Wednesday, August 24, 2011

Blythe switches to PV

There have been dramatic, astonishing and significant developments with the 1000 MW Blythe solar project in California, the biggest solar project in the world.  On 18 August 2011, the developer (Solar Trust of America, a joint venture between the German companies Solar Millenium and Ferrostaal) announced that the first 500 MW of the project will now be powered by PV, and not solar thermal as previously announced.

Dramatic news?  Absolutely – the Solar Millenium share price collapsed from around EUR 20, where it had been for around a year, to around EUR 4.

Astonishing?  Absolutely – the foundation stone for this project had already been laid, approvals had been secured, funding was in place, and construction of the solar thermal plant was under way.  A switch at such a late stage in a multi-billion dollar project is truly surprising.  One might well wonder at the quality of the decision-making process in the first place.

Significant?  Some might argue this is proof that PV has decisively triumphed over solar thermal in the mortal struggle for market share.  That would be taking things too far, but it does say that in SE California where water is scarce, at the present, PV gives cheaper power than parabolic troughs with air-cooled condensers and no thermal storage.

I analysed the Blythe installation in Cost of Solar Power (12).  According to my standard assumptions, I found the cost per peak Watt was USD 5.79 and the Levelised Electricity Cost was USD 315/MWhr.  I thought the costs were high and commented at the time that “the LEC for the Blythe project is 25% higher than for another big US project currently under construction – Ivanpah”.  The Ivanpah installation, which I analysed in Cost of Solar Power (10), uses heliostats and towers, and will be able to collect heat at a higher temperature than the parabolic trough technology proposed for Blythe.   That indicates a technological advantage for power towers over troughs when air-cooled condensers are used.

According to Solarbuzz data, the cost of PV modules in the USA has fallen from USD 4.75/Wp in mid 2008 to below USD 3.00/Wp today.  Solarbuzz comments that the “module price is 50-60% of the total installed cost of a solar energy system”.  Presumably this led the decision-makers at the Solar Trust of America to turn away from their tried and tested parabolic trough technology.  It must have been a painful decision.

What does this mean for Solar Millenium and their parabolic trough technology?  Well, that technology still offers the great benefit of thermal storage.  Moreover, where water-cooled condensers are possible, the peak power output will be higher than with air-cooled condensers as proposed for Blythe. 

My view is that there remains a place for solar thermal power with storage.  Several contenders are slugging it out for dominance – troughs, heliostats/towers, Compact Linear Fresnel Reflectors and dishes – without a clear winner in sight yet.  (I might also modestly draw attention to my evaporation engine, see www.sunoba.com.au.)  I also think that costs for solar thermal power station components will continue to fall as more large-scale installations are constructed.

I expect the battle between solar thermal and PV to continue.  It will be a tough fight.

Sunday, August 21, 2011

2011 Solar World Congress, Kassel

Several German visitors visit this blog occasionally, especially when I comment on the activities of Solar Millenium, for example in association with solar installations at Andasol, Ibersol, and Blythe.  Therefore, I’ll just mention that I’ll be attending the 2011 Solar World Congress in Kassel from 28 August to 2 September.

I’ll be presenting a poster paper entitled “Output of the Evaporation Engine (Sloping Canopy)”.  My session is scheduled for Thursday morning, 1 September.

The figure below contains my main findings (click to see full size image).  This shows the simulated output of my evaporation engine at a suitable site for a complete year.  Both horizontal and sloping canopies have been studied; the figure shows the output from a 1,000 m^2 canopy with canopy losses included, but not engine losses.  Engine losses were evaluated in my paper in the congress proceedings, and are expected to reduce the estimates in the figure by around 25-30%. 

The table below summarises my results for the 1,000 m^2 canopy, with all losses included.  If achievable in an actual system, these results would be highly pleasing.  It is noteworthy that the sloping canopy gives greater annual output than the horizontal canopy.  Moreover the output occurs in a more uniform fashion and can be achieved with a slightly smaller engine.


Units
Horizontal
Sloping
Peak power
kW
65
60
Annual output
MWhr/yr
74
94
Water consumption
m^3/yr
1,181
1,422
Cost per peak Watt
AUD/(Wp)
1.38
1.42
Levelised Electricity Cost
AUD/(MWhr)
173
128

I would be very happy to discuss these issues with delegates, developers and fellow enthusiasts.  More generally, I’d also be very happy to discuss the cost of solar power, as investigated in this blog since the start of 2011.

Monday, August 15, 2011

Yet more on LEC

As I mentioned in my last post, I’ll present a talk to the Sydney branch of the Australian Solar Energy Society next Tuesday.  The talk is entitled “Comparison of Levelised Electricity Cost for Large-Scale Solar Projects”.

I’m sure to be questioned about my methodology for estimating the Levelised Electricity Cost (LEC).  By way of explanation, I’m mainly interested in international comparisons rather than being able to give a definitive statement to potential investors or to the Department of Taxation.  So my methodology ignores inflation and taxation.  I have stated my assumptions in every post in which I have discussed the cost of solar power; they are

          there is no inflation,
          taxation implications are neglected,
          projects are funded entirely by debt,
          all projects have the same interest rate (8%) and payback period (25 years), which means that the required rate of capital return is 9.4%,
          all projects have the same annual maintenance and operating costs (3% of the total project cost), and
          government subsidies are neglected.

