Fuel cell field test demonstration of economic and environmental viability for portable generators, backup and UPS power system applications Programme type: FP7 Subprogramme Area: Portable generators, backup and UPS power systems Contract type: Joint Technology Initiatives - Collaborative Project (FCH) Object of document: D 4.1 Development of test protocol Work Package Leader: EP Contributors: JRC, UNIDO-ICHET, LUASA, EPS, Future-E Date of issue: 29-06-2011 Status of the document: draft approved by Internal Quality Management Number of pages: 44
CONTENT LIST EXECUTIVE SUMMARY... 4 1. INTRODUCTION... 5 1a. Overview of FITUP project... 5 1b. Description of the system and state of the art... 5 2. TEST PROTOCOL DEVELOPMENT... 9 2a. EU power grid characteristics... 9 2a.1 L1 grid failures... 9 2a.2 L2 and L3 grids failures... 13 2a.3 Conclusions... 17 2b. Test protocols adopted in previous projects... 18 3. TEST PROTOCOL GENERAL FEATURES... 20 3a. Test purpose... 20 3b. General test structure... 20 3c. Test procedures... 22 4. BENCHMARK TESTS... 24 4a. Test equipment and setup... 25 4a.1 Installation instructions... 26 4a.2 Climate chamber... 27 4b. Long-term durability benchmark test procedure (BENCH-LTD)... 27 4b.1 Base test procedure... 27 4b.2 Climate chamber tests... 30 4b.3 Hydrogen consumption evaluation... 30 4b.4 System stability under stress conditions... 31 2
4b.5 Grid restore during system start-up, grid failure during system shut-down... 31 4c. Additional benchmark tests (BENCH-ADD)... 32 5. ON-FIELD TESTS... 33 5a. Systems equipped with COMPRESSED H 2 CYLINDERS... 34 5a.1 Test equipment and setup... 34 5a.2 Non-TETRA systems test procedure (OF-CH2)... 36 5a.3 TETRA systems test procedure (OF-TET)... 36 5b. Systems equipped with ELECTROLYSER (OF-ELS)... 37 5b.1 Test equipment and setup... 38 5b.2 Test procedure... 40 LITERATURE REFERENCES... 41 ANNEX A - TEST PROCEDURE SCHEMES... 42 3
EXECUTIVE SUMMARY The purpose of this document is to present the test protocol developed for the fuel cell powered backup systems that will be tested in the framework of FITUP project. The aim of FITUP is in fact to demonstrate the good performance and reliability of this innovative kind of UPS system, through an extensive test campaign, carried on both in laboratory and real-world applications. In the introduction the main characteristic of the project and the state-of-the-art of the employed technology are reported, underlining the benefits that the project can bring to the rising fuel cell UPS industry. Chapter 2 resumes the documents used for the development of the protocol, including studies regarding duration and frequency of the grid failures in European electric grid and the results of previous research projects carried out in the same field of FITUP project. In the following three chapters the details of the protocol are presented, starting from its general structure (chapter 3), and going to the precise description of the two main branches of the protocol, regarding laboratory tests (chapter 4) and on-field tests (chapter 5). In fact, starting from the same principles, stated in chapter 3, different procedures were defined according to the location of the test, taking into account the constraints of real-world situations and the wider possibilities offered by laboratory environment. In this way, laboratory and on-field tests will prove different features of the systems, leading to the complete satisfaction of project s targets. In summary, all tests are variants of the following base scenarios: short term grid failures (A type), duration of 15 medium and long term grid failures (B type), duration of 240 (4 hours) catastrophic grid failures (C type), duration of 4320 (72 hours). The document is completed by a list of literature references and an annex containing graphic representations of the different test procedures. 4
1. INTRODUCTION 1a. Overview of FITUP project A total of 19 (plus 2or 3 awaiting confirmation) market-ready fuel cell systems from two suppliers, Electro Power Systems and Future-E, will be installed as UPS (Uninterruptible Power Supply) backup power sources in selected sites across the European Union. Real-world customers from the telecommunications industry will use these fuel cell-based systems, with power levels in the 3-12 kw range, in their sites. These units will demonstrate a level of technical performance (start-up time, reliability, durability, number of cycles) that qualifies them for market entry, thereby accelerating the commercialisation of this technology in Europe and elsewhere. The demonstration project will involve the benchmarking of units from both fuel cell suppliers according to a test protocol to be developed within the project. It will employ this test protocol to conduct extensive tests in field trials in sites selected by final users in Italy, Switzerland and Turkey. The performance will be logged and analysed to draw conclusions regarding commercial viability and degree to which they meet customer requirements, as well as suggesting areas for improvement. A lifecycle analysis using data from the project will be carried out to determine economic and environmental value proposition over incumbent technologies such as batteries or diesel generators. The system producers will use the results of the tests, and obtain valuable first hand feedback from customers, optimise their systems as needed, and demonstrate commercial viability. On the other hand, final users from the telecommunications industry will experience first-hand the advantages of fuel cells for their applications under real world conditions. The optimisation potential is expected from the production process itself, from the installation of a significant amount of fuel cell systems and from the testing. The project will also develop a certification procedure valid in all countries where the fuel cells will be either produced or tested under the expert advice of TÜV Süd. The dissemination of project progress will be geared mostly towards getting the word out to final users through presentations at specialised conferences, thus improving the visibility of market-ready fuel cells and pave the way for market penetration. The consortium consists of large and small entities which are fuel cell suppliers, end users and R&D centres (for data acquisition and analysis). The partners are located throughout the EU covering a range of environmental conditions, such as those found in Italy, Switzerland and Turkey. This spread will demonstrate the viability and enable the use of fuel cell-based UPS across the EU. 1b. Description of the systems and state of the art The main components of the fuel cell UPS systems are the following: PEM fuel cell(s) fuel cell auxiliary equipments (air blower, water pump, power electronics, ) start-up batteries (or ultracapacitors) 5
