The sun has been the source of most of our energy (except
geothermal). Wind & wave are secondary/tertiary effects of solar
energy. Even coal, oil, gas and other fossil fuels are merely stored
(i.e. time-shifted) form of solar energy locked in hydrocarbons. The
figure below shows the incident solar energy potential
Solar energy conversion refers to a number of mechanisms: solar thermal
(conversion to heat for solar cooking, or solar water heating, HVAC via
solar chillers, or very high temperatures for electricity conversion
via heat engines) and solar photovoltaic (PV, including
concentrated or thermo-photovotaics) are the two important categories.
Another important aspect is that solar energy conversion is modular,
i.e. can be done cheaply at at the distributed retail
(home/businesses), or community levels in addition to being provisioned
as large centralized solar farms, i.e. utility scale.
A Solar Photovoltaic (PV) system comprises the module,
wiring, and inverter system (that subsumes controls and grid-tie in
etc). Solar PV modules have no moving parts and last 30+ years (and are
warrantied for 25 years to produce 80% of original rated capacity at that time); and
inverters have 10-25 year warranties. Solar PV module costs have dropped
rapidly, as have balance of system (BOS) and installation costs. This
is depicted by the "learning" curve. As more forms / types and business
models for solar PV penetration happen the total number of units
installed and learnings from them will correspondingly grow. The measure
used here is capital cost per Watt-peak (eg: $/Wp) and is applied to
either a component cost (eg: module cost, inverter cost) or to the full
installed system. Today the installed costs for a solar PV system
(including inverters etc) has dropped to US$1.5-3.5/ Wp (note: sometime this does
not include costs for leasing the land/area for the system) in western economies, and US$0.8-1.5/Wp in India. The lower
end of installed costs are either in emerging markets like India or in
utility installations at scale. The higher end are for
distributed/retail installations. Interestingly, even high labor cost
economies like Germany and Australia have installed costs of $2/Wp or
less. For example a 4kWp system would cost under US$8000 installed. India has possibly the lowest installed costs of 76-80c/Wp at utility scale (as of 2015 end, as surveyed by CERC). Solar module costs at utility scale as of 2015 are appoximately 25-30% of total costs in USA, but 50-65% of costs in India.
Wayne Gretzky, the legendary (ice) hockey player once
summed up his secret-of-success: "I skate to where the puck is going to
be, not where it has been" ...Project out the trends of costs from the
above diagram, an imagine where the "puck" (i.e. costs) are going...
Think about the economic implications! We have to prepare for that world
we will be living in.
Coming back to solar.... Once the system is installed (i.e.
CAPEX incurred or costs are "sunk"), it has fairly low operational
costs (eg: periodic cleaning to reduce dust/soiling accumulation). What
is variable though is the (daily) "Energy Yield",
i.e. the normalized energy in kilo-watt-hours (or "units") per day
produced per kWp (kilo-watt-peak), i.e. kWh/d/kWp generated by the
system. Note that in the McKinsey graph later, they use annual energy
yield, i.e. kWh/year/kWp as the measure (multiplying the daily energy
yield by 365 will give the annual energy yield). The energy yield is a
function of the solar irradiance characteristics (function of latitude,
weather conditions (eg: cloudiness), dust / soiling / birds, and any
temporary or persistent partial shadowing), the operating temperature,
the solar PV technology (c-Si, poly-Si, CdTe etc), and how the solar
panels are wired w/ inverters (string vs microinverters) etc. [Note: A
grid tie system will generally not produce energy for safety (islanding)
reasons when there is a power cut, unless advanced inverters are used
to switchover and charge a battery system.This is a factor for
geographies with highly intermittent power supplies like India ]. The
energy yield in Bangalore, India is around 5-5.5 kWh/d/kWp (or about 4.5 kWh/d/kWp if you adjust for power cuts, uncleaned dust, aerosols etc) year round
while in Melbourne, Australia or London UK or New York, USA can vary
between 3.5 - 5 kWh/d/kWp, or an average of about 4 kWh/d/kWp. Here are
examples of annualized energy yield numbers: 3 kWh/d/kWp = 1095
kWh/y/kWp; 4 kWh/d/kWp = 1460 kWh/y/kWp; 5 kWh/d/kWp = 1825 kWh/y/kWp;
and 6 kWh/d/kWp = 2190 kWh/y/kWp.
