lable at ScienceDirect

Renewable Energy 68 (2014) 103e110

Contents lists avai

Renewable Energy

journal homepage: www.elsevier .com/locate/renene

The reliability of distributed solar in critical peak demand: A capitalvalue assessment

Kerry B. Burke a,b,*

aCommodities Carbon and Energy, Westpac Institutional Bank, Sydney, Australiab Physics Department, University of Newcastle, Callaghan, Australia

a r t i c l e i n f o

Article history:Received 25 May 2013Accepted 23 January 2014Available online 26 February 2014

Keywords:Photovoltaic generationElectricity marketPeak demand

* Corresponding author. Physics Department, UniveAustralia.

E-mail addresses: KerryBurke@westpac.com.au, Ke

http://dx.doi.org/10.1016/j.renene.2014.01.0420960-1481/� 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Generation is most valuable when demand is highest. As electricity can’t yet be cheaply stored, generationand transmission infrastructuremust be built tomeet the highest expected demand, plus amargin of error.Reliably producing power at times of critical demand not only offsets the need to use expensive liquid fuelssuch as diesel or condensate, but also removes the need to build backup power stations and transmissioninfrastructure that would only be used for a small fraction of the year. Under the most extreme demandconditions, solar has reduced the peak demand seen by retailers andwholesale energymarkets. This studycompares the capital cost of critical peak availability from gas turbines to the capital cost of critical peakavailability from distributed solar in the Australian National Electricity Market (NEM).When compared onthis basis, 10e22% of the cost of installing the solar system can be attributed to the capital value of criticalpeak generation. Northewest andwest facing PV isworth a further 3e6% of system installation costs whencompared to generally north facing PV. Finally, southern states, with longer summer days and more sun-shine in the afternoon are found to benefit more from peak supply of solar PV.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

The most recent work conducted in Australia on the value of PVgeneration has been performed so that regulatory bodies candetermine a fair price for household PV exports. The first regulatorywork on the value of PV exports is that prepared by Frontier Eco-nomics for the Independent Pricing and Regulatory Tribunal(IPART) of New South Wales (NSW) regulatory determination [1].Frontier’s terms of reference were to determine a “fair andreasonable value for the electricity generated by small-scale solarPV units and exported to the grid”. Frontier accomplished this byexamining the wholesale market value of the electricity exported,choosing FY10/11 as the historical year to evaluate the market valueof PV exports. This choice was made over FY9/10 because FY10/11was closer to a median year in terms of midday spot electricityprices. Solar PV exports valued in the FY9/10 year were worth $135/MWh, whereas solar PV exports valued in the FY10/11 year wereonly valued at $65/MWh. While it was not a conclusion of the

rsity of Newcastle, Callaghan,

rry.b.burke@gmail.com.

Frontier Economics study, the variability in these results highlightssolar PV’s value as a hedge against high prices.

Acil Tasmanproduced a similar analysis for the Essential ServicesCommission of South Australia [2]. Acil Tasman’s study modelledspot electricity prices for the year ahead rather than using historicaldata. The Queensland Competition Authority (QCA) chose not toregulate the price paid for solar PV exports for most of the state, onthe basis that a competitive market for retail electricity supplyexisted in Queensland (QLD) [3]. TheQCAdid, however, recommenda fair price for PV exports that was based on an average wholesaleenergy cost, similar to the methodology used in the other states.

The regulatory studies on the value of PV exports are oftenconstrained by their terms of reference. While the studies accu-rately calculate the costs avoided by a retailer accepting customers’PV exports, they do not examine the system level benefits of solarPV’s ability to supply critical peak demand.

Elliston et al. modelled both the technical feasibility [4] and theleast-cost optimisation [5] of a 100% renewable energy grid as areplacement for the National Electricity Market (NEM). This simu-lation is a valuable contribution to the Australian Energy policyliterature because it does take into account the reliability andtiming of generation sources. However, a simulation of this scaledoes make some simplifying assumptions necessary, and one

Fig. 1. Ausgrid network peak demand; actual and weather corrected. Reproduced fromSmith [12].

