This is a response by the Council for Scientific and Industrial Research (CSIR) to a media statement issued by Eskom on 10 January 2017, which makes reference to a methodology that was developed by the CSIR in 2015 to assess the fuelsaving effect of a new power generator in a constrained power system.
Value of a new power generator
Introducing a new power generator into the power system has following effects over the lifetime of
the new power generator:
Items 1 and 2 above have immediate cash effects on the existing fleet of power generators, and they affect the fuel and maintenance cost of this existing fleet and new power generators that are built
over the lifetime of the new power generator investment.
Item 3 is a long-term effect that stems from the change in required structure of complementary new power generators to meet expected demand when a single new power generator investment is introduced. This general logic is aligned with a number of publications that are often referenced in this discussion:
 Agora Energiewende, The Integration Costs of Wind and Solar Power, 2015, Available Online:
 International Energy Agency (IEA), Next Generation Wind and Solar Power, 2016, Available
 U.S. Energy Information Agency (EIA), Levelized Cost of Electricity and Levelized Avoided
Cost of Electricity Methodology Supplement, Technical Report, 2013,Available Online:
The U.S. Energy Information Administration (EIA) calls the lifetime value of a new power generator
per energy unit “Levelised Avoided Cost of Energy (LACE)”. The total lifetime cost of a new power
generator is expressed as “Levelised Cost of Energy (LCOE)”.
If the lifetime value (measured as LACE in R/kWh) of a new power generator is greater than its lifetime cost (measured as LCOE in R/kWh), then the new power generator can finance itself from
within the power system (tariffs will stay constant or can even decrease).
If the lifetime value (measured as LACE in R/kWh) of a new power generator is lower than its lifetime
cost (measured as LCOE in R/kWh), then the new power generator cannot finance itself from within
the power system at current tariffs and needs support (through tariff increases or from outside the
power system in the form of tax-based subsidies).
The difference between LACE and LCOE for different new-build options must then be compared in order to make a decision which new power generator to build.
It is worth noting that introducing any new power generator into an existing power system that has
very low existing operating costs (which can be the case for historic reasons, if the system is based
on fully depreciated, old plant that is cheap to operate, like coal or hydro) will require support, i.e.
increasing electricity tariffs. That is because a new power generator simply cannot compete with an
existing, fully depreciated fleet that carries basically only fuel cost. This is true for any new power
generator introduced into such a system, whether it is coal, renewables, nuclear or something else.
For example, the LCOE of Medupi and Kusile are between 1,05 R/kWh and 1,17 R/kWh, the tariff for
the recently procured baseload coal IPP project is 1,03 R/kWh. All of these numbers are higher than
the pure fuel cost of the existing Eskom fleet. The fuel saving value will therefore be lower than the
LCOE, and very likely for that exact reason the full LACE will also be lower than LCOE. As a consequence, electricity tariffs have to increase. The question is which new-build decision lets the electricity tariffs increase the least.
The logic is the same as to why a new car purchase cannot compete with the pure fuel cost of an old
car. If a new power generator needs to be built regardless, either because demand increases or because
existing plants are due to be decommissioned, i.e. supply decreases, then the fact that LACE is
lower than LCOE is irrelevant. A new power generator has to be built. The power generator for
which the difference between LACE and LCOE is least negative needs to be built in such a case.
In South Africa, fortunately, complicated LACE vs. LCOE considerations to assess individual investment decisions do not need to be made, because the country has embarked on an Integrated Resource Plan (IRP) approach to determine which and how many new power generators are cost optimal. An IRP optimises the entire new-build portfolio in one shot, as opposed to LACE vs. LCOE which assesses one individual investment decision and its effects on the power system over its lifetime.
CSIR Fuel-Saving Methodology
Scope of the methodology
Effect 1: is relatively easy to quantify if the appropriate hourly production data is available on aggregated level (i.e. per fuel type).
Effect 2: is more difficult to quantify, because the rescheduling of the existing fleet in a world with vs. a world without the new power generator must be assessed. Generally effect 2) is relatively small though.
Effect 3: is the most difficult to quantify because it requires a prediction over the entire lifetime of the new power generator in terms of its effect on the type of new power generators that need to be built.
The CSIR methodology quantified only effect 1). In doing so, the absolute minimum and undisputable
value of the new power generator (solar PV and wind in this case) was calculated.
This was clearly stated in the initial report:
“Generally speaking, the pure fuel-saver effect of renewables that was quantified in this study always
grossly underestimates the total financial value of renewable energy. The fuel-saver logic purely applies in the short-term and measures the effect of renewables on the already existing conventional
fleet. In the medium- to long-term, renewables together with relatively speaking inexpensive flexible
new-build options need to be compared with alternative non-renewables new-build scenarios.
Hence, this study underestimates the financial value of renewables not only because the methodology
and the cost assumptions were chosen conservatively, but more importantly because of the neglected
long-term effects of renewables on the power mix. This was however done on purpose, as the
study was meant to be based purely on actual data, without making assumptions on future developments. What the study therefore does is it establishes the floor below which the combined short- and long-term value of renewables in South Africa in 2014 will certainly not lie” .
Applicability of the methodology in a less constrained power system
In addition, the CSIR methodology furthermore only reliably calculates fuel-saving values in a constrained power system. This is because it is conducted as an outside-in analysis without knowledge
about the remaining reserve capacity of the existing coal fleet within a certain hour. The methodology
therefore assumes that whenever the Eskom diesel turbines are not operational, renewables
saved coal and only coal.
However, in a less constrained system, during many hours the diesel turbines will not be operational, but the renewables actually prevented them from being switched on in the first place, i.e. the use of diesel turbines was avoided. This is not measured by the CSIR methodology – hence it underestimates the diesel savings.
