Solar power on the island of Ta’u, a preliminary appraisal

A 1,400kW(p) solar PV array backed up by 6,000kWh of battery storage and a smart grid has been installed on the island of Ta’u in American Samoa. It’s reported that this system already allows Ta’u to obtain 100% of its electricity from renewable sources for 100% of the time, and this brief review suggests that it will in fact be capable of delivering 100% electricity for almost 100% of the time if and when it reaches full operation. However, the operator Solar Power (recently acquired by Tesla) foresees potential problems in integrating 100% solar energy with the grid and plans on a “phased approach” to identify and resolve them. Consequently the system will not supply 24/365 renewable electricity immediately, and its ability to deliver it in the future will be dependent on whether a grid fed by 100% solar energy can be made to work.

Ta’u, the easternmost island in American Samoa, has an area of 44.31, a population of 790 and a 2015 per-capita GDP of $11,289. Historically it has obtained its electricity from diesel units which generate 1,300,000 kWh/year, or about 3,600kWh/day. Peak load is 229kW and average annual per-capita consumption 1,650kWh/year. (Data from various sources).

Google Earth image of the volcanic island of Ta’u, latitude 14.25 degrees south, showing the location of the solar panels. Habitation is confined to the northern coast. Note the caldera rim feature in the southern part of the island.

The Ta’u solar installation:

Project funding was provided by the American Samoa Economic Development Authority, the U.S. Environmental Protection Agency and the U.S. Department of Interior. Costs, about which more later, are reported at $8 million.

The installation consists of:

  • 1,400 kW of solar pv (5,328 panels)
  • 6,000 kWh of storage in 60 Tesla powerpacks
  • A microgrid which “allows the island to stay fully powered for three days without sunlight and can recharge to full capacity in only seven hours.”

The image below shows the solar installation. The panels are fixed and oriented at a shallow angle to the north, presumably at or around 14 degrees to comply with fhe 14 degrees south latitude:

The Ta’u 1,400kW solar panel installation (image credit Solar City)

The image below shows the Tesla powerpacks, each of which is about eight feet high:

The Tesla powerpacks (image credit Inhabitat)

Performance Evaluation:

To evaluate the likely performance of the Ta’u installation I reviewed two questions: 1) what is the solar pv capacity factor going to be and 2) how large are seasonal variations in solar output? This turned out to be less problematic than I had anticipated because there are already a number of solar pv installations operating on other islands in American Samoa and the Sunny Portal website provides operating data on 20 of them. Averaging monthly output from the 18 installations that gave usable results, which range in size from 7kW to 35kW and contain 391 months of generation data between 2011 and 2014, gave the results shown in Figure 1:

Figure 1: Average monthly capacity factor for 18 solar installations in American Samoa

Capacity factors range from 14% in June up to 20% in September, October and December. Comparatively small seasonal variations of this type are typical of low latitudes. The average annual capacity factor is 18.0%, which closely matches the 17.7% capacity factor calculated from data supplied by the American Samoa Power Authority (3,800,000 kWh/year from 2,450kW of solar pv capacity). At a capacity factor of 18%, however, the 1,400kW Ta’u installation would generate 2,200,000 kWh/year, well in excess of Ta’u’s present consumption of 1,300,000 kWh/year, and as a result about 40% of Tau’s Annual solar generation would probably have to be curtailed.

The next question is whether the 6,000kWh of Tesla battery storage is enough to fill demand during extended cloudy periods. To evaluate it I took the 2012 daily generation data from the 28kW “Procurement” installation on Tafuna Island 130km to the west, scaled them up by 1,400/28 to replicate generation from the 1,400kW Ta’u installation and used them to run a “battery balance” assuming a constant daily demand of 3,600kWh. Figure 2 shows the daily generation data for the 1,400kW case. It fairly consistently exceeds the 3,600 kWh daily demand, usually by a substantial margin:

Figure 2: Ta’u daily solar generation estimated by scaling up 2012 generation from “Procurement” installation on Tafuna. The month divisions are approximate.

Figure 3 plots the battery balance, which assumes no charge/discharge limitations. The 60 Tesla powerpacks remain fully charged at 6,000kWh for over 80% of the time and become completely discharged on only four days (May 24 and 25 and June 4 and 5), i.e. for about 1% of the time. However, over 40% of the total solar generation has to be wasted or another use found for it:

Figure 3: “Battery balance” obtained from Figure 2 data.


Ta’u is the latest entrant in the growing field of “100% renewables” projects, and this brief appraisal suggests that it probably has a better chance of succeeding than some of the other projects that have been marketed under this mantra. The key, however, is whether the smart grid can be made to work with 100% solar generation and zero diesel backup. As discussed earlier Solar Power plans to adopt a “phased integration” approach under which solar energy is sent to the grid in small amounts to begin with and in increasingly larger amounts only after the grid has been shown to be capable of handling them, but this approach was adopted at King Island, Tasmania, eighteen or nineteen years ago and the goal of 100% renewables has still not been reached there. My guess is that in common with King Island and Gorona del Viento on El Hierro the island of Ta’u will never be able completely to get rid of backup diesel generation, or at least not at any time in the foreseeable future, but I would be happy to be proven wrong.


The total cost of the project is reported to have been $8 million. I don’t have enough backup cost data to check this estimate out, but here are some of the numbers I have been able to come up with:

  • According to NREL installed costs for 1,400kW of pv panels in the US (Ta’u is part of the US) were around $2,000 per installed kW in 2015, which works out to $2.8 million.
  • According to Electrek 54 Tesla powerpacks (5,400kW) and 10 bi-directional inverters (2,500kW) can be purchased in the US for $3.2 million excluding installation. Raising the price by 10% to include 6,000kW of powerpacks plus a few more inverters and adding, say, $1 million for installation gives $4.5 million.
  • $2.8 million plus $4.5 million is $7.3 million, and we still need to allow for the costs of shipping everything to Ta’u, which has no docking or airport facilities worth speaking of, and erecting it there. How much to allow? I arbitrarily doubled the $7.3 million estimate, giving a grand total of $14.6 million.

Now $14.6 million may be too high and $8 million may be really what the project cost, although it still seems a little on the low side. But even if $8 million is the right number it still works out to $10,000 (or almost one year’s GDP) for each of the island’s 790 residents.

Further updates will follow as and when additional information becomes available.

This entry was posted in Energy and tagged , , , , , . Bookmark the permalink.

113 Responses to Solar power on the island of Ta’u, a preliminary appraisal

  1. Willem Post says:


    Those battery packs will have losses.

    Round trip, AC to AC, in balancing mode, will be at least 15%.

    • Willem Post says:


      TESLA markets floor-mounted 100 kWh Powerpack @ $25,000, or $250/kWh*, lithium-ion cells made by Panasonic.

      TESLA utility-scale, turnkey, Powerpack battery systems would be 6000 kWh x $400/kWh = $2.4 million in the US mainland.

      Such systems need HVAC systems for ventilation and cooling in the tropics, for ventilation, heating and cooling elsewhere.

      Here is some data on the 7 kWh TESLA wall-hung unit:

      The TESLA 7 kWh Powerwall batteries actually store 10 kWh, of which 70%, or 7 kWh DC, is available once every day for 10 years, 3650 cycles.

      If 7.756 kWh AC is fed via an AC to DC inverter into the battery, 7 kWh DC is charged (charging efficiency 0.914).

      If 7 kWh DC is discharged via a DC to AC inverter, 6.400 kWh AC is the usable energy (discharge efficiency 0.914).

      The AC-to-AC efficiency 0.914 x 0.914 = 0.836 is likely less in the real world, due to other system energy losses, and due to degradation of performance over time.

      An AC-to-AC efficiency of 0.80 or less would be more realistic. See URL.

      • Peter Lang says:

        Willem Post,

        Could you offer an estimate for the FO&M and VO&M costs for the total PV+battery storage+ smart grid system for the Ta’u system?