electrolyser (only for models with autonomous hydrogen production). The systems without electrolysers produce electric power using pure hydrogen as energy source; therefore, they are connected to external hydrogen tanks. Systems equipped with the electrolyser need tri-phase electric power as an input to produce hydrogen, stored in an hydrogen reservoir. Figure 1 - General layout of a fuel cell UPS system During the last few years, backup power sources based on fuel cell technology have experienced a significant surge in interest, due to the fact that they can offer economic advantages to potential users, as well as technical and environmental benefits. These potential benefits may be briefly summarised as follows: autonomy - fuel cells are able to operate as long as there is available fuel, and therefore can have longer autonomy when compared to a battery. This is due to the inherent nature of hydrogen fuel cells, which separate the energy and power sources, as opposed to batteries remote monitoring - fuel cells can be fully monitored from one central location, alerting the operator when the system is in use and how much time it can run before refuelling is required, to ensure no downtime footprint - the space required for the same period of runtime is considerably less for fuel cells than for battery banks 6
fuels - the majority of these systems operate on hydrogen (in this instance the only emission is water), which can be generated from renewable sources (electrolysis) or fossil fuel reforming cost - over its lifetime these units can offer cost savings over existing technologies (taking into account maintenance, repairs, transport and disposal) reliability - in many cases, fuel cells are able to offer higher reliability and MTBF (Mean Time Between Failures), and there is low degradation of voltage over time; failures tend to be less critical and easily dealt with environmental - fuel cells using hydrogen produce only water vapour, presenting a clear advantage over diesel generators which generate pollutants and significant CO2 emissions; batteries present instead the issue of their recycling, especially with the large presence of lead, which requires special measures maintenance - fuel cells have very few moving parts which reduces the need for regular maintenance lifetime - longer lifetimes than batteries, in UPS systems requiring 50-200 hours of annual use fuel cells can meet 10-15 years of lifetime with present status of technology; batteries must be replaced approximately every 3-5 years due to self discharge regardless of their hours of real use. Several studies have also pointed out the economic benefits of fuel cells over these other alternatives. Lifecycle cost analyses reveal that fuel cells provide a more economic alternative over both batteries and diesel generators over their lifetime. Primarily this is due to the replacement schedule of batteries and the significant maintenance needs of diesel generators, adding up to significant OEM costs that overcome the larger initial capital investment in fuel cells. With these factors in mind, it is clear to see where the increasing interest in fuel cell-based solutions comes from. While there are several specific markets that fuel cells can enter, the telecommunications industry is perhaps one of the most attractive, due to their use patterns and overall increase in size worldwide. In a study for the U.S. Department of Energy, Battelle Memorial Institute analyzed lifecycle costs of emergency response radio towers, comparing fuel cells with a 2 kw battery-only backup (8 hours autonomy) and a 5 kw battery-diesel generator backup (52 hours, 72 hours, and 176 hours backup duration). PEM fuel cells can provide service at substantially lower total cost than current technologies (the higher cost for the fuel cell system for the 176-hour backup results from the cost of hydrogen storage tank rental) 1. Figure 2 - Net present value of total cost of backup power systems for emergency response radio towers 2 7
Several government facilities are already using or have tested fuel cells as backup power for communications and other systems in the US. The State of Maryland currently uses a fuel cell backup power system at a wireless communications microwave site at Elk Neck State Park near North East, Maryland. The Elk Neck tower, which consists of a single-channel microwave repeater radio tower, supports several tenants, including the State s E 911 communications system. During hurricane Isabel (2003) and its aftermath, the fuel cell system enabled critical radio communications over the microwave network for Maryland State Police and emergency medical response services until primary grid power was restored. In the US, the three major suppliers for backup power fuel cell systems, Plug Power, Reli-on and IdaTech, have installed more than a hundred fuel cell systems so far. In the EU, the manufacturers have made inroads into the marketplace, but could still greatly benefit from a project such as FITUP. So far, the level of sales in EU countries by EU suppliers appears to have been somewhat limited, although they have been increasing. The fuel cell suppliers participating in the FITUP project are ramping up production of their units. At the same time, widespread knowledge of this technology is still an objective to be achieved in order for potential users from a number of industries to consider using fuel cells, alongside batteries or diesel generators. Because the number of demonstrations of fuel cell-based backup power applications has been limited, there is a lack of data for users to access. Without references, many potential users are perhaps understandably reticent to try a new technology that provides reliability for their networks, e.g. in the case of the telecommunications industry. Furthermore, the set of criteria for commercialisation that are included in the call regarding reliability, durability, cost, cyclability and response times, have not, as a whole, been proven with a representative set of units. FITUP project s aim is to overcome these problems, carrying on an extensive test campaign that will supply data about the systems and spread technology knowledge. 8