K.B. Burke / Renewable Energy 68 (2014) 103e110104

weakness of the study is that it only includes north facing solar as ageneration source, neglecting to evaluate the time-of-generationbenefits from west-facing solar.

Molyneaux et al. [6] also modelled high renewable energy sce-narios for the NEM and mentioned the ability of solar PV to reducesummer peak demand. Solar generation output was modelled fromhistorical data on solar intensity in each capital city. No details weregiven as to how solar correlation with demand was established.

Passey et al. [7] examined the value of PV in West Australianelectricity networks. They specifically identify that the benefits ofPV are dependent on correlation between output and load. Simu-lated, rather than measured PV data was used for this study. Thisstudy did explicitly model west-facing PV systems and found thatthe west-facing systems had a higher present market value whenthe value of offset peaking generation was included.

Liu et al. [8] studied the optimal system size and elevation (butnot azimuth) of solar panels in various population centres ofQueensland using simulated solar production data.

Maine and Chapman [9] evaluate simulated PV outputs againstseveral individual days of historical prices in the SA wholesalemarket during 2004. Their simulations show that an ideal PV sys-tem under clear skies would generate power in the afternoon onsummer days when prices are high. They only evaluate the energycomponent of the solar PV without regard for option value. In theirintroduction they state that “.retailers can and do hedge thewholesale costs, giving them some insurance against the possibilityof unexpectedly high pool prices but that added complication doesnot alter this discussion fundamentally” (emphasis added).Crucially, retailers enter into hedging contracts because it lowerstheir risk of exposure to volatile wholesale prices. A reliable sourceof energy provides the same insurance value as a financial insur-ance product and should form part of a valuation methodology.

Watt [10], using NEM data from 2004, also found the west-facing PV arrays provide a better match to residential load pro-files than north facing systems. A major benefit of this study is thatit uses data from real PV systems. Only 15 systems were availablefor data collection at the time. Access was obtained to substationlevel load data and so a focus of the paper was on the value oftransmission cost avoidance. The paper found that PV outputmatched load better for commercial substations, rather than resi-dential area substations and hence was of more network value in acommercial area. A graphical comparison to NEM spot prices wasmade, but there was no calculation of the commercial value.

This paper aims to assess the value of distributed solar PV’savailability in times of critical peak demand, rather than calculatingan average value of energy. To compare availability, systems shouldbe assessed on the amount of power they can be depended on toproduce under peak demand conditions. Results in this paperextend on previous work by considering each of the four mainlandstates in the NEM, considering the impact of directional facing onsystem peak demand availability and using physical rather thansimulated results of distributed solar generation.

1.1. Critical peak demand

The top ten maximum demand days in QLD, Victoria (VIC) andSouth Australia (SA) all occured during summer. In the case of NewSouth Wales (NSW) the three highest demand days occured inFebruary 2011, but there are also three winter days in the top tenfrom July 2008 and 2007. Since 2008, there have been significantchanges to consumers behaviour, which reduce the size of expectedwinter peaks relative to expected summer peaks. Importantly, gasheating and reverse cycle air conditioners have become more pop-ular [11]. Barheatersuse 2e3 timesmoreenergy than a reverse cycleair conditioners used in heatingmode. The increased penetration of

reverse cycle air conditioners has had the effect of increasing sum-mer cooling peaks while decreasing winter heating peaks.

Ausgrid, the largest distributor in NSW has examined the factorsdriving peak demand in their own network [12]. Weather (tem-perature) drives much of the variability year to year. After adjustingdemand for weather variation their data confirms that underlyingwinter peak demand has been declining since 2006. Their data forsummer demand shows an increase since 2008 and a flatteningover the last few years. While evening demand has historicallybeen high inwinter in NSW, the future development of the networkand peaking generation capacity will focus on peak summerdemand.

Future peaking generation capacity investment in each statewillbe driven by the requirement to supply summer peak loads.