        I used the NREL LCOE Calculator value for FO&M of about $20/kW.yr and added 150% to include the storage and smart grid system. Your figures above suggest the batteries would have to be replaced at least once, perhaps twice, in 20 years. Inverters would have to be replaced. I haven’t a clue what the O&M costs would be for the smart grid. Include labour costs, and all other costs normally included in O&M.

    • Graeme No.3 says:

      The original plan for the Ta’u system, serving 790 people, was for 1,200 kilowatts of solar PV and 3 megawatthours of storage, and aimed at supplying 80% of electricity demand.
      Samoa has installed two 55 meters high turbines that can pivot at the base, and be lowered and locked in place in less than one hour. This unique collapsible design helps to avoid damage from the region’s NUMEROUS cyclones. There was also a tsunami in 2009.

  2. David E says:

    Great article Roger, if solar power can’t work well within 14 degrees of the equator, where can it?

  3. Syndroma says:

    I wonder if going 100% solar is easier than 100% wind.

    • gweberbv says:

      Of course it is (near to the equator). You may have weeks with low wind, but not weeks of low sunshine. Unless, of course, you have one of these giant volcano eruptions that happen once a few hundred years.

      • robertok06 says:

        … don’t forget the “occasional” pacific hurricane, Guenter!… a couple of months ago there has been a deadly Atlantic hurricane (can’t remember now which one it was) which crossed the island of Haiti from south to north… and went right on top of an area where several 100kWp-scale PV plants were located… I’ve tried to find informations about the fate of such plants… I can’t see anything else than total destruction of the panels as soon as 250-300+ kmh winds throw trees and debris onto them…

        A quick google search shows that…

        … “The Virginia-based National Rural Electric Cooperative Association (NRECA) is mounting an effort to repair a microgrid in Coteaux pummeled by Hurricane Matthew, leaving the 1,200 co-op members it served without power for an expected two months or longer.

        If the designers had known the storm was coming, they may have changed the solar system’s racking system, Waddle said.”

        This one here, mentioned in the text of the previous one…

        … says that they had restored power in 55 hours… but this PV system is placed on top of a brick building, a hospital… not free-standing PV panels in the middle of a patch of cleared forest.

        Some photos of another PV system here…

        “The microgrid’s generation system is still largely intact and only 25 percent of the solar panels were lost, according to a blog posted on EarthSpark’s site.”

        … so, “only” 25% of the panels have been damaged… let’s now imagine the same in the southern USA, where the electricity per capita is ~10x that of Haiti, one of the poorest countries on the planet.


        • Thinkstoomuch says:

          In addition being near the Equator is not the be all and end all of PV performance. There is this thing associated with any storm or rain for that matter. Its called clouds and other stuff.

          Which as usual Roger has found real world data to allow for and the numerous other factors involved. Thank you Roger!

          Now if you allow for a year’s worth of GDP to buy all them batteries, I think it may actually work.

          I am probably more optimistic than most on this project’s 100% renewable working. For how long, I dunno.

          But then again if you throw enough money at anything you can get a few people to the moon. On the other hand how many since the government money dried up (more important mirages to chase) and where are my moon tourist shuttles? Almost a half century later.

          For that matter you can power a toy on Mars with solar. Though if you want something bigger you need a more reliable power source.

          Thank you all,

          • Roger Andrews says:

            Willem, David et al. The question here isn’t how well solar works and how much it costs but whether it can be made to work at all with no spinning reserve.

          • Alex says:

            Well, you could always use a DC motor to spin up a shaft at 3000/3600 rpm, and stick an AC generator on the other end.

            But you can also use the electronics to create nice sine waves at 50/60 Hz regardless of the current draw (within limits). A UPS for your computer (desktop, with a AC input) does that – though of course, the “spin” it can maintain is minimal.

            Thinking this through (disclaimer: I’m not a specialist in power electronics) – with spinning reserve, a load increase will increase the resistive force on the generator which causes the shaft to slow down. This reduces the frequency, which is detected by the sensors, which increase power to the boiler/turbine. The sub-second response is provided by the spinning inertia of lots of drive shafts.

            If the power is supplied by a battery, increased current draw will cause the voltage to drop, but shouldn’t alter the 50/60Hz, which is set by the electronics. The electronics can increase the current to compensate.

            If the supply is a mix of spinning and battery, then if the frequency drops, the inverter needs to drop it’s frequency to match so as to remain in phase. This is what a grid tie inverter will do. You could also however set the output “sine wave” to be a few degrees ahead of the grid sine wave – to “drag it forward”.

          • Willem Post says:


            If the battery capacity is large with respect to load changes, then inertia is inherent in the system, I.e., no synchronous-condenser system is needed to maintain frequency, voltage, etc., within set limits.

            Islanders may be used to some flicker, which would be unacceptable in a modern economy.

  4. Joe Public says:

    Thanks Roger.

    “The installation consists of:
    1,400 kW of solar pv (5,328 panels)
    6,000 kWh of storage in 60 Tesla powerpacks”

    Except ….. for some reason the MSM fail to mention the full project description, which lists 1,410 kW of Solar panels and 6,000 kWh of battery storage. Also, three new 275KW Cummins Diesel Generators and a 480V switchgear.

    So maybe someone anticipates Ta’u might not obtain 100% of its electricity from renewable sources for 100% of the time.

    • The idea behind the diesels may be that they will ultimately be powered by biofuels. This was the plan at King Island a couple of years ago but I don’t know how much progress has been made since then, if any.

      And El Hierro’s main export crop is bananas. I assume you can make biofuels from bananas.

  5. Alex says:

    Excellent article. I’ve long thought that islands are a great place to test these concepts – partly because the natural resource is good, and partly because shipping costs raise the price of diesel. The Falkland Islands would be a good site for a “UK” experiment, obviously based on wind rather than solar.

    Perhaps the next step is (with Government subsidies of course – at a GDP of less than $12k it can’t afford it) would be to replace the vehicles with electric vehicles. They don’t need Teslas on that island.

    The next step – useful for everyone on a larger scale – is to see what they can do with surplus electricity. The marginal cost is zero, but it’s only available when the batteries are full and it’s sunny. Is there an application that can take lots of current, but not all the time.

    On a global scale, that application might have to be production of liquid fuels. But for a small island, one application that could work is biorock production:
    Could this be a way to build a harbour for the island – or at least a breakwater out at sea? Once the conducting frame is built and put in place, and laid out as a break water, excess current is simply diverted there. If a KWh can make 1kg of biorock, then a surplus 2MWh/day could make an average of 2 tons of reef per day. After a decade they’d have a useful reef to protect the landing area.

    • Greg Kaan says:

      The Falkland Islands has long had wind farms and uses them for diesel reduction. Wind generated output is curtailed for the sake of stability so the grid is never close to 100% renewable

      Graeme No3 has commented on this here in the past with information from a friend who runs the grid there.

        • Graeme No.3 says:

          I’ve lost touch but they had cut their use of diesel fuel by 30% without wrecking their system (unlike South Australia).
          The big difference is that the change was controlled by an experienced (and savvy) electrical engineer, not by an Arts Graduate with zero knowledge captive to green enthusiasts (they make the most noise).

          • Alex says:

            No Arts Graduates! – that’s why we don’t read about it in the Guardian.

            I assume they [Stanley area] could cut their diesel usage by a lot more, but they’d end up spilling a lot of wind power. Hence they’d have needed to build a pumped storage reservoir, which whilst cheap per MWh of storage, would be a bit extreme for a population of 2,000 people.

            That could change now as from about this year batteries have become contenders for day to day storage. (Still nearly two orders of magnitude off for season to season storage).

          • Graeme No.3 says:

            the trick was to ‘spill’ some possible wind generation in order to get more stable input. Part of the gains resulted from using some of the waste heat from the diesel generators to provide a district hot water heating service for Stanley.
            Pumped storage was looked at on a secondary island and proved illusory.
            Bateries are fine for believers in AGW who have lots of money and are very poor at simple arithmetic. Useless when larger amounts or continuous supply is essential.

