Abstract
Glossary
- AGR - Annual Growth Rate
- BoP - Balance of Plant
- CCP - Combined Cooling and Power
- CHP - Combined Heat and Power
- DG - Distributed Generation
- JTI - Joint Technology Initiative
- kWe - kilowatt electric
- kWth - kilowatt thermal
- LPG - Liquefi ed Petroleum Gas
- PPA - Power Purchase Agreements
- R&D - Research and Development
- SECA - Solid Energy Conversion Alliance
- SGIP - Self Generation Initiative Programme
- UPS - Uninterruptible Power Supply
Summary
Large stationary fuel cells are units above 10kWe operating in grid tied or
off grid operations, as CHP, CCP or operating as electricity generators. Over
the past fi ve years we have seen MCFC and PAFC become commercial, under
favourable subsidy condition;, the settling into three distinct size blocks
(10-20 kWe, 200-300 kWe, and above 1MWe) each targeting diff erent
applications; and the increased R&D eff ort in SOFC start to pay off .
During the last twelve months the companies involved in this space have been
operating in something of a business-as-usual mode with slightly increased
sales, the predicted increase in average unit size to the MW level realised
and the continued concentration on key markets such as California and
Connecticut in the US. Although we have yet to see the hockey stick curve of
units sold / MW installed kick in it is looking more likely that this will
happen around 2011-2012, around the proverbial next corner.
In terms of company developments, whilst no companies have exited this space
Siemens are purported to be selling off their SOFC business unit and HydroGen
have laid off two thirds of their workforce.
Market developments have seen the increased focus on distributed generation by
legislators and business planners alike with an increased number of units
being sold into offi ce blocks and schools. Also data centres, and their
bigger brother server farms, are starting to look at the technology for CCP
applications, with some potentially serious implications for fuel cell
companies.
Market Growth
Whilst our 2007 and 2006 surveys reported substantial market developments,
this is now translating into something of a business-as-usual scenario for the
sector. The number of units adopted has hovered around the level of 50 per
annum for the last three years whilst the number of installed MWe has doubled
during this period
Whilst this market has so far avoided the boom-and-bust model of some of other
technologies it does raise concerns over the industry' s learning curve ratio.
If we disaggregate the systems involved into fuel cell stacks and BoP then we
see a much steeper learning curve for the stacks than BoP. As units are
getting larger, many stacks are being linked in series to produce them so this
50 systems fi gure cloaks a much higher stack production fi gure, more than
double that of 2007. The problem is that the needed cost down is not just a
function of stack costs but includes the integrated system costs, which appear
to be experiencing a much slower cost reduction ratio.
Looking forward we expect to see a slight market upturn next year with the
order books for the major players such as FuelCell Energy and UTC Power
already fairly full. We forecast the anticipated hockey-stick S-Curve to start
after 2011. If we turn to MWe installed we can see the step change in
installed capacity this year. This has happened due to the predicted increase
in average unit size. As the average size unit is now 1MW, as shown in Graph
4, although the number of systems sold is steady the MW installed is
increasing.
This year the market segmentation has become even clearer with 3 clear size
classes now available. The 10kW units are being targeted and developed for
data centres (more on this in the special section of the report later), 250 -
400kW for offi ce blocks, hospitals, prisons etc and 2MW+ class units for
power plants and, potentially, server farms. The exception to this is Fuji
Electric which is producing 100kW PAFC units which it aims to sell into medium
size supermarkets and community centres etc. During 2008 the number of verifi
ed shipments of units in the 10kW bracket was less than 5% of the total.
Turning to electrolyte mix: again this year we see a fairly even balance of
systems shipped with MCFC and PAFC taking the lion' s shares. Note that Graph 5
represents electrolytes of systems shipped, not MWe installed by electrolyte
type, which is shown in Graph 6 and highlights the dominance of large (>1MW)
MCFC systems.
Interestingly this is one application where we could see alkaline fuel cells
coming back into the frame. AFC Energy has a deal with Akzo Nobel to ship a 50
and a 200 kWe unit, contingent on further development. At present the
company' s technology has been tested for 5000 hours on a number of cells.
If we look at region of the fuel cell stack manufacture, as shown in Graph 7,
it is clear that North America is very strong with around two thirds of the
current, 2008, market. This market split has been fairly consistent over the
past few years and is anticipated to remain so unless companies such as Rolls
Royce and Nuvera, both European companies, decide to manufacture their stacks
in Europe and not at their North American plants. At present the Rest of the
World category has really only one serious player, ElectroCell (Brazil), which
is developing PEM technology for both small and large scale applications. BHEL
(India), which is also developing PEM technology, is currently at development
scale testing.
Over the next decade we expect to see the dominance of North America stack
production increase with the expansion of the current producers outstripping
the projected growth of those in Asia and Europe. This does not take into
account any new stack producers opening up in these regions, or any of the
main players leaving the sector.