2. Data methodology

Monthly solar installation data is available from the Clean En-ergy Regulator [13] with geographic resolution at the postcodelevel. This data is available because system installers must registerthe details of a system in order to receive the tradeable small scaletechnology certificates (STCs) those solar systems are eligible for.STCs are currently worth approximately a third of the total systemcost, so the incidence of installers failing to register systems can beexpected to be very low.

STCs may be created and registered with the Clean EnergyRegulator up to 12 months after the installation of a system, so thedata will underestimate the total capacity of residential systemsinstalled in the most recent months. However, installers aremotivated by cash flow concerns to register STCs promptly. Fig. 2shows that the lag in installation reporting lasts only a fewmonths, and that installations not counted due to registration lagare a small component of total installed systems. No attempt hasbeen made to estimate the installed, but not yet reported systems(Fig. 1).

Fig. 3 shows the state breakdown of solar installations. In-stallations booms can be seen leading up to the end of eachfinancial year (June) as households rush to get installations ahead ofthe annual reductions in federal incentives byway of the number ofSTCs issued for each system.

The boom in residential solar installations has been accompa-nied by the creation of an online community of solar enthusiastswho share real time data on solar system generation. This data isavailable at http://PVoutput.org/live.jsp. Five and 10 min resolutiongeneration output data is available for several hundred individualsystems. The data includes the postcode that each system isinstalled in, allowing aggregation by state. Also included are the

Fig. 2. Solar PV installations. [13].

K.B. Burke / Renewable Energy 68 (2014) 103e110 105

elevation and the facing of each system to the nearest semi-cardinaldirection, allowing aggregation by facing. This data does notcontain all solar installations; however, at approximately 800 in-stallations in summer 12/13, it is a large sample.

From the online data, power output as a percentage of ratedcapacity can be calculated for each system for every half hour overthe last two summers. Multiple steps of data quality assurance areapplied. Any system that has an obviously inappropriate systemsize relative to its maximum reported output is excluded from thesample. Any system only reporting for part of a day is excluded forthat day. The percentage power output for each system is aggre-gated on either a state level, or by state and facing, to determine anaverage output efficiency of solar systems installed in eachgrouping for each half hour over the last two years.

By combining the state level output ratings with the totalinstalled capacity as published by the Australian Clean EnergyRegulator (CER), a strong estimate of the solar power generation ineach state can be obtained. This analysis is only made possible bythe growth of enthusiast data sharing. While some systems arereporting live data for the 2010/2011 summer, the number is toosmall to be considered a statistical sample of thewhole state. This isnot available publicly from any other source, as distributed solargeneration is behind-the-meter and appears to the system as ademand reduction if it is offsetting local use, rather than beingexported.

Distribution network businesses have a larger data set of solarsystem generation. This would be of considerable public policyinterest and could be published in an anonymous or aggregatedform.

Fig. 3. Cumulative distributed solar installations [13].

Demand data for each state is obtained from the AustralianEnergy Market Operator (AEMO). The specific data series used isDemand and Non-scheduled Generation. AEMO operates the Na-tional Electricity Market Dispatch Engine (NEMDE), which controlsall generators to meet demand. Some smaller generators are clas-sified as non-scheduled as they are unable to receive instructions tovary their output. For the purposes of operating NEMDE, AEMOcounts these generators as negative demand. The data series De-mand and Non-scheduled generation represent total metered de-mand, as it includes demand supplied by the non-scheduledgeneration. For the sake of brevity the AEMO data series “Demandand Non-scheduled Generation” will be referred to as metered de-mand in this paper. Metered demand does not include generationfrom distributed solar PV as it is behind the meter.

The PV energy production data obtained from PVoutput.org andthe CER has been added back to AEMOmetered demand in order toshow the solar PV effect on peak demand for each of the highestdemand days of the last two summers in each NEM region. The sumof metered demand and the demand supplied by solar PV will bereferred to as native demand.

3. Gas turbines

3.1. Nameplate construction cost

The Australian Energy Technology Assessment 2012 [14] as-sesses the cost to build an open cycle gas turbines at $723/kW, notincluding the cost of interest during construction, finance costs ortransmission costs. Evaluating these additional costs is importantto assess for the purposes of comparison, as they are all costs thatdistributed solar PV is able to avoid.

