Vous êtes sur la page 1sur 11

international journal of hydrogen energy 34 (2009) 7114–7124

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Lessons learned from infrastructure


operation in the CUTE project

K. Stolzenburga,*, V. Tsatsamib, H. Grubelc


a
PLANET GbR Engineering and Consulting, Donnerschweer Strasse 89/91, 26123 Oldenburg, Germany
b
BP International Ltd, Chertsey Road, Sunbury on Thames, Middlesex TW16 7LN, United Kingdom
c
Vattenfall Europe New Energy GmbH, Überseering 12, 22297 Hamburg, Germany

article info abstract

Article history: The project CUTE (Clean Urban Transport for Europe) comprised the operation of 27 fuel
Received 13 November 2007 cell buses and their hydrogen infrastructures for two years under everyday conditions.
Received in revised form This paper focuses on the experiences with on-site hydrogen infrastructure operation in
16 June 2008 CUTE. Every site had a hydrogen refuelling station (350 bar rated pressure), and most of the
Accepted 21 June 2008 facilities included a plant for generation of the fuel (steam methane reforming or water
Available online 19 September 2008 electrolysis). Qualitative and quantitative performance indicators such as system avail-
ability, root causes for downtime and hydrogen losses are discussed. Problems during the
Keywords: operating phase are outlined and challenges for future work are highlighted.
Hydrogen in transportation In general, the station units and on-site electrolysers performed well. The on-site steam
Hydrogen infrastructure reformers did not. Hydrogen compressors and dispensing equipment turned out to be the
On-site hydrogen generation most critical components of the station units.
Water electrolysis ª 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights
Steam methane reforming reserved.
Hydrogen refuelling stations
Assessment framework
Technology validation
Performance indicators

1. Introduction Stuttgart, and Stockholm. CUTE was the most ambitious field
trial of hydrogen and fuel cells in public transport worldwide.
1.1. Project outline and scope of this paper Transportation with its ever-increasing demand for crude
oil contributes to climate change significantly. On the other
The CUTE project aimed at pollution-free local public transport hand, hydrogen – from renewable energy sources – and fuel
and at reducing the dependence on crude oil. Its main objective cells have been identified among the portfolio of technologies
was to demonstrate fuel cell buses and their hydrogen supply that can address the emerging energy and climate change
infrastructures, proving their reliability in regular service. 27 problems. From this perspective, CUTE can be regarded as
buses were running in nine European cities, three per site. This a milestone on the way to sustainable mobility by displaying
ensured a wide range of operating conditions in terms of the feasibility of these novel technologies at significant scale.
climate, topography, and traffic. The nine cities were: Amster- The project comprised two phases: Design and construc-
dam, Barcelona, Hamburg, London, Luxembourg, Madrid, Porto, tion took place from November 2001 until May/December

* Corresponding author. Tel.: þ49 441 85051; fax: þ49 441 88057.
E-mail address: k.stolzenburg@planet-energie.de (K. Stolzenburg).
0360-3199/$ – see front matter ª 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2008.06.035
international journal of hydrogen energy 34 (2009) 7114–7124 7115

