentiful supplies of coal.  Coal power is an established technology.  It is ‘dispatchable’  in that generation can be ramped up and down in response to demand (within limits).  New super-critical coal fired power stations have materially lower CO2 emissions per MWH than the older technologies in use in Australia (particularly compared to brown coal).

Wind power is relatively new, but does have well over a decade of operational experience in Australia.  Capital costs and operating performance have improved significantly – such that wind power now has a slower levelised cost of production than coal or gas fired generation.  For example, BNEF estimates the levelised cost of a new supercritical coal fired station at A$134-203/MWH, significantly higher than wind at A$61-118/MWH.

Wind’s main weakness is that it is not dispatchable.  Production cannot respond to demand and is often quite low during heat wave type conditions.  Peak production for most windfarms occurs at night (often late at night) which doesn’t necessarily align well with the pattern of electricity demand.  It is these characteristics that give South Australia the dubious distinction of simultaneously being the state with the greatest number of high price peak events, as well as the state where power prices are most frequently negative!

One of the terms that is bandied around, and often misused, is ‘base load’.  Base load generation is generation that is operating at near full capacity all the time.  It might ramp up or down gradually in response to changes in demand, but it is running flat out pretty much all the time.  The figure below illustrates this for a typical daily demand profile.  Good baseload generation is cheap, and in particular, has low marginal operating costs.  This is important because baseload generation is the biggest share of total generation and, hence, you want it also to be the cheapest.

This is why brown coal generators (CO2 emissions to one side) make the best base load generators.  Their fuel is extremely cheap and they can pump out power day and night at low cost.

Gas doesn’t make great base load generation.  While Combined Cycle Gas Turbine (CCGT) technology has boosted efficiency from perhaps 30-40% to 50-60% - gas is expensive.  A gas price of $12-14 per GJ implies a pure fuel cost for a CCGT plant of $80-$100/MWH.  High fuel costs, combined with an inability to ramp up and down quickly (CCGT are more like coal plants in this regard), are the reasons that Pelican Point, despite being both one of the newest and most efficient gas fired power stations in SA, is half mothballed, while older and less efficient gas fired plants continue to operate.

Where thinking about base load power needs to change, is around optimising the generation mix aligned with policy objectives.  For example,  a diversified portfolio of wind and solar can form cheap base load power.  

Wind and solar are natural complements.   Wind tends to have a dip in production during the day, but produces a lot at night (see below for an example of the daily production from two wind farms).   Solar delivers during the day (volatility is actually quite low).  Solar produces more in summer and correlates well with air-conditioning demand.  Wind produces more in winter. 

If Australia’s wind farms where more evenly distributed by State (rather than clustered in SA/Vic), then overall production would be much less sensitive to weather patterns in a particular area. 

Furthermore, the key attribute that makes a combined wind and solar portfolio attractive for base load power requirements is that they are cheap.  They are cheaper than new build coal.  They are cheaper than gas fired generation. 

And on top of this, using solar/wind for base load electricity is going to put us on track to meet our Paris commitments. 

Source:  Joep Vaesessn, Co-locating wind and solar.

Sounds straightforward but what’s the problem? 

The problem is how do we meet peaks in demand or situations where NEM-wide wind or solar production is lower than average.  To this we need some form of peaking capacity.  In particular, we need plant that is able to deliver large amounts of power on a rapid response basis … preferably within seconds/minutes and definitely not within the hours it takes a coal or CCGT plant to ramp up.  We need plant that can operate efficiently in scenarios where they go from zero to maximum output multiple times per day (something coal/CCGT plants are physically not designed to do).

So what are our options?

Peaking Economics

The choices of technologies that can perform these roles are:

  • Open cycle gas turbines.  This is the technology that underpins gas peaking plants.  It is an established and proven technology.  Capital costs are quite low (US$0.8-1 million per MW).   Plants can be small or large (providing flexibility to distribute production across the grid).  Their main negative is fuel costs – at a gas price of $8-12/GJ – variable operating costs are $100-$140/MWH.  Returns (and hence levelised costs) of gas peakers are extremely sensitive to utilisation rates as the return on capital needs to be recovered over a small base.   For example, if a plant has a 10% capacity factor (equivalent to operating at full capacity 3.5 hours even weekday), then levelised costs are $250-300/MWH.  By contrast, at a 20% capacity factor, capital costs can be recovered over a much larger base and levelised costs are perhaps $150-170/MWH.  It is also important to recognise that gas fired plants face significant policy uncertainty – be it a carbon tax or a emissions intensity scheme.  Any of these would boost the effective levelised cost.
  • Batteries.  While batteries, according to Wikipedia, might have first appeared in Iraq (250BC – 224AD) – batteries with the cost, efficiency and durability to act as a viable store of electricity are a very new phenomenon.  For batteries to out compete gas peakers requires two things - lower capital cost and cheap power to charge the battery.   For example, a Tesla Power Pack has the capacity to work as a 4 hour peaking plant.   At today’s price of circa $600/KWH this equates to a capital cost per MW of peak output of A$2.5 million – roughly double the cost of a gas peaker.  Assuming charging power costs $50/MWH and providing for a return on capital gives a levelised cost of the order of $400/MWH.  However, batteries are relatively new and it is reasonable to expect very significant falls in capital costs in short order.   Taking $300/KWH as a near term target – which seems highly achievable – this would reduce the cost of a battery peaker to be roughly in line with the gas fired equivalent (ie A$1.25 million per MW of peak output).  Layer on top of this the potential to charge the battery with cheap surplus power (and, one of the natural by products of increased wind and solar penetration is going to be periods of surplus generation – the negative price outcomes in SA are testament to this) gives the potential near term levelised cost around $170/MWH.  That is, you can be quite confident over the next few years that batteries will overtake gas peakers in cost terms.  Batteries have the further advantage that they can be deployed in small or large scale and, in addition to acting as a peaking power plant, they can reduce transmission or distribution costs which will further boost their competitiveness relative to gas peakers.
  • Pumped hydro[1].  Pumped hydro is a proven power storage technology (for example, the Tumut 3 station as part of the Snowy scheme).  It has the potential to be quite low cost – particularly when measuring cost per MWH (i.e. hours of storage) rather than peak capacity (as for a given generator only the size of the dam needs to be scaled to add more storage).  However, despite having a long history, there don’t seem to be a lot of sites naturally suited to pump hydro (Genex in Queensland as a possible exception to this).   The cost of the Snowy Scheme - $9 billion in today’s dollars – wouldn’t necessarily give confidence about the capacity of pumped hydro to offer low cost peaking capacity.

For the power wonks – email me for a copy of the spreadsheet that outlines the levelised cost calculations quoted above.


In summary:      

  • The energy debate needs a reconsideration of what baseload power is and what are the best sources of cheap baseload power.
  • A diversified and balanced portfolio of solar and wind has the potential to deliver the lowest cost base load power for Australia.  However, to foster this, policy frameworks (particularly the RET) needs to change to incentivise developers to create uncorrelated sources of power.
  • For those that think gas in the answer – have a hard look at the viability of a gas peaker vs a battery as that technology develops.  The analogy I would have is - are you shopping for an electric typewriter in 1980?  If so, investing heavily in gas might leave you short-term happy but long-term disappointed.














[1] I should apologise for excluding other potential technologies such as solar thermal, compressed air storage or heat based energy storage such as hot sand or liquid silicon technologies.   At this stage – these seem too early or too expensive or both to be viable energy storage approaches in the near future – but that might reflect my ignorance rather than real potential of these approaches.   I have also excluded nuclear power as a potential source of base load.

[2] Yes my first computer was a Mac 128.