The most recent gas turbine power station to be built wasMortlake Power station, with a nameplate capacity of 550 MW. Itcost $810 million to build [15], including the cost of an 83 kmburied gas pipeline. The 450 MW Braemar 2 power station wascompleted under budget in 2009 for $530 million [16]. Thisincluded the cost of the necessary gas pipeline. The 640 MWUranquinty power station was completed in 2009; Origin Energypurchased Uranquinty from ERM power for $700million [17].

Colongra power station was completed in December 2009 andillustrates an important point about gas turbines. When operatingat maximum output, gas turbines typically use more gas thantransmission pipelines can sustainably supply. The pipeline leadingto a gas turbine also acts as a storage bottle. Under peak turbineoperating conditions, the gas pressure will drop in the pipeline andthe pipeline will be repressurised later under low gas demandconditions. Because Colongra was located so close to existing sup-ply pipelines, a large diameter pipe was needed to provide enoughstorage to operate Colongra at full power. Even with the largerdiameter pipeline, Colongra can only operate for 5 h at peak powerbefore depleting the pipeline pressure, at which point it must burndiesel as fuel [18]. Colongra was built for $500million [18] and thepipeline was built, owned, operated and maintained by Jemena fora contract value of $76 million [19], demonstrating that gas infra-structure costs are in important part of the cost of critical peakpower availability.

3.2. Temperature effects of generation

Gas turbines, as with any heat engine, have a maximum effi-ciency that is determined by the difference between the hot sideand the cool side of the turbine unit. Under very hot summerconditions, power station electrical output is often less than thenameplate capacity because of the reduced ability to reject wasteheat to the ambient air. This reduction in power station output

K.B. Burke / Renewable Energy 68 (2014) 103e110106

often occurs simultaneously with high demand as the hot tem-peratures also drive a large air-conditioning load. AEMO assessesthe expected power production capability of each generator underheat wave conditions [20]. This summer availability figure has beenused to compare the value of gas to solar rather than the nameplatecapacity as it more accurately reflects the costs of meeting peakdemand.

3.3. Losses

Power generated at a gas turbine must be transported to con-sumers. Some power is lost along the way and a centralised gen-eration asset must be oversized tomeet both customer load and theenergy losses incurred in transport to the customer. Line losses areclassified as transmission losses when they occur along the highvoltage lines leading into a regional demand centre, or as distri-bution losses when they occur along the lower voltage lines thatlead from the high voltage transformers out to residential andcommercial premises. Transmission losses are priced on a timeaveraged marginal basis and distribution losses are priced on anaverage basis.

AEMO publishes distribution loss factors annually for eachcustomer type in each distribution area [21]. Larger customers,with higher voltage connections, typically have lower averagedistribution losses. For the purpose of this study, the loss factors forcustomers connected to a low voltage line have been used. Theserepresent residential users and small businesses, reflecting the typeof customer that has been the target of various governmentincentive schemes, such as the solar bonus feed-in tariffs, and thefocus of the installation industry to date.

In Victoria, loss factors for low voltage lines have been split intoshort line and long line customers, which have been averaged forthe purposes of this paper. Across the three largest states, higheraverage loss factors apply to the distribution companies coveringregional areas. The typical loss factors vary from 5% to 9%, with anaverage of 7%. These figures are the losses incurred under typicalload scenarios.

On a hot day, when the system is transporting more power, linelosses will rise in proportion to the square of the power trans-mitted. No data is available on the losses under high load condi-tions, so the average values have been used to translate thegeneration capacity of centralised gas generators to the capacityavailable to customers. This will overstate the cost effectiveness ofgas generators in meeting peak demand, but improving this esti-mate is not possible without detailed network power flowmodelling (Table 1).

Table 1Distribution loss factors.

