Final Report Summary - FLUIDCELL (Advanced m-CHP fuel CELL system based on a novel bio-ethanol Fluidized bed membrane reformer)
Executive Summary:
FluidCELL is a 52 months project focussing on developing a high efficient PEM fuel cell micro Combined Heat and Power cogeneration system (net energy efficiency > 40% and overall efficiency > 90%) for decentralised off-grid applications. The new m-CHP will be based on a novel bio-ethanol fluidised bed catalytic membrane reformer and the most advance technology at the fuel cell level. The main focus of FluidCELL is to develop a new catalytic membrane reformer for pure hydrogen production (≈ 3.5 Nm3/h) from bioethanol reforming in order to intensify the hydrogen production process through the integration of reforming and hydrogen purification in one single unit. The novel reactor will be more efficient than the state-of-the-art technology due to an optimal design aimed at circumventing mass and heat transfer resistances. Besides, the design and optimization of the subcomponents for the BoP for the integration of the membrane reformer to the fuel cell stack has been also addressed.
After more than four years of intense collaboration between the partners, FluidCELL arrived at the end. Main achievements are detailed hereafter.
Novel catalyst and membranes for ATR membrane reactor:
An active and highly stable catalytic formulation was searched for oxidative steam reforming of ethanol in fluidized bed. Different catalysts were tested and the most suitable formulation (Pt-Ni/CeO2-SiO2) was employed for a 400 h-stability test in the fluidized bed. A very interesting behaviour was observed: after an initial catalyst deactivation, a new stationary condition was reached with ethanol conversion and H2 yield of almost 80% the equilibrium value. The above promising formulation was properly characterized to correlate the physiochemical properties with activity as well as stability. 7.4 dm3 of filler (CeO2-SiO2) and 2 dm3 of Pt-Ni/CeO2-SiO2 catalyst were prepared for the catalytic tests in the prototype.
In addition, Pd-based membranes by direct deposition of ultra-thin (< 2 µm) and thin (< 6 µm) dense metal layers onto long ceramic tubular supports have been developed. The H2 permeance of the 0.46 µm thick membrane is extremely high (1.53 x 10-5 mol m-2 s-1 Pa-1) but shows a relatively low H2/N2 ideal perm-selectivity (close to 50). When the thickness increases from 0.46 to 1.17 or 1.29 µm, the H2 permeance reduces to 0.9 x 10-5 mol m-2 s-1 Pa-1 but H2/N2 ideal perm-selectivity increases to around 2,500. In these membranes, defects in the porous support are significant which can explain the difference in selectivity. Despite the high permeation and long-term stability of these Pd-Ag supported membranes, the selectivity of the membranes is not high enough for the membrane reactor. Thin film (≈5 µm) Pd-Ag membranes developed by direct simultaneous Pd and Ag electroless plating (ELP) were selected for the prototype to ensure membrane stability under fluidization regime and ideal H2/N2 perm-selectivy above 8,000. Selective layers were deposited onto asymmetric alumina 50 cm long porous supports with thicker wall (10/4 mm o.d./i.d.). Overall 40 membranes were manufactured.
Membrane reactor a lab-scale:
A fluidized bed multi-membrane reactor (for testing of 5 membranes) was designed and constructed by TU/e for SMR/ATR of methane at different operating conditions (p, T). First, membrane- catalyst interaction and integration strategies for the different components (i.e. sealing) were investigated. Afterward, membranes by TECNALIA have been integrated in the reactor and tested in fluidized conditions showing that equilibrium conditions can be achieved with hydrogen recovery up to 66%. Experimental results were used to validate a fluidized bed membrane reactor model allowing the overall membrane reformer design and its main characteristic. Additionally, new sealing techniques have been studied for ceramic supported membranes.
Design and manufacturing of novel ATR reformer:
A novel bio-ethanol autothermal membrane reactor has been designed, manufactured and tested. The reactor was designed for an output of 3.2 Nm3/h with a Hydrogen Recovery Factor (HRF) of 76.8 %. The reactor employs 37 membranes of 400 mm length. The compact design aims to meet the targets for efficient hydrogen production either as a stand-alone reactor or integrated within the micro-CHP system. Operating conditions of the reactor are 12 bar(a) at 500 °C. Steam is used as sweep gas flowing through the permeate side of the membranes. The reactor is designed to operate at partial load down to 40 % of its nominal value.
During the tests in WP6, a lower hydrogen permeance was observed in the membranes when compared to the values obtained in lab scale tests in WP5. Despite the lower hydrogen permeation, the pilot scale bio-ethanol ATR membrane reformer based on in-situ hydrogen extraction had never before been constructed nor tested at relevant conditions as implemented in FluidCELL. This makes such fuel processor first of its class at pilot scale. A detailed model of the membrane reactor was developed by POLIMI, calibrated against experimental data from lab-scale reactors, and used to simulate the off-design performance of the prototype.
Fuel Cell Stack:
Several Fuel Cell components have been extensively tested in single cells, coupled with modelling for operation understanding, and concomitantly in short stacks (8 cells) including measurements of local current distribution. All tests have been conducted in wide ranges of operating conditions selected by the consortium to help dimensioning and designing system operation. Additional work has been conducted with the analysis of experimental data and modelling on fuel cell components.
Following the validation of the components with the short-stack tests, the fuel cell prototype (120-cells stack) has been assembled and tested at CEA on a test bench for conditioning and validation of performance in some specific conditions, particularly at low current or low hydrogen flow. The PEMFC model developed by POLIMI was useful to simulate the recirculation of high/medium/low-purity hydrogen, depending on the membranes selectivity, and to find the optimal operation in each condition
Integration and validation in the CHP system:
The m-CHP system was fully integrated and ready for testing: Unfortunately, it was not possible to test the fuel processor and the fuel cells stack as a combined unit but only separately. The main reason is that the membrane reactor was not able to produce the desired quality and amount of hydrogen for the minimum time to have a stable operation system and the hydrogen flow rate was too low for feeding the stack module.
Life cycle assessment:
The environmental impacts of the FluidCELL system have been calculated. The FluidCELL system was also compared to other m-CHP systems using other PEM FC technologies, conventional m-CHP with various fuels and systems where the electricity and heat are produced separately by conventional grid/source or renewable source.
The LCA results obtained showed a very high contribution of the bioethanol production and delivery for all indicators. The sensitivity analyses on bioethanol showed that the use of second generation bioethanol is an interesting way to reduce the overall impacts and that depending on the distance from bioethanol production place to the use place, the less impacting bioethanol can correspond to various dilution. The sensitivity analyses on the system design showed that the younger the system, the better the efficiency, the lower the environmental impacts, and that increase the fuel cell area enable to decrease the impacts while doubling the fuel processor membrane area does not. Based on the results of the detailed LCA, key leads for environmental performance optimization were identified: all the elements improving the efficiency in absolute and over time can improve the environmental performance, the type of bioethanol to use should be selected considering the feedstock used, the origin, the transportation, the infrastructure could be decreased trying to reduce the amount of Pt, Pd and Ru use (while keeping the same efficiency), or reducing the weight of the big metallic pieces.
See FluidCELL_Publishable_final-report-18102018_v01.pdf
Project Context and Objectives:
According to the Future of Rural Energy in Europe (FREE) initiative , rural areas represent 90% of all territory in the EU-27 and over 50% of the population (around 46% of population in the world ). They generate 43% of Europe’s gross value and support 55% of all employment. From these, there are at least 30 million homes and businesses which will probably never have access to the natural gas grid and instead largely rely on high carbon intensive energy sources. Paradoxically, rural communities have a higher carbon footprint per person than they need to and often higher than their urban compatriots. This is, among others, because of the fuel sources, as well as the inefficient conventional technologies used.
Major concerns on anthropogenic CO2 emissions and related greenhouse effect have pushed several governments to support greenhouse gas emission reduction policies. EU, for example, set a very high target for reduction of greenhouse gas emissions by 40% compared to 1990 levels within 2030 (at least 80% by 2050), together with an increase by 27% of energy efficiency and renewables share in the energy consumption . Stationary fuel cells offer a clean and efficient source of electricity in systems ranging from 1 kW up to 1 MW or more . With an appropriate fuel processing technology, fuel cells are able to tap into established or accessible sources of fuels such as natural gas but also various other fuels including biofuels and bio-gases.
Distributed power generation via Micro Combined Heat and Power (m-CHP) systems (cogeneration unit with a maximum capacity below 50 kWe ), has been proven to over-come disadvantages of centralized generation since it can give savings in terms of Primary Energy consumption and energy cost (residential applications are about 30% of the total consumption) and CO2 emission reduction. The main advantage is that m-CHP systems are able to recover and use the heat that in centralized systems is often lost. In a distributed generation scenario, fuel cells systems could lead to particularly high efficiencies (electrical efficiency up to 60%, first law efficiency in cogeneration of more than 90%), thereby attaining considerable primary energy saving whilst avoiding transmission losses. However, wide exploitation of these systems is still hindered by high costs and low reliability due to the complexity of the system.
FluidCELL aims at developing a high efficient PEM fuel cell micro Combined Heat and Power cogeneration system (net energy efficiency > 40% and overall efficiency > 90%) for decentralized off-grid applications. The new m-CHP will be based on a novel bio-ethanol fluidised bed catalytic membrane reformer and the most advance technology at the fuel cell level. Taking advantage of, a) bio-ethanol, representing a non-toxic, high energy density, easy handling and commercially available worldwide as a renewable energy supply, b) the fuel cell, that combines high efficiency, low emissions, and low noise, and finally c) Catalytic Membrane Reactor (CMR) achieving, high hydrogen conversion rates, lower working temperatures and smaller physical footprint, FluidCELL targets giving an answer to the large number of off-grid decentralized energy consumers that actually depend on expensive and high polluting sources such as LPGs, bottle gas, heating oil or solid fuels.
This new high efficient PEM fuel cell mCHP system is based on:
- The design, construction and testing of an advanced catalytic membrane reactor for pure hydrogen production (3.5 Nm3/h) from bio-ethanol reforming. The reactor components (catalyst, membranes, heat management etc.) were deeply investigated and optimized. The novel reactor is more efficient than the state-of-the-art technology due to an optimal design aimed at circumventing mass and heat transfer resistances.
- The design and optimization of the subcomponents (BoP) for the integration of the membrane reformer to the fuel cell stack.
The main idea of FluidCELL was to apply the concept of process intensification in the production of hydrogen. FluidCELL develops a novel more efficient and cheaper membrane reactor by intensifying the process of hydrogen production through the integration of reforming and purification in one single unit. While traditional reformers include several steps for producing H2 with enough quality to feed the fuel cell stack (see Figure 1), the new concept addressed in FluidCELL reduces this process to one step (Catalytic Membrane Reactor, see Figure 2). In addition, there is a reduction of the other components in the reformer (heat exchangers) and in the BoP (auxiliary elements) when integrating the membrane reformer to the stack and building up the m-CHP system. In addition, there is a reduction of the reforming temperature.
Figure 1. Layout of PEM m-CHP unit using traditional reforming (TR) for fuel processing.
Figure 2. Layout of PEM m-CHP unit using membrane reformer (MR + sweep gas) for fuel processing.