2. TEST PROTOCOL DEVELOPMENT The FITUP project proposes to conduct an EU-wide set of field trials that demonstrate the technical viability and economic maturity of backup power systems based on fuel cell technology. This three year demonstration project will constitute the largest such undertaking of its kind in Europe and will place EU companies on par with those active in the US and thus at the forefront of fuel cell technology worldwide for these applications. It will increase the visibility of fuel cells as a potential alternative to conventional backup power sources (batteries and diesel generators) and prove to potential customers in different industrial sectors their advantages. In order to achieve these objectives, an appropriate test protocol for the UPS systems is needed. The purpose of this document is to describe the test protocol and explain the reasoning behind its definition. In fact, the protocol has been developed starting from project objectives, and taking into account grid failures pattern in EU s power grid and known test procedures, previously created during other projects and industrial activities. 2a. EU power grid characteristics In order to understand how to test the UPS units in a realistic way, it is necessary to investigate the characteristics of the electrical grid failures, especially in terms of duration and frequency. In principle, three levels of supply should be considered: L1 - High level voltage grid (> 110 kv) L2 - Mid level voltage grid (> 4/6 kv and < 110 kv, typically 20kV) L3 - Low level voltage grid (< 4/6 kv, typically 400 V) Adequate and rather detailed data are available for L1 and L2 grids, but not for L3. However, while the grid is designed to allow supply at various nodes to reduce the risk of lack of supply, in reality the grid capacities are often so strained that there is a hierarchical effect, i.e. a failure on the higher level affects in most cases the lower level. 2a.1 L1 grid failures In order to obtain information about frequency and duration of the electrical grid failures in European countries, data has been taken from the last full statistical yearbook published by UCTE (former Union for the Co-ordination of Transmission of Electricity, now part of ENTSO-E network) in 2009, which contains 2008 data 3. After processing it results that during 2008 there were approximately 100 L1 grid failures in the UCTE territory (24 countries, continental western EU plus Switzerland and the western Balkan countries less Albania). This correlates also with a more detailed research by M. Rosas Casals 4, who reports a total of 908 major events for the period from 2002 through 2008. 9
In the following pictures, the number of reported L1 grid interruptions and their duration is shown. Both these elements (number and duration of interruptions) have to be analyzed, as they can influence the test design. Figure 3 - Number of L1 grid failures in UCTE countries during 2008 10
Figure 4 - Duration distribution of L1 grid failures in UCTE countries during 2008 As can be seen from the above data there is no clear distribution pattern. This is confirmed also by the average duration, which is 105.4. Excluding the minimum and the maximum of the failure duration distribution (respectively 0 and 1949 ), the average is 88.8 and the new extremes are 1 and 848. The distribution of events reasons is reported below: 11
Figure 5 - Reason distribution of L1 grid failures in UCTE countries during 2008 Legend: R4 - overload (including planned ones) R5 - false operation R6 - failure in protection devices or other elements R7 - outside impacts (animals, trees, fire, avalanches ) R8 - very exceptional conditions (natural or artificial disasters) R9 - other reasons Table 1 - Average duration per reason of L1 grid failures in UCTE countries during 2008 From this data it appears that the more technical issues can be handled quicker, but still lead on average to approximately 1 hour of power absence. External induced problems last longer, on average almost 2.5 hours. However, also the other categories contain actual cases with restoration periods longer than 200. It is important to note that the maximum value of 1949 or 32.4 hours occurs during an R9 case, which is not related to a major catastrophe. Therefore the specification of 72 hours of power supply for a governmental emergency response network may actually be quite reasonable. 12
2a.2 L2 and L3 grids failures For the lower level grids the statistical data are typically reported in form of Reliability Indexes, e.g. SAIDI (System Average Interruption Duration Index) and CAIFI (Customer Average Interruption Frequency Index). In a report of the Project UNDERSTAND 5, which among others led to a White Paper on Security of Electricity Distribution, a grid failure statistic study of CREER is quoted, regarding the period between 1999 and 2004. Table 2 - SAIDI data from UNDERSTAND project 13
Table 3 - CAIFI data from UNDERSTAND project Using these data it is possible to calculate average event duration (disregarding Latvia and Lithuania, as the data seem inconsistent). It is possible to clearly identify two event types, characterized by an approximate length respectively of 60 (±15 ) and 130 (±25 ). Table 4 - Average grid failure duration data from UNDERSTAND project 14