State Distribution co. Average

VIC Jemena 5.4%CitiPower 5.7%PowerCor 8.5%SP Ausnet 8.1%United energy 6.3%Average 6.8%

NSW Endevour 8.0%Essential 9.4%Ausgrid 6.3%ActewAGL 5.1%Average 7.2%

QLD Energex 6.3%Ergon East 7.8%Ergon West 35.7%Av. (Ex Ergon W) 7.3%

SA ETSA 8.0%All state average 7.3%

3.4. Effective peak capacity costs

To calculate an effective peak capacity cost, the final headlinecost of each gas turbine, including its pipeline has been inflated toMarch-2013 $AUD. The effective hot day capacity was reduced byboth the applicable transmission loss factors and the state averagefor distribution loss factors. These adjustments are necessary tomeaningfully compare the cost of availability to PV, which operatesat the load sitewithout transport losses. The average of the effectivecapacity cost of the last four gas turbines built is $1.38/W (Table 2).

4. Solar PV

The Australian Energy Technology Assessment 2012 [14]assessed the installed cost of a fixed axis solar farm at $3380/kW(AC) and a single axis tracking solar farm at $3860/kw (AC). Thisrelies on an assumedmodule cost of $1700/kW. Module prices havesince declined to half that value [22]. The rapidity of cost declines inthe PV industry makes the Australian Energy Technology assess-ment out of date, despite its recent publication date. Costs of utilityscale solar are also likely to be different to distributed solar.

The cost of rooftop solar PV is variable, but can be estimatedfrom a survey of installer quoted prices. Solar Choice, a quotecomparison website, publishes a monthly index of the installationcost of a residential solar systems [23]. Solar Choice quotes priceswith the discount from STCs included. They also quote STC valuesfor each region and for the purpose of this paper; the quoted STCvalue was added back to the index prices to determine a subsidy-free installation price.

Systems were substantially more expensive for the smaller1.5 kW size, as expected when fixed costs, such as labour, arepresent. For the purposes of this paper, the 3 kW system size is usedto value the cost of distributed solar PV installations. PV systems arecommonly quoted at the DC rating of the panels. This neglectslosses in the inverter and this is taken into account later bymeasuring the output of the panels in AC units after the inverterlosses have taken effect (Table 3).

4.1. Distributed solar in peak demand

Like gas turbines, Solar PV does not generate at its rated capacityunder peak demand conditions. Solar PV is commonly faced northand peak demand occurs in the afternoon when the sun is strikingmost PV panels at an angle. Solar panels also operate at lower ef-ficiency under high temperature conditions. To examine the peakreliability of solar PV, production has been evaluated under the fivehighest demand conditions in each NEM region. This empiricalmethodology captures both the angle effect and the temperaturereduced efficiency effect. Using empirical data also captures theeffects of aerosols or clouds that are harder to model. Electricityproduction values on PVOutput.org are reported in AC units, so thismethod also captures the inverter losses.

Shown in Fig. 4 is the native demand and metered demand forthe highest demand day in each state over the last two summers.The time window of 7ame7pm in Fig. 4 has been chosen to bestshow the time periods with solar generation and the demand peak.In summer, demand continues to decline after 7pmwith no seconddemand peak. Under the most extreme demand conditions, solarhas reduced the peak demand seen by retailers and wholesaleenergymarkets. In QLD and SA, the time of peak demandwithin theday has also been pushed back. While PV generation is strongestbefore peak demand occurs, PV still contributes significantly togeneration at the times of critical peak demand. The amount thatPV reduces peak demand relative to the installed solar capacity (inMW-DC) will be referred to as the rating.

Table 3PV system costs.

Capital city Nominal system size

1.5 kW 3 kW 5 kW

Adelaide, SA $3.3/W $2.8/W $2.6/WBrisbane, QLD $3.3/W $2.9/W $2.6/WMelbourne, VIC $3.5/W $2.7/W $2.4/WSydney, NSW $2.6/W $2.3/W $2.3/W

Table 2Recent OCGT construction costs.