This general objective is directly related to the development of a novel catalytic membrane reactor (CMR) for hydrogen production with:
- Improved performance (high conversion at low temperature for the autothermal reforming reaction).
- Enhanced efficiency (electrical efficiency of > 40 % compared to conventional 34 %).
- Lifetime ambition (>40,000 hours) under CHP system working conditions.
- Extremely reduced CO2 emissions compared to conventional fossil fuels.
- Good recyclability of its individual components and safety aspects for its integration in domestic CHP systems.
The technical objectives on component level needed to achieve these goals with the bio-ethanol Catalytic Membrane Reformer based m-CHP system are the following:
- Application of advanced, active and selective, catalysts under moderate (< 500ºC) conditions and reduced cost.
- Application of new hydrogen permeable membrane materials with improved separation properties, long durability, and with reduced cost, to be used under reaction conditions.
- To assess the large-scale production of the membrane development.
- Understand the fundamental physico-chemical mechanisms and the relationship between structure/property/performance and manufacturing process in membranes and catalysts, in order to achieve radical improvements in membrane reactors.
- To design, model and build up novel more efficient (e.g. reducing the number of steps) bioethanol catalytic membrane reactor configurations based on the new membranes and catalysts for small-scale pure hydrogen production (3.5 Nm3/h of hydrogen).
- To validate the new membrane reactor configurations, and design a semi-industrial Reforming prototype for pure hydrogen production (3.5 Nm3/h).
- To improve the cost efficiency of membrane reactors by increasing their performance, decreasing the raw materials consumption and the associated energy losses.
Other technical objectives are related to the integration and validation of the bioethanol reformer into the PEM fuel cell CHP system and the proof of concept of the new m-CHP:
See FluidCELL_Publishable_final-report-18102018_v01.pdf
Project Results:
4.1.3. Main S&T results/foregrounds
4.1.3.1. Industrial specification of Fuel Cell CHP-System
One of the first steps in FluidCELL was the assessment of the requirements for the introduction of the novel PEM fuel cell m-CHP system, including the innovative multi-fuel processor, in the market. Therefore, state-of-the-art reactors, PEMFC, CHP systems were identified and in-deep assessment of the components and process parameters was carried out.
Nowadays the environmental policies, focused on carbon reduction and a rational use of primary energy, are leading to a development of various μ-CHP technologies in order to improve efficiency in energy conversion to useful heat and electric power compared with the separate generation.
Different technologies are available for μ-CHP such as Fuel Cell (FC), Internal Combustion Engine (ICE), Stirling engine, micro Gas Turbine (μGT), Organic Rankine Cycle (ORC), Thermo-Photovoltaic (PV-T). Table 1 summarizes the main characteristics of each system.
Table 1. m-CHP system comparison.
Fuel cells based micro-CHP systems are more interesting and studied despite the higher costs than other technologies. Indeed, fuel cells, which convert fuel directly into electricity without combustion, appear to offer very low emissions and high efficiency; furthermore, there are no moving parts which results in very low noise levels.
Table 2 summarizes the available information on the state-of-the-art and the available system (the information is based on internet search and direct contact with vendors at industry fairs). It can be noted that the costs of these systems are higher than normally predicted and can be over 14,000 €/kWe depending on the technology used. Nevertheless, many systems are available by now and several fuel cells based micro-CHP systems have been installed. In Europe both German Callux project and European Enefield project have been installed hundred 1 kWe systems, while in Japan thanks to Ene-Farm project about 50,000 1kWe systems have been installed so far.
The most available systems presented in the previous table have an electrical production of around 1 kWe, lower than the one being developed within the FluidCELL project that is a 5 kWe. Nevertheless, the comparison of these systems is a good guideline for defining the context in which the FluidCELL is being developed and understands the possibilities of commercial success.
As far as the electrical efficiency is concerned, the strength of fuel cells over other system is evident: some FC based systems could achieve values over 40% with also a great total efficiency around 90%.
Some companies have already a commercial product, while others have systems in field trial. Some others plan to have a commercial product from next years. Thus, the FC technology can be considered commercially available by now.
Table 2. Fuel Cell based CHP system comparison.
The systems shown in the previous table are all fed with natural gas or biogas. Few companies (Helbio, Tropical S.A. Prototech and Zsw-Bw) have products suitable to be fed with bio-ethanol, based on PEM Fuel Cell (LT and HT) with size from 1 to 20 kWe. However, no information about systems’ costs or electrical efficiency was available.
The main components of these systems are: fuel, fuel processor, fuel cells, inverter, BOP, controller. The fuel processor must be suitable for supplying pure H2 or syngas of adequate quality for feeding the stacks involved. However, there are few commercial fuel processors having the size of interest for the project. In addition, they are very expensive due to also the use of precious metal (palladium, platinum...). The BoP includes all the remaining components, such as the demineraliser, pumps, valves, etc. For reasons connected to reproducibility, cost and quality, it is essential to use commercial components and not to develop them or require suppliers to develop them ad hoc. Reference systems in the market use fuel cells with polymer membranes (LT-PEM and HT-PEM) as well as ceramic ones (SOFC).
From the collected data it is possible to think that reducing the system cost achieving cost bellow 5,000 €/kWe in mass production (target of this project), could make this promising technology compatible with the current CHP market. Furthermore, the net electric efficiency target higher than 40% using bioethanol and an overall efficiency higher than 90% make this system more attractive than other for an energy, economic and environmental saving.
Economic feasibility analysis
An economic evaluation has been carried out to quantify the real economic feasibility of the FluidCELL system, compared also with commercially available CHP systems. Economic indexes (i.e. Pay-Back Time (PBT), Net Present Value (NPV) and Internal Rate of Return (IRR)) of different cases have been calculated, or solutions to maximizes that indexes have been investigated. an investment results interesting if NPV>0, IRR is highest, and PBT is lowest. Variables considered, assumption and methodology used for the analysis are discussed in the FluidCELL public report: Industrial and market introduction requirements for bio-ethanol fuelled Fuel Cell CHP systems .
It is well known that the energy market (both natural gas and electricity) can be different from country to country, depending on the primary energy used to produce these commodities, which is strictly related to the context and resources of the location. To better understand the impact of micro cogeneration, it is a good practice to evaluate the specific situation of the individual country in which the interventions are performed.
On the other side, cogeneration systems have maximum PBT when all electrical and heat energy produced is used. Furthermore, as efficiency is normally higher when working at full capacity rather than with reduced loads, CHP full power is always considered both in winter and in summer. If the heat and electricity produced are used completely when working at full capacity, it must be considered that installations cover the energy demand of different dwellings, in which a back-up boiler is surely required. The results of a 5 kWe size FluidCELL system which supplies energy for several dwellings are summarized in Table 3 and Figure 4. The analysis consists to find the minimum dwellings number which make interesting the investment, minimizing PBT
Figure 4. FluidCELL system PBT across Europe.
Table 3. FluidCELL system economic indexes and saving per year.
The same indexes have been calculated also for the 1 kWe ICE, 20kWe ICE and a μGT14. The last two have very good indexes, but buildings with several tens dwellings (over hundred in some cases) are necessary. The technical features that the FluidCELL system must have, to be industrially attractive, are summarizes in Table 4.
Table 4. FluidCELL CHP target.
Production costs [€/ kWe] 5,000
Electrical efficiency [%] 40
Total efficiency [%] 90
Lifetime (at least) [year] 10
Availability [hours/year] 7,500
Maintenance per year 1
starting time hours 3
Regulation [%] 25 - 100
CO content < 20 ppm
Catalyst lifetime 5 years
Membrane lifetime 5 years
Main conclusion from the analysis are detailed hereafter:
- FluidCELL system cannot supply only one average dwelling even reducing its size (it’s hold good for all countries).
- It becomes competitive if it can supply several dwellings, also when compared with different systems (as well as when compared with other fuel cell systems),
- The system is not competitive in the market if the natural gas and electricity costs are low, but it becomes very suitable if the electricity to natural gas price ratio is highest.
- Around 5 years PBT is however still a long time if the estimated lifetime is 10 years. In order to start large sales volumes, it is necessary to drop below 3 years. This means a reduction in specific price.
- Until production costs cannot be reduced, it is indispensable to make use of government grants.
- Good lifetime and good availability are essential.
- The recycling of precious materials and the use of standard components seems to be a way to reduce production costs.
4.1.3.2. Novel catalytic materials
Literature surveys of ethanol steam reforming catalysts revealed that rare-earth oxides based catalysts are less prone to deactivation with respect to other common supports (i.e. Al2O3) employed for the above reaction . In particular, bimetallic formulations (Pt-Ni/CeO2) displayed carbon formation rates among the lowest found in the recent literature , which has been selected as the first-generation catalysts. However, despite the very interesting catalytic performances of the above formulation, the results of fluidization tests performed at TU/e revealed very weak mechanical resistance. Therefore, to render fluidizable the selected catalyst, silica gel was chosen as mechanical support. Such material is characterized by a very high specific surface, which is expected to enhance active species dispersion and, consequently, the specific activity of the final catalyst. The second-generation catalysts were prepared by wet impregnation. Before preparation, SiO2 gel powder (Sigma-Aldrich) was calcined in air at 600 °C for 3 h (dT/dt=10 °C/min) in a muffle furnace. The impregnation procedure started with the salt precursor dissolution in bi-distilled water at ambient temperature: Ce(NO3)3∙6H2O, Ni(NO3)2·6H2O and PtCl4, supplied by Strem Chemicals, were selected as CeO2, Ni and Pt precursors, respectively. Once dispersed SiO2 in the aqueous solution of CeO2 precursors, the impregnation was carried out for 1 hour at 80 °C and then drying overnight at 120 °C and calcination, carried out at the same conditions of the support, occurred. Two following impregnations were performed for the active species and Ni was deposited earlier than Pt on ceria/silica support. The whole procedure (impregnation, drying and calcination) was repeated until reaching the desired amounts of CeO2 and active species. The final loads are equal to 20%wt for CeO2 with respect to silica, 3wt% for the noble metal and to 10wt% for the non-noble metal with respect to ceria. The values for metal loads have been optimized in the previous studies . To develop the proper formulation for the three generation catalysts, dedicated tests in the fixed bed reactor were carried out at 500 °C, H2O/C2H5OH ratio of 4, O2/C2H5OH ratio of 0.5 and at a WHSV (weight hourly space velocity, referred to ethanol flow-rate) of 4.1 h-1, in order to optimize CeO2/SiO2 ratio. All the bimetallic catalysts prepared at different CeO2 content exhibited excellent stability: during 100 h-reaction, the samples maintained a stable catalytic activity without deactivation and the average hydrogen yield was 41%. However, for higher times, the faster drop in C2H5OH conversion was recorded over the PtNi25Ce sample, which deactivated after almost 115 h. In the other two cases, 100% of ethanol conversion was measured for almost the same hours but the growth of C2H5OH concentration was more pronounced over the PtNi40Ce with respect to the catalyst containing 30% of CeO2. The spent catalysts were characterized by thermogravimetric and XRD analysis to investigate the factors resulting in their different stability performance. Figure 5 displays the dependence of carbon formation rate from the Ni crystallite sizes: the lowest dimension was recorded for the PtNi30Ce catalyst which showed the minimum carbon selectivity. Also, other authors observed that coke deposition is strongly dependent from the nickel crystallites, proving that the largest Ni crystallites caused the greater carbon deposition. In fact, low Ni surface area retained high residence time of coke precursors, thus increasing the probability of more stable carbonaceous deposits formation. Moreover, to verify the effectiveness of the developed catalysts for oxidative reforming of ethanol, their stability performances were compared, in terms of carbon formation rates, with the data found in the recent literature .Other studies were rarely focused on stability tests higher than 100 hours. Moreover, for WHSV higher than that applied in the present work, total conversion was only observed for C2H5OH concentrations lower than 10% and/or r.a. values double with respect to stoichiometric feeding conditions. At comparable operative conditions (temperature, space velocity and water/ethanol ratio), the catalysts developed in the present study assured CFR values three order of magnitude lower than that found in other works.