In order to validate these data, other examples were collected from literature. Austria (2008, 2009) Regarding unplanned interruptions, E-control GmbH, the Austrian Energy Regulator, reported SAIDI of 43,69 for 2008 and 36,65 for 2009 6, 7. These values represent an aggregate and weighted average. The disaggregation for 2009 leads to a more complex picture, as shown in the graph below. Figure 6 - Planned and unplanned grid failures in Austria during 2009 As can be seen there were nearly 100 interruption events, including both L1 and L2 grids. Some of them were planned (in green), but the majority of events was of unplanned nature (in blue). Maximum duration of an unplanned event was 3052. It is also worth noting that planned grid shutdowns are quite often followed by extensions of an unplanned nature. Furthermore, the 2009 E-control report contains a comparison with other countries: 15
Figure 7 - comparison between Austrian and other countries SAIDI values between 2007 and 2009 Italy In a newsletter of a small town in Northern Italy owning its own electricity plant and some L2 and L3 grid, a CAIFI of 2.8 failures per customer and year and a SAIDI of 78.5 are reported for 2005 8. In 2004, a larger electrical utility in the region (Etschwerke) was quoted to have a CAIFI of 3.3 failures per customer and year, and a SAIDI of 123, whereas Italian national average was CAIFI 2.9 failures per customer and year and SAIDI 153 (significantly contrasting to CREER data shown earlier). Switzerland Switzerland did not show up as having L1 failures in 2008 in the UCTE statistics, but it has incidents sometimes. The following events are listed in the annual report 2002/2003 of EW Höfe AG 9, a local/regional electricity provider: 20 supply interruption in January caused by an L1 grid failure in North Eastern Switzerland, and two interruptions of unknown duration, probably on a L2 grid sector. This would mean three interruptions in total, falling in-line with statistical data from the CREER study for that year. Bosnia I Hercegovina Data are available for this country from a Quality of transmission report 10, 11, and give the following SAIFI (System Average Interruption Frequency Index) and SAIDI: 16
Table 5 - SAIFI and SAIDI due to interruptions in the TRANSCO BIH network Table 6 - SAIFI and SAIDI including outages of MV feeders caused by interruptions in the distribution net As can be seen the interruptions are much more frequent there, but are quite often caused by planned shutdowns, probably for repair and extension of a highly damaged network. However, the average duration again roughly fits into the 60 /130 pattern. 2a.3 Conclusions Grid stability is improving in Europe, however there are still some outages. Major weather problems and unexpected consequences of repair and maintenance works can affect even generally stable and highly developed grids. Apart from very short failures (on which complete data are not available), it is possible to classify the outages in the following way: short term (less than 1 hour) 40-60% of cases on L2 and L3 networks, happen also on L1 grid. Average duration of 15-20 medium term (1 hour) 20-30% of cases on all grid levels long term (2 hours) 20-30% of cases on all networks, less frequent than medium term ones catastrophic (more than 24 hours) possible once a year in all grids at least regionally. 17
The obtained results have been used as a basis for the development of a mix of grid failure simulations that represent the reality. However, in order to test systems long-term durability within FITUP project time limits, the frequency of the simulations in the actual test procedure will be higher than in real world. Part of the grid failure simulations will be performed with system starting at room temperature, since this is the case in large part of real world events. Anyway also start ups with warm system (i.e. with system temperature higher than ambient one) will be operated, representing events in which planned shutdowns are followed by unplanned problems. Finally, also 72 hours grid failure simulations will be performed, even if statistically very rare, because some final users involved in the project required that the UPSs demonstrate the capability to sustain the load for that amount of time. 2b. Test protocols adopted in previous projects During the development of the test protocol, existent reports concerning test procedures of this kind of systems were taken into account. These reports were produced during previous research projects and industrial tests, and some of the suggested procedures have been included in the FITUP protocol after proper modification. The Lucerne University of Applied Sciences and Arts - Lucerne School of Engineering and Architecture conducted field tests with a fuel cell UPS since January 2006 12, 13. The project took place in collaboration with the industrial partners Swisscom AG and American Power Conversion Corporation, respectively as end user of UPS systems in the telecommunications sector and market leader of UPS systems. In this project, lead-acid batteries were replaced by a PEM fuel cell system, and the delayed start-up behaviour of the fuel cell was bridged with supercapacitor technology. The system was connected to an existing and working telecommunication base station, installed on the roof of the Lucerne School of Engineering and Architecture in Horw. Hydrogen was provided by two tanks. The field test was performed during a period of three years and a half, with approximately 350 system start-ups. The test protocol was developed together with the final user. Short term tests included a sequence of five grid failure simulations of 5 and a sequence of two simulations during 20, and were performed monthly; long term tests during 4 hours were instead performed 2 or 3 times a year. Excellent results confirmed the functionality, reliability and performance of the system. The environmental impact of lead-acid batteries and of the fuel cell system was also investigated: without recycling processes, the fuel cell leads to a reduction of CO 2 emissions close to 90% within a life cycle of 10 years. Considering lead-acid batteries recycling projects the reduction is anyway more than 80%. In a publication of a CRES Project 14 regarding a 5 kw PEM FC and electrolyser backup power system, designed by the project code HELPS, the authors report that a system being able to cope with a 5 hours interruption every 48 hours, would cover 99,6% of all possible grid power failures. With Vodafone as main sponsor and Electro Power Systems as product manufacturer, Environment Park performed tests of an Electro7 unit, connected to a radio access site located in an equipped van 18