          • OpenSourceElectricity says:

            Hmm you are aware that the last three blackouts in south australia (before the last one)were caused by the northern power station, which was too big for the South Australian grid so whenever there was a major failure in this power station this caused a blackout of the whole state?
            These blackouts happened without storm and multiple grid failures in short succession.
            Now some brilliant mind came to the idea to bring this power station back online to prevent blackouts……

          • Greg Kaan says:

            Please provide some source to back up those claims about the Northern coal plant being “too big”

            There was a partial blackout (about a third of South Australia) back in 2005 when a substation fault caused Northern to reduce output which then caused Pelican Point to trip. Both plants were fined for not meeting performance specifications and were rectified to prevent future events of this nature.



            I would contend that a partial balckout is far less damaging and easier/faster to recover from than a full blackout as occurred on the 28th of September this year

          • singletonengineer says:

            Open Source Electricity (December 17th, 12:49pm) mentioned the former Northern Power Station.

            At 2 x 260MW units, it was a very much smaller presence in the SA Region of the NEM than the Heywood Interconnector, which is currently expanding from 460 to 650MW.

            Even the wind farms that failed immediately prior to the Heywood tripping totalled over 400MW

            Now what’s too large?

          • OpenSourceElectricity says:

            @singletonegineer, source is:
            Event one: 2 December 1999 13:11
            Cause: Trip of both units at
            Northern Power Station
            Event two: 8 March 2004 11:28
            Cause: Runback of both units at
            Northern Power Station
            Event Tree: 14 March 2005 06:39
            Cause: Runback of both units at
            Northern Power Station

            If you run a unit of the size of norrthern power station in a grid, and want to keep it stable, you need to have a idle capacity in excess of 520 MW running in the system. Which they don’t have, they even did not have any frequency control reserves this august.
            The same with Heyworth interconnector. Whatever you transport by this interconector must be available either on another interconnector or in capacity roning indle, but with full spped. Otherwise there is no n-1 redundancy in the grid.
            So if you would run the grid as it is usual in central europe, the single Heyworth connector would be counted as no interconnector forpower trade, it’s just suitable as backup. Only if there is a second dual circuit interconnector with e.g. 2x600MW, you can transfer 2x 600MW over both parallel interconnecors, being safe then when either losing a circuit or loosing a whole interconnector. Or loosing a mayor power station like northern power station.
            As you could see from all 3 events there were local errors causing the loss of power generation, not a multitude of errors caused by weather conditions.

          • Greg Kaan says:

            Thanks for that link OSE – I had not realised that the AEMO had released a 3rd preliminary report on the September blackout.

            You didn’t mention that the 3 previous blackouts were only partial, nor did you bring up the period for the frequency drift prior to separation – all in table 10. In the case of the Northern plant incidents, the drift took around 2 seconds, giving the grid operators sufficient time to shed load and avoid a state wide blackout. The 0.6 second drift period of the recent blackout was too short for similar action to be taken. A glance at the system inertia at the time of the blackouts (in the same table) provides the reason why,

            As for the second 650 MW interconnector that you propose, where will it run to, will there be sufficient excess generation at the other end and most importantly, who pays for it to do nothing for most of the time?

          • gweberbv says:


            I think you misinterpret the recent blackout in Australia.

            The one and only cause were the fault settings of the wind turbines. For a low wind penetration it might make sense to have settings that throw wind out of an unstable grid so that conventional generators can stablize it using well-known procedures. But if wind is a main pillar of the supply the turbines need to operate with settings that are much more tolerant to an unstable grid.

            About 450 MW out of 900 MW of wind production dropped out within a few seconds. And that in a system were total supply was about 1800 MW with only 330 MW of thermal generation.

            By the way: What had happened if – for what reason ever – one of the two interconnectors had shut down?

          • OpenSourceElectricity says:

            @Gweberb – if the Heyworth interconnector would have collapsed, the same blackout would have happened.
            And the settings were not to let other power generators take care of the grid, they were to
            a) protect the equipment against overheating,and
            b) nobody expected to have anything like a working grid outside the wind farm when such a high number of voltage drops happened in the grid in such short succession.
            There is no example for a similar sequence in the records so far. Its a classical black Swan event.
            Difference to the nothern power station is, that external, unexpected conditions trigegred the blackout, while in case of the northern power station internal events were sufficent to cause the blackout. A internal event in a wind turbine would have resulted in a loss of 1…4MW power generation, a event at the conection of a wind farm in a loss of 10…100MW power generation, which could be handeled.
            Problem with the northern Power station, and most likely Torrens in the south Australian Grid is, that as long as the existing interconnectors are used for trading (importing) power, there is nothing which stops a partial or full blackout if one of these big power plant fails. There is simply nothing which can stop a grid collaps than throwing of mayor parts of loads faster than the Heyworth connector shuts down. There is a cause why east and west germany were connected by _3_ power lines for the start, each of them sufficent to balance out the loss of the biggest power plants in the grid alone. Today the grid connections here are significant stronger.

          • Greg Kaan says:

            And when the wind drops out, it’s not just one turbine representing a tiny amount out of a relatively small total from a wind farm that is affected. The output from the whole farm collapses and furthermore, it is likely that a large number of wind farms will fall off at the same time since they are sited for maximum average wind strength so they are deployed in clusters.

            This has happened several times recently and the effect on South Australia has largely been economic due to the availability of the Heywood Interconnector (and MurrayLink) for grid support. When Hazelwood is shut down in Victoria, those interconnectors may well not be able to supply South Australia and the damage will be material. And this could easily happen during this summer while Hazelwood is still in operation since the Victorian and South Australian weather systems are highly correlated and the majority of wind farms of both states are geographically proximate. Maximum demand (due to airconditioning loads) will often occur at the same time in both states and calm, hot evenings (after PV generation has stopped) are not unknown.

            The argument against large generators is philisophical beat up against any centralised resource (I guess that goes along with your “Open Source” moniker). Too bad it is also the most effective way of maximising returns from resources.

  6. Alex says:

    Regarding costs,
    has a bit more information about the difficulties of installation. I suspect also Tesla may have sold the power packs more cheaply for good PR.

    The island does have a landing strip – probably a WW2 relic – but I’d have thought they could use something like a Hercules transport plane?

  7. jim brough says:

    John Donne said,
    ‘No man is an island”
    Which means that we all rely on others.

    Solar Ta’u is not possible without the work of others.

    Solar for Ta’u is a good solution to providing its electricity but it comes at the expense of CO2 emissions elsewhere to make the cells and install them. Saves some diesel emissions but we should understand that there is no such thing as CO2- free electricity, or carbon-neutral, or beyond zero carbon.

  8. gweberbv says:

    I think the battery system is oversized simply as all these islands *really need* to keep their diesel backups operationable because of things like this:
    So, it does really no harm when the diesel generators are running for let’s say 10% of a typical year. And relying a little bit more on diesel would significantly lower the battery costs.

    • singletonengineer says:

      That comment essentially admits defeat before the race starts, the race being to achieve 100% renewables with battery backup.

      However, as demonstrated above, the race was always fouled by the presence of enough new Cummins diesel power to do the job without either wind or batteries.

      So, since zero percent wind has been adopted for design of the diesels, a mere 10% diesel input is almost trivia.

      IMHO, this is a typical island renewables system – unaffordable without external heavy funding, overpromised, overhyped, overdesigned and underperforming.

      • OpenSourceElectricity says:

        So you would operate a nuclear power plant, a coal power plant etc to power the island without and backup (diesel or similar)??

      • gweberbv says:


        as long as our takeaway coffee cups that are thrown into the trash bin the next minute are made (partially) from oil, I don’t see any point in trying to avoid at all costs the use of oil in the electricity sector.

        • singletonengineer says:

          My point was an still is that a claim of 100% renewables has been made. That denies the existence of those three brand new Cummins diesels.

          Indeed, it is sensible and probable for at least one diesel to tick over 24/7/365 in order to provide frequency support and related services which are generally not possible in renewables-only AC power systems.