Figure 5. Carbon formation rate as a function of Ni average crystallite sizes for the different catalysts. Figure 6. Results of stability tests on CeO2-SiO2 based catalysts at 450°C, H2O/C2H5OH=3 and WHSV=4.1 h-1.
Once optimized the CeO2-SiO2 ratio, a series of bi-metallic and tri-metallic catalysts (Pt-Ni, Rh-Pt-Ni, Rh-Ni, Pt-Ni-K, Pt-Ni-Cs) were tested under more stressful operative conditions (450 °C, H2O/C2H5OH ratio of 3 and in the absence of water); WHSV was fixed to 4.1 h-1. The results of stability tests are shown in Figure 6 in terms of ethanol conversion (X). It was found that Pt-Ni and Rh-Pt-Ni catalysts maintained initial catalytic performances for at least 240 min without apparent deactivation. Conversely, similar C2H5OH profiles were recorded over the Rh-Pt-Ni and Pt-Ni-Cs samples, with an activity reduction from 100 to 97% in the time range of 0-1900 min. The K-containing catalyst displayed the worst performances among the prepared catalysts with a very fast conversion decrease to 93%. The Pt-Ni conversion profile was higher than that recorded over the other samples, with an X value of 95% after 5,000 min. It is also interesting to note that the bimetallic catalyst containing Pt also assures an initial hydrogen yield better than the other samples (26.5%) and very close to the value predicted by thermodynamic equilibrium. The unsatisfactory behaviour of the Pt-Ni-K catalyst can be ascribed to its low surface area and a subsequent worse active species dispersion, which reduced ethanol conversion in reforming reactions. Therefore, a negative effect of support basification, Rh addition as well as Pt substitution by Rh on the long-term stability and H2 yield of Pt-Ni/CeO2-SiO2 catalyst was observed. Based on the above discussion, the Pt-Ni catalyst was selected as the most interesting sample in terms of catalytic performances and employed for further tests.
The third-generation catalyst was also tested in a fluidized bed reactor, properly designed to perform catalytic tests at operative conditions close to that of the prototype. A very long stability test (400 h) was performed over the Pt-Ni/CeO2-SiO2 catalyst at 500°C, H2O/C2H5OH mol ratio of 4, O2/C2H5OH mol ratio of 0.5 and 4.1 h-1 under fluidized bed conditions. Ethanol was completely converted for more than 150 h; for longer TOS, only a partial deactivation was observed which resulted in ethanol conversion as well as hydrogen yield lowering from 100% to 80% and 43.5 to 36%, respectively. In fact, after almost 300 h of TOS, the system reached a stationary condition with no more activity decay. The test was performed for further 100 h and the system was found to hold the plateau condition. The results shown in Figure 7 suggest that carbon formation does not follow a constant trend during the test; conversely, it is supposed that coke deposited during test progressively deactivate the most reactive catalytic sites, involved in both reforming and by-products formation pathways. As a result, catalyst activity remains unchanged with no further increase in catalyst deactivation. To verify the above assumption, a further test in the fluidized bed reactor was performed over the Pt-Ni/CeO2-SiO2 catalyst: at regular intervals, the test was stopped and a sample was withdrawn and characterized by TGA to study the variation of CFR with TOS. However, to accelerate the achievement of the stationary condition, WHSV was triplicate: the results of stability test are shown in Figure 8. At WHSV=12.3 h-1, the system reached a stationary condition after almost 150 h; CFR was quite high at the beginning of the test and displayed a decreasing trend with TOS. In particular, the carbon formation rates after 175 and 200 h were almost the same, proving that the net rate of carbon formation (i.e. the difference between carbon formation and carbon gasification rate) was equal to zero. Also, other authors studied the evolution of coke during reforming reaction in fluidized bed reactors over Ni-based catalysts . However, a markedly different behaviour was observed during operation: almost steady performances were followed by a severe drop in ethanol conversion and finally a slow decrease in ethanol conversion until total deactivation.
Figure 7. Stability test over the Pt-Ni/CeO2-SiO2 catalyst in the FBR, 500°C, H2O/C2H5OH=4, O2/C2H5OH=0.5 WHSV=4.1 h-1.
Figure 8. Trend of C2H5OH conversion and CFR over Pt-Ni/CeO2-SiO2 in the FBR, 500°C, H2O/C2H5OH=4, O2/C2H5OH=0.5 WHSV=12.3 h-1.
Once selected the final formulation for the pilot plant test, two different silica gel materials were employed for the preparation of TU/e and HYGEAR catalysts. In fact, for TU/e tests, a particle size distribution between 150 and 250 µm was required. Conversely, for the tests in the pilot reactor, lower particle sizes (90-115 µm) was necessary. At that hand, two batch of supports, having a pore size of 60 Å, were provided by Sigma-Aldrich. However, concerning the materials for the pilot reactor, UNISA supplier was only able to provide a product with a particle size distribution wider than that required (53-115 µm). Therefore, it was necessary to preliminary sieve the silica gel material for HYGEAR catalysts. The total amount of material prepared was 1.4 dm3 of CeO2/SiO2 and 0.14 dm3 of Pt-Ni/CeO2-SiO2 for TUE tests and 7.4 dm3 of filler (CeO2-SiO2) as well as 2 dm3 of Pt-Ni/CeO2-SiO2 catalyst for the prototype tests. Figure 9 reports some steps of catalyst preparation.
Figure 9. Samples after drying post CeO2 (a), Ni (b) and Pt (c) impregnation.
4.1.3.3. Membranes development
The Department of Energy of the United States of America (DOE) has defined the following targets for membranes for hydrogen separation by 2015: a purity of 99.99% for a H2 flux of 100 Nm3/m2·h at 6.90 bar of differential pressure, a life time higher than 5 years and a cost lower than 1,000 US$/m2 . Among the membranes for H2 separation, Pd based membrane shows the highest permeability and exclusive selectivity for H2 due to the unique permeation mechanism. In order to achieve DOE’s target, very thin Pd membranes (less than 5 µm) are required. Thin Pd based membranes are generally supported on porous substrates including stainless steel and ceramic materials of planar or tubular configuration , . However, pure Pd membranes are often damaged by hydrogen embrittlement due to the α-β phase transition of palladium hydride, which occurs at below the critical temperature (293 °C) and pressure (2 MPa). In comparison to Pd, Pd-Ag alloy membranes are stronger against hydrogen embrittlement (PdH α-β transition at low temperature) and have higher H2 permeability up to 70% (Pd77Ag23) . Pd membranes are very sensitive to sulphur poisoning leading to the decrease of membrane performance , .
There are several technologies reported in the literature for the deposition of Pd and Pd-alloy dense layers on supports, such as physical vapor deposition (PVD, including magnetron sputtering, thermal evaporation, or pulsed lased evaporation), chemical vapor deposition (CVD), electroless plating (ELP), and electroplating (EP)23, . The strength and weaknesses of each of these technologies are also reported on these references.
Membrane activities in the frame of FluidCELL aimed at developing suitable membranes for the prototype (i.e. high permeation and selectivity, long term stability). Both PVD and ELP technologies as well a combination of both technologies were applied for the development of thin film (<6 µm) and/or ultra-thin film (<2 µm) Pd-based membranes. In addition, pore filled membranes were also studied.
Pd-Ag films are not dense when directly deposited on porous supports by PVD-MS. The H2/N2 ideal perm-selectivity selectivities are very low and not suitable for H2 purification. Ultrathin Pd-Ag membrane layers were also prepared by a combination of PVD-MS and ELP techniques, showing a very high H2 permeance (8 x 106 mol m2 s1 Pa1) at 400 °C and 1 at with H2/N2 ideal perm-selectivity around 500. Detailed information and characterization of these membranes can be found in the work of Fernandez et al.
The dependence of the Pd-Ag membrane thickness as a function of the plating time is shown in Figure 10-a 30. The orange and green points refer to the membranes prepared in references [ ] and [ ] respectively. The red dashed line represents a 1.0 µm/h metallic layer growth rate. In the zoom-in graph presented at the right bottom part, the correlation between the membrane thickness and the plating time is presented showing an almost linear dependence. Membrane thicknesses obtained with longer plating times (see orange and green dots) seems to follow this correlation. There may be a small decrease at longer plating times indicating that the membrane growth becomes somewhat slower probably due to the decrease in the concentration of the reactants along the time or to a decrease in the specific surface area for the autocatalytic Pd deposition.
a)
b)
Figure 10. a) Dependence of the time of plating with the thickness of Pd-Ag membranes prepared by ELP; b) H2 permeance and H2/N2 ideal perm-selectivity as a function of the membrane thickness at 400 ºC and 1 bar30.
The H2 permeance and H2/N2 ideal perm-selectivity as a function of the membrane thickness at 400 ºC and 1 bar transmembrane pressure difference is shown in Figure 10-b 30. The H2 permeance of the 0.46 µm thick membrane is extremely high (1.53 x 10-5 mol m-2 s-1 Pa-1) but shows a relatively low H2/N2 ideal perm-selectivity (close to 50). When the thickness increases from 0.46 to 1.17 or 1.29 µm, the H2 permeance reduces to 0.9 x 10-5 mol m-2 s-1 Pa-1. On the other hand, 1.17 and 1.29 µm thick membranes showed similar hydrogen permeances. However, the H2/N2 ideal perm-selectivity was almost double for the thicker membrane. This may indicate that after around 1 hour of plating time the entire surface of the support was covered by a Pd-Ag layer. In these membranes, defects in the porous support are significant which can explain the difference in selectivity.
The ultra-thin membranes developed in the present work show one of the highest H2 permeances reported in the literature for supported membranes (see Table 3 in reference [31]. Despite the high permeation and long-term stability of these Pd-Ag supported membranes (Figure 11), the selectivity of the membranes is not high enough for the membrane reactor of the FluidCELL.
Figure 11. H2 and N2 long-term permeances (mol m-2 s-1 Pa-1) and H2/N2 ideal perm-selectivity of the 1.29 µm thick membrane at 400 °C and 1 bar pressure difference.
