The Third Story - Renewable Base Supply

A commonly heard criticism is that "renewables can't do baseload" due to the intermittency in the wind and solar resources. Based on the data and analysis work OzEA has done over the last year, we show this criticism to be largely missing the point. While the cost and timeframe of a transition to 50% renewables presents many issues for exploration and reasoned argument, we here make the prima facie case that wind and solar can effectively displace existing fossil fuel energy supply.

This Third OzEA Story is a different sort of document to the papers, reports or blogs that readers may anticipate. The Open Science approach to public-interest science that we road-test has a special place for the "stories" that describe where thinking has developed to and set the scene for upcoming work. While seeking to minimise the technical load, we appreciate that for many this remains a somewhat technical read. Please note your questions and comments as you proceed, and if they are not resolved by the end (and many will not be), please consider joining the Open Discussion. Please note that here we are looking at a 50% scenario, perhaps by around 2040, and that OzEA will later this year specify the intermediate 20, 30 and 40% scenarios that together provide the basis of an evolutionary pathway.

Before handling the complexity of an electricity system with many thousands of kilometres of transmission lines, it is necessary to abstract and simplify to a solid core, upon which important layers of detail can be added back. In this Third Story we tell of this solid core, and of how renewable electricity can in time become the dominant source of supply in the Australian electricity system.

Introduction

Stripping away the spatial complexity of the transmission and distribution networks, and stripping away the workings of the electricity market in dispatching individual generators, what remains is the core requirement that supply and demand be arranged to match at all times. In the system we have now, most of the control exists on the supply side, and especially with gas turbines that can be ramped up and down as need be.

Onto this currently functioning system we seek to impose 50% renewables with (tentatively) 35% Wind and 15% Solar. These renewable electricity supply sources are intermittent, although Concentrating Solar Thermal (CST) can be built with thermal storage that provides some control (allowing, importantly, the solar energy captured during the day to be dispatched as electricity into the evening).

Figure One: Siting of simulated renewable plant. [click through for full pdf]   Of the 11 solar sites for which we have data, six (Rockhampton, Wagga Wagga, Mildura, Alice Springs, Tennant Creek and Kalgoorlie) are used in what follows. In reality the NT sites are proxies for central QLD. There are 30 wind sites "from Broome to Cooktown", although here the sites north of Geraldton are not considered (and we also examine a no-WA case). In what follows we consider 2003 demand data for the National Electricity Market (NEM), which at that time included only QLD, NSW, VIC and SA.

Here we focus on how, with 50% Wind and Solar, the core requirement that supply meet demand can be upheld, and we do this simply in terms of the supply side configuration. That is, we ignore the demand management aspects. When it comes time to include some control on the demand side, this will further improve the situation for renewables.

A somewhat historical and semantic point will aid understanding. Historically, the sources of supply are matched conceptually to distinct load phases; that is, coal power running all the time accounts against a 'base-load', efficient but somewhat costly Combined Cycle Gas Turbines (CCGT) pairs with an 'intermediate-load', and cheaper but less efficient Open Cycle Gas Turbines (OCGT) operate as needed to provide further supply into the mix at times of high demand. Of course there is some Hydro and other aspects, but this conceptual pairing of three supply technologies with the three levels of load is key. The linguistic difficulty arises because 'supply' technologies have become conflated in language with 'loads'. For example, coal power is not a load, and thus not a base-load; rather it is a source of supply adding into the total supply. Simply by being careful with the language, correctly referring to load as load and supply as supply, can greatly aid clear thinking.

This story has words and pictures - each depending on the other. With the figures shown you click through to the pdf, and then use your pdf viewer to zoom in and pan around. The various 'ribbon plots' in particular provide a graphical view of the data and models, and you should spend a little time examining these, expanded to full screen and panning along the time base in each. Thus the story will unfold.

Note also the key terms 'demand remainder' and 'demand remainder profile'.

Imposing 50% wind and solar

For much of the last year OzEA has been working on the data and methods for simulating large-scale and widely distributed wind farms and CST plant. Recently this became sufficiently developed to allow a start on explicit 50% renewable scenarios, using wind speed and solar irradiance data at the sites shown in Figure One above. Constrained by the solar data in particular, we work for now with real-world wind speed and insolation data from the year 2003.

Starting with a naive model, based on placing large (very large) CST plant at six selected locations where we have data, and wind farms at thirty selected locations "from Broome to Cooktown", OzEA imposes 50% renewables into the supply mix as shown schematically in Figure Two. Ignoring (for now) the use of buffering mechanisms such as the Pumped Hydro Storage (PHS), the upper panel shows the demand being met solely by coal and gas powered generators, with the lower panel (naively) including 50% renewable supply, and showing the consequent changes in the fuelled supply requirements.