Some solar calculators (eg: http://www.energymatters.com.au/climate-data/)
gives average solar irradiance in kWh/m^2/day (the web site seems to
have a typo - should be kWh and not kW), plotted on a monthly basis.
Once you have the solar irradiance, and the solar module/system
efficiency (eg: 15%), you can work out that a 1kWp system needs 6.67
m^2/kWp; therefore dividing the data by 6.67 gives a energy production
yield (i.e. kWh/day/kWp) on a month-by-month basis. NREL has a solar
resource map / data for international sites (eg: India: http://www.nrel.gov/international/ra_india.html and http://rredc.nrel.gov/solar/new_data/India/ ).
The India solar resource map (with solar irradiance in kWh/m^2/day) is
reproduced below, as are maps for Australia (1 MJ = 0.278 kWh. So 24
MJ/m^2/day = 6.67 kWh/m^2/day for instance). The colors are not directly
comparable across the maps (use caution!).
If you divide the (daily) Energy Yield by 24 (i.e. hours in a day), you get a measure called "Capacity Factor" (Cf),
which estimates what fraction of rated capacity you are actually
getting out of the system. For instance in Bangalore, the Cf is about
21-22%, and for Melbourne it is 15-16%. The capacity factor is used to
compare renewables with each other or with traditional energy generation
systems. For example Nuclear and Hydro tend to have high capacity
factors (80-90%+), and wind energy ranges from 30-45% (function of
location, turbine height, scale etc). Solar at 15-25% capacity factor
therefore is quite low. As an example of the economic impact of this,
consider if you are building a utility solar plant and plan a
transmission line to match the rated peak of the solar production, the
transmission capacity will be unused 75-85% on average (while the cost
of capital has to be borne)! This is why transmission lines in utility
scale solar are sometimes undersized, and if there is excess solar
production on the short term (that the line or grid cannot absorb), it
is curtailed (and wasted unless it can be stored temporarily and
time-shifted). What is remarkable, is that despite the low capacity factor of solar,
the impact of reducing costs, and modularity (that allows both
distributed, community-scale and utility-scale deployments), its growth
rate is tremendous and inexorable. The figure below shows that the
installed power normalized by the capacity factor is still growing to
overtake all other forms of renewable energy before 2020, and continue
tremendous growth beyond! This is why we should pay attention to Solar
since it, combined with wind (that is also growing rapidly) will change
the landscape of our renewable energy mix.
Before we get too excited, we should reflect that all
renewables still form a fairly small fraction of total energy capacity
installed, and energy produced on average, and has a large geographic
variation. This also points to a large upside in growth potential with
declining costs and ability to harness, package and manage the energy
generated by renewable options.
In conclusion, the final measure I will introduce is Levelized Cost of Energy (LCOE), i.e. cost/kWh, and the concept of grid parity. The
LCOE is a "cost" measures that "levelizes" or "flattens" the capital /
operating costs, and normalizes it over the energy yield i.e. kWh
estimated to be produced by the system, assuming a discount factor. We
saw that a 1 kilo-Watt-peak (kWp) solar PV system can offer an average
Energy Yield of about 4 - 5 kWh/day/kWp in many parts of the world. We
annualize the actual energy yield & discount it by the cost of
capital or discount rate for the denominator. Note that the discounting
is done over N periods (i.e. a fixed time horizon). Similarly the fixed
and variable costs of the solar system are discounted and put in the
numerator. This gives a "levelized" cost, i.e. cost / kWh or cost / unit
that can be compared against other forms of energy (eg: coal-fired
electricity, diesel-generated electricity, natural gas, hydro, wind etc).