Membranes for the prototype
Thin film (≈5 µm) Pd-Ag membranes developed by direct simultaneous Pd and Ag electroless plating (ELP) were selected for the prototype to ensure membrane stability under fluidization regime and ideal H2/N2 perm-selectivy above 8,000. Selective layers were deposited onto asymmetric alumina 50 cm long porous supports with thicker wall (10/4 mm o.d./i.d. provided by Rauschert Kloster Veilsdorf) than in previous developments to improve the mechanical properties compared to the 10/7 mm o.d./i.d. supports (a higher torque could be applied in the connections when using the thicker support). In addition, to reduce the number of sealings in the reactor prototype, it was decided to scale up the length of the membranes from 15-20 cm (original plan) to around 40 cm (Figure 12). Overall, 40 membranes (37+ 3 spare parts) were manufactured.
Cost analysis
A cost analysis of the semi-industrial production of Pd-based membranes considering thin films (< 6 µm) onto ceramic and metallic supports developed by ELP and PVD and ultra-thin film onto ceramic supports was performed. Depending on the production method and type of support applied (ceramic or metallic), a cost per square meter of membrane ranging from 5,000 - 10,800 euro/m2 was calculated. Higher costs are related to the use of the metallic supports instead of ceramic supports. Recycling was not considered in this analysis. When the recycling of both the Pd-Ag waste and the membranes (selective layer and supports, mainly for the metallic supports) is considered in the cost analysis, we could estimate that the cost of the membranes could be between 2,000 - 5,000 euro/m2.
4.1.3.4. Lab-scale bio-ethanol catalytic membrane reformer
The experimental tests have been carried out at the TU/e using the permeation setup under reactive conditions (Figure 13).
Figure 13. The lab facility at the TU/e.
This experimental set-up can perform reactive permeation experiments with a maximum of five membranes. There are three thermocouples, and two pressure sensors distributed over the reactor height. A vacuum pump can be switched on to decrease the partial pressure of the permeate side, as low as 10 mbar. The system does further include feed connections to pure components as O2, H2, N2, CO, CO2, CH4 and pressurized air. The components can be fed through a mass flow controller, which depending on the type of mass flow controller can have a maximum flow of 2 L/min (CO) up to 22 L/min (N2). The retentate can then be analysed by FT-IR. The FT-IR can measure concentrations (after calibration) for water and ethanol, which was not possible at a lab scale. The retentate then passes through a cooler where the water and ethanol are condensed and separated in a flash column. The retentate is connected to a Sick® analyser, which can measure concentrations of the dry retentate. A similar analyser is also connected to the permeate side, which can measure hydrogen concentrations in the range of 0-100% and COx concentrations on a ppm accuracy. The membranes were prepared and delivered by TECNALIA. However, some difficulties were encountered when placing the graphite ferrules. All the membranes used for the experiments had an external diameter in the same range (10.38 mm). The graphite ferrules ordered for these experiments had an average diameter of 10.4 mm. Theoretically, it should be possible to seal the membranes with these graphite ferrules. However, the graphite ferrules started to break when they were stretched the to the desired size. This was solved by gently warming up the graphite ferrules with a heating gun, no visual ruptures could be observed prior to placing the graphite ferrules. For the new tests in future and for the prototype reactor it is suggested to use ferrules of the right size. TUE has already paid the company for the tools to prepare ferrules with different sizes, which can now be ordered also by the partners of FluidCELL.
The membranes connectors were placed by closing with a torque wrench at 11 Nm. Closing the connectors at the same torque (up to14 Nm) as used in earlier experiment has resulted in breaking of the ceramic support. Helium permeation tests were performed after closing the membranes, and the results can be seen in Table 5.
Table 5. Membrane dimension and He leakage test.
The membrane permeation tests were performed at 450 and 500 °C, in excess of hydrogen (to maintain overall reactor concentrations more or less equal to the inlet concentrations), and at different feed pressures. The lab scale tests were performed at hydrogen fractions in the range of 40-100%. There was observed that the slope of the permeation tests does not differ that much as a function of concentrations. The hydrogen fraction ranges for the pilot scale tests were set to be in the range of 25-100%, to assess the permeation at lower concentrations. The membranes were tested individually and all together during the permeation tests. The membranes performed relatively better individually, in comparison to operation with more than one membrane.
The membrane assisted fluidized bed model merged the concept of reaction kinetics with bed hydrodynamics and gas separation. The complete description of the model is presented in our paper C. Ruocco et al .
a)
b)
c)
Figure 14. Nitrogen free retentate compositions (%vol) for multi-membrane reactive experiments at a) 450, b) 500 and, c) 550 °C. By varying feed pressure on x-axis (bar(a)).
The dry retentate composition was continuously analyzed by a Sick ® analyzer. Furthermore, during these experiments it was also possible to measure steam and ethanol compositions due to the presence of an FT-IR analyzer. Due to the rather high nitrogen dilution, the retentate composition will be compared by using nitrogen and steam free compositions.
The optimum numbers of CSTR for the emulsion and bubble phase were found to be 6 and 18 respectively. From Figure 14 we can observe the same as from the performed lab scale experiments, namely higher methane compositions at low temperatures (which indicates low SMR conversions) and low methane compositions at high temperatures (which indicates high SMR conversions). Further, it can be observed that the model predicts the composition accordingly to the experimental results, even if the nitrogen is excluded from the composition (which should make existing deviations even larger). The large CO deviation at 450 °C is not observed, which was the case during the lab scale experiments. Further, it can be observed (just as in the lab-scale results) that the hydrogen fraction increases significantly at higher temperatures (Figure 15) due to the higher CH4 conversion. However, during these experiments no larger deviations are observed when operating at 550 °C and 4 bar(a). The permeate pump connected to the experimental set-up at this scale performs better than the one utilized during single-tube membrane experiments, and is thus able to pump higher gas flow rates. This is beneficial to maintain a low, and constant, permeate pressure which is required in the model as the permeate pressure is there given as a constant value.
a)
b)
c)
Figure 15. Hydrogen permeation, model vs. experimental results at a) 450 °C, b) 500 °C, and c) 550 °C by varying feed pressure on the x-axis.
The HRF during these standard experiments are tabulated in Table 6. From the table we can clearly observe an increasing trend of the HRF with increasing temperature. The HRF is defined as the maximum pure hydrogen obtained relatively from the ethanol present after combustion. This is mainly due to the higher methane conversion at higher temperatures, which enables more hydrogen to permeate through the membrane.
Table 6. HRF vs. T (feed pressure: 4 bar(a)).
By then analyzing the retentate composition from FT-IR at multiple experiments, it can be concluded that no ethanol is present in the retentate stream, as expected from the kinetics study performed and discussed in C. Ruocco et al35.
The permeate concentration was frequently checked during experiments on the CO, CO2, and H2 concentrations. The permeate CO and CO2 concentrations resulted to be in the range of 100 – 300 ppm, whereas the hydrogen concentration was measured constantly at values of >98%, the remaining ±2% is expected to be nitrogen throughout all the experiments.
All the experiments of the study have been plotted in one single parity plot, in which every component is highlighted separately in Figure 16.
Figure 16. Parity plot for all measured and predicted value of the retentate stream.
Components CH4 and CO result to have multiple predicted points with an error larger than 10%, this, however is explained due to the relative high read-off error for these components when utilizing the SickTM-analyzer. The analyzer measures concentrations round off up to one decimal, which gives high deviations for components having low concentrations. From the parity plot above there can be concluded that the proposed reaction kinetics, proposed reaction route and corresponding hydrodynamics are suitable for the description of the system.
4.1.3.5. Design and manufacturing of novel bio-ethanol catalytic membrane reformer
WP6 dealt with the design of the pilot scale membrane reactor for ATR of ethanol. The reactor is intended for simultaneous production and purification of hydrogen in one single step. It uses palladium-based membrane technology developed in WP4. The ATR reaction is sustained in a fluidized bed of particles of catalyst mixed with inert filler. Table 7 summarizes the reactor design specifications.
Table 7. Specifications and definitions on the pilot reactor.
Parameter units Value
Nominal hydrogen output Nm3/h 3.2
Minimum partialization (in terms of EtOH processed) % 40
Sweep gas (nominal) Nm3/h 1.3
Sweep gas (maximum) Nm3/h 1.8
Operating temperature °C 500
Operating pressure barg 11
Permeate pressure mbarg 200
Hydrogen Recovery Factor % 76.8
Hydrogen recovery factor (HRF) is defined as the ratio between the flow of hydrogen permeated over the maximum production capacity given the feed of oxygen needed for temperature control. The calculation of HRF is shown in equation (1).
HRF=F_(H_2,perm)/(6F_(C_2 H_5 OH,feed)-2F_(O_2,feed) ) (1)
A schematic view of the reactor flow configuration is shown in Figure 17. The feed enters the reactor through the distributor in the bottom and flows vertically through the fluidized bed. The first section before reaching the membranes is intended for oxidation and pre-reforming. The membrane reactor section takes place across the height level where the gas is in contact with the membranes. The oxidation supplies the required heat for the endothermic reforming reaction. When hydrogen reaches the membrane, this is permeated across the membrane. Sweep gas flows on the permeate side of the membrane counter-current to the reforming stream. This gas is used for reducing the partial pressure of hydrogen on the permeate side.
Figure 17. Schematic view of the ATR-MR used in the model.
Table 8. Stream compositions at inlet/outlet of reactor.
Pipeline Fuel & steam Air Sweep Permeate Retentate
Comp. (%molar)
EtOH 21.7
H2 71.2 4.7
H2O 78.3 100 28.8 33.7
CO 0.8
CO2 29.7
CH4 3.9
N2 79 27.2
O2 21
Flow (Nm3/h) 3.71 1.61 1.29 4.44 4.68
The characteristics of the membrane tubes used in the pilot reactor are listed in Table 9.
Table 9. Membrane characteristics.
Support Ceramic porous
Outer diameter 10 mm
Inner diameter 4 mm
Effective length 400 mm
Sealing Swagelok union with graphite ferrule
Number of tubes 37
The design of the reactor is shown in Figure 18. The reactor holds 12 thermocouples at different radial and axial positions. An external heat source (electrical) is used to start up the system. The design of the manifold allows the distribution of sweep gas along the permeate side of the membranes and collection of the permeated hydrogen in a single outlet. The design gives the possibility of removing the inner tubes to evaluate the tightness of each membrane. If during operation the impurities were too high, it could be that a tube was leaking. By opening the top flange, it is possible to find out which tube is leaking. This is done without unmounting the membrane manifold.
Figure 18. Reactor design.
Figure 19 shows the Process Flow Diagram (PFD) of the assembled system including the membrane reactor and auxiliaries. The system, as shown in Figure 20 is skid mounted for ease of transport to the testing facility of ICI in Italy.
Figure 19. Process Flow Diagram of the membrane reactor prototype.
Figure 20. Assembled membrane reactor test setup.