Figure Two: Supply meeting demand. [click through for the pdf]   A. Demand in time showing the 'base', 'intermediate' and 'peaking' phases for no-renewables; B. The Demand Profile for Panel A (over the full year); C. Supply meeting demand for the naive 50% renewables scenario; D. The Demand Remainder Profile for Panel C (over the full year).    The full-year versions of A. and C. are given in two parts: first half, and second half. Here the 'base' (dark grey) has been defined as the 97-100% of the time region (on the demand / remainder profile), and the 'peak' (red) as up to 20%. These cut-offs are simply selected for illustrative purposes.

While it may initially be surprising that the effect of the renewables is to displace the coal base-supply (Panel D), this in fact makes a lot of sense. One inflexible source of supply is being displaced by another, while the supply components that provide the control are maintained (and indeed expanded in this naive first attempt). It can always be possible to include a high penetration of renewables if there is also enough supply available from gas turbines to "fill the gaps" - and this is the real essence of the "renewables can't do base-load" criticism. To build a useful 50% renewable electricity scenario OzEA is working to understand how these gas requirements can be kept in check.

Comment on the naive 50% renewable scenario

It is striking that the neat appearance of the current matching between the demand and the fuelled sources of supply (Figure 2, top panels) changes into a somewhat messy or jagged mix when the renewables are included (lower panels). This is not as surprising or worrying as it may first seem. The current system has been built and refined over some fifty years, and thus contains much internal order. Conversely, this blunt imposition of 50% renewable supply has not been accompanied by changes to the way demand is structured and supply buffered. As the 50% scenario is refined, both here and in the months ahead, expect the fuelled supply trace to become smoother.

Panel D requires particular attention, observing especially that peak demand at ~28.5 GW has reduced in the Demand Remainder to only ~23 GW. That is, after the expense of 50% renewable infrastructure, the amount of fuelled supply infrastructure needed has been reduced by only ~20% (the so-called "capacity credit"). Such a low capacity credit is most unhelpful in seeking to make an economic case for renewables; however, as we will see shortly, it is not difficult to develop the scenario to achieve a much improved capacity credit.

Examining the full temporal view (i.e. full versions of Panels A and C as linked in the Figure legend), the night-time lows in the demand are close to identical night after night in traditional mode, thus enlarging the 'base' over the 'intermediate' phase. Mechanisms such as off-peak hot water, and the overnight 'charging' of PHS facilities, act to produce this neat situation. In contrast, the 50% renewable supply often meets much of the demand, thus enlarging the 'intermediate' supply phase over the 'base'. Of course, with 50% renewables, old and new demand shaping mechanisms will be focused on buffering the demand and supply dynamics into harmony.

Note also that while raw demand peaks in the afternoon or early evening, the demand remainder often peaks in the evening or early in the day (before the days solar energy catch provides supply). While these peaks will be managed with the storage mechanisms introduced next, it remains noteworthy that a systematic change has occurred in the timing of the peak requirement for fuelled supply.

An improved scenario

Part I. Applied storage mechanisms for buffering supply

To improve the renewable supply dynamics we first include two supply-side storage mechanisms:

Centralised Pumped Hydro Storage: The existing electricity system depends on ~20 GWh capacity of Pumped Hydro Storage (PHS), and here we tentatively increase this to 30 GWh for the first-pass 50% renewables scenario. Peak output is 4 GW (variable between 0-4), with charge-up occurring at (and only at) 1.5 GW for at least three hours. These constraints roughly enforce engineering realities. See this comment for more detail. See the second ribbon of Figure Three for an example of the peak shaving produced by this PHS model.

Thermal storage within the CST farms: observe that the CST supply often counters the peak demand during the day, but leaves a demand peak in the early evening. As thermal storage is part of the promise of CST, we include a simple and modest thermal storage model into the simulated CST plants. We suppose the CST plants have 6 hr of thermal storage, and can use this to regulate output over the course of the day -- until 6 hours after the last solar input [see this comment, and linked worksheet, for details]. This buffering of the solar energy, and the consequent smoothing of the Demand Remainder, is shown in the third ribbon of Figure Three.