2003, depending on the site. The operating phase then lasted Given that Germany is a ‘‘frontrunner’’ in the European
for two years in each city. The project followed a comprehen- Union in the amount of hydrogen research and development
sive approach, covering also environmental and economic projects, as Hekkert et al. observed, the situation is likely to be
studies, training of staff as well as education and dissemina- similar throughout the EU. This would explain the lack of rele-
tion programmes. vant publications, at least from this region. On the other hand, it
This paper concentrates on aspects of the on-site hydrogen raises further questions as to whether there is a solid back-
infrastructure. More information on the overall project and its ground to the view expressed in many of the above studies that
results can be found in the final project report [1]. Saxe et al. hydrogen (refuelling) technology seems to be ‘‘in place’’.
carried out an energy system analysis of the fuel cell buses [2]. This paper provides a factual basis of results and learnings
A life cycle assessment study was performed by Faltenbacher from refuelling infrastructure operation. The findings are not
[3]. A report on the attitude towards hydrogen fuel cell buses materially impacted by the circumstance that the infrastruc-
was compiled by Saxe et al. [4]. ture was designed to serve fuel cell buses rather than cars. To
The scope of this paper is not to give a full account of all the contrary, as buses from captive fleets put higher daily
operating details of each of the sites but to provide an over- loads on the stations, this can help to make the strong and
view of findings and learnings, and to compare performance weak points of the installations more visible.
figures. Furthermore, conclusions for future complex multi-
site demonstration activities are drawn. 1.2. Hydrogen infrastructures: objectives and
Girón [5] presented a thorough evaluation of the hydrogen considerations
refuelling installation in Madrid.
Numerous studies of hydrogen infrastructure for the The main project objectives in the infrastructure domain
transport sector have been published. Many of them focus on comprised:
strategies for the build-up of infrastructure for large numbers
of fuel cell cars, often in a particular country or region, and on  Building and operating a hydrogen supply system in each of
the cost involved (for example Melaina for the US [6] and the nine cities following individual approaches
Mercuri et al. for Italy [7]). They frequently aim to contribute to  Ensuring a reliable fuel supply to the buses in terms of
overcome the chicken-and-egg problem of ‘‘no hydrogen quantity, pressure and purity
infrastructure, therefore, no demand for hydrogen vehicles’’  Learning about the strengths and weaknesses of the
and ‘‘no hydrogen vehicles, therefore, no incentive to build installations and their components.
a hydrogen infrastructure’’. Apart from this rather global
perspective, there is also some literature concerned with Fig. 1 shows a generalised schematic of the CUTE infra-
developing infrastructure at a project level and with vehicles structure facilities. Hydrogen was generated on site or supplied
other than cars, in particular buses, such as Shayegan et al. [8]. by truck from external sources. It was compressed, stored, and
Agnolucci [9] has recently presented a paper that reviews on demand dispensed to the buses. Table 1 gives an overview of
about 30 quantitative studies with regard to the approaches the nine installations. Details are explained in the following.
that authors take in terms of infrastructure cost and layout. Most of the infrastructure facilities were located in a bus
He claims that technological issues can be accommodated depot. Some bus operators owned and managed the infrastruc-
within the existing technical knowledge and industrial prac- ture themselves. Others partnered with a third party, usually an
tice, and that cost is the main obstacle. Other authors take energy company, and only purchased the refuelled hydrogen.
a similar stance. Baykara [10] judges most hydrogen energy Hydrogen refuelling installations could hardly be bought
technologies as matured enough for implementation. The off the shelf from one supplier, in particular, when they
view that technology is not a major issue also seems to be included a unit for on-site hydrogen generation. Therefore, it
shared by authors who have modelled hydrogen infrastruc- was a key consideration at most CUTE sites to select an energy
ture at the project level, such as Joffe et al. [11] and Vidueira company/turn-key supplier that took responsibility for the
et al. [12].1 design and erection of the entire plant and chose appropriate
However, there is little evidence in the literature that components according to the local requirements.
hydrogen infrastructure technology is in fact as mature as it is When the plant was situated in the bus depot, it was
assumed to be. Detailed experiences, learnings and quanti- important that there was no or little interference with day-to-
tative findings from the operation of hydrogen refuelling day business during installation and operation. Accordingly,
infrastructure do not seem to have been published recently. components were pre-assembled as much as possible and the
Hekkert et al. [13] have investigated research projects and major units arrived skid-mounted or containerised at the site.
patents related to hydrogen technology for mobile applica- Space was scarce at most depots, so solutions with compact,
tions in Germany over the period 1974–2000. One of their key modular design had to be found.
findings is that infrastructure and refuelling attracted rather Most sites opted for on-site generation of hydrogen (Table 1),
little interest in terms of the number of projects carried out many of them with the alternative to employ external backup
and of patents filed. delivery when necessary. Amsterdam, Hamburg and Stock-
holm solely used renewable energy for hydrogen production.
1 This power came from remote sites via the public grid. Barce-
Vidueria et al. showed the feasibility of PV autonomous supply
of hydrogen for the CUTE buses in Madrid. This system was not lona had a photovoltaic array on the station roof but this could
implemented, though. Reforming of natural gas was employed only contribute a small share of the power consumed. Most of
instead [5]. the power supply was based on the Spanish grid mix. All other
7116 international journal of hydrogen energy 34 (2009) 7114–7124

Fig. 1 – Generalised schematic of the CUTE hydrogen infrastructure facilities.

sites entirely relied on the power grid mix. In the cases of Madrid 120 kg to be supplied by the infrastructure facility daily. This
and Stuttgart, this also applied to the delivery of natural gas corresponds to 1335 Nm3 (normal cubic metres). An hourly
(hydrogen generation from steam reforming). production rate of about 55 Nm3 could cover this load.
London was the only site with external supply and on-site
storage of liquid hydrogen (LH2). The hydrogen was vapour-
2.1. On-site electrolysis
ised before entering the bus tank. The London hydrogen
station was located far from the bus depot to enable placing it
Table 2 summarises the design data of the units employed in
on the premises of a conventional petrol station.
CUTE. They all were pressurised alkaline water electrolysers.
The volumetric energy density of hydrogen is only about
Such electrolysers are produced in small numbers by several
30% that of natural gas, resulting in a reduced vehicle range. In
manufacturers. They display high load flexibility. Operating
order to (partly) offset this drawback, an increased tank
temperature is in the range of 70–80  C.
pressure compared to natural gas vehicles was necessary.
This, in turn, also required a step forward in refuelling tech-
nology. CUTE was the first major trial which followed the 2.2. On-site steam reforming
350 bar concept, compared to 200 bar for natural gas (and
previously also for hydrogen). Steam reforming of methane (natural gas) to produce
hydrogen and, as a by-product, carbon dioxide has been
a standard industrial process for decades, with production
2. Characteristics of production units and rates up to 100,000 Nm3/h and efficiencies of more than 70%.
station units Small-scale units as required for CUTE were only at prototype
stage during the planning phase of the project.
The maximum load that had to be covered in terms of refu- Table 3 summarises the design data of the reformer units
elling was about 40 kg per bus per day, therefore, requiring in CUTE. These small devices possessed significantly lower

Table 1 – Characteristics of the CUTE infrastructure facilities.


Site Location of the facility Hydrogen supply External backup Power Refuelling
supply foreseen source type
(only relevant for
sites with on-site
generation)

Amsterdam Bus depot On-site electrolysis No Green Overflow/booster


Barcelona Bus depot On-site electrolysis Yes Grid mix þ Overflow/booster
PV on site
Hamburg Bus depot On-site electrolysis Yesa Green Overflow
London Conventional station, External (LH2) Grid mix LH2 vapourisation
60 min. from bus depot
Luxembourg Bus depot External Grid mix Overflow
Madrid Bus depot On-site steam Mix of on-site Grid mix Booster
methane reforming generation and
plus external external delivery
Porto Bus depot External Grid mix Overflow/booster
Stockholm Bus depot On-site electrolysis No Green Overflow/booster
Stuttgart Industrial site close to bus depot On-site steam methane reforming Yes Grid mix Overflow/booster

a External hydrogen delivery was not foreseen initially but only made possible when problems with the on-site production unit occurred.
international journal of hydrogen energy 34 (2009) 7114–7124 7117

Table 2 – Design data of the production units based on water electrolysis.