The pilot scale autothermal membrane reformer of FluidCELL was successfully tested and validated in WP6. The reactor was tested at HyGear (Arnhem, The Netherlands), before it was shipped to ICI Caldaie (Verona, Italy) for the integration of the complete FluidCELL system in WP8. Initially, permeation tests were performed under non-reactive conditions, and later, test under ATR conditions. During the ATR tests, the temperature in the reactor was controlled fully by the ATR reaction, without the use of the electric heaters, which are used only for heating during start up, neither was any diluent inert gas used. During the ATR tests, several operating conditions were tested, varying parameters such as operating temperature, feed flow, water to ethanol ratio, and sweep flow. The main objective of the test campaign was to characterize the performance of the membrane reactor. The ATR tests were performed using the fully automated mode. During the tests, the controls of the reformer were fine-tuned for optimal response.
Temperature and pressure control accuracy were widely satisfied. With tests closer to nominal conditions a temperature distribution with lower than +/- 4 °C were observed, which would be practically inconceivable in conventional fixed bed ATR.
The results from the test campaign were overall satisfactory. Hydrogen recovery factors (HRF) of 26 % and separation factors (SF) of 59 % were achieved with the first tests. In terms of hydrogen production factor (H2Prod), levels of 44 % were achieved. Hydrogen productivity was in all cases higher than the theoretical hydrogen production of a traditional ATR at the same feed and operating conditions, reaching up to 3.4 times higher with the membrane reactor.
With regards to hydrogen permeation flux, the tests at pilot scaled confirmed the experience obtained at lab scale. Such result means that the hydrogen output is lower than that predicted at the beginning of the project with the design of the reactor in mid-2015. Studies on the phenomena behind this reduction in permeance were very recently submitted for publication to a special issue of Chemical Engineering & Processing on Process Intensification, by project partners TUe and Politecnico di Milano. Based on previous experience at lab scale, it is expected that hydrogen permeation values would further increase by means of a more effective membrane activation.
During the tests at HyGear much knowledge was obtained on the performance of such fuel processor. The reactor of FluidCELL is worldwide first in its kind, capable of producing pure hydrogen from bio-ethanol by means of autothermal reforming in a fluidized bed membrane reactor.
At the end of the testing campaign, the fuel processor was packaged for transport taking appropriate measures to safeguard the fragile membranes, and shipped to ICI Caldaie.
4.1.3.6. Fuel Cell Stack
A state of the art about PEMFC for stationary µCHP application has been proposed by POLIMI showing technologies for low and higher temperature technologies, showing operation features or degradation issues. The SoA report about the µCHP PEMFC systems allowed to define the status and get available data about the commercial fuel cells implemented on systems, showing few existing information or product directly adapted for the FuildCELL needs, like available data in specific conditions, thus confirming relevance to select and test fuel cell components.
Beginning of the project, experimental activities have been conducted in order to prepare the test of selected fuel cell components. Single cell equipment has been adapted for operation with polluted fuel. MEAs and components have been selected as first reference for testing. Small size MEAs have been provided for single cell tests at UPORTO, with start-up, conditioning and testing conditions, so as to enable comparing single cell and stack information. MEAs have also been made for tests in CEA PEMFC short stack of 8 cells. Sensitivity studies were conducted on the single cell and on a short stack (8 cells) made with reformate type reference MEAs (Figure 21).
Wide range of conditions has been defined with the consortium with regard to reformer developed and to system application specifications. Single cells tests have shown, after some adaptation of the testing device, similar performance as same MEAs in stack. All Electrochemical measurements gave information about the impact of conditions mainly related to fuel and air feeding. Simulation of single cell behaviour was conducted to help defining best suited components to be tested for selection. Results showed the influence of the fuel composition, like performance losses due to presence of some CO in the hydrogen produced. On the stack, dead end operation has been tested to check stack voltage compared to fuel circulating mode.
Figure 21. Single cell and 8 cells stack (including the S++ device with 480 segments for the measurement of local current densities) installed on the test stations at UPORTO and at CEA.
During the middle of the project, a deep investigation has been conducted in close collaboration between CEA and POLIMI, in the frame work of the PhD of S. Foresti, dedicated to the analyses of fuel cell data including in-situ measurements of the distribution of local current density focusing on the impact of fuel feeding conditions with mixtures of hydrogen, inert gases like nitrogen, methane, carbon dioxide and small amounts of carbon monoxide, all these components being possibly produced by the reforming process or the fuel feeding strategy like recirculation or dead-end. These results have been published .
Figure 22. Protocol applied to get data for CO/H2-inerts/H2 maps (left side) for different sets of operating conditions for the stack temperature and the fuel and oxidant pressures, flow rates and relative humidity. Example of a map showing the impact of fuel composition on the voltage (right side).
Before the end of the project, additional work has been conducted with the analysis of experimental data, from single cell and stack testing, and modelling on fuel cell components performance.
The electrochemical characterization was completed for the benchmark Membrane Electrodes assemblies (MEAs) provided by CEA. Polarization curves and potential-time history for poisoning tests were used to analyse CO tolerance. The impact of applying internal or external air bleeding to reduce anode pollution when fuel is containing CO was also explored in single cell tests.
Figure 23. Potential-time history at 0.5 A cm-2 with internal air bleeding (left-hand side) and direct air bleeding (right-hand side).
Additionally, a transient multi-dimensional non-isothermal multiphase model for simulating PEM fuel cells operating on pure hydrogen and reformate fuel was developed and validated against experimentally obtained results.
All information needed for the integration of the stack into the ICI system have been detailed and provided to the work package in charge of developing and operating the system.
Main final actions concerned the preparation and delivery of the fuel cell prototype (Figure 24). Electrodes and Membrane Electrode Assemblies were manufactured at CEA with lab scale pilot lines. Test of an 8-cell stack was conducted in several conditions for the validation of the MEAs before making the prototype. End-plates of the prototype stack was specifically designed and ordered.
Figure 24. Fuel cell prototype 120-cell stack tested at CEA and delivered to ICI (left side). Comparison of the polarisation curves in nominal conditions get on CEA test bench and measured at ICI with different conditions for the temperature and gases humidity (right side).
Stack prototype has been assembled, conditioned on a test bench and delivered to ICI in May 2017. ICI could implement the stack on a system but not with the reformer. Few data were obtained operating the stack at ICI but only under pure hydrogen,
Few data could be obtained at ICI, particularly no operation was conducted on the system including the reformer. A polarization curve could be obtained under pure hydrogen with conditions slightly different compared to the tests conducted at CEA before delivery, with lower temperature and probably different humidity for gases; thus, the performance was lower (estimated -7% on the stack power at nominal current point).
As a conclusion about the fuel cell activities in the project FluidCELL, investigations conducted at single cell and short stack levels on test benches allowed to improve the knowledge about the operation of PEMFCs under reformate fuel containing various amounts of contaminants in different sets of operating conditions which is giving valuable information for this technology. The very detailed studies with specific mixtures and in-situ measurements coupled with modelling allowed to further understand and improve the system management with the specific design considered. In addition, the selected core components were manufactured, validated and assembled in a 120-cell stack prototype delivered to ICI for being implemented in the system.
4.1.3.7. Integration and Proof of Concept of the CHP system
The optimization of the system was carried out in two steps. The first step consisted in the definition of the layout to be adopted for the prototype. Two configurations of the membrane reactor were considered in order to enhance the hydrogen permeation across the membranes: the vacuum pump layout and the sweep gas layout. The main output of the simulations are (i) maximize the heat recovery from the exhausts (ii) reduce the water excess to the reformer are fundamental to achieve the target efficiency (>40% electric); (iii) the efficiency of the vacuum pump layout is more sensitive to the reactor pressure than the sweep gas layout; The vacuum layout requires less membranes but has a lower electric efficiency due to the parasitic consumption of the vacuum pump, therefore the sweep gas layout was preferred for the prototype. When defining the layout not only the Pd-membranes area and the number of PEM cells for the stack were considered, for example the arrangements of flows in the heat recovery to avoid extremely high temperatures, which requires expensive heat exchangers, as well as no additional ethanol to the retentate combustor to avoid a liquid/gas burner.
The second step was the detailed simulation of the membrane reactor and fuel cell stack to achieve the best performances of each component, taking into account also off-design operation.
About the membrane reactor a 1D dynamic model was developed by POLIMI including the bubble-emulsion phases, the reaction kinetics of ethanol autothermal reforming, and the gas permeation across the Pd-layer and its support (which resulted one of the most limiting factor in the operation of the sweep gas layout). The model was validated against the experimental data at lab-scale and used to simulate the partial load operation of this unit. An example of the parametric analysis performed with the membrane reactor model is depicted in Figure 25: the gas diffusion through the thick support is a strong limiting factor in hydrogen permeation.
Considering the selectivity issues, a modelling for the optimization of the PEM operating conditions as function of the hydrogen purity has been carried out. In principle, Pd-based membranes can produce high-purity hydrogen (target= 99.99%, i.e. selectivity=10,000). However, the selectivity may decrease and low-purity hydrogen (e.g. 99.9% with 100ppm CO) could be obtained. A methanator is therefore added along the hydrogen cooling line to prevent CO-poisoning of the FC anodes.
Figure 25. Comparison of permeated hydrogen between the ideal (dotted line) and the phenomenological (bars) models for different configurations of the membrane reactor (vacuum/sweep gas layouts).
The quality of hydrogen affects the cell voltage and the overall system performance; therefore, FC control strategies were investigated. Build-up of inert and poisoning species in the hydrogen recirculation loop can be limited by venting a fraction of the anodic off-gas, whose amount can be optimized. These circumstances were investigated by simulating the FCs subsystem that includes the stack, the air and hydrogen blowers and the air humidifier. An experimental campaign on an 8-cells stack (cell area 220 cm2) was performed to characterize the FC operation with reformate gas. The impact of fuel composition, containing up to 20% of inert gases (N2, CO2, CH4) and CO up to 40 ppm, and operative conditions (pressure and humidity) was analysed on the overall stack performance as well as on the current density distribution along the cell surface. Experimental data constituted a valuable source for the validation of the model developed by POLIMI. Main results from the model show that (i) voltage losses can be minimized by adjusting the amount of hydrogen vent; (ii) the most critical regions of the PEM fuel cells are the cathode inlet, due to water deficiency and the anode outlet due to inert gases and CO concentration while hydrogen is consumed.
Figure 26. Integrated m-CHP system.
FluidCELL reactor unit arrived in ICI end September 2017. All the other parts of the system (i.e. Fuel cell, inverter, recirculating pumps, humidifier, controller, and BoP in general) were already set and ready for the connection with the pipes and wires from the reactor.
The system required different easy connections to be done (e.g. mechanical connection of piping, electrical, pneumatic, gases) correlated with a user manual for its properly start up and operation procedure.
As soon as the final integration with the rest of the system began, some unexpected actions appeared essential for the correct installation and operation of the reactor. The main issues to be solved are reported hereafter:
Reactor prototype electrical connections;
High leak in the membrane reactor after the ATR test carried out at HyGear;
Unexpected low H2 permeance in the prototype compare to lab scale;
When test for the membrane activation restart in manual mode, new problems arise and the system was out of control;
Optimization of the reactor prototype working conditions;
The cooling flowrate was too low;
The ethanol flow rate was too low.
