Karpowership – Is Eskom Taking us for a Ride Again?

Eskom is on the verge of signing a 20-year deal for Karpowership to provide 1220MW of ship-based power for 20 years at 3 locations around South Africa.  The contract does not include the land side costs of connecting to the grid i.e., switchgear, transformers and distribution lines.  It also probably does not include the wharfage costs payable to Portnet.

Main picture: Gas turbine

I am in favour of the use of gas turbines, but I fear the Karpowership deal is overpriced by about 50%.  For this specific solution, did Eskom get competing bids for other ship-based gas turbines as well as equivalent land-based turbines?  I doubt it.  On that basis alone, the project should not go ahead.

Although Eskom is loath to give specifics of the costs, apparently the cost of the electricity is 150 cents/kWh.  This is excessive and I object strongly.  My generous estimate of the cost is 98c/kWh.  Is this excess cost due to lack of competitive bids or because Eskom requires 41% black ownership of the deal which is a farce as no Black consortium can put up their share of the finance or contribute technically to the solution.  Leaving that aside for others to chew on, I will only look at the cost of the project.

For the purposes of calculation, I will assume a 1000MW plant which allows the numbers to be simply scaled up or down to compare to other installations.  I have also assumed a 35% thermal efficiency and a 50% plant utilisation rate.  On the financial side, I have assumed an exchange rate of R15/$ and an interest rate of 8%.

Calculation of Costs

Amortisation, Staff and Maintenance Costs

A comprehensive article[1] on construction costs gives a total plant cost of $530/kW in 2008 dollars.  The dollar and the technology are both very stable, so I have assumed a 10% rise to give a current cost of $583/kW.  In today’s Rands, a 1000MW plant would cost R8.6bn.  This is not unreasonable as scaling it up to Medupi size would yield a cost of R42bn and it is far, far simpler.  In another comparison, Wikipedia quotes the combined cost of the Ankerlig and Gourikwa gas turbine plants at R3.5bn in 2006/7 when the exchange rate was R6.50/$.

The number of kWh generated per month = 50% * 24hrs * 30days * 1,000,000kW = 360×106 kWh/month.

Paying off the capital cost over 20 years in 2021 Rands requires R8364/mth. per R1million for a total of R72million/month.

The amortization costs per kWh are:  R72x106 / 360×106 kWh = R0,2/kWh = 20cent/kWh

Gourikwa which produces 740MW has a staff complement of 45.  Scaling that up gives us a staffing level of 61 which, at a generous cost to company of R750,000pa or R62,500pm, gives us a monthly cost of R3,8million/month.

The staff costs per kwh are:  R3.8×106 / 360×106 kWh= R0,01/kWh = 1cent/kWh

To calculate the maintenance costs, I am going to assume that the total value of the maintenance actions over the 20-year life is equal to half the capital cost – probably a very generous assumption. 

This gives a monthly maintenance costs of R8.6×106 / 20 yrs. / 12mths per yr. = R35.8million/month.

Therefore, the maintenance cost per kWh is:  R35.8×106 / 360×106 kWh= R0,099/kWh = 10cent/kWh

Fuel Costs

Liquid natural gas is currently quoted as $2.63/MMBTU (Metric Million British Thermal Unit) (ex Henry Hub).  Let us presume that the cost of transport and storage adds 50% to the cost giving us $3.95/MMBTU – again very generous.  One MMBTU = 293kWh so the fuel cost would be 20c/kWh. 

Given the thermal efficiency, the plant would actually require 1/0.35 = 2.86 times more gas to give us a fuel cost of 57cent/kWh.

Summary of costs

Amortisation       20

Staff                      1

Maintenance       10

Fuel                      57

Total                     98 c/kWh

It might be a simplified analysis but there is enough fat built into the areas that are difficult to analyse quickly to give me confidence that it is an overestimation, if anything.  The question, then, is why is Karpowership so expensive.  We need competitive bids of land and ship-based systems to feel confident that we are not being taken for a ride yet again.

Can it be Done?

Open cycle gas turbine (OCGT) power stations are very simple to construct.  They essentially just consist of a bunch of power packs, typically of 150MW size.  The heart of it is the gas turbine which is just a beefed-up jet engine with the big fan stripped off.  Very little construction has to be done – this is essentially limited making the foundations.  Compare this to a coal power station where there are massive boilers, coal and ash handling equipment and cooling systems which, in the case of Medupi and Kusile, are huge dry cooling systems and not the simpler cooling towers which use a lot of water. 

We also have experience of successfully and quickly commissioning OCGT’s in the recent past.  Construction on the 2000MW Ankerlig and Gourikwa plants commenced in the beginning of 2006 and were completed in about 18 months.  In addition, the chances of messing up are minimal as the power packs are merely replicas of units used in multiple sites around the world.  They are virtually produced as production line items and are brought in complete (but in sections).

The other advantage is their modularity.  They are like Lego blocks that you can bring online as the project proceeds and can easily be expanded.  Also, breakdowns only affect 150MW at a time and not 800MW if a Medupi set needs maintenance/repair or up to 4800MW if there is wet coal or the conveyor belt breaks.


[1] TR122-09_GBL_Study_of_Equipment_Prices_in_the_Power_Sector.pdf

Leave a Comment.

*