from the Vodafone Disaster Recovery department. Test objective was to investigate whether the fuel cell UPS could technically substitute lead acid batteries as back up for the TLC equipment. The results obtained during the trial at Environment Park showed that Electro7 can cope with the typical network scenarios that a mobile network operator has to manage, being a valid alternative to traditional UPSs 15. The definition of the test protocol was also based on existent procedures for fuel cell stacks, in particular the ones developed in the FP5 RTN FCTESTNET thematic network, currently validated in the FP6 STREP FCTESQA 16. Both projects, scientifically coordinated by JRC, produced standardized documents, used as base structure for the protocols of this project. 19
3. TEST PROTOCOL GENERAL FEATURES 3a. Test purpose This document describes the test protocol that will be followed by the fuel cell UPS systems during FITUP project that includes both laboratory benchmark experiments and on-field tests, in locations defined by final users. The main purpose of the tests is to demonstrate that the fuel cell UPS systems satisfy the following objectives, stated in the project: reliability higher than 95% durability higher than 1500 hours and 1000 cycles (real-time or extrapolated) response time smaller than 5ms The reliability will be statistically verified at the end of the whole test campaign, that will involve 19 to 22 systems. Final users also specified that their usual reliability requirements are higher than 99%, and they expect that the tested systems will meet them. Response time will be controlled at each system start-up. The requirements in terms of durability will be instead completely checked only for 4 systems, two for each manufacturer involved in the project. These systems will be operated for more than 1500 hours, and will carry on more than 1000 on-off cycles. The data collected from these tests will be used to predict the aging of the other systems. The long-time tests will be operated in laboratory, due to the logistic complexity of the hydrogen tanks replacement in real-world applications. In addition to the mere verification of project objectives, during the tests additional data will be collected in order to investigate system performances and obtain useful information for manufacturers and final users. The described protocol could also be used as a standard performance test for fuel cell UPSs, both in laboratory and in real world applications. 3b. General test structure As said before, the protocol should include a mix of simulated failures that represent the actual grid operation. However, in order to reach the purposes of the tests within the project period, it will be necessary to expose the systems to a failure frequency higher than normal. In fact, the time available for the benchmark activities is 24 months (from M7 to M30), while the onfield tests will last a maximum of 23 months (from M13 to M35). 20
The grid failure simulations will be divided in three main groups, each one characterized by a different duration and representing a statistical class of failures that actually happen in electrical grids (see paragraph 2a.3): short term grid failures (A type), duration of 15 medium and long term grid failures (B type), duration of 240 (4 hours) catastrophic grid failures (C type), duration of 4320 (72 hours). B and C type simulations have a larger duration compared to the actual grid failures of the same class. The duration of B failures has been fixed in order to create test procedures respecting the given number of on-off cycles and hours of operation. C simulations satisfy instead a particular demand coming from some end users, that would like to check if the systems can match TETRA applications requirements. However, since the above said durations are larger than real world ones, the resulting tests are more stressing for the systems than usual functioning. As a consequence, if the systems will be able to withstand the programmed tests, they will prove to be capable of working in harder conditions than normal operation in Europe. During laboratory tests, each kind of grid failure simulation will be performed at different power levels (50%, 75% and 100% of total system power). This will not be done in on-field tests: in fact, in realworld applications it is not possible to choose the load applied to the systems. Moreover, two different start-up conditions will be employed: warm start up (1 type), duration of the off time before the cycle 1 cold start up (2 type), duration of the off time before the cycle 60 (minimum) The characteristics of the simulations and their frequency are indicated in the following table. Table 7 - test cycles characteristics Cycle name A1 A2 B1 B2 C1 C2 Cycle duration 15 15 240 240 4320 4320 Off period before cycle 1 60 min. 1 60 min. 1 60 min. It is important to note that 60 is the minimum interval between a cycle and the following A2, B2 or C2 cycle. Therefore, it is necessary to leave the system off at least for one hour before any 2-type cycle can be performed, but is also possible to increase this waiting time as desired. A graphical representation of the different cycle types can be found in Annex A, figures 11-16. 21
3c. Test procedures Five test procedures were defined, taking care of different test objectives, locations and system characteristics: 1. LONG-TIME DURABILITY BENCHMARK test (BENCH-LTD) 2. ADDITIONAL BENCHMARK tests (BENCH-ADD) 3. ON-FIELD test for systems with COMPRESSED H 2 CYLINDERS (OF-CH2) 4. ON-FIELD test for TETRA systems with COMPRESSED H 2 CYLINDERS (OF-TET) 5. ON-FIELD test for systems equipped with ELECTROLYSER (OF-ELS) Every procedure has been constructed combining the cycles listed in table 7. The precise description can be found in chapters 4 and 5. In the following table, the assignment of the test procedure to each system is showed. Table 8 - list of UPS systems installations Installation Location End-user Fc provider Code Test procedure Lucerne/Horw Swisscom EPS A1 OF-CH2 Lucerne Swisscom Future-E A2 OF-CH2 Zizers Swisscom EPS A3 OF-CH2 Davos Swisscom Future-E A4 OF-CH2 Ennetbürgen BKPNW Future-E A5 OF-CH2 Dallenwil BKPNW EPS A6 OF-TET ON FIELD Alpnach BKPNW Future-E A7 OF-TET Lucerne cantonal police to be defined in 2012 A8 to be defined in 2012 Settimo Milanese WIND EPS A9 OF-ELS Milan WIND EPS A10 OF-ELS Milan WIND EPS A11 OF-ELS Istanbul Turkcell EPS A12 OF-ELS Istanbul Turkcell Future-E A13 OF-ELS 22