Different values of Ethanol flow rate have been tested (higher than 12 Nl/min) but the system was not able to reach stable conditions delivering enough amount of H2 with the appropriate purity, and keep them for a time long enough to perform the analysis (Figure 27). After these tests it was decided to stop the activity because the smell of alcohol gave indication that there was a release of ethanol somewhere in the system, which may lead to unsafe operation. During the last month of activities some extra efforts have been done in changing the hardware (H2 recirculating system, humidifier installation, electrical connection) and the software of the stack module in order to perform some test with pure H2 from bottles instead from fuel processor. The tests done in ICI were highly influenced by the problems arising with system changing conditions (like new H2 humidifier that caused flooding in the stack) and they affected the final results. Since FC stack tests with H2 from bottles were out of the scope of the project and the same tests were already performed by CEA on the stack, this activity was suspended. The data analysis was carried out by POLIMI with the support of the models in order to understand the phenomena at the origin of the poor performance of the system.
Figure 27. Feed, retentate and permeate flow rates in one of the tests.
The technology of the fluidized bed membrane reactor is very promising for future investigations and studies and this project has been useful to understand where these technologies can be improved. In particular, a deeply analysis of the fluidized bed behaviour could help knowing the flexibility of the system and how it responds to the unexpected operating conditions. Improving the membrane strength would results in better long-term reliability and more flexible operations. The slow gas diffusion through the thick porous support limits the effectiveness of the sweep gas layout. Also, the software and the logic stands behind of it should be more flexible and manageable for having a better control on the reaction parameters. More precision on the reactor data collection could help in knowing exactly which conditions the catalyst is subject to. All the work done for developing and building the BoP and the control system has been very useful to improve knowledge and experience about CHP system and membrane reactor. The membrane reactor prototype will be opened to assess the status of the membranes and catalyst.
4.1.3.8. LCA and safety analysis
LCA analysis
The aim was to analyse the environmental profile of the FluidCELL system and identify key leads of improvement, based on the LCA methodology. The functional unit is defined as “Provide 2 German or Italian typical dwellings with useful heat and electricity over one year, in an off-grid situation”.
The LCA performed for this project covered the entire life cycle of the considered systems from the extraction and processing of all raw materials through the end-of-life of all product components (i.e. from “cradle to grave”). In this LCA, the IMPACT 2002+ LCIA method was considered and endpoint (damage-oriented) indicators were presented.
Environmental profile of the FluidCELL system
The LCA showed that the most important contributor to the impacts is the bioethanol production and delivery for the five environmental indicators considered (Figure 28), which is mainly related to bioethanol production part (delivery corresponding to 1 to 20% of this life cycle stage impacts depending on the indicator considered). The bioethanol production impacts are driven by the rye grain cultivation: fertilizers and pesticides production, on field emissions due to fertilizers application, agricultural machinery diesel consumption, energy and water for irrigation and land occupied by the crop. When working on improving the environmental performances of the system, the first action to consider consists of increasing the efficiency of the entire system at the maximum, in order to decrease the amount of bioethanol consumed for the same quantity of energy produced.
The remaining impacts are related to the equipment production, maintenance and end-of-life but the maintenance is the most important. The impact of the equipment is mainly due to the rare metals used: platinum and ruthenium used in the fuel cell electrode, platinum used in the catalyst of the fuel processor and palladium used for the fuel processor membrane. The other important contributors to the equipment impacts are the heavy metallic pieces (large pieces from the BoP as e.g. housing).
Key leads for the optimisation of the environmental performances of the FluidCELL system
Based on the results of the LCA and the different sensitivity analyses performed, some levers can be identified to lower the environmental impacts:
All actions of design that can increase the global efficiency should be explored, since bioethanol production remains the main contributor to the global impacts of the system (increasing the size of equipment should be explored depending on the equipment, maintaining the efficiency over the lifespan of equipment).
Encourage the use of low impacting bioethanol, by selecting a crop used as feedstock that has a high yield, low agrochemical inputs, with no or low irrigation and with no deforestation, or by selecting second generation bioethanol, or by exploring the possibility of bioethanol production on site made from biomass wastes available on site.
Decrease the impacts related to bioethanol delivery by finding a close bioethanol production place to reduce the delivery distance and ensuring to choose the right bioethanol dilution depending on the delivery distance from bioethanol production plant to the off-grid dwellings.
For the additional heat and electricity needed to fulfil the dwellings needs, having a heat pump is an interesting option but exploring wood pellet stove could be interesting too.
Decrease the impacts of infrastructures (m-CHP and heat pump production, maintenance and end-of-life): by ensuring recycling of metals at the end-of-life, by reducing the mass of these elements but always maintaining the efficiency of the system, and by trying to lighten the heavy metallic pieces and ensure that the micro-CHP is easily dismountable at the end-of-life to be able to recycle as much as possible the different pieces.
Figure 28. Climate change, resources consumption, impacts on ecosystem quality and human health and freshwater withdrawal related to the production of heat and electricity to fulfil the needs of 2 Italian or German dwellings over one year. Negative values correspond to an environmental credit and are due to the recycling of some materials in the end-of-life (EOL) stage. Indeed, the recycling provides secondary material on the market that will enable to avoid primary material use.
Safety Assessment
Based on the design of the membrane reactor as well as from the integration of the reformer into the fuel cell CHP-system a safety assessment was performed. Public deliverable D9.5 “Safety assessment of the m-CHP” summarizes the results of the assessment. The deliverable describes the safety standards and directives employed for the assessment and certification of the system. The most relevant results from the HAZOP are outlined, also describing the scoring used for the risk assessment, and the methodology used is described. Moreover, an analysis of the potential failure modes and effects is given for the CHP integration.
The fuel processor of FluidCELL was developed by HyGear. The technology here used incorporates a few chemical processes at elevated temperatures and pressures, and with both flammable and toxic gasses. Therefore, conventional standards, directives and procedures had to be applied to ensure safe design of the system. HyGear has vast experience in these type of processes, therefore internal procedures were already in line with the required standards. The fuel processor was CE certified.
For the fuel processor, the main directives applied were the Machinery Directive (2006/42/EC), the Pressure Equipment Directive (2014/68/EU), the EN 60204-1 Safety of machinery, and the EN 60079-10-1 Explosive atmospheres. A conformity assessment according to module G of PED directive was carried out for the pressurized assembly. This assessment is verified by an external notified body in order to certify that appropriate safety measures had been implemented. Moreover, an inspection for commissioning (Keuring voor Ingebruikneming) was performed by the NoBo.
An analysis of Potential Failure Mode and Effects has been carried out. To each item or function a value for Severity, Probability and Detection has been associated and reported in the table below. Based on the average values of Severity, Probability and Detection, it was evaluated that the probability of failure of the stack part was low and even if something happened it would not have a high severity. A risk analysis was done, and adequate mitigation measures were taken where needed. The safety assessment was essential to bring forward a system with safe and robust operation.
See FluidCELL_Publishable_final-report-18102018_v01.pdf
Potential Impact:
4.1.4. Potential impact
European countries are implementing environmental policies to cope with the EU 20 % emission reduction target set for 2020 (from 1990 levels) . Recently, m-CHP systems received growing interest and experienced further development, thanks to their superior performances in energy conversion. The target of this project is to develop a 5 kW fuel cell m-CHP system based on bioethanol.
Before considering the above-mentioned technology, it is useful to summarize the technologies supporting m-CHP systems. The following technologies can be integrated into a m-CHP system:
➢ ICE (Internal Combustion Engine)
➢ Stirling Engine
➢ μTG (micro Gas Turbine)
➢ ORC (Organic Rankine Cycle)
➢ PV-T (Thermo-Photovoltaic)
Table 10 summarizes the main features of these technologies.
Table 10. m-CHP systems comparison.
Features ICE Stirling μTG ORC PV-T
Electrical Efficiency 25-30 % <20 % 25-33 % 14 % 25 %
Overall Efficiency 80-85 % 65-85 % 70-85 % 80-85 % 80-85 %
Cost €/kWe 6,000 10,000 1,500 --- ---
Availability Commercial Pre-Commercial Commercial Under research Under research
The FluidCELL system is based on a m-CHP Fuel cell unit that runs on bioethanol; the system can be broken down into its main components/subsystems; these are:
➢ Fuel
➢ Fuel Processor
➢ Fuel Cells
➢ Inverter
➢ Balance of Plant (BOP)
➢ Controller
Fuel
Bio-ethanol is gaining strength in the U.S. market. In 2016, it witnessed record production along with unprecedented demand and a growth in exports. Moreover, with the new U.S. administration, it seems that many of the regulatory obstacles impeding the further ethanol growth, may be eliminated. Ethanol is also popular in Brazil and Canada, which represent the two top export consumers for the U.S. market. Ethanol price is significantly lower than that of gasoline on an equivalent heating value. Additionally, ethanol contains low sulphur and metal elements and it is CO2 neutral since it is produced from biomass. Ethanol has a relatively high hydrogen content, it can easily be carried and stored; finally, ethanol blend contains water, which is useful for steam reformers. On the other hand, there are some compounds which could poison or damage the reformer catalyst, including carbon formation. Indeed, bio-ethanol blend includes one-, two- and three-carbon unbranched alcohols: methanol, ethanol and propanol, four- and five-carbon branched alcohols: isobutyl alcohol and the two isomers of pentanol (also known as amyl alcohol) 2-methyl 1-butanol (active amyl alcohol) and 3-methyl, 1-butanol (isoamyl alcohol). Furthermore, ethyl acetate and, the di-ether, 1,1-diethoxyethane could be present as well.
Fuel processor
The Fuel Processor provides pure H2 or suitable syngas to the stacks and it is therefore the core of a m-CHP fuel cell based-system. Under current development there are systems reforming natural gas, gasoline, diesel, and renewable fuels. Companies like Precision Combustion Inc., Innovatek, WS-Reformer-GmbH provide small sized prototypes that are of interest for the project, although they are not yet fully automated, and an operator presence is still necessary. No price information is available; nevertheless, it is known that they are very expensive. Due to the small number of fuel processors produced, a reasonable price can be in the order of thousands € per Nm3/h of hydrogen supplied. Thus, this technology is not yet commercial at all and only research laboratories systems are available.
The fuel processor must supply pure hydrogen or the best suitable syngas to feed PEM stacks. Indeed, the fuel cell behaviour is influenced by the CO content in the syngas which cannot exceed 1 % for HT-PEM and 20 ppm for LT-PEM. Thus, particular attention must be paid to design and build the fuel processor keeping in mind the aim is a commercial product in a competitive market. Each component and the final assembly should not be labour-intensive to avoid quality instability and a subsequent cost increase. Additionally, high costs are incurred due to the noble metals required as catalysts for the fuel processor, as an example, palladium costs 27 €/g . Further attention should be paid to recover the catalyst once the fuel processor has reached the end of its life. Finally, the fuel processor should tolerate a small amount of impurities such as methanol, gasoline (denaturant), butanone, propanol, methyl butanol and thiophene.
TUE and TECNALIA have been developing small scale fuel processors based on membrane reactors. A first prototype of 2 m3/h of pure hydrogen using metallic supported membranes has been constructed and is under test. The 20 m3/h version is being constructed.