Amsterdam Barcelona Hamburg Stockholm
3
Production range (Nm /h) 15–60 15–60 30–60 15–60
Installed power (kW) 400 400 390 400
Rated power demand (stack and pumps) (kWh/Nm3) 4.8 þ/ 0.1 4.8 þ/ 0.1 4.8 þ/ 0.1 4.8 þ/ 0.1
Resulting efficiency (stack and pumps) (%) 63% 63% 63% 63%

Note that the installed powers given are the sums of the nominal powers of all consumers in the installations. They are not the powers to be
actually supplied in practice. This is illustrated by the following calculation for the power demand for the electrolysis process at rated capacity:
4.8 kWh/Nm3  60 Nm3/h ¼ 288 kW, thus substantially lower than the installed power.

efficiencies than industrial plants. Operating temperature was developed that covers all elements of CUTE as outlined above.
in the range of 900  C. It includes quantitative performance indicators but also
The rated capacities of the two reformer units differ by addresses qualitative aspects, specified for example in the
a factor of two: form of ‘‘guiding questions’’.2
A selected number of performance indicators is covered in
 In Madrid, it was decided to rely on regular road supply of this paper:
hydrogen and on-site production in parallel. The design
capacity could, therefore, be determined below the rated A. Availability of the on-site production unit; defined as the ratio
demand of the fuel cell buses. The idea behind this of time that the unit was operational to the duration of the
approach was to accomplish long periods of reformer evaluation period; ‘‘operational’’ meaning that the unit
operation at full load and a small number of start–stop was in operation or could have been operated if it had been
cycles, even when not all buses were in service. required; calculated by summing up all reported downtime
 Stuttgart selected a larger unit in order to be able to cover hours and subtracting them from the total evaluation
the load under any circumstances. This decision was based period.3
on the expectation that the reformer could be operated B. Causes for downtime of the on-site production unit; categories
about as flexible as an electrolyser, with short start-up and of cause include scheduled maintenance, failure and
shut-down periods, not significantly affecting lifetime or repair of major sub-units (such as compressors or
efficiency. controls), and safety concerns (meaning that the station
unit was technically operational but for safety concerns
remained out of service); defined as the ratio of downtime
2.3. Station units related to the individual categories to total downtime;
determined separately for systems with electrolysers and
Refuelling of gaseous hydrogen can be accomplished in two steam reformers.
ways: By pressure differential (overflow refuelling) and by C. Electrical efficiency of on-site hydrogen generation (for sites
a compressor charging the vehicle tank (booster refuelling). with electrolysers); defined as the ratio of the energy
Both modes can be combined in a sequence (overflow refuel- content of the hydrogen produced to the total power
ling followed by booster mode). Most CUTE sites used the consumed for its production; power consumption based
combined method (see Table 4). All sites with gaseous supply on end energy; hydrogen energy content based on its net
needed a compressor for filling the station storage, denoted as calorific value.
‘‘storage compressor’’ in Table 4. In some cases these storage D. (1) Thermal efficiency of on-site hydrogen generation (for sites
compressors were actually part of the on-site production unit. with steam reformers); defined as the ratio of the energy
Nevertheless, they are listed in Table 4 for completeness. content of the hydrogen produced to the energy content
Note that the hydrogen on-site storage in Stockholm was of natural gas consumed for hydrogen production;
only 95 kg. The three buses could thus not be refuelled back- based on the net calorific values of both the gases.
to-back. The storage size was limited by the local approval D. (2) Overall efficiency of on-site hydrogen generation (for
authority because the station was located close to residential sites with steam reformers); defined as the ratio of the
houses. energy content of the hydrogen produced to sum of the
energy content of natural gas and of the electrical end
2.4. Performance indicators

2
For clarity and accuracy, the analysis distinguishes between The CUTE Assessment Framework was not published.
the on-site hydrogen production units on the one hand, as However, many of its elements were adopted by the PREMIA
installed at six out of nine sites (see Table 1), and the station project for their Assessment Framework which is publicly avail-
able [15]. An outline of the CUTE Assessment Framework can be
units installed at all sites on the other hand. Reference to
found in [1].
regular external hydrogen delivery is only made as far as it 3
The evaluation period is not for all sites identical with the two
affected the operation of the station units. years of operation. Only months with complete data sets could be
To capture the results from this complex project in an evaluated. In particular, for the first months of the operating
appropriate way, an Assessment Framework [14] was phase, such complete data sets were not available in some cases.
7118 international journal of hydrogen energy 34 (2009) 7114–7124

3.1. Performance of the hydrogen production units:


Table 3 – Design data of the production units based on
steam methane reforming. electrolysis