          My concern is about an intentionally false and misleading statement.

          Re coffee cups: paper ones are both cheap and readily available, easy to shred and make good mulch. I don’t understand why plastic ones are still so common.

          • Alex says:

            They don’t need the diesels ticking over for frequency stabilisation. That’s what batteries and electronic timing is for.

            They do need the diesels there in case the batteries empty and they need electricity. That was a key point from “UK 2050 Electricity Supply Part 3”: You’d need silly amounts of storage before you can ditch the backup capacity. On a tropical island, the amount of storage might be a bit less silly, but silly none the less.

            The other issue is whether there will always need to be a diesel run on a regular basis, as diesels may not start if called upon to do so after a year of sitting idle.

          • singletonengineer says:

            Alex, I can understand a single diesel as backup, plus a spare as standby, but 3 x 33%? That is either overkill or a decision has been taken to adopt a full belts and braces approach, i.e. the decision-maker didn’t have faith in one of either the PV or the batteries or the control system or the electronics (includes the inverters).

          • Alex says:

            What the UK model showed is that when your storage is empty, you still need the full capacity. For Ta’u, I guess that means 3 diesels. If the solar panels get blown over, or a volcano nearby erupts, or the inverter blows up and the spare has been damaged, and a replacement takes a fortnight to arrive ……, then there’s a few weeks back on diesel. Maybe they reckon they need two at a minimum, so want a spare.

            They might have given it more thought if diesel costs a lot, but diesel generators are cheap.

  9. OpenSourceElectricity says:

    Iteresting point would be how Diesel and other fuel gets on the island so far, and how reliable. As I remember I did read that El Hierro in the 1990’s had weeks without electricity because diesel could not be delivered.
    If Diesel fo that island would come by Hercules planes or similar, it could cost a fortune.

    • OpenSourceElectricity says:

      The document Roberto referenced gies some hint to the diesel power costs – 38-29ct/kWh depending on diesel price development as average for all islands, and a peak ot prices in the Pacific due to long transport distances. So this island wich is also small hand has nearly no harbour, should be at the top end of diesel prices per kWh, so well above 38-29ct/kWh. Maybe 50% higher? 44-57ct/kWh for Diesel today? That would be 528.000-684.000$/year at the moment with diesel generators as cost for electricity. Compared to this 8 Million is not cheap, but also not completely unafordable.

  10. roberthargraves says:

    Batteries can supply 2000 kWh/day for 3 days of clouds, or 105 W/capita average power.

    Current consumption is 188 W/capita.

    Peak solar power is 1772 W/capita.

    North America average power consumption is 1500 W/capita, EU 700, China 400.

    American Samoa GDP is $11,289/capita. Andrews’ prior coarse estimate of $5GDP/kWh for 188W/capita predicts GDP/capita $8,234 — pretty close!

    Power plant investment of $8M = $10,126/capita — a whole year of earnings.

    To install ThorCon fission power plants at $1.2B/GW at the North American level of power consumption (1500 W) would cost $1,800/capita.

  11. Javier says:

    Ta’u inhabitants clearly cannot pay for this and it remains to be seen if they can pay to have the batteries (and panels, and inverters) substituted when needed. So a solution it is not as we better do not scale up. If they are charged the full cost of that electricity they will probably have to move to another island. So once again we are talking about a subsidy paid by many for the benefit of a few. That appears to be the mantra of renewable energy.

    • Alex says:

      But their diesel is also currently subsidised. The costs of a diesel solution are impacted more by transport costs than those of a renewables solution. Hence the level of subsidies will be reduced.

      • Thinkstoomuch says:

        Ok lets compare some silly numbers.

        In HI the average price of Diesel is “In November 2016, the regular gasoline price in Hawaii averaged $2.882 per gallon, which was 71.
        5 cents per gallon (33.0%) higher than the national average for the same month.”. Even if it costs twice as much to get it there that is $600k to $700k. If the federal government is paying for all of that it is roughly what 12 year supply for the cost of this system. How long does a power wall last again?

        Next people are acting like 110,000 gallons is a huge number. To an individual it is, granted. It is between 10-20 tanker truck loads. Not really all that much.

        How hard is it to get to the island I don’t know. But consider a warship conducting a under weigh replenishment at sea. A US Cruiser or destroyer takes 2 or three times that in a couple of hours at sea. Not sure how they do it on islands but in a real sense we are talking a moderate size landing craft would be needed.

        I would bet commercial companies have made it cheap and easy to do it easily. Small hose, relatively, is to transfer liquids. On the other hand bulky stuff on a one time delivery every few years …

        A probably stupid post but … best I can do.


        • OpenSourceElectricity says:

          Well google MAps shows some kind of reef protecting some kind of tiny harbour, and a landing acility which might be suitable for a average size Yacht. So I would estimate Diesel comes in Canisters or barrels. one boat trip to a small freight ship remaining outside at sea pert ton.
          Reasonable for a tiny island of that size, a “real” harbour also would cost millions.
          Don’t know how they transfer material in reality.

        • OpenSourceElectricity says:

          “Harbor” + boat there. As it seems most transport is done by aircraft today.

        • OpenSourceElectricity says:

          “harbor” facility there:, most transport seems to be done by aircraft today.

        • Graeme No.3 says:

          Petroleum fuel (Avgas, motor gasoline, low sulphur diesel, high sulphur marine fuel, kerosene etc.) is imported into Pago Pago by tanker, approx. 6 times a year and the higher volume usage is stored in bulk tanks. From there to Ta’u it would have to go in drums by boat.

  12. robertok06 says:

    I’ve just bumped into this…

    … about powering islands with a mix of intermittent renwewables.

    It is behind a paywall… but this…

    … is not.

    Enjoy the reading.

  13. JerryC says:

    Maybe I missed it, but what do they do at night? All diesel? Obviously they aren’t discharging the batteries every night.

  14. Simon cove says:

    A few points and questions:

    1) What is the cost of the diesel and generators etc partic in the island environment? Likely more labour needed to manage these than solar and batteries. job losses and decreased labour costs?
    2) I suppose the solar panels (at least some of them) may last longer than 20 years which may decrease costs
    3) likely that electricity usage will increase so overcapacity may not be a bad initial plan?
    4) batteries however likely won’t last that long so… these likely will cost more
    5) Whether carbon leads to alarmist level of warming or denier levels of warming it appears by at least this study that wind power reduces carbon emissions. I certainly At least think man is having a massive impact on other species life on earth and we should be very much researching it as much as possible and acting accordingly.

    Wind power key to curbing greenhouse emissions, study finds

    I’d be interested on views on this.

    6) related to 5, there was an interesting quote from a scientist on the guardian science podcast this week. essentially the scientist believed this week that science works by evolution and essentially is not linear. the example he used was that 10 billion spent on a log burning cooker would improve it massively but not produce an induction hob or microwave oven as the latter at least came out of the communications industry.. So I guess that all science is good and whilst wind and solar tech may not solve the energy problem… it likely will solve something else. however, this does not mean that a master plan Manhattan style project to solve the world energy and climate and other environmental issues would not half be a good idea!!

    • Beamspot says:


      First of all, congratulations for a great post.

      Now, let me do some back of the envelop calculations of my own.

      Mean temperature in that island is in the 70 – 90 ºF (about 26ºC), that according to Arrhenius, will mean less than 8 years of life expectancy for the batteries:

      But since they have excess power production just when the temperature is high due high insolation, and there are coolers, chillers that can keep batteries cooler, below 20ºC if necessary, and since depth of discharge is low, I will do a best guesstimation of 10 years battery life.

      At 4M$, 1300MWh/Year, that implies 300 to 350$/MWh of electricity extracted from the batteries only as battery costs.

      Add 20 years to replace the remaining infrastructure, O&M costs, CAPEX, insurances, and taxes, and the costs for the inhabitatns will rise easily to 600$/MWh.

      Now, try to fit this costs into your personal budget, and compare them to your particular electricity bills wherever you live.