Fuel Cells
A Fuel Cell based m-CHP system requires stable, reliable and cost competitive stacks. At present, this technology is making its way through a commercial product, mature, resistant and with feasible costs. Fuel Cell manufacturers are continuously trying to develop technologies to provide a more cost-effective reduction to commercialize competitive products in market. Companies such as Ballard or Serenergy foresee a cost of around 500 €/kWe for the whole stack in the short period. But costs are still high, and not at all competitive; for example, a PEM stack costs around 1,300 €/kWe for Ballard and 1,800 €/kWe for PowerCell.
The most commercially ready technologies available within fuel cells are LT-PEMs, HT-PEMs and SOFC; the last two are still a step behind the other. FluidCELL project uses the LT-PEM, which is the most mature technology available at acceptable prices, even if there are several uncertainties about its stability and duration. Using LT-PEM, poses stringent requirements on the fuel processor because this kind of cells needs pure hydrogen or, according to stacks manufacturers, CO content in syngas up to 20 ppm, even if experimental data have shown that CO values less than 5 ppm give the best electrical performance. Thus, a close restriction on syngas composition is underlined.
Power Conditioning
Power conditioning has already been largely used in solar applications, as a result, it is a well-known technology, available on a large scale and on a wide size range, and finally, it can be easily adapted to the fuel cell technology. Therefore, this component does not pose any significant problem.
Balance of plant
The term Balance of plant identifies all the remaining m-CHP system components such as pumps, blowers, heat injectors, demineralizer and analysers. All of these components must be chosen from commercially available products since they offer lower costs and higher reliability than ad-hoc built ones. Failure of one BoP component causes the whole system malfunction, especially when a fluid management component (like pumps or valves) failure occurs. Experience proves that statistically, the main system problems come just from these components. Thus, particular attention must be paid towards ordinary maintenance of each element to avoid problems in normal operation.
Controller
The development of a fully automated CHP system is the industrial final aim. Therefore, the controller plays a key role in managing the system and seems to be one of the most critical components. To analyse the correct system behaviour, several variables must be acquired and monitored.
Country assessment - most attractive countries
Some requirements for the market introduction for the m-CHP system is hereafter discussed. The aim is to find the right match both in terms of size and configuration for each country as well as establishing economic viability for the FluidCELL system. The following key variables change from one country to another:
➢ Standard and regulation
➢ Government grant
➢ Cost of electricity
➢ Cost of natural gas
➢ Dwellings average consumption of electric energy
➢ Dwellings average consumption of hot water
Investment and O&M costs are assumed constant across of the countries being studied. Standards and regulations tend to differ across countries but all of them come from European Directives. Thus, it was decided to only consider European Standards and not those of the individual countries, assuming that, in the near future, all member states will implement and comply with the various European Directives regarding micro cogeneration. In particular, reference has been made to the CHP Directive 2004/08/EC (On the promotion of cogeneration based on useful heat demand in the internal energy market) and its correlated documents (2007/74/EC and 2008/952/EC).
Moreover, government grants differ among countries. For example, the Italian grant (ESC Energy Saving Certificate) corresponds to an average remuneration of 120 €/toe (0.022 €/kWh) for electricity produced by a high energy efficiency system. In Netherlands the grant is given by the price difference between fossil and renewable power production, up to a max price of 0.15 €/kWh. Switzerland provides grants to programs that are most efficient based on the kWhe consumption avoided . Instead, Portugal government grants are fiscal benefits on Individual Income Tax for high energy class level homes, which are an indirect subsidy for high efficiency systems installed. Thus, for the analysis the Italian grant for all countries has been assumed, only for Netherland will be applied their own grant (the max price of 0.15 €/kWh applicable for CHP from biogas). This is due to lack of information on the other countries government grants and the difficulty to apply their own grants to Portugal and Switzerland. The choice of the Italian grant instead of the Dutch one is due to the lowest impact that it should have in the economic analysis. Energy costs and consumptions of sample countries across Europe, will be presented in next paragraph, while for the bioethanol cost, it has been assumed the same price for all countries because its market isn’t enough widespread like that of gas and electricity. The price assumed is the quoted price at the beginning of June (2018) of 1.462 $/gal (0.034 €/kWh).
European countries – Cross section
The European countries taken into account as sample are the participants to this project, representing different Europe areas (Italy, France, Spain, Portugal, The Netherlands, Switzerland), plus other countries like Germany, Finland, Poland, Austria, Greece, Ireland and Sweden, considered interesting for the electricity to natural gas price ratio (Table 11). This index identifies countries where systems like the one studied in this project could be much profitable due to high electricity price and low natural gas price. Some of these countries are also interesting for the dwelling average energy consumption rate (Table 12), which is very high. Therefore, households might be more sensible towards economic savings arising from m-CHP systems.
The following tables summarize data on energy cost and consumption, useful for an economic evaluation, obtained from the European database. Data includes all taxes and levies and it is referred to the first semester (S1) of 2017.
Table 11. NG and electricity price (2017) .
Country Natural Gas 2017 S1 [€/kWh] Electricity 2017 S1 [€/kWh] EE/NG Ratio Spark spread [€/kWh]
AUSTRIA 0.067 0.195 2.893 0.128
FINLAND 0.051 0.193 3.784 0.142
FRANCE 0.064 0.169 2.645 0.105
GERMANY 0.061 0.305 4.989 0.244
GREECE 0.056 0.194 3.457 0.138
IRELAND 0.063 0.231 3.647 0.167
ITALY 0.070 0.214 3.043 0.144
NETHERLANDS 0.076 0.156 2.047 0.080
POLAND 0.042 0.146 3.494 0.104
PORTUGAL 0.077 0.228 2.955 0.151
SPAIN 0.067 0.230 3.442 0.163
SWEDEN 0.121 0.194 1.597 0.072
SWITZERLAND 0.097 0.152 1.567 0.055
Table 12. Unit consumption per dwelling by end uses (2014 Report).
Country Electricity [kWh/y] Space heating [kWh/y] Water heating [kWh/y]
AUSTRIA 2,675 15,468 2,326
FINLAND 4,187 16,980 2,442
FRANCE 2,793 12,432 1,841
GERMANY 2,210 13,723 2,210
GREECE 3,024 10,816 1,047
IRELAND 2,576 15,835 3,666
ITALY 2,185 9,380 1,169
NETHERLANDS 2,752 11,688 2,460
POLAND 2,358 11,791 2,358
PORTUGAL 1,996 1,980 1,986
SPAIN 2,232 4,793 2,912
SWEDEN 3,722 12,677 1,977
SWITZERLAND 3,275 13,194 2,681
Since within the same country a lot of variability can occur in consumption of electricity and heat, the financial analysis will account for dwelling average data about energy consumption. Indeed, in Italy, the heating consumption difference between Sicily and Alpine location is remarkable. Moreover, variables such as family size, energy class, dwelling size and seasonality can vary greatly as well. As a result, the data presented is semi-quantitative and every application must be considered individually to obtain the maximum benefit in consumption reduction. In this respect, the Energy Service Companies, ESCO, can provide ad-hoc configuration for each situation.
Comparing FLUIDCELL with existing m-CHP systems
To provide an overview of the context in which the FluidCELL system is found, a comparison with the main technologies for m-CHP systems is now discussed. Table 13 provides a comparison among the systems based on electrical and thermal performances and efficiencies. Table 14 summarizes the yearly amount of electricity and thermal energy output for each system.
Table 13. Systems powers & efficiencies.
System EE Power (Size) EE Efficiency Tot. Efficiency TH Power
[kWe] [%] [%] [kWt]
ICE 1kW 1 26.3 85 1.8
ICE 20kW 20 30 85 36.7
Stirling 1 15 85 3.3
μTG 30 26 85 47.3
FC 1 40 90 1.3
FluidCELL 5 40 90 6.25
Table 14. System costs & yearly energy produced.
System Investment Cost O&M Cost Electricity Thermal energy
[€/kWe] [€/year] [kWhe/year] [kWht/year]
ICE 1kW 6,000 86 7,500 13,750
ICE 20kW 2,000 1,716 150,000 275,000
Stirling 10,000 114 7,500 24,375
μTG 1,700 3,104 225,000 354,545
FC 20,000 117 7,500 9,375
FluidCELL 10,000 587 37,500 46,875
Financial Analysis
To perform the financial analysis for the FluidCELL system, three parameters have been considered: pay-back time (PBT), net present value (NPV) and internal rate of return (IRR). The rationale for establishing economic feasibility requires NPV>0, high IRR and low PBT.
To perform this analysis the following costs have been taken into account:
➢ Natural gas and electricity for separate generation
➢ Operation and Maintenance costs of back-up boiler
➢ Cost of Natural Gas for back-up boiler
➢ Government grants
Moreover, the investment cost considered the system cost, the bio-ethanol consumed by the system and the O&M costs. For the evaluation of the economic indexes the cash flow is discounted with an inflation long term forecast rate of 2 % and a nominal real risk-free rate of 4.5 %.
A service lifetime of 10 years and an availability of 7,500 hours per year have been assumed. Moreover, it was assumed the system will operate 3,924 hours in winter season and 3576 hours in summer. Since dwelling energy consumption data is an average value, a simplification on system partial load has been introduced with two periods of regulation, winter load and summer load.
The financial analysis prioritized payback time (PBT) reduction in order to let the system become more attractive for the costumer. For doing that, it is assumed that during all seasons the system has the same (maximum) load tuned on minimum thermal requirement by the dwellings. In this way, all the electrical and thermal power is used and electricity from the grid as well as a backup boiler is also required.
Even though sharing one FluidCELL system among a number of dwellings higher than 10 may not be feasible, it provides valuable information about FluidCELL financial performances in this optimistic setting. To assure the shortest PBT, the system must run 24/7 at full workload in both summer and winter and all the energy produced must be used. Thus, several dwellings are required.
Determine the cost of the investment
Table 18 includes the costs for the FluidCELL components provided by all the partners of this project for three different case scenarios, that is, for 1, 100 and 1,000 units production per year. While for the Pd based membranes TECNALIA provided their own cost estimates, the price for the other components was calculated from the prototype with historic cost evolution from ICI purchasing department for new products while moving from prototype to large sales volume.
The starting point for the costs analysis is to divide the components in three types of product:
➢ Market-standard components based on mature technology, the cost of this commodity doesn’t change so much with the quantity
➢ Market-specific components based on mature technology, study-time have been spent by the supplier to reduce significantly the cost when quantity increases.
➢ Immature component that has required a lot of study time to develop, it is just a prototype phase.
In the tables below are presented how the quantity of production affects the costs evolution for each component’s typology described above (cost evolution based on ICI purchasing department unit).
Table 15. Market-Standard components ratio cost evolution/quantity.
Production quantity impact for a mature technology
Standard components (such as electrotech components)
Quantity Ratio Product cost /
single standard component cost
1 100.0
10 99.0
100 96.5
1,000 95.0
Table 16. Market-Specific components ratio cost evolution/quantity.
Production quantity impact for a mature technology
Specific components (sheet metal, electrical board...)
Quantity Ratio Product cost /
single specific component cost
1 100
10 70
100 50
1,000 30
Table 17. Market-“Immature” components ratio costs evolution/quantity.