Madrid Stuttgart
The average availabilities (indicator A) of the electrolysers in
3
Production range (Nm /h) 30–50 50–100 Amsterdam and Stockholm were well above 90%, as expected
Installed thermal power (natural gas) (kW) ca. 280 ca. 500 (Table 5). In Hamburg, material problems caused leakage from
Rated thermal energy demand (kWh/Nm3) 4.86 4.65
a pipe in the lye loop. This reduced the unit’s availability
Resulting thermal efficiency (%) 62% 65%
below 70%.
Installed auxiliary electrical power (kW) 45 89
Leakage in Hamburg originated from crack corrosion on
Note that the installed powers given are the sums of the nominal the oxygen side of the lye loop. The pipe material (AISI 316L)
powers of all consumers in the installations but not the powers to
had been used under similar conditions previously (30% KOH
be actually supplied, as explained in Table 2.
concentration, lye saturated with oxygen, temperature about
80  C). However, under the operating conditions of this
energy consumed for hydrogen production; based on particular unit, the use of AISI 316L turned out not to be
the net calorific values of both gases. appropriate. It took considerable time until a suitable substi-
E. Availability of the station unit; defined as for the production tute type of steel (Douplex SAF 2205) could be identified.
unit (see indicator A) Moreover, the new material was laborious to handle. Several
F. Causes for downtime of the station units; defined as for the pipes in the unit had to be replaced. Pipes with an inner
production unit (indicator B) diameter of less than 20 mm could not be welded and had to
G. Specific power consumption of the station unit (for compres- be connected using special fittings. Larger pipes had to be
sion, dispensing etc.); defined as the power consumption nickel coated after welding to prevent corrosion.
per unit hydrogen dispensed; power consumption based The lye leakage in Hamburg was dominant in terms of
on end energy; hydrogen energy content based on its net downtime of all four electrolysis units, accounting for 75% of
calorific value. all downtime hours (indicator B; Table 6). Hydrogen
H. Specific hydrogen losses; defined as the sum of the amounts compressors caused about 8%; all other root causes contrib-
of hydrogen generated on-site and delivered from external uted less than 5% each.
sources, minus the amount dispensed, divided by the sum The electrical efficiency of the production units (indicator
of the amounts of hydrogen generated on-site and deliv- C; Table 5) in Amsterdam and Hamburg was determined as
ered from external sources. 60% and 58%, respectively. Note that these figures cannot be
directly compared with those in Table 1. The latter include
power for electrolysis process and pumps but not other
auxiliaries, as the formers do. In Barcelona and Stockholm
only the entire site (including the station unit) was metered,
3. Results and learnings from operation so indicator C could not be determined.

Over the two years of operation at each site, the 27 CUTE buses
covered a distance of 841,123 km, providing service for more 3.2. Performance of the hydrogen production units:
than 4 million passengers during 62,545 operating hours. The steam reforming
buses were refuelled more than 8900 times and consumed
over 192 tonnes of hydrogen. The average availabilities of the reformer units were rather
About 78.5 tonnes of hydrogen were generated on-site via low (indicator A, Table 5). Downtime in Stuttgart was mainly
electrolysis and 42.5 tonnes based on steam methane caused by a single issue. Process temperatures turned out to
reforming. More than 56% of the amounts generated on site be higher than those anticipated from numerical simulations
were derived from renewable energy. when the unit was designed in a compact manner. Even part

Table 4 – Characteristics of the refuelling station units. Amsterdam, Barcelona, and Porto had one compressor that could
operate in both modes, i.e. charge the station storage and boost vehicle tank pressure.
Site Hydrogen storage Refuelling type Compressor type Rated capacity (Nm3/h)
size (kg)
Storage compressor Booster compressor

Amsterdam 250 Overflow/booster Hydraulic 300


Barcelona 150 Overflow/booster Hydraulic 300
Hamburg 400 Overflow Diaphragm 62
London 3,200 LH2 vapourisation Cryogenic pump 900
Luxembourg 500 Overflow Diaphragm 60
Madrid 360 Booster Diaphragm/diaphragm 50 2,400
Porto 172 Overflow/booster Hydraulic 300
Stockholm 95 Overflow/booster Diaphragm/hydraulic 60 525
Stuttgart 282 Overflow/booster Hydraulic/hydraulic 100 5,380
international journal of hydrogen energy 34 (2009) 7114–7124 7119

Table 5 – Results by indicator for the on-site hydrogen production units.


Indicator Availability Electrical Thermal Overall Remarks
efficiency efficiency efficiency
(electrolysis) (reforming) (reforming)

A C D (1) D (2)

Amsterdam 99% 60%


Barcelona Not available Not available Availability: incomplete data basis;
performance probably similar
to Stockholm.
Efficiency: On-site hydrogen
production unit and station unit
were not metered individually.
Hamburg 68% 58%
London Not applicable
Luxembourg Not applicable
Madrid 18% 43% 38%
Porto Not applicable
Stockholm 95% Not available On-site hydrogen production
unit and station unit were not
metered individually.
Stuttgart 57% 39% 35%

of the supporting construction was damaged. Appropriate problems also was long delivery time for spare parts. A full
materials that could withstand the actual level of temperature account of operating the reformer in Madrid has been pub-
had to be identified, components built or ordered, respec- lished by Girón [5].
tively, and replaced. About half a year of operating time was About 44% of all downtime of the two production units
lost. based on steam reforming were caused by the reformers
In Madrid there was a series of problems affecting almost themselves (indicator B; Table 6), about 21% by hydrogen
every major component of the production unit. Due to design compressors, some 11% by other types of compressors, and 7%
errors, the reactor tubes had to be replaced twice after by cooling units.
cracking and bursting, respectively. Leaking valves in the The reformers could not be started up and shut down as
hydrogen purification unit (employing pressure swing easily as expected. Due to the specific (compact) design of the
adsorption) caused more downtime. This was due to a design units, the heating phase was long and energy consuming
problem concerning the packing. Issues with the hydrogen (Stuttgart: about 50 h from ambient to rated temperature
compressor further reduced availability severely. Part of the using nitrogen as a heating agent that was ‘‘piped’’ through

Table 6 – Share in downtime hours of the on-site electrolysers and steam methane reformers (indicator B), and share in
downtime hours of the station units (indicator F); categorised by cause; across all sites.
Indicator Production unit: on- Production unit: on- Station Remarks
site electrolysis site reforming unit