Completed Nameplatecapacity (MW)

Hot day capacity(MW)

Headline cost($m Mar-13)

State average distributionloss factor

Transmissionloss factor

Peak demandcapacity cost ($/W)

Mortlake Aug-12 566 518 815 6.8% 0.3% 1.69Colongra Dec-09 724 600 625 7.2% 1.4% 1.14Braemar 2 Jul-09 519 495 579 7.3% 5.3% 1.33Uranquinty Jan-09 664 640 775 7.2% 3.4% 1.35

K.B. Burke / Renewable Energy 68 (2014) 103e110 107

Table 4 shows the solar rating on each of the five highest de-mand days in each mainland NEM state from the last two summers(Nov 2011eMar 2013). Queensland, the most northern state, oftenmarkets itself as “the sunshine state” in tourism campaigns. How-ever it is Victoria and South Australia, the twomost southern states,which have the highest minimum top-5 solar ratings.

There are several likely reasons for the higher peak effectivenessof solar PV in southern states. Unlike NSW, VIC, and SA; QLD doesnot have daylight savings time in the summer. Queenslanders finishwork, return home and cook dinner when the sun is lower whencompared to VIC, SA and NSW. VIC and SA also sit to the west of thecentre of their time zones. In VIC and SA, solar noon occurs around1:25pm local time, compared to w1pm in NSW and 11:50am in

Fig. 4. Native demand and metered demand in the four mainland NEM

QLD. Finally the southern states VIC and SA have proportionallylonger days in summer, with sunset occurring as late as 8:30pmlocal time in the middle of summer. The full detail of seasonal solarposition can better be captured by simulated PV generation studies.The key point here is that while northern latitudes may give ahigher energy production from a solar system, the longer summerdays of southern latitudes allows for better critical peak supply bysolar PV.

The worst contribution by solar PV to critical peak supplyoccurred on the 11th-Jan-2012 in QLD. As can be seen in Fig. 5, solargeneration actually pushed the timing of themetered demand peakback by 1.5 h. With metered demand occurring later in the day,solar was less effective, although solar still reduced demand by 21%of its rated capacity. Future high penetration rates of rooftop solarwill shift the metered demand profile so that critical peak demandoccurs in the early evening, rather than mid-afternoon.

4.2. The capacity value of solar PV

As established earlier in this paper, generation capacity avail-ability, when delivered via transmission infrastructure by cen-tralised peaking generators such as gas turbines, costs $1.38/W.Solar PV is not fully available in the late afternoon due to the

states during the highest demand days of summer 11/12 and 12/13.

Table 4Solar rating at peak demand on the five highest demand days of summers 11/12 and12/13.

QLD Rating NSW Rating VIC Rating SA Rating

9-Jan-12 37% 18-Jan-13 32% 12-Mar-13 32% 17-Jan-13 48%11-Jan-12 21% 8-Jan-13 30% 29-Nov-12 40% 18-Feb-13 50%4-Dec-12 38% 30-Jan-12 24% 4-Jan-13 50% 12-Mar-13 45%10-Jan-12 46% 30-Nov-12 29% 24-Jan-12 54% 23-Jan-12 58%17-Dec-12 41% 11-Jan-13 42% 18-Feb-13 50% 24-Jan-12 50%Minimum 21% Minimum 24% Minimum 32% Minimum 45%

Table 5Capacity value of solar PV.

Minimum solarrating

Capacity valueof solar

Proportion ofsystem cost

QLD 21% $0.29/W 10%NSW 24% $0.33/W 14%VIC 32% $0.44/W 16%SA 45% $0.62/W 22%

K.B. Burke / Renewable Energy 68 (2014) 103e110108

angle of the sun. However, it does generate at least a minimumfraction of its rated output under high demand conditions. Thisminimum reliable generation level (Table 4) can be valued at thecost of gas turbine availability and used as the capacity value ofsolar PV as shown in Table 5. This valuation does not take intoaccount the fact that gas turbines do not start with 100% reli-ability, hence the capacity value of solar PV will be under-estimated. Comparing the reliable capacity to gas turbine capacityalso does not take into account the value of avoided transmissionand distribution upgrades, which would a subject worthy offurther study.