The LCOE is a measure of "cost". Lets turn to "revenue" or "value" yield i.e.
how much the energy yielded converts into dollars of value. A unit
(kWh) of energy generated by the system can have an economic value
determined by the local utility price, or diesel cost or policies (eg:
feed-in tarriff or net metering). For example, in (averaged) net
metering, with a tiered tarriff structure (eg: California) where a heavy
residential user pays over 20 US cents/kWh at the margin, offsets or
saves that amount via solar. This is the "monetized" value of the solar
energy yield, i.e. by multiplying 5 kWh / d / kWp x 20 cents / kWh = 1
USD / d / kWp. In finance terminology, this is the "cash flow" from
solar on a day-by-day basis. Since solar PV produces energy over years,
the future "cash flows" has to be discounted by the "cost of capital" to
get a discounted cash flow (DCF). If we compare the actual solar discounted revenue yield to LCOE,
we can ask the question which is greater. If the revenue yield is
greater, we have achieved or exceeded economic parity / break-even.
One specific simplification is called "grid parity",
where you can compare the LCOE with the marginal cost of electricity
offset from the grid. For example if the marginal cost of (prior) tier 4 pricing
in California is 34 c / kWh (also see McKinsey graph below, please note it is a little dated already) and LCOE is
13 c / kWh, we say that solar generation is cheaper at the margin, and
therefore has achieved grid parity in that location or region. Companies like SolarCity, SunRun, Vivint offer a third-party owned service where a homeowner can get solar for their home with zero up front costs, as long as they can sign a 20 year PPA around 13-15 c/kWh in 2015. Financial
analysis are also done for payback periods, NPV, IRR and other metrics.
These financial metrics essentially compare revenue and costs, and ask
how quickly we recover costs (payback period), or the rate of return
(IRR) or net value (discounted revenue minus discounted costs, for NPV).
McKinsey has a nice graph (a little dated now) that plots installed cost of solar ($/Watt-peak), solar (annual) energy yield kWh/kWp and the retail power price. It shows how different countries (size) and pricing structures indicate whether they have achieved grid parity or not. For example Germany has low energy yield (left of the graph), but due to a high retail price (or equivalently feed-in tarriff), it has achieved grid parity. India has much better solar resource, and is a large market, but its retail prices are lower. Notably, this graph shows that China, India (at their base rates) are still away from grid parity due to a combination of cost of capital and low retail price, and therefore, either the solar cost has to drop (to $1/Wp installed) or there needs to be policy/subsidy support to stimulate the market. But recent market information suggests the $1/Wp installed cost point has been reached at utility scale. The graph also shows that at current rate of cost declines in which year one can expect the "cross-over" i.e. grid parity in different markets. The graph also shows approximate grid parity price of 15 c / kWh today (the transition between light blue and dark blue). The grid parity price in markets like India is now 9-10c / kWh at the retail level (i.e. Rs.5.8 - 6.5), and around 5 c/kWh at wholesale (Rs.3.5-4/kWh). Recently two companies (SunEdison and SB Energy) have bid Rs. 4.63/kWh (or 7 c/kWh) in India to win several 100 MW of solar in reverse auctions, i.e. we are very close to grid parity in India today at both the wholesale and retail levels.
To summarize in layman's terms, as the capital cost of
solar rapidly declines, the Io in LCOE numerator declines rapidly. The
annual costs Ai tends to be low for solar. The LCOE formula is also a
function of N (the amortization period). Energy yield is variable
function of the efficiency of the installation, local irradiance that
falls on the system, and any non-linear effects due to shadowing,
soiling etc. It also assumes that solar production is not curtailed or
wasted due to inverter, grid / load availability constraints. In some
cases, there may not be a feed-in tarriff policy in some regions, i.e.
any solar production that doesnt offset local demand is donated to the
grid for free.
Subject to these subtle and operational issues, we can see that innovations
to reduce capital costs (Io) or increase energy yield (Et) are the key
to bringing solar to grid parity without subsidies. Also note the important role played by the "cost of capital" or discount rate i (or r).
Financial engineering innovations have been a big part of solar
companies work to make solar affordable. For instance in the US,
financial innovations have allowed the cost of capital to drop from 20%
in 2008 to 4-8% today. This has a huge impact on afforability of solar.
We now need to make similar innovations at the technical and financial
levels to bring solar affordably to the emerging markets and the poor.
No comments:
Post a Comment