B B F

Transformer and rectifier 3.6% Not applicable


Water conditioning 0.5% 0.8%
Cooling 2.2% 6.9% Cooling of production unit excluding hydrogen
compressors.
Air or process gas compressor 0.6% 1.5%
Electrolysis stack/Reformer 0.3% 43.8%
Hydrogen purification 0.1% 11.4%
Hydrogen compressors 8.1% 20.8% 47.1%
Production unit 7.6% Lack of fuel due to stop of production.
External delivery 4.1% Lack of fuel due to trailers arriving late or
repairs.
Storage 0.1%
Dispensing 6.2%
Control and electronics 0.8% 0.4% 7.2%
Safety devices and alarms 75.0% 0% 0.8% Hydrogen leaks of dispensing equipment
including leaks considered under ‘‘Dispensing’’.
Safety concerns 0% 0% 13.7%
Maintenance 3.0% 3.5% 2.3%
Miscellaneous 5.8% 10.9% 10.9%
7120 international journal of hydrogen energy 34 (2009) 7114–7124

the reformer) and frequent thermal cycling increased wear The instructions for installing, operating and maintaining
and tear (Madrid, in particular). nozzles, hoses and breakaways were revised and extended, in
Avoiding start–stop cycles was an issue particularly in order to ensure correct handling and to restore the confidence
Stuttgart, where, unlike Madrid, the production capacity of the of the personnel operating the equipment. In some cases, staff
reformer unit exceeded the maximum daily demand of the had simply been scared by unexpected component behaviour.
buses and where, unlike Madrid, trucking out surplus Regular leakage tests were introduced or their frequency
hydrogen was not foreseen. Therefore, the Stuttgart reformer increased, respectively.
was operated continuously, usually at a low production rate in The major safety concern, however, was the fact that one
order to match fuel consumption as closely as possible, i.e. to of the turn-key station suppliers questioned the durability of
avoid overproduction as much as possible. a component of the nozzle in a hydrogen environment. This
A consequence of operating at part-load in Stuttgart caused some operators to reduce the maximum pressure
was a severe loss of thermal efficiency (indicator D (1); during refuelling temporarily or, as mentioned, shut down
Table 5). The numerous start–stop cycles for repairs and the infrastructure installation entirely. Expert advice was
test runs in Madrid had a similar effect. The actual obtained from the infrastructure suppliers and operators,
average thermal efficiency for Stuttgart was 39% compared and the statements were discussed in detail. Consequently,
to 65% rated. In Madrid, only 42% were accomplished on the concerns could be resolved and operation got back to
average instead of rated 62%. When additionally consid- normal.
ering the power needed for operating the reformer units, Some early safety concerns originated from misunder-
efficiency reduces by another 4–5% (indicator D (2); Table standings in communication between the nine sites and
5). On the other hand, when the Madrid reformer could be resulted in over-reactions in the attempt to prevent any risk to
operated at stable conditions for longer periods, over 65% personnel. A Task Force for Safety and Security was setup in
efficiency was reached, thus better than what had been order to improve the flow of information in terms of speed and
stated as rated [5]. accuracy, also in cooperation with technology suppliers. The
Another consequence of continuous hydrogen production work of this Task Force turned out to be a major element in
in Stuttgart was that surplus hydrogen had to be vented to the preventing disruption throughout the operating phase of
atmosphere whenever the storage of the station unit was full. CUTE.
Therefore, only part of the hydrogen produced from steam Non-availability of the station due to downtime of the
reforming could be utilised (see Section 3.4). production unit (therefore, lack of fuel; see Table 6) mainly
occurred in Stockholm and initially also in Hamburg. In
3.3. Performance of the station units Hamburg this was resolved by enabling external backup
supply. Downtime under ‘‘external supply’’ represents fresh
On average, the station units were available for 87% of the trailers arriving late and repairs to the docking station.
time over the two years of operation (indicator E; Table 7). All The specific power consumption per unit of hydrogen
units achieved availabilities of 80% or better with the excep- dispensed (indicator G; Table 7) ranged from 0.06 kWh/Nm3 to
tion of Barcelona. 0.82 kWh/Nm3. The wide range and high upper threshold is
Table 6 displays the root causes for downtime in terms of partly due to the fact that at some sites nightly illumination of
their share in total downtime hours (indicator F). It shows that the facility grounds was metered together with the actual
the hydrogen compressors were the most critical component operation of the station unit. This was discovered late within
with regard to downtime. The compressors broke down for the operating phase, so no additional meters were installed. It
various reasons or they affected hydrogen purity. A major follows that the figures are not fully comparable between the
factor for the comparatively low availability of the station unit sites. Still, it can be noted that the installations with external
in Barcelona (indicator E; Table 7) was compressor issues. delivery of gaseous hydrogen (Madrid, Luxembourg and Porto
The second most critical components were the dispensers, at 200 bar pressure) required significantly less power per unit
namely their nozzle, hose, and breakaway coupling. They did of hydrogen dispensed than those with on-site generation.
not, in fact, cause a great deal of downtime hours due to failure This could be expected because part of the energy needed for
or repair. However, in the wake of incidents at some sites, their compression was deployed at the central production facility
safety was discussed. This made some station operators to close rather than locally. At the locations with on-site generation,
down their facility, thus reducing availability, and became the the station units received the hydrogen at only 10–15 bar from
main contributor to downtime induced by safety concerns the production units and, thus, required more energy for
(Table 6). In summary, ‘‘dispensing’’ and ‘‘safety concerns’’ compression (except Madrid, as mentioned4). The London site
accounted for about 20% of all station unit downtime. with external delivery of liquid hydrogen consumed energy
Incidents included hydrogen leaking from nozzles and for vapourisation before dispensing.
hoses or from the connections between them, respectively.
These resulted in minor modifications of the nozzles and
replacement of hoses. It must be noted that hoses for high
4
In the context of indicator G, Madrid can be regarded as a site
pressures were on the market at the time but that hoses
with external hydrogen supply at 200 bar. One reason is that, due
specifically designed for hydrogen have not yet been devel-
to the low availability of the production unit, the site had to rely
oped. At some occasions, breakaway couplings popped off on external sources most of the time. Second, the Madrid
without obvious reason. It was then found that they had not production unit had its own hydrogen compressor and supplied
been mounted correctly. the gas to the station at 200 bar.
international journal of hydrogen energy 34 (2009) 7114–7124 7121