The current subsidy for solar PV (the small scale technologyrenewable energy certificates scheme) is theoretically based onthe amount of energy produced by panels in various locations, andis practically implemented as a subsidy per rated kW. Table 5shows that a significant proportion of the value of distributedsolar is related to the ability to reliably generate energy in criticaltime periods. This critical capacity value of distributed solar PV isnot reflected in current subsidy structures or Feed-in-Tariffvaluations.

4.3. West facing solar PV

Figs. 4 and 5 demonstrate that they key to using solar PV toeffectively reduce system peak demand is to maximise the pro-duction of power in the late afternoon. Fig. 6 shows the averagerating of solar systems during the top five demand days in eachstate, grouped by the semi-cardinal facing of each system. Fig. 6shows solar systems oriented westward have substantially betterenergy production in themid afternoon,when demand peaks occur.

Given the increasedmid-afternoon generation,west-facing solarPV has more critical peak capacity value than generally north facingsolar PV. To estimate this increased value, the rating of solar PV wasrecalculated using only west and northewest facing systems.

For example, solar reduced peak demand by only 21% of its ratedcapacity on 11-Jan-2012 in QLD (Fig. 5). If all of the installed solar

Fig. 5. The worst contribution of solar to a critical demand peak.

was facing West or North West, then solar PV would have reduceddemand by 33% of its rated capacity on that day. Table 6 shows theextra capacity value of northewest andwest facing solar PV relativeto the current distribution of system facings.

5. Conclusions

5.1. Limitations of this study

This study makes no attempt to calculate the avoided costs ofdelayed transmission and distribution infrastructure upgrades.Such a calculation is not possible without substation level load datafrom distribution companies. Previous research supplied to IPART[1] suggests that solar PV is not well correlated to residential loadprofiles, and may not provide any reliable supply in local peakdemand conditions. The corollary of late peaking residential de-mand is early-peaking commercial demand. Full availability ofsubstation load data would allow a more localised study to becompleted, which would be able to calculate the value of anyreduced need for transmission and distribution upgrades.

This study makes no attempt to calculate the value of energyproduced by distributed solar PV. The value of critical peak capacityis additional to the value of energy produced as calculated byregulatory bodies.

5.2. Capturing market value

At current demand levels, additional centralised peaking gen-eration is not needed. Part of the reason for this is that growth insolar PV has started to supply the demand peaks.

Current and previous policy settings such as feed-in-tariffs, highpeak hours regulated energy tariffs and the renewable energycertificates schemes have primarily incentivised energy productionfrom north facing solar PV. Solar PV could provide additionalmarket benefits if they were partially west facing, matching theiroutput better to native system load profiles.

It will be possible to economically capture this benefit forlarger customers with interval meters installed. A customer withsolar PV and an interval meter would have a demand profile un-correlated to peaks in system demand. A retailer could supplycustomers with solar PV and interval meters more cheaply, as theywould buy a larger percentage of their power at off-peak prices.Further studies would be needed to determine if distributors couldbenefit by delaying network upgrades in areas with daytimepeaking demand.

Centralised solar PV project developers could also model thepotential benefits of west-facing solar. However unlike distributedsolar PV, centralised solar PV can be disrupted by a single cloud, andtherefore may not have a minimum reliable generation level incritical peak conditions.

5.3. Generalisation

Distributed solar PV can reliably supply power during peakdemand in the NEM because the NEM is a summer-peaking grid,

Fig. 6. Solar Capacity Factor in each state for unshaded panels on the top five demand days by facing.

K.B. Burke / Renewable Energy 68 (2014) 103e110 109

where the importance of air-conditioners on hot afternoons out-weighs the importance of heaters on cold evenings. The result thatsolar PV (particularly west-facing) can supply critical peak de-mand will generalise to other grids that reach critical peak de-mand in the afternoon, rather than the evening. Local factors, suchas the timing of demand and solar radiation intensity and therelative costs of solar PV and peaking centralised generation willdetermine the magnitude of the capacity benefit solar is able toprovide.