Istanbul Istanbul Vodafone (to be confirmed) Vodafone (to be confirmed) EPS A14 OF-ELS Future-E A15 OF-ELS Istanbul UNIDO-ICHET EPS B1 BENCH-LTD Istanbul UNIDO-ICHET Future-E B2 BENCH-LTD BENCHMARK Petten JRC EPS B3 BENCH-LTD Petten JRC EPS B4 BENCH-ADD Petten JRC Future-E B5 BENCH-LTD Petten JRC Future-E B6 BENCH-ADD 23
4. BENCHMARK TESTS These tests will be performed in a laboratory environment and have the main purpose of supplying reference data, that will be compared with the information coming from the on-field measurements. Moreover, benchmark tests will include features that cannot be performed in on-field tests. In the following table, the list of the systems that will undergo to laboratory benchmark tests is reported; only systems not equipped with electrolyser are involved in these tests. Table 9 - list of benchmark systems Locations End-users FC provider Code Test procedure Istanbul UNIDO-ICHET EPS B1 BENCH-LTD Istanbul UNIDO-ICHET Future-E B2 BENCH-LTD Petten JRC EPS B3 BENCH-LTD Petten JRC EPS B4 BENCH-ADD Petten JRC Future-E B5 BENCH-LTD Petten JRC Future-E B6 BENCH-ADD The controlled test variables (test inputs) during benchmark tests will be: external grid power availability (controlled through a controlled power switch) load applied to the UPS systems environment temperature (only during climatic chamber tests). The resulting measured variables (test outputs) will be instead: voltage and current at fuel cell system terminals voltage and current at start-up batteries (or ultracapacitors) terminals voltage and current at electronic load terminals. Table 10 - list of measured variables (test outputs) during benchmark tests Variable Symbol Max uncertainty Sample rate Voltage at fuel cell terminals V fc 100 mv 1000 S/s - 10 S/s Current at fuel cell terminals I fc 1.05 A 1000 S/s - 10 S/s Voltage at battery/ultracap terminals V bat 100 mv 1000 S/s - 10 S/s 24
Current at battery/ultracap terminals I bat 1.05 A 1000 S/s - 10 S/s Voltage at electronic load terminals V load 100 mv 1000 S/s - 10 S/s Current at electronic load terminals I load 1.05 A 1000 S/s - 10 S/s 4a. Test equipment and setup The benchmark test system is composed by the following parts: an industrial pc used to control the tests and collect the data through an acquisition board, equipped with necessary signal conditioning devices a temperature sensor used to monitor the environment temperature three couples of cables that carry the voltage signals to the acquisition board three shunts used to measure the currents an electronic load that absorb the power produced by the UPS an ac/dc converter used to power the load using AC power, before the grid failure simulations a controlled switch that breaks the connection between the AC/DC converter and the grid an optional hydrogen sensor, used to detect gas leakage in the test environment. 25
Figure 8 - Setup for benchmark tests (BENCH-LTD and BENCH-OF) 4a.1 Installation instructions The installation procedure of the test equipment will vary from site to site, depending on the characteristics of the place, but there are some basic rules to follow in order to ensure the correct working of the system. The hydrogen sensor must be placed in the higher point of the system room in case of indoor installations (outdoor installations does not need this device). The ambient temperature sensor should be placed as close to the fuel cell as possible, in order to have a correct information about the temperature close to the UPS. The shunts, used to measure the current at fuel cell and batteries terminals, should be installed as close as possible to these devices. The shunt that measures the current going to the load should be instead placed immediately after the insertion of the backup systems on the DC bus, i.e. as far as possible from the load, in such a way that the ohmic losses on the bus are taken into account as loads. It is also necessary to place the shunts in a way that the heat produced by the shunts (around 10 W) can be dissipated. The voltage measurement points should be placed before shunts connections, excluding in this way the voltage drop on the shunts from the measurement. 26
The measurement cables should be shielded and as short as possible, in order to reduce electrical noise. The fuel cell system and start-up batteries (or ultracapacitors) must be installed and connected to the required supplies (air, hydrogen, ) according to manufacturer indications. 4a.2 Climate chamber An integrated environmental test system consisting of a walk-in climate chamber (ambient temperature in the range -40 C/+60 C, humidity up to 95%) is available at JRC. This chamber will be used to perform tests in extreme climate situations (tropical environment and freezing condition). During the preparation of a test inside the climate chamber, it is important to avoid any substantial modification of the tested system. Therefore, it is strongly recommended to connect the test equipment to the system in a way that it can be moved in and out from the chamber without any component disconnection. 4b. Long-term durability benchmark test procedure (BENCH-LTD) This test will completely satisfy the target of 1500 hours of system operation and 1000 on-off cycles, and will be done only in laboratory during benchmark activities, since it is very difficult to carry it on without an hydrogen line, due to the high number of hydrogen tanks needed. It will be applied to one system for each producer at each R&D centre in charge for benchmarking (4 systems in total, 2 at JRC and 2 at UNIDO-ICHET), in order to obtain data for the same procedure in different locations (and therefore different environmental conditions). Obtained data will be the basis for an extrapolation method that will be used to predict the performance of the systems that will run less hours. 4b.1 Base test procedure After system installation according to paragraph 4a, the test procedure can start with the turning on of the UPS, that will enter in stand-by mode. Each grid failure simulation will be carried on using the following sequence: setting of the grid failure simulation duration in the software turning on of the electronic load to the desired power value activation of the grid failure simulation through the software, that will automatically start the high frequency data logging (1000 S/s) and disconnect the grid from the load. After 10 seconds, the data logging frequency is reduced to 10 S/s at the end of the simulation, the software automatically connects the load to the grid. The data logging frequency is increased again to 1000 S/s for 10 seconds around this event. A total of 1548 operating hours and 1005 on-off cycles will be done. The characteristics of the simulations and their frequency are indicated in the following table. 27