      And keep in mind that they are really gifted: they don’t have seasonality, that is the killer point for almost all renewables, particularly for PV.

      Just to do things interesting, cost of Lithium batteries bottomed by the end of 2015, and this year they have skyrocketed. Lithium carbonate, battery grade rised from 6500$/Ton by oct 2015, to 22500$/T on oct 2016. Cobalt rised about 30% in the same period.

      I bet things will be far worse whithin two years, when Tesla Model 3 hits the marked and prices for all electric and hybrid cars will also rise, adding pressure to the automotive and mining industries.

      Its time to begin to do some sports, specially bicycle: this is the future of human transportation…

      Best regards, and Merry Xmas.


      • Greg Kaan says:

        Again, thanks for injecting some reality into the discussion of lithium based battery storage.

        I don’t agree with the low depth of discharge assumption, though, as Roger’s analysis does not seem to allow for any nighttime discharges. With a daily average demand of 3600 kWh (ex cooling requirement for the batteries), nightly demand could be in the region of 1500 kWh (allowing for 25% nighttime demand and factoring 830 W cooling each battery over 12 hours of non-PV production time to give a nice round number) producing a minimum daily cycle of 25%

        My figure is obviously rubbery (I think it understates the battery cooling load) but illustrates the concept.

      • gweberbv says:


        let’s assume that your assumptions are slightly pessimistic, then we might find that these guys would not have to pay much more than people on the Hawaii islands:
        (But of course they are not paying full costs – be it diesel or PV. Where should they take the money from?)

  15. Svend Ferdinandsen says:

    I heard of the project not long ago, but some diesel generators were mentioned.
    I believe they serve as reserve and most important to keep the 50 (60) Hz and power for the inverters to work against.
    I wonder if it is possible in larger scale to make a battery and inverter function as the traditionally spinning reserve that keeps the frequency and balance in the net.
    The “Smart grid” is only a fancy word. Let us just call it a small independent net like in other islands.

    • Alex says:

      You can emulate spinning reserve with a battery system.

      I doubt the diesels will be needed for spinning reserve – more for insurance. It could be they have new generators because:
      1. They’re cheap – in the UK capacity auctions they provide capacity at £22.50 per KW per year.
      2. The old diesels may not have have been suitable to infrequent usage.
      3. They do want a back up in case the proverbial hurricane removes the panels. (I assume the panels are designed to withstand hurricanes – but could be more impacted by flying debris)

      As for harbours and costs, the Guardian article is useful:

      There’s lots of discussion on costs, but bear in mind, solar is cheaper than diesel in most of the world. Add to that the high shipping costs of diesel – compared to the one off costs of shipping the equipment – and this is going to work out a lot cheaper than their current solution.

      As both solutions are subsidised, the effective discount rate for the investment is the US Government’s rate, making solar even more attractive. .

      • Svend Ferdinandsen says:

        You can emulate spinning reserve with a battery system.

        Of cause you can, it is done with small inverters to supply small equipment. But could you make a battery and inverter to make a cold start in a larger net, and has it yet been done?
        It seems they are not quite sure themselves.

        Regarding “waste” of excess power: Thats normal for every supplier. You need to have power to the max need so normally you run on less than max. It is a bit funny that it is concidered a problem that you can not use all that “free” energy the PV’s can deliver, when it is not a problem that the diesels have to run at less than max.

        • OpenSourceElectricity says:

          Yes, you can start a grid with full load with a iverter. You can not do this with a Diesel, it would say “blubb” and stop operation. With a diesel you must gradually increase load.

          • singletonengineer says:

            Surely, starting a black islanded 1400kW system with a couple of hundred customers at full load using only battery supplies would be unnecessarily extreme, especially when diesels are available.

            I imagine that the recommended procedure would be to start one or more diesels and then to introduce load incrementally, ultimately drawing from the batteries and/or PV panels once the system is synchronised, loaded, and stable.

            That is, unless the goal is to fry the inverter(s).

          • OpenSourceElectricity says:

            Singletonengineer, you never designed and tested sunbstantial power supplys, correct??
            The Inverters and batteries are nothing more than a classical UPS, which does exactly what? Take over 100% of load in a fraction of a second when mains power is switched of by some cause.
            You don not “fry” the inverters. They are designed for the load and overload situation when they take over power. And they can stabilize the frequency with all the power capacity of the batteries. I calculate this in MWs if you like, but the Diesel has about 30 MWs inertia available, the 6MWh BAttery power are several digits more, and can be pumped in the grid to keep the frequency exactly stable. Only limits are the fuses, which exist in exactly the same size at the diesel systems. You can kill any diesel generator by switching on too much load too fast. You do not manage this with inverters. No Way.

          • singletonengineer says:

            Let’s set aside for the moment the Appeal to Authority, which is a discredited debating technique used by OSE when he affirmed that a UPS is not an inverter.

            In plain language, batteries are DC and in this installation, the load is AC. That which converts DC to AC is what – a magic wand?

            Besides which, we aren’t really considering a UPS – if the task is to start the system at full load, then what we are discussing is an emergency power supply or standby generator.

            So OSE and I are indeed not discussing the same thing.

            Now let’s also note that OSE wrote: “Yes, you can start a grid with full load with a iverter.” [an inverter].

            So an inverter is not a UPS but a UPS is an inverter. A Corgi is a dog but not all dogs are Corgis.

            Clear as mud.

            Meanwhile, the diesels are, contrary to affirmation, more than capable of restarting a black system in much the same way that a UPS/Emergency Battery Supply/Inverter can do: initially off-line, then incrementally matching increasing load to increasing available power, in order to reduce the magnitude of the transients that would be present if the full load was re-energised instantaneously (“Start[ed] with full load).

            OSE drew attention to circuit breakers or fuses. Fair enough, but irrelevant to “Full load”, except that the fuses would probably all blow immediately due to those pesky transients, also known as inrush and starting currents.

            If, however, the batteries were considered to operate as support for the PV in an OPERATING system that would leave the diesels to handle re-start from black, which is precisely what OSE stated above is not possible.

            Explainer: No, I am not an electrical engineer and I doubt that OSE is either. But I have spent 30+years as a professional engineer in all facets of the power generation industry. Perhaps a fully qualified electrical engineer will sort out the terminology used by both of us, but my position remains essentially unchanged and that is that the three diesels are able to back up 100% of the anticipated load and handle black starts. If they were a 100% functional duplication of the PV/battery system, why would a reasonable designers have so little faith in the PV/battery system that he would also specify the provision of a brand new, complete, 100% capacity, traditional diesel powered system, ie a direct replacement of the pre-existing system.

            The PV and batteries are thus simply additional capital cost added to a conventionally designed replacement diesel system. This is a waste of taxpayers’ money.

          • Alex says:

            Is it a bit like a car? If I turn on the diesel whilst it’s engaged to the wheels, it will stall. So I need a clutch to engage the load gradually. With a battery it’s a bit easier. Just press the accelerator, and some torque is provided from zero.

          • gweberbv says:

            singletonengineer and OSE,

            I really doubt that on this island you will find anything that justifies to make the whole electric system blackout-proof. If they have some applications that really need to run through, these devices will probably have their own USV.

            The purpose of the batteries is to store PV production for use during the night (and on a cloudy day), thus reducing diesel consumption to a minimum. However, if something goes really wrong with the PV system and/or the sky stays dark/cloudy for several days, of course you need to have the diesel backup. And as diesel generators (in contrast to their fuel) are dirt cheap, why not have a few of them sitting around?

          • OpenSourceElectricity says:

            Sigletonengineer, I am electrical engineer, and I can tell you that the inverters in a UPS and in the system here in question are the same, since the tesla system is also used as UPS.
            If fuses blow with inrush and starting currents, you have a design problem in such a tiny island grid as we are having here.
            And what I tell you about the behaviour of Diesel and Inverter are from practical testing.
            The Diesel has to increase power gradually, and the additional rotating masses attached to the synchronus generator do not add enough Joule to bridge the gap between a sudden rise of power demand and the “slowly” (few seconds maximum) rising output of the diesel engine.