Further commentary on my LEC methodology has been given in the posts on Real cost of coal-fired power, LEC – the accountant’s view and Cost of solar power (10).

For completeness, I’m now going to give a full model including taxation and inflation.  The model will be applied to the Kogan Creek installation I described in Cost of Solar Power (7).  The capital cost of this project was AUD 105 million and the annual output is stated to be 44 GWhr/yr.

WARNING: This is not exciting stuff and I just can’t think of any way to make it easy reading!  Basically, the requirement is to study the construction and entries in a spreadsheet.  In essence, future Profit & Loss accounts and future balance sheets are presented (in future $), which are then discounted back into present $.  Changes in the sum of NPAT plus equity are calculated, both for individual years and on average over years since the start of the project.  This leads to a quantity called average TSR (present $), which is truly the bottom line; the goal is for this to hit a shareholder-specified target such as 0.05.

Here are the assumptions for the full model:

cost
[AUD]
105,000,000
interest rate
[-]
0.075
depreciation period
[yrs]
25
annual output
[MWhr/yr]
44,000
tax rate on profits
[-]
0.3
debt fraction
[-]
0.5
{annual O&M costs}/{total cost}
[-]
0.03
inflation rate
[-]
0.025
estimated LEC
[AUD/MWhr]
320

Here are the financial details for the first three years (all figures in AUD millions):

Year
0
1
2
3
sales (future $)

14.080
14.432
14.793
O&M (future $)

3.150
3.229
3.309
EBITDA (future $)

10.930
11.203
11.483
depreciation (future $)

4.200
4.200
4.200
EBIT (future $)

6.730
7.003
7.283
interest payments (future $)

3.938
3.623
3.308
pre-tax profit (future $)

2.793
3.381
3.976
taxation (future $)

0.838
1.014
1.193
NPAT (future $)

1.955
2.367
2.783
NPAT (present $)

1.955
2.309
2.649
* cumulative NPAT (present $)

1.955
4.264
6.912
depreciated assets (future $)

100.800
96.600
92.400
borrowings (future $)
52.500
48.300
44.100
39.900
equity (future $)

52.500
52.500
52.500
* equity (present $)

52.500
51.220
49.970
sum of items marked *
52.500
54.455
55.483
56.883
average TSR (present $)

0.037
0.028
0.028

Acronyms:

O&M              Operations and Maintenance expenses
EBITDA         earnings before interest, taxation, depreciation and allowances
EBIT               earnings before interest and taxation
NPAT             nett profit after tax
TSR                 total shareholder return

Note that the annual depreciation figure is used to reduce borrowings, which means that interest payments decrease with time.

The last line is the key to it all – in words it is the cumulative nett profit after tax plus the equity, all expressed in today’s $, divided by the initial investment, and then expressed as an average over the total numbers of years that have elapsed.  That is the number that should eventually hit the target specified by the investors.  In this example, it will be 0.050.

The spreadsheet then continues uneventfully until year 13 when the borrowings hit zero.  Here are four years from the spreadsheet around that time.

Year
12
13
14
15
sales (future $)
18.474
18.936
19.409
19.895
O&M (future $)
4.133
4.236
4.342
4.451
EBITDA (future $)
14.341
14.670
15.067
15.444
depreciation (future $)
4.200
4.200
4.200
4.200
EBIT (future $)
10.141
10.500
10.867
11.244
interest payments (future $)
0.473
0.158
0.000
0.000
pre-tax profit (future $)
9.669
10.342
10.867
11.244
taxation (future $)
2.901
3.103
3.260
3.373
NPAT (future $)
6.768
7.239
7.607
7.871
NPAT (present $)
5.158
5.383
5.518
5.570
* cumulative NPAT (present $)
43.990
49.373
54.892
60.461
depreciated assets (future $)
54.600
50.400
46.200
42.000
borrowings (future $)
2.100
0.000
0.000
0.000
equity (future $)
52.500
50.400
46.200
42.000
* equity (present $)
40.013
37.475
33.514
29.725
sum of items marked *
84.003
86.849
88.406
90.186
average TSR (present $)
0.050
0.050
0.049
0.048

Thereafter, not much dramatic happens.  The installation earns a bit more than 5% annually after tax, as expressed in today’s $, and the value of the asset continues to depreciate until there is no residual value after 25 years.  For the parameters that have been chosen, the average TSR (present $) after 25 years is 0.050.

I therefore conclude that, with taxation and inflation included and for parameter settings as given in the first table, the LEC is around AUD 320 per MWhr.  Note that the eventual LEC depends on the assumptions that have been made.  This needs to be systematically explored with the spreadsheet, but I’ll give one example.  If I make only one change, namely to set the debt fraction to be 0.65, then LEC = AUD 287 per MWhr ensures that the average TSR (present $) turns out to be 0.05.

My original estimate of the LEC for the Kogan Creek development was AUD 295 per MWhr.  In view of the analysis and discussion above, I think that’s a reasonably robust estimate, and certainly useful for the purposes of international comparisons.