Table 11 - BENCH-LTD procedure cycles characteristics Cycle name A1 A2 B1 B2 C1 C2 Cycle duration 15 15 240 240 4320 4320 Off period before cycle 1 60 min. 1 60 min. 1 60 min. n cycles at 50% power 100 100 30 30 1 1 n cycles at 75% power 175 175 65 65 1 1 n cycles at 100% power 100 100 30 30 0 1 Total number of cycles 375 375 125 125 2 3 Total on hours 93.75 93.75 500 500 144 216 % of total on hours 6.06 6.06 32.31 32.31 9.31 13.96 The most of A and B cycles will be performed repeating 30 times the following sequence of simulation groups, named standard routine. Every group of the sequence contains less than 8 hours of total test time (sum of on and off periods, counted from the first start up to the last shut down of the day), therefore it can be contained in a standard working day. However, it is not compulsory to perform a group per day, but they can be arbitrarily arranged in time, as long as the waiting times between cycles are respected. Groups indicated by odd numbers share the same cycle sequence, but are performed at different power levels. The same holds for groups marked with even numbers. In Annex A graphical representations of odd- and even-numbered groups cycle sequences are reported in figures 17 and 18. Table 12 - BENCH-LTD standard routine Group 1 2 3 4 Cycle order B2 3 A1 3 A2 B1 B2 3 A1 3 A2 B1 N of A1 3 0 3 0 N of A2 0 3 0 3 N of B1 0 1 0 1 N of B2 1 0 1 0 Total time 288 406 288 406 Power level 75 % 75 % 75 % 75 % 28
Group 5 6 7 8 Cycle order B2 3 A1 3 A2 B1 B2 3 A1 3 A2 B1 N of A1 3 0 3 0 N of A2 0 3 0 3 N of B1 0 1 0 1 N of B2 1 0 1 0 Total time 288 406 288 406 Power level 50 % 50 % 100 % 100 % Every 6 times the standard routine is performed, a C cycle must be carried on. The order of C cycles is arbitrary, but all the cycles enumerated in table 11 must be done. At the end of the 30 repetitions of the standard routine, a final routine is performed as below: Table 13 - BENCH-LTD final routine Group 1 2 3 4 5 Cycle order B2 3 A1 3 A2 B1 B2 3 A1 3 A2 B1 B2 3 A1 N of A1 3 0 3 0 3 N of A2 0 3 0 3 0 N of B1 0 1 0 1 0 N of B2 1 0 1 0 1 Total time 288 406 288 406 288 Power level 75 % 75 % 75 % 75 % 75 % Group 6 7 8 9 10 Cycle order 3 A2 B1 B2 3 A1 3 A2 B1 B2 3 A1 3 A2 B1 N of A1 0 3 0 3 0 N of A2 3 0 3 0 3 N of B1 1 0 1 0 1 N of B2 0 1 0 1 0 29
Total time 406 288 406 288 406 Power level 75 % 75 % 75 % 75 % 75 % 4b.2 Climate chamber tests Climate chamber tests in extreme environmental conditions will be performed at JRC, at the beginning (after 1 repetition of the standard routine), at mid-term (after 16 repetitions of the standard routine) and at the end (before the final routine) of the whole long-term test campaign: start-up in freezing environment (-15 C) the system will start and work for some minutes, until normal operating temperature is reached tropical environment continuous operation (40 C) the system will start and work for 1 hour at maximum power. The cycles carried on during climate chamber tests are additional to the ones enumerated in table 11. The climate chamber tests will be carried on in the following way: the UPS is placed inside the chamber the environment conditions are set when the temperature inside the chamber reaches the desired value, the test can start using the same procedure mentioned above. 4b.3 Hydrogen consumption evaluation The evaluation of the hydrogen consumption of the systems (and therefore the estimate of their efficiency) will be carried on through gravimetric measurements of the hydrogen tank used to feed the systems. In fact, this method assures the highest level of accuracy. These measurements will be operated both at JRC and UNIDO-ICHET, using a balance characterized by a range of 150 kg and an accuracy of 1 g. A special procedure was defined: 6 A2 cycles will be performed at 6 different constant power levels, starting at 100% of total system power (6 kw) and going down to 1 kw using 1 kw steps. The procedure is graphically represented in Annex A, figure 19. Each grid failure simulation is performed as indicated in paragraph 4b.1, but during the 60 of system inactivity between subsequent cycles, the hydrogen tank used to feed the system will be weighed. The tank should be placed directly on the balance, in order to weigh it without moving any part of the system. If this will not be possible, maximum care should be used during the disconnection of the tank from the system in order to avoid changes in tank weight not depending on hydrogen consumption. The whole procedure will be completed using a single hydrogen tank (50 l, 200 bar, approximately 10 Nm 3 ). The procedure will be done at the beginning (after 1 repetition of the standard routine), at mid-term (after 16 repetitions of the standard routine) and at the end (before the final routine) of the longterm benchmark test. 30