            The inverter can increase power output within one pulse from zero to maximum, which means within a few milliseconds or less. So swithching on full load does not matter.

            Even more a properly designed inverter – be it located in a UPS or in “something else” which also involves a battery a Inverter and a mother generator for the grid frequency and grid control to keep a ismand grit running, is designed to survive any currents which the fuse allows. A fuse allows significant higher currents above nominal currents for short times, and the inverters are designed for this. And the batterys deliver these currents easily.
            Which solves the problems of the inrush and starting currents unless there are very special conditions in the grid – which would then also usually blow some fuse, or some overtemperature switch.
            The “difference” between a island system with battery backup and a UPS is, in case of a offline UPS, that there is no control of grid voltage and switch back to grid, and in case of a online UPS no double conversion or other support in case of brownouts and other grid failures wich are no total lack of grid power.

          • Greg Kaan says:

            The Inverters and batteries are nothing more than a classical UPS, which does exactly what? Take over 100% of load in a fraction of a second when mains power is switched of by some cause.

            OSE, that is not the same as black starting a load and you know it. Inrush currents would greatly increase the current requirements over the normal load case and no one in their right mind would over provision the inverters for that capability (plus the batteries probably couldn’t handle it).

            They would black start the grid in an isolated segment and switch in additional segments in sequence – the same as if using synchronous generators.

  16. climanrecon says:

    I hope these islanders have been smart enough to get paid for hosting this system, rather than being hoodwinked into paying anything for it, this is clearly just a marketing event for the solar/battery industry, who should themselves pay full advertising rates, and shame on them if they try to extract any planet-saving subsidy from anywhere.

    • Simon cove says:

      According to links from roberto6. The solar and batteries look cheaper than Diesel I think

      • singletonengineer says:

        The Guardian’s article mentioned current subsidies as hundreds of thousands of dollars per year, just for diesel fuel. Add maintenance. Compare with cash flows necessary to operate the new system.
        Capital $8M to $15M.
        Operating +Maintenance =?
        Discount rate =?

        My wild guess is that just to recover the capital, even if zero maintenance and operating costs and zero diesel usage this still represents represents 20 to 40 years for capital payback.

        During which time the Tesla powerwalls will need replacement a couple of times.

        This isn’t about economics. This project is an entry in a beauty competition.

  17. Rob Slightam says:

    There are two small harbours with breakwaters on the north of the island.

  18. OpenSourceElectricity says:

    Well fossil fuel power on such a island seems to be by factor 10 above the wholesaleprice on a continent, due to transportation issues. Same issues drive up the prices for any construction on such a island.
    To bring in a “U-Battery” on the island, most likely the construction of a harbor would be neccesary whoch costs more than the whole solar power supply. And which is not needed for just 800 Inhabitants later on.
    Thorcon stated on EnergyPost that they hope to deliver Power at 60€/MWh, not that impressive compared to other numbers on the market today. And since they do not have a prototype yet it’s a “hope”. Not a price in a contract. Or a price of power delivered to any grid.

  19. Peter Lang says:

    My interest is in the costs presented in a way that is properly comparable with alternatives. the most widely used and understood is LCOE. But it has to be presented in a way that is compartable across technologies; i.e. including all costs for a systems that meet the same requirements. In this case I’d like to see the cost for the PV+battrery and smart grid to provide 100% electricity meeting the reliability requirements. The proposed system cannot meet the reliability requirements, so more cost should be included.

    I’ve had a go at estimating LCOE for the system system using the NREL LCOE calculator . I don’t have figures for O&M costs for PV, batteries and smart grid for 20 years service so I’ve assumed $50/kW-yr. I used 10% discount rate for this high risk project (commercial operators would not invest in such a system on a purely commercial basis). Inputs I used are:

    Capacity 1,400 kW
    Battery storage 6,000 kWh
    Demand 1,300,000 kWh p.a.
    Peak load 229 kW
    Capital Cost $8,000,000
    Contingency $2,000,000
    Total cost $10,000,000
    CF, June 14%

    Ammorttisation period 20
    discount rate 10%
    Unit cost $7,142.86 $/kW
    CF 14%
    Fixed O&M 50 $/kW.yr
    Variable O&M 0 $/kWh
    LCOE 72.2 $/kWh

    I’d welcome comments and improved better figures.

    • Peter Lang says:

      Appologies for the typos; I posted before I’d checked. The LCOE should read 72.2 c/kWh or $722/MWh

    • Alex says:

      Regarding the discount rate – this project is viewed as a way of reducing subsidies, so (1) it’s Government funded, and (2) it’s a cost reduction, rather than revenue enhancement. That would argue for a lowish cost of capital – which is not quite the same as the discount rate, but probably the figure needed.

      O&M costs: Diesel has been bidding into the UK capacity auctions at £22.50 (or even less) per year. I would assume that covers Capital depreciation and O&M based on zero usage. Can solar + batteries not beat that?

      The Guardian says:
      “Five of those fifteen locals – previously low skilled, odd-job men on the island – have now transitioned to full-time jobs as solar power technicians managing the grid.”

      Call that $100,000 per year including over heads and spares – even if most of their work is probably cutting back vegetation. OK – that actually comes to $70/KW/year. (A UK diesel operation might have 5 staff, but that would be for 20-50MW of capacity)

      $722/MWh is probably similar to the diesel costs. Indeed, that would be suggested at the end of the Guardian article:
      “Although the islanders’ power bills remain the same – around US$80-100 per month for an average household – Malae said the reliability and self-sufficiency of the new system had been a cause of celebration on this remote outpost, with neighbours Ofu and Olosega islands planning to follow suit by Christmas.”
      (Still spending over 10% of GDP/capita on electricity).

      • singletonengineer says:

        Robertoko’s second reference, above, provides an estimate for all island systems globally of 38 eurocents per kWh, ie say $0.40 US.

        • Peter Lang says:


          Thank you for your comment and for the linked paper. I’d suggest their estimates are not relevant for the case of 100% solar which is what Roger’s post is about.

          The paper estimates the renewable energy penetration (wind and solar) with energy storage (Scenario 2) to be 71%, whereas Roger’s analysis is for 100% solar. As you know the extra 29% makes an enormous difference – perhaps doubling or tripling the LCOE to achieve the extra penetration while meeting system reliability requirements. At 71% RE penetration the paper estimates LCOE to be 28.5 EURct/kWh for Scenario 2.

          However, the paper is clearly RE advocacy as demonstrated by their emotive language and ridiculous claims. For example, from the Abstract: “huge potential for RE”, “expensive diesel”, “abundant RE resources”, “enormous potential”. How is a theoretical maximum of 53 TWh of electricity supply an enormous potential? The authors assume annual electricity consumption is 53 TWh which is just 0. 2% of 2014 global electricity consumption (22,433 TWh)

          My very rough lowball estimate above is $7.22 c/kWh for the solar system compared with the papers estimate of EUR ct 38/kWh for diesel. My estimate is low ball because it does not include owner’s costs, and IDC nor the cost to meet the system reliability requirements.

          The fact solar is not even competitive against diesel on an island demonstrates how far it is from being competitive with conventional power generation on mainland.

          • singletonengineer says:

            Comment understood. This is another project where opinion has outvoted analysis.

            The sad part of this is that those who will receive the benefit are not paying the bills – the public are, ie the people of the Solomons generally plus USA’s taxpayers. (“American Samoa Economic Development Authority, the U.S. Environmental Protection Agency and the U.S. Department of Interior.”)

            Since solar are now stated by many to be mature technologies, how can Australian MRET’s, regulated feed-in tariffs, subsidies and so forth be justified? The time for these rorts to come to an and and for them to stand on their own feet.

            I wonder what the islanders would say if offered the difference between a least cost system and the whistles-and-bells hybrid system as cash in their hands? But it is OPM (Other People’s Money), so it doesn’t matter!

        • Peter Lang says:


          Thank you. I agree with all points in your last comment.