This is acceptable in a constrained power system situation (like 2015), because the number of hours
without diesel usage is small, but in a less constrained power system (like 2016), the methodology
must be expanded to measure the effect of diesel turbines not being switched on because of renewables.
Type of renewable power generators that were assessed by the methodology
Finally and most importantly, the operational renewable power generators in South Africa are those
of Bid Windows (BW) 1 and 2 for solar PV and wind (and some BW 3 projects that became operational
recently). BW 1 and 2 for solar PV and wind are by far the most expensive BWs that South Africa
will ever see for these type of power generators.
This directly results in a high average tariff paid to solar PV and wind in 2016 of 2.03 R/kWh (as per
Eskom’s media statement of 10 January 2017: R12.2 billion / 6.0 TWh = 2.03 R/kWh). The weighted
average tariff for new solar PV and wind in BW4 and BW4 Additional is roughly 0.80 R/kWh or 60%
less. The weighted average tariff for new solar PV and wind in BW4 Expedited is 0.62 R/kWh or 70% less (all in April-2016-Rand). This is summarised in the table below and based on information from
the Department of Energy .
Sunk cost of operational power generators vs. new investment decisions
The Power Purchase Agreements (PPAs) for BWs 1, 2 and 3 are signed by both parties, an Independent Power Producer (IPP) on the one side (as the seller) and Eskom’s Single Buyer Office (SBO) on the other (as the off-taker). These contracts have a 20-year lifetime, and although these tariffs are high, they need to be treated as “sunk costs” and are essentially the school fees that had to be paid in order to achieve the very low solar PV and wind tariffs of BW4 forward.
Clearly the decision on whether or not new solar PV or wind (or any other power generator for that
matter) should be implemented must not be taken on the basis of sunk cost, but on the basis of a
comparative analysis of different new-build options (which is what an IRP does).
While the operational solar PV and wind projects (of BWs 1 and 2) triggered tariff payments of
roughly R12-billion in 2016 and produced roughly 6 TWh in the same year, the entire BW4 solar PV
and wind projects (BW4, BW4 Additional and BW4 Expedited) will trigger tariff payments of merely
R6,6-billion per year while they will produce more than 9 TWh/yr. That means 45% less annual payments for 50% more energy compared to the currently operational solar PV and wind projects.
These new projects will therefore be almost cost neutral from a pure fuel-saving perspective (i.e. the
entire cost of the new car is similar to the cost of the fuel only for the existing car).
Of the R12-billion in 2016, almost 43% of that (R5,1-billion) was paid for solar PV projects of BW1
alone. These projects produced only 1,4 TWh in 2016, which is 23% of the total 6 TWh. This aspect in
itself shows how distorted the current total cost of solar PV and wind are by the very first (expensive)
projects. These expensive projects of the first BWs are not representative of the latest procured
projects of BW4 and their costing must not inform new investment decisions.
Summary on the applicability of the CSIR fuel-saver methodology and the meaning of the results
CSIR’s fuel-saver methodology measures only short-term effects
The methodology developed by the CSIR measures only the immediate fuel saving effect on
the existing fleet and not the long-term effects on new investments that an IRP will measure.
Methodologically it therefore always grossly underestimates the lifetime value of a new
Even these short-term effects are measured incorrectly if the methodology is applied without
The methodology was developed to be applicable during a time of a constrained power system
with high diesel turbine usage and load shedding. When the power system is less constrained
and diesel turbines are not operational most of the time, the methodology underestimates
the diesel fuel savings. Thus, it needs to be adjusted to give the correct fuel-saving
In 2016 only the most expensive solar PV and wind projects were operational
Only BW1 and BW2 projects of solar PV and wind were operational in 2016 (and a few BW3
projects for a few months). These are by far the most expensive projects – and they will be
for the next 18 years (the residual time period of the 20-year PPAs).
The true value of the expensive first BWs lies in the cost reduction they achieved
The real value in the projects of BWs 1-3 lies in the fact that without them the country would
not have been able to bring the cost for new solar PV and wind down to the latest achieved
0,62 R/kWh – which is 40% cheaper than new coal .
New projects of all BW4 are almost cost neutral from a pure fuel-saving perspective
While the operational solar PV and wind projects (of BWs 1 and 2) triggered tariff payments
of roughly R12 billion in 2016 and produced roughly 6 TWh in the same year, the entire BW4
solar PV and wind projects (BW4, BW 4 Additional and BW4 Expedited) will trigger tariff payments
of merely R6,6-billion per year while they will produce more than 9 TWh/yr. That means 45% less annual payments for 50% more energy compared to the currently operational solar PV and wind projects. These new projects will therefore be almost cost neutral from a pure fuel-saving perspective (i.e. the entire cost of the new car is similar to the cost of the fuel only for the existing car).
It would be very beneficial for the quality of the discussion if Eskom could publish the hourly raw
data that underlies the power system analysis, such that researchers can conduct independent peer
reviews and additional power system analyses.
This publication of power system data is common practice by Transmission System Operators (TSOs)
across the globe. For example, the European Network of Transmission System Operators for Electricity (ENTSO-E) has been publishing (on its transparency platform ), in real-time the 15-minute to hourly production and availability data of all power supply sources on an aggregated level (i.e. per fuel type) along with demand and transmission interconnector data in all European countries, markets and balancing areas. Planned and unplanned outages of large units are even reported on unit
Without this transparency about the operational details of the power system it is very difficult
to comment on final results of any study for any researcher, market participant and/or decision
 CSIR: Financial Costs and Benefits of Renewables in South Africa in 2014,
page 4, 10 February 2015.
Send your comments to email@example.com