It is interesting to note that this application has the highest sensitivity of
all the diff erent fuel cell sectors to company / commercial development. At
present this application has only ten companies capable of producing stacks
for products that are not classed as late stage R&D. Of these ten less than
half can be considered commercial and only around three quarters are focusing
solely on this sector. Going forward there is another ten or so companies
potentially coming on stream in the next fi ve years taking the market up to
around 20 companies. With the low number of companies operating in this sector
any fall out, in terms of company closure or sell off , has real implications
for the growth of the entire market. HydroGen for example, which has had a
positive year in terms of its market development with Samsung, has
subsequently also had to lay off two third of its staff and is in talks to
secure short term fi nancing. At the start of FY08 it was expected that within
twelve months Hydro- Gen would see the sale and shipment of two, 2MW units to
Samsung for use as power plants in Korea. Although this is only two systems
this 4 MW represents around 8% of the total installed capacity of 2008 as
shown in Graphs 1 and 3. SOFC units are coming increasingly close to market
with Phase 1 SECA targets being met. In contrast to this the statement given
by Siemens as to why it has put up for sale its SOFC business unit, which was
working on MW class fuel cell systems, is that it does not see this business
unit making the desired profi t targets by 2010. This short time horizon for
profi tability of a technology that is still in late stage R&D, primarily
funded by government money, is not only highly implausible but signalling, we
believe, some other reason for Siemens to once again withdraw from fuel cell
development.
Economics
When looking at the cost structure of large stationary fuel cells there are at
present realistically only FuelCell Energy and UTC Power who have costs that
are not one-off deals. Next year sees the launch of the 400kW PureCell from
UTC Power. The company has provided a clear market signal regarding the unit
by forward pricing it at US$1 million installed cost, or US$2500 per kW.
According to the company' s own calculations this brings the unsubsidised
electricity price in at US$12 cents per kW/h.
FuelCell Energy is currently reporting electricity prices from its DFC units
at US$15 cents per kW/h with a future targeted installed cost of less than
US$2000 per kW, unsubsidised. The graph below is taken from a FuelCell Energy
investor presentation showing the historic and projected cost reductions from
its units.
This all adds up to a massive market for power. What this hides though is the
uses of the power. As well as providing electricity to the servers a similar
amount is used on cooling them. A recent article on server farms in Fortune
magazine highlighted this saying that "a rack of "blade" servers can get as
hot as a seven-foot tower of toaster ovens. It gets hot enough that for every
dollar a company spends to power a typical server, it spends another dollar on
air conditioning to keep it cool." It is no surprise then that new server farm
developments, which unlike offi ce based data centres are somewhat location
independent, have a number of key criteria including access to cheap
electricity and increasingly cooler or at least more temperate climates.
Taking this into account Iceland is now actively marketing itself as an
excellent location for new developments in this market place.
As this is a large and growing demand for power then some companies are
starting to look at fuel cells as a potential segue around the issue of grid
saturation. At present most of the work is being done in the smaller rack
mountable UPS units such as those being developed by APC and Hydrogenics,
FuturE and Rittal. These units are designed to replace battery banks to
provide UPS power in enclosed spaces where fl oor space is at a premium and
emission requirements very strict. The units in the market place at present
are all PEM units and run off bottled compressed hydrogen, providing UPS
function in times of grid outage. This is just power provision and can be
classed as the low hanging fruit of this sector. The big money, big challenges
and big opportunity come from providing primary combined cooling and power
(CCP) to the server rooms.
Traditional cooling systems either burn a fuel such as natural gas or use
electricity to drive an electric motor used for operating a compressor used
for raising the pressure of refrigerant vapours. Absorption chillers on the
other hand use heat to provide the cooling. Although not as effi cient as the
traditional systems, when the heat is a waste or low cost product then
absorption chillers are being marketed as an eff ective solution. It is for
this reason that fuel cells operating as CHP units are now being seriously
looked at for CCP, linking an absorption chiller to the unit. To date CFC
Solutions, FuelCell Energy and UTC Power all have demonstration installations
of this technology linking.
Technology Fit Issues
Large stationary fuel cells off er great potential in this sector as they are
modular, so increased demand can be met a unit at a time, can run off a
natural gas grid, avoiding any issues of grid saturation of overload, are
(fairly) temperature independent, and can produce high or low grade waste
heat. But there exist two real issues: the fi rst is a technological one of
proven reliability, and the second is of the players being able to actually
meet any market demand.
1. Reliability - power availability to data centres, and especially to
server farms, has to be very reliable. Five nine availability is the norm with
power available 99.99999% of the time. Any outage can cost a company many
millions of dollars which is why generators and UPS systems are also
commonplace. For fuel cells to operate in this area the up time availability
would need to increase from its current level. Current projects, of which
there have not been that many, have reported availability of between 85 to
95%. This could be a function of the fuel cell not being designed or optimized
to work with chiller technology or at a more basic level an increase in stack
reliability is still needed for this sector.
2. Manufacturing Capability - the second market based issue is manufacturing
capability. New server farms can have up to a reported 125 MW of power demand
each. If only ten percent of this was to be met by fuel cells this would
represent 32, 400 kW stacks operating in series, or 42, 300 kW units. Current
manufacturing capability of FuelCell Energy is reported as being 25 MW per
annum and Ansaldo only 3 MW. With FuelCell Energy already contracted to supply
stacks to POSCO Power, CFC Solutions and systems to Marubeni, as well as
selling its own DFC units in the market place, any new market demand could see
a delay in supply of the units by up to a couple of years. Total manufacturing
capability in this sector at present is a maximum of 60MW per annum and is
supply constrained.
| 
View Cart