    • OpenSourceElectricity says:

      It’s a states project, interest rate is 2% then. Capital cost shrinks from 800.000 to 160.000 in the first year then.

  20. sod says:

    Well, i think that it is obvious, that these kind of projects are a real alternative to island by now.

    It is also obvious that most islands should start the cheap part of the project (solar, perhaps even some wind) NOW, as they would start saving money at once.

  21. singletonengineer says:

    OSE, please clarify what you are trying to say.

    The figures for capital cost printed thus far on this thread range between 8, 10 and 15 million dollars.

    800,000 what?

    What do you mean by “capital cost shrinks”?

    Besides which, what country provides finance to its projects at 2% over a decades-long period? I’d be surprised if there is any.

    • gweberbv says:


      even Italy is below 2% for their 10 year bonds. German 20 year bonds are below 1%.

      • singletonengineer says:

        Today’s government bond rates are not comparable to the discount rate which is appropriate for economic evaluations.

        Indeed, there is no direct link between the two. For example, the current discount rates applied by NSW and Victorian governments to projects is around 7%. The federal government bond rates are below 2%. The difference between the two is the Principal’s decision as to how to allocate limited funds, risks and opportunities.

        My apologies for not presenting this more elegantly, but the truth is that, in both public and private businesses, the base cost of capital at today’s rate has little or no bearing on the discount rate which must be used in evaluating projects.

        For example, private corporations typically need their money to “work harder” and, often, to achieve much shorter payback times than a government might expect. Under current economic circumstances, private discount rates in the range 10 to 15% and payback times of 4 or 5 years might be typical, with some even tighter.

        Now what does that do to this project?

        • OpenSourceElectricity says:

          Well, the 5-7% you add to the interest rate are called “risk” and”profit” in usal calculations.
          “risk is project dependent, ant is very low in this case, since the likelyhood of the sun never shining or the people stopping to consume electricity is very low. This is different with many other projects. In usual, more risky calculations we have 1 or 2% for this powition.
          “Profit” – there is no real need for a state to make profit, so usually the investments are calculated without profit to get the real costs of the investment, and then the projects are compared by the amount the returns are above the investment+interest+risk.
          If you compare the spendings of a investment with 7-8% profit added on top of the capital costs every year, while you add zero profit on the “do nothing” alternative you compare appels with coconuts.

          • singletonengineer says:

            Nonsense. That does nothing to link discount rate over the life of a project with instantaneous government bond rate. They are fish of a different species.

          • Peter Lang says:

            The World Bank and other aid agencies have traditionally used discount rates of around 10% for making decisions about which projects to invest in.

            “What discount rates are used by the World Bank
            Transport Sector 12% (14% in Peru, 15% in Philippines)
            Energy Sector 10% or 12%
            Education Sector 10% and cost effectiveness
            Environmental Projects None found with economic evaluation
            Health Sector 10% (mostly cost effectiveness assessment and few economic evaluations)
            PREM 4% (Argentina Documentation System)

            “Traditionally the World Bank has used 10% to 12% as the discount rate for all Bank-financed projects. This rate is but a rationing device for WB funds and should not be construed to reflect the cost of capital in borrowing countries. Task Managers are free to use higher or lower rates where warranted, as long as they provide a sound justification. A discount rate of less than 10% might be difficult to justify as most research has shown that the cost of capital for developing countries is higher than 10%”
            Handbook on Economic Analysis of Investment Operations ”

            The discount rate has been reduced recently for various reason (not economic) and they include a disclaimer. I’d suggest the current low growth, low inflation, low interest rate environment is temporary. Interest rates have not been as low as they are now in the past 3000 years – but the reasons for the low rates are another subject for another day. Therefore, we should use the long term rate of 10% for projects that are being advocated for reducing GHG emissions.

          • OpenSourceElectricity says:

            @ singetonengineer, since I do procects and calculations for various entities since more than 20 years, I can tell you that you write nonsense.
            @Peter Lang I do projects with the World Bank. They are doing business in developing countries with high interest rates, and not in industrialised countris with low interest rates.
            And they are dealing with high level of corruption usually, and with a significant likelyhood that the project will not be finalised, thus leading to significant higher
            -interest rates
            – risk rates.
            But no profit as obligatory requirement.

          • Peter Lang says:


            Thousands of people work on WB projects. That comment is meaningless. I worked on pre-feasibility, feasibility, design and construction of world bank energy projects and worked on projects on every continent except Antarctica during me career. I gave you authoritative information form WB. I have many others I could give you that all show discount rates of 10%. Australian Government Chief Economist’s Australian Technology Technology Assessment Report which compares electricity generation technologies for federal government policy analysis: .

            I explained that current low growth, low inflation, and low interest rates are and aberration and not relevant for long term projections such as thosefor justifying policies to reduce GHG emissions. Calculation of SCC go out 300 years to get the numbers to justify GHG mitigation policies.

            Your comment to SingletonEngineer is uninformed. He is correct. Apparently, uou do not understand what you are talking about.

          • OpenSourceElectricity says:

            Peter. I know very well about what I am talking and I know where you ans singletonengineer are wrong.
            Calculating 10% in developing coutries where worldbank works is correct in average. It is nonsense in developed countries. The same rate in germany Austria etc, is 3% and fits to the situation since many years.
            The reference you provided did not lead to any discount rates I could find. But to disclaimers that there is no conclusion allowed from their model data to real life project costs, and also that financing costs vary a lot betwenn counties, although they only included industrialised countries in their overview, excluding the developing countries with their totally different conditions.

          • Peter Lang says:


            You can’t even read. You’re a waste of time.

          • OpenSourceElectricity says:

            I can not read what is not there.

          • Peter Lang says:


            “To ensure consistency in the comparison between technologies, and as a result of consultations with the Stakeholder Reference Group, a discount rate of 10 per cent has been applied to all technologies.” [p22]

            “Typically, the discount rate can be used to account for some of these differences in risk with a higher discount rate applied to the ‘riskier’ projects. For ease of comparison, however, a common discount rate of 10 per cent is applied for all technologies.” [p24]

            Clearly your reading and research skills are not up to much. I certainly will not trust you on anything you state in future.

          • singletonengineer says:


            We are still waiting for your clarifications of unintelligible comments made two days back. Did you read and understand them?

            1. “OSE, please clarify what you are trying to say.”
            2. “The figures for capital cost printed thus far on this thread range between 8, 10 and 15 million dollars. 800,000 what?”
            3. “What do you mean by “capital cost shrinks”?”

          • OpenSourceElectricity says:

            @ singletonengineer – good

        • gweberbv says:

          Thank’s for the explanation. Now I understand.

          However, but such standards the only decent investment on these islands would be to relocate the population to the mainland.

        • OpenSourceElectricity says:

          @ Peter, the text you reference is not on the webpage you reference. correct your link,
          And the text ypu reference explains that the discount rate is not choosen based on the characteristics of the technology in view or on the situation in the country where the project is located, but as a not further justified value for comparison only.
          We do not discuss here for comparison only, but the discussion is about the value of a specific technology at a specific place. Randome numbers do not help with this discussion.
          You could simply check your “discount rate” assumptions, using the nrel-calcuulator Robertoko referenced. Fill in the known construction costs of a solar power project, and see if you get the agreed power purchase price (+Tax credit in the US), if this does not fit together your discount rate is wrong, nit the reality. (It does not make any sense to use a discount rate which the economy does not use in practice for such projects)

          @ singletonengineer, the project costs are stated with 8 Million$, so I take this number for calculation. The other numbers are wild guesses, but would make the deviation only bigger.
          And if you remove very high profit rates included at random in the capital cost calculation, the capital costs shrink to a realistic value. Profit and capital costs are two different things.

          • singletonengineer says:

            Nonsense again from OSE, who has refused to be rational and has failed on numerous occasions to answer reasonable questions. Indeed, I very much doubt that he has the qualifications and experience that he claimed upthread.