Figure Three: Storage Mechanisms: [click through for full pdf]   Top Ribbon: the naive scenario (as in Fig 2.) showing demand in time, the CST and wind supply traces, the demand remainder, and a nominal breakdown of the fuelled supply requirement into base, intermediate and peaking components; Second Ribbon: application of the centralised PHS storage model (described above); Third Ribbon: application of the CST storage model (described above); Bottom Ribbon: application of both storage models concurrently.    Note that the time base has been restricted to the April-September half of the year (which is the interesting part) in order to present a single ribbon rather than two halves.

The supply-buffering provided by the CST thermal storage (Fig. 3; Ribbon 3) acts to substantially smooth the demand remainder, and thus contribute to our aims. The PHS model is seen to work well at peak shaving when applied alone (Fig. 3; Ribbon 2), but not so well in combination with the CST storage (Ribbon 4) as the OzEA heuristic for implementing the PHS struggles with a less up-and-down trace to work on.

Part II. Annealing site capacities to tune the renewable supply

Some influence can be had on the renewable supply, and thus the demand remainder dynamics, through the choice of renewable plant location and size. As in Figure One, we are limited by the data as to the locations considered, but within these can shift capacity around in order to search for better dynamics. OzEA has implemented a simulated annealing approach to do this (detailed here and here). Note that no particular optimality is claimed; the procedure simply gives answers that improve in the direction required.

Seeking to reduce both the peak value of the demand-remainder and its variance, Figure Four shows an example (lower ribbon) of a scenario produced by an annealing run - that has the demand-remainder peaking at 15.5 GW (compared to 17.1 GW for the naive case). Note that the annealing runs do not 'converge' on a unique solution, but rather produce one or another interrelated patterns of site capacities. As chance relationships between peak demand times and output from the simulated renewable plant can drive some selections, it will be interesting to see which sites are consistently scaled up / down when we proceed with analysis of the 2004 and 2005 data.

Figure Four: An example of a refined 50% renewables scenario. [click through for full pdf]   Top Ribbon: reference, being a repeat of final ribbon from Fig. 3; Bottom Ribbon: renewable supply dynamics for a 50% renewable scenario where the capacities of the renewable plant have been chosen to produce improved dynamics in having the renewable supply match with the historical demand.

It is of interest to consider the exclusion of the WA sites, and we find the demand remainder peak is increased by only a couple hundred MW. Of course this 'result' is based on a high-level analysis incorporating only the current OzEA selection of simulated plant, for 2003 data, and using the storage mechanisms described. It may yet be determined that a transmission link to WA is desirable; however, for now we note that WA inclusion does not appear critical to the prospects of integrating 50% renewables into the NEM.

Winding back the abstractions and simplifications

The overall and simplified view of renewable and fuelled supply combining to meet demand, with the aid of some storage, is the core concept. It allows us to understand the implications of the intermittency of the wind and solar resources. In time the focus needs to become one of cost and timelines; for now the focus remains on the intermittency and developing the current analysis.

The simplification of looking just at 2003 needs to be unwound, and OzEA has 2004 and 2005 data in hand, albeit with the continuing issue of sparse solar data. The plan is to step forward a year at a time, taking stock each time of the new aspects and issues that arise.

The abstraction of ignoring the transmission will be unwound also, at least in part, by considering each NEM state as here they are considered in combination. At the same time, transmission dynamics between states will be added in. That is, transmission will be considered, but only in terms of interstate flows. This approach can later be broadened by breaking the system into smaller regions again.

It remains premature to attempt a costing of a 50% renewable scenario; however, OzEA can usefully and sensibly include the costs of the fuelled generators and transmission infrastructure. That is, we hold the costs of the renewable plant aside, while using the costs associated with the remaining infrastructure to inform scenario development. In particular, rather than shaping the demand remainder by seeking to lower its maximum value, we will use the cost of the infrastructure (and fuel) needed in addition to the renewable generators as the objective.

Wrap up

The core of the story is represented in Figure 5 to the right. The key is to note that the requirements for intermediate and peaking supply (taken as Combined and Open Cycle Gas Turbines respectively) as seen in Panel A, are not all that different from those in Panels D and E. In this 'evolution' from A to D, much of the coal base supply has been replaced by renewable supply.

Figure Five: Demand Remainder profiles summarising the development through Figures 2, 3, & 4 above. A. No renewables. A traditional demand profile, broken into base, intermediate and peak components - from Fig. 2; B. Naive 50% renewables inclusion (no storage mechanisms) - from Fig. 2; C. Inclusion of PHS model (Fig 3, Ribbon 3); D. Additional use of 6 hr CST storage (Fig 3, Ribbon 4); E. Demand Remainder Profile for refined 50% renewables model (Fig 4. lower Ribbon).