Table 7 – Results for station units (indicators E and G) and for the entire facilities (indicator H).
Indicator Availability of Specific power Comments Specific hydrogen Remarks
the station unit consumption of the losses
station unit (kWh/Nm3)

E G H

Amsterdam 97% 0.50 7%


Barcelona 59% Not available On-site hydrogen production 20% Hydrogen contamination
unit and station unit were not by compressor
metered individually.
Hamburg 86% 0.58 11% Hydrogen contamination
by compressor; otherwise: 9%
London 98% 0.49 69% Lower demand than planned
for caused high level of boil-off
Luxembourg 91% 0.16 29% Metering problem
Madrid 86% 0.06 9% Losses due to hydrogen
purification (PSA) not included
Porto 80% 0.16 9%
Stockholm 96% Not available On-site hydrogen production 8%
unit and station unit were
not metered individually.
Stuttgart 99% 0.82 46% High losses due to surplus
production and venting;
otherwise: 6%

3.4. Hydrogen losses hydrogen loss figure for Madrid in Table 7 excludes this loss
mechanism and is based on the amount of hydrogen actually
About 274.5 tonnes of hydrogen were supplied during the delivered to the station unit by the production unit.
operating phase, some 121 tonnes produced on site (see Further losses occur due to occasional purging of system
above) and 153.5 tonnes coming from external sources. components (for example after maintenance work) and due to
However, only 192 tonnes were refuelled to the buses. Thus, background leakage. Therefore, sites with no or few problems
more than 80 tonnes were lost. during the operating phase, such as Amsterdam and Stock-
When assessing the individual values for hydrogen losses holm, display 5–10% losses.
in Table 7 (indicator H), it must be borne in mind that the on- Sites with severe component failures suffered a higher
site generation units need some of the produced amounts to level of loss. For example, in Hamburg the storage had to be
regenerate the purification modules periodically. emptied once after rupture of the compressor membrane and
Hydrogen generated by water electrolysis at first is satu- subsequent hydrogen contamination. In doing so, about
rated with water and, therefore, must be dried to accomplish 400 kg of hydrogen were vented. Excluding this particular
the specified purity level (here: 99.999% or better). Usually two event would reduce the loss factor to less than 9%. The high
drier columns are installed (or two sets of drier columns): figure for Luxembourg is probably due to a metering issue.
While one is operating (at rated pressure, i.e. about 10 bar in This could not be fully verified, though.
the CUTE units), the drying agent in the second is regenerated. Special circumstances must be considered for London and
This requires part of the dry hydrogen which is depressurised Stuttgart:
to increase its adsorption capacity. It carries away the mois-
ture and is vented to the atmosphere. Typically 5% of the  The liquid hydrogen storage in London was designed for
generated hydrogen is consumed in this way.5 a daily withdrawal of 120 kg for refuelling the buses. The
Downstream steam reformer tubes, pressure swing actual consumption pattern, however, was about 60 kg, five
absorption units (PSA) serve a similar purpose but also purify days a week on average. For this reason, substantial boil-off
the hydrogen from, e.g. carbon compounds being leftovers of liquid hydrogen occurred. The level of losses could have
from the reformer reactions. This process, including regen- been lower, if the design consumption pattern had been
eration of the adsorber beds, is much more complex compared realised in practice.
to mere hydrogen drying. In industrial reformer plants, typi-  The main loss mechanism in Stuttgart was caused by the
cally 15% of the hydrogen is lost in this way.6 Note that the fact that the reformer could not start and stop hydrogen
generation as flexibly as originally projected. Instead of
5
It must be noted that this hydrogen loss mechanism does not intermittent operation, the reformer had thus to be oper-
reduce the rated amount of hydrogen supplied by the production ated continuously, at the lowest possible production rate,
unit to the station unit, for example 60 Nm3/h in case of the
even at times when the hydrogen storage was full, as
electrolysis CUTE installations. The production units, both
mentioned above. This resulted in the excess hydrogen
reformers and electrolysers, were designed to be capable of
generating the rated amounts plus those needed for regeneration. vented to the atmosphere. Over a period of six months when
6
After all, this tail gas can be employed energetically to provide the reformer was in repair and the site relied on external
heat for the reformer reaction. supply, hydrogen losses amounted to only 6%.
7122 international journal of hydrogen energy 34 (2009) 7114–7124