            “very high profit rates” – OSE are absolutely wrong about this, as has been demonstrated. Where does OSE get this nonsense from? Does he know the difference between capital cost and profit? It seems clear that he does not.

            Contributors cannot expect their comments to be persuasive when their numbers are picked out of their heads. OSE has been provided with both authoritative references and rational discussion, yet he continues to make irrational and unjustifiable assertions.

            I wonder…is the intention to derail this discussion?

            Back to the two paragraphs of the most recent post:
            Paragraph one. OSE agrees that his “other numbers” (not fully explained) are wild guesses.
            Paragraph 2: OSE arbitrarily and without evan an attempt at explanation reduces the project’s capital costs from at least $8M, possibly more than $15M, to $800,000 on the pretext that this is “reasonable”.

            I’m out of here.

  22. singletonengineer says:

    King Island’s new abbatoirs didn’t happen. So, no biofuel.

  23. Kees van der Pool. says:

    Singletonengineer and Peter, thank you for your lucid comments, much appreciated.

    Regarding Open Source: I think he used to participate on this blog under the name of ‘hfrik’ with comments that did not make any sense. As OSE, they still don’t.


    • OpenSourceElectricity says:

      Kees thank you for you valuabel technical and engineering contribution.

      • Kees van der Pool says:

        Any time, Open, and from my side, thank you for your insights on inertia and inverters:

        “It would be nice if kees would finaly understand whi “instant” for the inertia of synchronus genertors is not as instant as he thinks, but needs a substantial time in millisecond till the needed angle between stator and rotor field is developed”
        October 7, 2016 at 9.32 am


        “@ Kees – problem is that you do niot understand that the inverters can do the same thing [as inertia] just
        – faster
        – with more energy.
        They do not have to wait till a speed difference between field and iron has created a certain angle difference which results in a power output. And they are not restricted to use just a tiny speed difference, they can extract much more energy from the rotation”
        November 11, 2016 at 8:49 am


        • singletonengineer says:

          How many of these inverters that can emulate inertia were installed on the hundreds of wind turbines in SA?

          I have been told that the answer is: Nil. Not one.

          Having an answer is a long way from implementing a solution.

          And why was there none? At least in part, because wind enthusiasts and their corporate backers have nil interest in solving the problems that they cause.

          • Kees van der Pool says:

            Hi Singleton,

            Inertial response as proposed by OSE through inverters is impossible. There is nowhere near the power available and it is not instantaneous.
            Moreover, despite OSE’s contention that the inertial response of a motor/generation is not instantaneous, it really and truly is.

            One certainly hopes that it was not missing inertia causing the mishap in SA. It is relatively easy to add – one common way is by means of synchronous condensers (synchronous capacitors), essentially a synchronous motor/generator running at gridfrequency. Sometimes, generators from decommissioned fossil power plants are re-purposed as synchronous condensers. It increases reactive power in response to a system voltage drop while also supplying the local system with significant short circuit support.


          • Greg Kaan says:

            OSE seems to have confused synchronous inertia with ramping. Inertia has no lag since it is the tendency for the synchronous generators to continue supplying the same frequency and voltage via the mechanical rotational inertia of the generator/turbine assembly.

            Inverters can react more quickly than a synchronous generator can ramp but the problems are the energy source backing up the inverter plus the nature of their power delivery. The doubly fed and full inverter wind turbines can ramp output very quickly if they have synthetic inertia circuitry to detect the frequency drop in the grid that they are connected to but the effect is to raise voltage rather than maintain frequency. The increased voltage then reduces the current draw from the synchonous generators which increase speed so the frequency support from inverters is indirect.

            Additionally, the wind turbine rotor slows dramatically when providing synthetic inertia (since the rotating energy of the rotor is the source for the increased power output) and eventually the the rotor stops at which point, the turbine cannot generate any power. So there is a longer period of reduced or zero power output from the turbine after providing synthetic inertia while the rotor is brought back up to speed by the wind (the recovery phase).

            A battery powered inverter is limited by storage technology and cost and has an even longer recovery period where it is parasitic.

            At best, the inverter power sources can extend the period for grid operators to shed load. This is a worthwhile capability but let’s not pretend they can hold the grid up against significant generation shortfalls.

          • Kees van der Pool says:


          • Werner says:

            Typical overload factors for inverters is 3-5 times for 5 seconds. Batteries can deliver much higher power output above 3C, which would result in many MW in this example.
            A synchronus generator with a load angle of 19,2° at nominal load would go unsynchronised at a overload factor of 3 no matter how high his inertia is.
            Better have some distance to the synchronus generator when this happens, it induces very high mechanical loads.
            By the way, there is no such direct output of a generator as frequency it is always voltage and current. Synchronus generator increases current depending on load angle. Dropping frequency causes the load angle gradually to increase, since rotor keeps rotating with original frequency. This increases the induced current, which leads to increased power output, which stabilises frequency.

            @ Greg, the steam turbine would also take some time when you extract so much energy that it stands still. Just that before you reach that point the grid frequency would be quite far away from 50Hz.
            Both turbines then have to be accelerated again by either steam or wind, if they have been curtailed before the power input can be increased, a coal power station can increase input by 4%/minute when well designed, the wind power generator can do about 4%/second.
            If both have been running at maximum output it takes a time of lower output to reaccelerate. MAin difference remaining: the maximum power of the coal power plant is fixed, the maximum power of the wind turbine is depending on wind.

          • Kees van der Pool says:

            Thanks Werner!

  24. stone100 says:

    What is the electricity used for? If much of it is used for cooling, perhaps underground cold stores (perhaps linked to a heat pump system) might be more effective than electricity storage.

  25. stone100 says:

    There was a discussion above about discount rates and the cost of capital for funding energy projects. A 10% discount rate may be appropriate if the financiers are at risk of losing their money. It makes sense to have a 10% discount rate if a project is being built by a private company and there is some risk that the company will fail to complete the project and so suffer a total loss. Once a project has been completed and is working fine then it makes sense for the system to be bought outright by the government or a utility company. The cost of capital for that purchase will be very much lower. If it is the government buying the system, then the cost of capital is just long term treasury rates and that is just a consequence of the central bank’s monetary policy choice. If it is a utility company buying, then there is the added political risk as to whether they will continue to have a market for the electricity but it is still very far from the discount rate appropriate for financing the initial construction. This is why it is so very misguided that our government has a lifetime discount rate financing approach to nuclear power. They should instead guarantee to buy outright nuclear power plants as soon as they are built and working well. We would then have a gilt-rate discount rate after the construction phase (ie typically no more than inflation). By all means then lease the power plants back to utility companies (eg EDF) if that’s who we want to run them.

  26. singletonengineer says:

    Werner, you perhaps are confusing the weighted average cost of capital (WACC) and discount rate.

    Discount rates are whatever the financier/owner decides that they will be, in order to obtain an attractive return on investment. If, in my experience, the government can borrow at X% but decides that the discount rate used for project evaluations is 2X%, then that is what it is. This is for a number of reasons, including that available capital is limited. Only the most attractive projects will get approved.

    Besides which, if today’s capital cost is X%, what is the rate going to be in 5 years’ time? 10? 20?

    There is no simple link between the discount rates and WACC and cost recovery rates and so forth.

    I have, in a stable AAA-rated government organisation become used to 7 or 8% plus a requirement that projects reach break-even within as little as one year and an upper limit of perhaps 10 years.

    Besides which, several writers here have argued that only the interest must be repaid. The project must also return the capital.

    It seems to me that argument at the lower end of the range is doomed to fail – simply, there are always better projects to invest in than ones returning 2 or 3 or 4%.

    8 to 10%, even for government projects, and especially given the long project duration and thus lengthy exposure to market forces, seems to me to be much closer to reality. Anything less amounts to shifting the risk onto the taxpayer as well as being commercially unrealistic.

  27. Wookey says:

    Solar in a sunny place like this is about 35-45g CO2/kWh. Don’t know how much the batteries add. The previous diesel setup will be more like 800g CO2/kWh.

Leave a Reply