5.4. Conclusion

The ability of distributed solar PV to reliably supply energyduring critical peak demand conditions in summer afternoons isworth 10e22% of installation cost, when centralised gas turbinesare the alternative. Northewest and west facing PV can reliablysupply more peak demand when compared generally north facingPV. This additional supply is worth a further 3e6% of the systemcosts. Southern states, with longer summer days and more sun-shine in the afternoon are found to benefit more from peak supplyof solar PV.

Table 6Extra capacity value attributable to West and North West facing solar PV.

Improvement in minimumcritical peak supply

Additional capacityvalue

Additional value(% of system cost)

QLD 12% $0.17/W 6%NSW 6% $0.08/W 4%VIC 11% $0.15/W 5%SA 6% $0.09/W 3%

Declaration

The author has received no funding to complete this work. Theinitial idea for the research arose out of analysis work conducted forthe author’s employer, Westpac. However this article wascompleted in personal time. Any statements that can be interpretedas opinions are the author’s.

Acknowledgements

The author would like to acknowledge PVOutput.org forproviding the individual system level data that made this paperpossible. The author would also like to acknowledge Robert Smithfor the use of Fig. 1.

References

[1] Frontier Economics. Market value of solar PV exports; 2012. Melbourne.[2] Tasman Acil. The fair and reasonable value of exported PV output; 2013.

Melbourne.[3] Queensland Competition Authority. Estimating a fair and reasonable solar

feed-in tariff for queensland; 2013. Brisbane, QLD.[4] Elliston B, Diesendorf M, MacGill I. Simulations of scenarios with 100%

renewable electricity in the Australian National Electricity Market. EnergyPolicy 2012;45:606e13.

[5] Elliston B, MacGill I, Diesendorf M. Least cost 100% renewable electricityscenarios in the Australian National Electricity Market. Energy Policy 2013;59:270e82.

[6] Molyneaux L, Froome C, Wagner L, Foster J. Australian power: can renewabletechnologies change the dominant industry view? Renew Energy 2013;60:215e21.

[7] Passey RJ, Watt ME, Snow M, Outhred HR, Spooner T. Study of grid-connectphotovoltaic systems: benefits, opportunities and strategies. Prog Photovol-taic Res Appl 2009;17:419e31.

K.B. Burke / Renewable Energy 68 (2014) 103e110110

[8] Liu G, Rasul MG, Amanullah MTO, Khan MMK. Techno-economic simulationand optimization of residential grid-connected PV system for the Queenslandclimate. Renew Energy 2012;45:146e55.

[9] Maine T, Chapman P. The value of solar: prices and output from distributedphotovoltaic generation in South Australia. Energy Policy 2007;35:461e6.

[10] Watt M, Partlin S, Oliphant M, Outhred H, Macgill I, Spooner T. The value of PVin summer peaks. In: European PV Solar Conference, Paris; 2004.

[11] Australian Bureau of Statistics. Environmental issues: energy use and con-servation; 2011. Canberra:.

[12] Smith R. Crouching demand, hidden peaks: what’s driving electricity con-sumption in Sydney? Energy efficiency: towards the end of demand growth.Oxford: Academic Press; 2013. p. 325.

[13] Clean Energy Regulator. Data reports e renewable energy target; 2013.[14] Bureau of Resources and Energy Economics. Australian Energy technology

assessment; 2012. Canberra.

[15] Origin Energy. ORG completes 550 MW Mortlake power station; 2012. ASX/Media Release.

[16] Arrow Energy. Braemar 2 power station fully operational; 2009. ASX/MediaRelease.

[17] Origin Energy. Uranquinty power station comes online; 2009. ASX/MediaRelease.

[18] Delta Electricity. Annual report; 2010. Sydney.[19] NSW Treasury. Colongra Gas pipeline and storage facility e PPP; 2009.[20] Australian Energy Market Operator. Generation information; 2012.[21] Australian Energy Market Operator. Distribution loss factors for the 2012/

2013 financial year; 2012. pp. 0e24.[22] Bloomberg New Energy Finance. Australia LCOE update: wind cheaper than

coal and gas; 2013.[23] Solar Choice. Solar choice PV system price index April 2013; 2013.