recommended by manufacturers. This could be of benefit for


4. Conclusions all components of the on-site hydrogen infrastructure, at least
during the early stages of operation.
The four on-site production units with electrolysers performed Damage is an issue in particular when it leads to contami-
well in terms of availability (indicators A and B). In Amster- nation of the hydrogen, for example, due to defective seals.
dam, Barcelona and Stockholm no major issues had to be Development work to meet this challenge is ongoing. For
addressed. The leakage/material problem encountered in example, so-called ionic compressors are being tested.7 They
Hamburg was dominant in terms of downtime hours. It could compress the hydrogen using an ionic liquid that has no vapour
not be foreseen based on previous experiences and, therefore, pressure and, therefore, cannot deteriorate hydrogen purity.
highlights the necessity and value of demonstration under CUTE was the first major project that trialled 350 bar
everyday operating conditions. It also demonstrates that great refuelling. Moreover, the stations were designed in a way that
care should be taken while choosing component materials enabled non-engineers to take care of refuelling. A short
even when only slightly modifying operating conditions training was sufficient. No meters had to be monitored during
compared to units that were previously designed and built. the process, etc. The fact that the buses were refuelled almost
The actual energy efficiency data for the entire production 9000 times without major incidents can clearly be seen as
units (indicator C) cannot directly be compared with the rated a success. 350 bar refuelling is feasible and has become
figures for stacks and pumps in Table 2, as explained. More- a standard procedure. The incidents as outlined in Section 3.3
over, at two out of four sites, the production unit was not were addressed appropriately. Still, there is some way to go to
metered individually. Still, the two available values for accomplish dispensing equipment fit for the use for retail
Amsterdam and Hamburg (for the entire production units) are customers. In order to establish hydrogen as a ‘‘trustworthy
not much lower than the corresponding rated ones (for stacks fuel’’ in the public arena, preventing any risk of irritating
and pumps only). This is a positive outcome. customers by unexpected noises or events or even small
Both on-site steam methane reformers clearly did not fulfil hydrogen leaks must be further taken care of. This includes
expectations in terms of availability (indicators A and B). This precise and proven procedures for the installation and main-
holds true for the reformer tubes, i.e. the core of the unit, as tenance of dispensing equipment.
well as for auxiliaries, as the case of Madrid has shown. The specific power consumption of those station units
Concerning the reformer tubes, also here care must be taken (indicator G) with on-site hydrogen generation was rather
in choosing component materials appropriately and designing high in some cases. Here, it would have been desirable to
the processes. Manufacturers have learned their lessons in meter the actual power demand of the compressors and of
this respect. They upgraded the CUTE prototype installations further major consumers individually in order to identify
during the trial and will consider their learnings for the next optimisation potentials in a systematic manner and exclude
generation of small-scale reformers of compact design. the demand for illumination, etc. This remains a task for
Energy efficiency (indicator D) under part-load conditions future projects.
turned out to be far below rated figures. Also start–stop cycles Given that all CUTE infrastructure facilities were proto-
were cumbersome and energy consuming. There certainly is types their level of performance was satisfying on the whole.
potential to improve both. On the other hand, it is question- From an overall perspective, enhanced system integration,
able whether small-scale reformers – given their high process simplification, and standardisation of the infrastructure
temperatures – can match the flexibility that state-of-the-art facilities are required, especially for plants that comprise on-
electrolysers of the same production capacity display. This site generation and station units. Although all CUTE cities had
must be considered when selecting hydrogen on-site genera- a turn-key supplier for their on-site hydrogen infrastructure,
tion technology and defining rated capacities. the major components usually came from individual manu-
Contingency arrangements for external hydrogen backup facturers. This often caused a poor integration of modules, for
delivery turned out to be vital. example, separate controls for hydrogen production units and
With regard to availability of the nine station units (indica- stations, or in a mismatch between components. System
tors E and F), the hydrogen compressors turned out to be the development should also consider, to a greater extent, the
most critical component. Also the hydrogen compressor in the special needs associated with variable load patterns, inter-
Madrid production unit caused a significant share in down- mittent operation, and part-load conditions.
time; see indicator B in Table 6. This may be surprising at first A basic level of standardisation for hydrogen refuelling
sight because such units have been in use for decades in facilities will also enable turn-key suppliers to choose
industry. However, industrial plants usually operate contin- components from a range of manufacturers and, therefore,
uously and, thus, also the compressors as part of the should help to reduce the investment cost and footprint,
production process rarely have to perform stop–start increase efficiency and advance overall performance.
sequences. At the CUTE sites on the other hand, in particular In order to improve system availability, infrastructure
the booster compressors were started only for some minutes suppliers and operators need to develop clear concepts of how
in order to refuel the buses up to rated level. Also the storage to react rapidly to unforeseen problems with the installations,
compressors stopped whenever the on-site tank was full especially in crucial ramp-up phases of operation. This need is
overnight or during weekends. Frequent cycling due to inter- illustrated by the safety concerns regarding dispensing
mittent operation increases wear and tear, and thus, suscep-
tibility to damage. Therefore, it is advisable to shorten 7
One ionic compressor is operated in Berlin in the ongoing
inspection and maintenance intervals compared to what is HyFLEET:CUTE project [16].
international journal of hydrogen energy 34 (2009) 7114–7124 7123