The cycles included in the hydrogen consumption evaluation procedure are additional to the ones enumerated in table 11. 4b.4 System stability under stress conditions Since the benchmark tests will start before the on-field ones, they will confirm that the chosen operating patterns do not damage the systems. In particular, the final users asked for a laboratory trial of the system stability under stress conditions test before on-field application. This test is composed by two sequences. The first one comprises 5 cycles composed by 5 start-ups and 5 shut-downs, alternatively repeated every 10 minutes (i.e. each cycle implies 5 repetitions of the succession composed by 10 minutes system on and 10 minutes system off). Between subsequent cycles the waiting time is one hour. 25 on-off cycles are performed in total. The second sequence is similar to the first one, but the duration of on and off times is 5 minutes instead of 10. Therefore, it is composed by 5 cycles made of 5 start-ups and 5 shut-downs, alternatively repeated every 5 minutes (i.e. each cycle implies 5 repetitions of the succession composed by 5 minutes system on and 5 minutes system off). Between subsequent cycles the waiting time is one hour. 25 on-off cycles are performed in total. In Annex A, graphs representing the cycles that form both sequences are reported (figures 20 and 21). 4b.5 Grid restore during system start-up, grid failure during system shut-down The final users asked also for a laboratory trial of the Grid restore during system start-up, grid failure during system shut-down test before on-field application. This test is composed by 5 grid restores during system start-up, followed by 5 grid outages during system shut-down (20 on-off cycles performed in total). The first part of the test (5 grid restores during system start-up) is automatically operated by the test software repeating 5 times the following sequence, starting with electronic load on and switch closed (i.e. load fully powered by the AC/DC rectifier): the switch opens the connection with the grid the start-up batteries power the load while the fuel cell starts up as soon as the current supplied by the fuel cell is equal to the 50% of the current drained by the load, the switch is closed again the system shuts down (the load powered again by the AC/DC rectifier) a 10 pause is observed before a new grid disconnection. The second part of the test (5 grid failures during system shut-down), always performed by he automatic software, is instead carried out repeating 5 times the sequence below, starting with electronic load on and switch open (i.e. fuel cell system supplying 100% of the current drained by the load): the switch is closed 31
the system starts the shut down procedure (the fuel cell does not supply power, but auxiliary equipment is still working) after 10 from the grid restore, the switch is opened again, i.e. the grid is disconnected the system starts up when the load is fully sustained by the fuel cell (100% of the current drained by the load supplied by the cell) the switch can be closed again. 4c. Additional benchmark tests (BENCH-ADD) The remaining 2 systems at JRC, one per manufacturer, will be tested after the end of the BENCH-LTD tests. The precise definition of the procedure is not included in this document, since the partners decided to leave it open for further development. In fact, the procedure will be decided taking into account the data resulting from BENCH-LTD tests, when they will be available. In this way, it will be possible to include in the procedure particular operating patterns, with the aim of investigating remarkable aspects underlined during previous tests. The BENCH-ADD procedure will be fully described in a separate document released after the end of the BENCH-LTD tests. 32
5. ON-FIELD TESTS The main target of these tests is the evaluation of UPSs behaviour in case of grid failure, in all the possible environmental conditions experimented in real-life applications. These tests must confirm the functionality, reliability and good performance of the system. In order to obtain this result, grid failure simulations will be performed in all the on-field installations of the project. Different test procedures were defined and assigned according to systems architecture and application. The attribution of the procedures to the systems is showed in the following table. Table 14 - list of on-field installations Locations End-users FC provider Code Suggested test Lucerne/Horw Swisscom EPS A1 OF-CH2 Lucerne Swisscom Future-E A2 OF-CH2 Zizers Swisscom EPS A3 OF-CH2 Davos Swisscom Future-E A4 OF-CH2 Ennetbürgen BKPNW Future-E A5 OF-CH2 Dallenwil BKPNW EPS A6 OF-TET Alpnach BKPNW Future-E A7 OF-TET Lucerne cantonal police to be defined in 2012 A8 to be defined in 2012 Settimo Milanese WIND EPS A9 OF-ELS Milan WIND EPS A10 OF-ELS Milan WIND EPS A11 OF-ELS Istanbul Turkcell EPS A12 OF-ELS Istanbul Turkcell Future-E A13 OF-ELS Istanbul Istanbul Vodafone (to be confirmed) Vodafone (to be confirmed) EPS A14 OF-ELS Future-E A15 OF-ELS 33
5a. Systems equipped with COMPRESSED H 2 CYLINDERS The controlled test variables (test inputs) during on-field tests of systems equipped with compressed hydrogen cylinders will be: external grid power availability (controlled through a controlled power switch). The resulting measured variables (test outputs) will be instead: voltage and current at fuel cell system terminals voltage and current at start-up batteries (or ultracapacitors) terminals voltage and current at electronic load terminals. Table 15 - list of measured variables (test outputs) in on-field tests of systems equipped with H 2 cylinders Variable Symbol Max uncertainty Sample rate Voltage at fuel cell terminals V fc 100 mv 1000 S/s - 10 S/s Current at fuel cell terminals I fc 1.05 A 1000 S/s - 10 S/s Voltage at battery/ultracap terminals V bat 100 mv 1000 S/s - 10 S/s Current at battery/ultracap terminals I bat 1.05 A 1000 S/s - 10 S/s Voltage at load terminals V load 100 mv 1000 S/s - 10 S/s Current at load terminals I load 1.05 A 1000 S/s - 10 S/s 5a.1 Test equipment and setup The test system is composed by the following components: an industrial pc used to control the tests and collect the data through an acquisition board, equipped with necessary signal conditioning devices a temperature sensor used to monitor the environment temperature three couples of cables that carry the voltage signals to the acquisition board three shunts used to measure the currents an ac/dc converter used to power the load using AC power, before the grid failure simulations. It is often already installed a controlled switch that breaks the connection between the AC/DC converter and the grid an optional hydrogen sensor used to detect gas leakage in the test environment. 34