Including a high level of renewables can always be achieved if enough gas turbines (and gas) are used to fill in the gaps. What is striking here is that even at this rudimentary stage in scenario development we have the intermediate supply requirement (taken as CCGT, and according to a rough cut of the demand remainder profile) within a few percent of the current requirment. Of course this is a high-level view with many caveats.

Further improvement can be anticipated. While some gains were made above in seeking to drive the peaks of the demand remainder down through site selection, and noting the CST thermal storage as the biggest contributor so far, the obvious way to improve the current situation (Figure 5, Panels D & E) is to shape the demand remainder profile from the base side. This is precisely what has occurred to date in order to achieve the profile seen in Panel A. Finding ways to shift use towards the times when renewable electricity is plentiful will bring down the intermediate supply requirement under renewables scenarios at the 50% and other levels.

The work presented here, whereby 50% renewable supply (in terms of total energy delivered over a year) is placed within the existing system, and tuned in order to reduce the need for flexible fuelled generators, has focused on the supply dynamics, and has not included the nations straight hydro resources. Evolution of demand-side dynamics, and inclusion of hydro, can be expected to provide significant synergies in developing more detailed 50% renewable scenarios. It is within this demand-buffering domain especially that the ecological niche metaphors of The Second Story apply.

The logic of focusing on 50% renewable electricity has two major strands. First, it allows for working backwards into 40% and 30% scenarios (and even 20% by 2020), thus developing evolutionary pathways that can inform decision making now. Second, with 50% renewables and 50% fuelled supply, the intermittency issues with wind and solar can be addressed in a confined way. Specifically, we require sufficient storage and demand-side-elasticity so that in worst-case renewable supply dynamics, the fuelled supply can maintain the system. As a mental picture, a period of a week where renewable supply is limited and patchy may require the full complement of intermediate CCGT and peaking OCGT supply be run full-time, partially meeting demand at times while providing charging for storage at others. While a costly and inefficient way to meet demand on an ongoing basis, such occasional periods are balanced against those when peaking plant is hardly used.

Finally, the focus here has been mostly on the base and intermediate supply requirements. Peaking supply requirements are made especially difficult to consider at this point by their rapid evolution, which will be explored as OzEA steps first through the years 2004 and 2005, and on from there as the necessary data is sourced.

Discussion lead in

In what is presented here, longish as it may be, many issues have been brushed over. What follows is the Open Discussion where you can question and develop both what has been presented, and what has not. Especially appreciated will be comments addressing these issues:

A high-renewable-penetration electricity system needs the capacity to use electricity when the renewable supply is plentiful. Of course, the NEM system will price electricity cheaply when supply is plentiful, and thus engender this use. It will be helpful for OzEA to develop more concrete expectations of this demand elastically, both in terms of the day-to-day cycles, and the more spasmodic periods of plentiful supply that come with a high-penetration of wind and solar.

A fundamental aspect of the 50% renewables approach is the use of Gas Turbines to provide supply to meet much of the demand remainder. The efficiency of gas-powered-supply depends heavily on how rapidly and how often the turbines need to ramp-up and ramp-down. In turn, this depends on both (i) how the renewable supply varies in time, and (ii) the mechanisms that buffer supply (via storage) and demand (via demand management). It is thus important to note that the demand remainder profiles shown in Figure 5 tell only part of the story. In moving forward OzEA needs to deal explicitly and intelligently with these buffering aspects, especially in relation to the efficient use of gas resources.

Hydro. Distinct from the PHS infrastructure, Australia has some large established Hydro resources that 'fill' (i.e. charge) from run-of-river (i.e. rain). To date OzEA has not put efforts into understanding this resource, and thus not incorporated this Hydro into current analysis. The difficulty is that in periods of drought the resource can be run down. It is now becoming more pressing to understand this resource, and to determine parameters for "reliable hydro".

Finally, OzEA has used (and will continue to use) a data-driven bottom-up approach. In time a meteorological perspective can coalesces with the data processing, and bring understanding to (i) compensatory weather patterns within the catchment, and (ii) causal connections between demand and the renewable resources. With this knowledge it becomes easier to address the question of "what is the worst case" for distributed renewable supply in the Australian context.

The OzEA stories are the work of Dr Francis Clark, with some input from Prof. Barry Brook. While Barry takes care to catch and comment on the rough edges, the bottom line is that responsibility for this material sits principally with Francis.

[Jump to bottom]

[Show Full Lists]

 

  TOP   |   REFRESH PAGE

fc - May 2011