equipment and how they were resolved, as discussed in transparent to them and misunderstandings regarding the
Section 3.3. Accordingly, agreements with component meaning of individual indicators and their data bases must be
manufacturers and local contractors need to be in place. They avoided as much as possible. Again, diversities regarding
must consider the special demands stemming from multi-site vocational training background, language and culture must be
and multi-country projects, such as language and culture. The considered.
work of the Task Force for Safety and Security is an element of Most of the CUTE sites have continued operation of their
how this can be addressed. It shows that rapid and efficient buses and infrastructure for a third year in the successor
communication is a major success factor in complex project HyFLEET:CUTE. HyFLEET:CUTE involves the opera-
demonstration projects such as CUTE. tion of 47 hydrogen powered buses, fuel cell buses as well
The facilities in Amsterdam and Barcelona possessed as buses powered by hydrogen internal combustion
(almost) identical components. However, the performance engines, in 10 cities on three continents [16]. As regards
displayed by their station units was remarkably different. This infrastructure, one aim is to refine the performance
points to the need for ‘‘fleet trials’’ of hydrogen infrastructure analysis.8
units, i.e. installations that share the same technology and are
operated concurrently at different sites in order to explore
their durability under diverse operating conditions.
Furthermore, hydrogen losses and their root mechanism Acknowledgements
must be addressed. Most CUTE sites displayed a level of loss
between 5% and 10% (indicator H). No operator of a conven- The authors would like to thank all those at the nine project
tional refuelling station would or could allow this for petrol or sites who collected and supplied the operating data.
diesel. On the other hand, it must be acknowledged that The CUTE project was co-funded by the European
hydrogen is a gaseous fuel, e.g. of low viscosity, that has Commission under its Fifth Framework Programme for
different properties than liquid fuels. Moreover, very high Research, Technological Development and Demonstration
purity/purification requirements must be met for the use of Activities (contract no. NNE5/2000/113).
hydrogen in fuel cells. The question remains open and is to be The HyFLEET:CUTE project is co-funded by the European
answered in the future whether a certain level of hydrogen Commission under its Sixth Framework Programme for
loss must be accepted also in ‘‘mature’’ commercial Research and Technological Development (contract no. TREN/
installations. 05/FP6EN/S07.52298/019991).
The technical analyses regarding critical components and
performance issues are well in line with statements from the
bus and station operators when consulted about their views references
on advances and issues arising from the trials. Bus operators
that had previous experiences with natural gas powered
vehicles and refuelling installations pointed out that there [1] Binder M, Faltenbacher M, Kentzler M, Schuckert M. Clean
Urban Transport for Europe, Deliverable No. 8 Final Report;
were no fundamental differences between natural gas and
2006. <www.fuel-cell-bus-club.com/>.
hydrogen infrastructures.
[2] Saxe M, Folkesson A, Alvors P. Energy system analysis of the
The significance of some of the numerical findings of this fuel cell buses operated in the project: Clean Urban
analysis is hampered by the fact that the raw data were not Transport for Europe. Energy 2008;33(5):689–711.
obtained in a harmonised manner across all sites. Issues with [3] Faltenbacher M. Modell zur ökologisch-technischen
evaluating power consumption due to a lack of separate Lebenszyklusanalyse von Nahverkehrsbussen (Model of the
meters for the production units at some sites (indicator C) and ecological and technical life cycle analysis of public transport
buses). Ph.D. Thesis. Stuttgart University; 2006.
for individual major consumers inside the station units
[4] Saxe M, Folkesson A, Alvors P. A follow-up and conclusive
(indicator G) document the importance of developing report on the attitude towards hydrogen fuel cell buses in the
a coherent concept for data acquisition and evaluation and CUTE project – from passengers in Stockholm to bus operators
translating it into hardware requirements at a very early stage in Europe. Int J Hydrogen Energy 2007;32(17):4295–305.
of a project. This holds true in particular when performance [5] Girón E. The hydrogen refuelling plant in Madrid. Int J
shall be benchmarked between several facilities from various Hydrogen Energy 2007;32(10–11):1404–8.
manufacturers or even between individual demonstration [6] Melaina MW. Initiating hydrogen infrastructures:
preliminary analysis of a sufficient number of initial
projects. Under CUTE, the planning period was very tight, so
hydrogen stations in the US. Int J Hydrogen Energy 2003;
development of the Assessment Framework on the one hand, 28(7):743–55.
and system design and installation on the other hand took [7] Mercuri R, Bauen A, Hart D. Options for refuelling fuel
place in parallel and not consecutively. In future projects, an cell vehicles in Italy. J Power Sources 2002;106(1–2):
assessment framework should be defined and agreed by all 353–63.
relevant partners before any hardware or data acquisition [8] Shayegan S, Hart D, Pearson P, Joffe D. Analysis of the cost of
hydrogen infrastructure for buses in London. J Power Sources
equipment is ordered.
2006;157(2):862–74.
Another important factor for significant inter-site bench-
[9] Agnolucci P. Hydrogen infrastructure for the transport
marking is that there should be one person at each site who is sector. Int J Hydrogen Energy 2007;32(15):3526–44.
responsible for the capture of all data. These persons in charge
must be trained in a joint workshop before the start of oper- 8
The HyFLEET:CUTE Assessment Framework is to be published
ation. The objectives of the data collection procedure must be soon on the project website [16].
7124 international journal of hydrogen energy 34 (2009) 7114–7124

[10] Baykara SZ. Hydrogen as a fuel: a critical technology? Int J Germany heading for an early lock-in? Int J Hydrogen Energy
Hydrogen Energy 2005;30(5):545–53. 2005;30(10):1045–52.
[11] Joffe D, Hart D, Bauen A. Modelling hydrogen infrastructure [14] Clean Urban Transport for Europe. Assessment
for vehicle refuelling in London. J Power Sources 2004; Framework for CUTE (Phase 2), Version B; 2006
131(1–2):13–22. [unpublished].
[12] Vidueira JM, Contreras A, Veziroglu TN. PV autonomous [15] Pelkmans L, van Poppel M, Govaerts L. PREMIA assessment
installation to produce hydrogen via electrolysis, and its use framework for hydrogen demonstrations, 2006.<www.
in FC buses. Int J Hydrogen Energy 2003;28(9):927–37. premia-eu.org/reports.htm>
[13] Hekkert MP, van Giessel J-F, Ros M, Wietschel M, [16] HyFLEET:CUTE. website. <www.global-hydrogen-bus-
Meeus MTH. The evolution of hydrogen research: is platform.com/>.

Vous aimerez peut-être aussi