Production quantity impact for an immature technology
Reference used: ICI burner designed and developed with ad hoc features
Quantity Ratio Product cost /
Prototype cost
1 1
10 0.753
100 0.452
1,000 0.070
Additionally, the cost for the Fuel Cell is based on an estimate from the 2017 report by the Department of Energy (USA) that set a cost of 50 $/kWe when mass production is considered. The most cost-relevant component appears to be the Fuel Processor subsystem.
Table 18. FluidCELL Cost per Unit.
Price per 1 unit [€] Price per 100 unit [€] Price per 1000 unit [€]
Description Company Component/
material Cost Labor
Cost Total
Cost Component/
material Cost Labor
Cost Total
Cost Component/
material Cost Labor
Cost Total
Cost
Catalyst UNISA 2,500 2,500 5,000 250 1,250 1,500 200 200 400
Pd Membrane TECNALIA - - 13,000 - - 3,400 - - 2,500
Fuel Processor HYGEAR 65,000 15,000 80,000 40,000 10,000 50,000 20,000 5,000 25,000
Fuel Cell CEA - - 35,000 - - 10,000 - - 250
Integration ICI 35,240 123,200 158,440 26,599 36,960 63,559 16,382 6,160 22,542
291,440 128,459 50,692
FluidCELL System Analysis
To decrease the payback time (PBT) the final user has to fully consume all the electrical and thermal output of the system and run the system at full load all the time. Furthermore, efficiency is higher when working at full capacity rather than with reduced workloads. Therefore, to maximize efficiency the system has to cover the energy demand of multiple dwellings, thus, it requires a back-up boiler. With this consideration and merging data from Table 11 (NG and electricity price in different European country) and Table 12 (Unit consumption per dwelling by end uses) the following Table 19 can be obtained.
Table 19. FluidCELL Financial Analysis.
Country Dwellings PBT [Years] NPV [€/Y] IRR Saving[€/Y]
AUSTRIA 14 9 2,049 1% 7,253.15
FINLAND 13 >10 - 16,224 - 7% 4,858.38
FRANCE 26 >10 -2,645 -1% 6,637.87
GERMANY 18 5 33,327 11% 11,352.16
GREECE 40 9 195 0% 7,010.16
IRELAND 14 7 12,588 5% 8,634.26
ITALY 17 8 4,492 2% 7,573.22
NETHERLANDS 19 >10 - 8,058 - 3% 5,928.54
POLAND 16 >10 - 25,245 -11% 3,676.11
PORTUGAL 19 6 16,558 6% 9,154.57
SPAIN 17 7 16,001 6% 9,081.62
SWEDEN 19 6 22,607 8% 9,947.32
SWITZERLAND 17 8 5,095 2% 7,652.33
Figure 29 visualizes how the different countries performed based on PBT. The current analysis yields a cost of 10,000 €/kWe. Germany appears to be the most attractive country according to all three financial indicators. Across the countries examined, the average Payback Time is 8 years while the life of the FluidCELL system is 10 years.
Figure 29. FluidCELL PayBack Time.
Main conclusions
➢ FluidCELL has to provide energy to several dwellings, from the analysis emerged 19 dwellings are associated on average with the system. Multiple dwellings have been introduced due to the necessity to let the system always operate at full workload, while also utilizing all of the electrical and thermal energy produced.
➢ The system cost is 10,000 €/kWhe.
➢ The lowest PBT for the FluidCELL system has been achieved by Germany which scored 5 years PBT, still, the average PBT for the FluidCELL system is 8 years.
➢ To achieve an attractive 3-year PBT, the individual component costs have to decrease.
➢ Currently, the Fuel Processor is the most critical cost component for FluidCELL, accounting for 49 % of the system total cost.
➢ Since production costs are still high, it is advisable to make use of government grants.
➢ If possible, to further reduce production costs, it is advisable to increase the use of standardized components.
➢ Further work is required in order to reduce investment cost through research in optimization and industrialization of fuel processor and system integration
See FluidCELL_Publishable_final-report-18102018_v01.pdf
List of Websites:
The address of the public Website of the Project as well as relevant contact details.
Project public website with further information of the about the project and consortium and main contacts details are detailed hereafter:
https://www.fluidcell.eu/
Project manager: Dr. José Luis Viviente
e-mail: joseluis.viviente@tecnalia.com
Technical manager : Prof. Fausto Gallucci
e-mail: F.Gallucci@tue.nl
Dissemination manager: Associate Prof. Giampaolo Manzolini
e-mail: giampaolo.manzolini@polimi.it
Exploitation manager: Dr. Leonardo Roses
e-mail: leonardo.roses@hygear.com
List of all beneficiaries with the corresponding contact name and associated coordinates
Nº Participant short name Contact name E-mail
1 TECNALIA José Luis Viviente joseluis.viviente@tecnalia.com
2 TU/e Fausto Gallucci f.gallucci@tue.nl
3 CEA Sylvie Escribano sylvie.escribano@cea.fr
4 POLIMI Giampaolo Manzolini giampaolo.manzolini@polimi.it
5 UNISA Vincenzo Palma vpalma@unisa.it
6 UPORTO Adelio Mendes mendes@fe.up.pt
7 ICI Carlo Tregambe carlo.tregambe@icicaldaie.com
8 HYG Leonardo Roses leonardo.roses@hygear.com
9 QUANTIS Violaine Maguad violaine.magaud@quantis-intl.com
See FluidCELL_Publishable_final-report-18102018_v01.pdf
FluidCELL is a 52 months project focussing on developing a high efficient PEM fuel cell micro Combined Heat and Power cogeneration system (net energy efficiency > 40% and overall efficiency > 90%) for decentralised off-grid applications. The new m-CHP will be based on a novel bio-ethanol fluidised bed catalytic membrane reformer and the most advance technology at the fuel cell level. The main focus of FluidCELL is to develop a new catalytic membrane reformer for pure hydrogen production (≈ 3.5 Nm3/h) from bioethanol reforming in order to intensify the hydrogen production process through the integration of reforming and hydrogen purification in one single unit. The novel reactor will be more efficient than the state-of-the-art technology due to an optimal design aimed at circumventing mass and heat transfer resistances. Besides, the design and optimization of the subcomponents for the BoP for the integration of the membrane reformer to the fuel cell stack has been also addressed.
After more than four years of intense collaboration between the partners, FluidCELL arrived at the end. Main achievements are detailed hereafter.
Novel catalyst and membranes for ATR membrane reactor:
An active and highly stable catalytic formulation was searched for oxidative steam reforming of ethanol in fluidized bed. Different catalysts were tested and the most suitable formulation (Pt-Ni/CeO2-SiO2) was employed for a 400 h-stability test in the fluidized bed. A very interesting behaviour was observed: after an initial catalyst deactivation, a new stationary condition was reached with ethanol conversion and H2 yield of almost 80% the equilibrium value. The above promising formulation was properly characterized to correlate the physiochemical properties with activity as well as stability. 7.4 dm3 of filler (CeO2-SiO2) and 2 dm3 of Pt-Ni/CeO2-SiO2 catalyst were prepared for the catalytic tests in the prototype.
In addition, Pd-based membranes by direct deposition of ultra-thin (< 2 µm) and thin (< 6 µm) dense metal layers onto long ceramic tubular supports have been developed. The H2 permeance of the 0.46 µm thick membrane is extremely high (1.53 x 10-5 mol m-2 s-1 Pa-1) but shows a relatively low H2/N2 ideal perm-selectivity (close to 50). When the thickness increases from 0.46 to 1.17 or 1.29 µm, the H2 permeance reduces to 0.9 x 10-5 mol m-2 s-1 Pa-1 but H2/N2 ideal perm-selectivity increases to around 2,500. In these membranes, defects in the porous support are significant which can explain the difference in selectivity. Despite the high permeation and long-term stability of these Pd-Ag supported membranes, the selectivity of the membranes is not high enough for the membrane reactor. Thin film (≈5 µm) Pd-Ag membranes developed by direct simultaneous Pd and Ag electroless plating (ELP) were selected for the prototype to ensure membrane stability under fluidization regime and ideal H2/N2 perm-selectivy above 8,000. Selective layers were deposited onto asymmetric alumina 50 cm long porous supports with thicker wall (10/4 mm o.d./i.d.). Overall 40 membranes were manufactured.
Membrane reactor a lab-scale:
A fluidized bed multi-membrane reactor (for testing of 5 membranes) was designed and constructed by TU/e for SMR/ATR of methane at different operating conditions (p, T). First, membrane- catalyst interaction and integration strategies for the different components (i.e. sealing) were investigated. Afterward, membranes by TECNALIA have been integrated in the reactor and tested in fluidized conditions showing that equilibrium conditions can be achieved with hydrogen recovery up to 66%. Experimental results were used to validate a fluidized bed membrane reactor model allowing the overall membrane reformer design and its main characteristic. Additionally, new sealing techniques have been studied for ceramic supported membranes.
Design and manufacturing of novel ATR reformer:
A novel bio-ethanol autothermal membrane reactor has been designed, manufactured and tested. The reactor was designed for an output of 3.2 Nm3/h with a Hydrogen Recovery Factor (HRF) of 76.8 %. The reactor employs 37 membranes of 400 mm length. The compact design aims to meet the targets for efficient hydrogen production either as a stand-alone reactor or integrated within the micro-CHP system. Operating conditions of the reactor are 12 bar(a) at 500 °C. Steam is used as sweep gas flowing through the permeate side of the membranes. The reactor is designed to operate at partial load down to 40 % of its nominal value.
During the tests in WP6, a lower hydrogen permeance was observed in the membranes when compared to the values obtained in lab scale tests in WP5. Despite the lower hydrogen permeation, the pilot scale bio-ethanol ATR membrane reformer based on in-situ hydrogen extraction had never before been constructed nor tested at relevant conditions as implemented in FluidCELL. This makes such fuel processor first of its class at pilot scale. A detailed model of the membrane reactor was developed by POLIMI, calibrated against experimental data from lab-scale reactors, and used to simulate the off-design performance of the prototype.
Fuel Cell Stack:
Several Fuel Cell components have been extensively tested in single cells, coupled with modelling for operation understanding, and concomitantly in short stacks (8 cells) including measurements of local current distribution. All tests have been conducted in wide ranges of operating conditions selected by the consortium to help dimensioning and designing system operation. Additional work has been conducted with the analysis of experimental data and modelling on fuel cell components.
Following the validation of the components with the short-stack tests, the fuel cell prototype (120-cells stack) has been assembled and tested at CEA on a test bench for conditioning and validation of performance in some specific conditions, particularly at low current or low hydrogen flow. The PEMFC model developed by POLIMI was useful to simulate the recirculation of high/medium/low-purity hydrogen, depending on the membranes selectivity, and to find the optimal operation in each condition
Integration and validation in the CHP system:
The m-CHP system was fully integrated and ready for testing: Unfortunately, it was not possible to test the fuel processor and the fuel cells stack as a combined unit but only separately. The main reason is that the membrane reactor was not able to produce the desired quality and amount of hydrogen for the minimum time to have a stable operation system and the hydrogen flow rate was too low for feeding the stack module.
Life cycle assessment:
The environmental impacts of the FluidCELL system have been calculated. The FluidCELL system was also compared to other m-CHP systems using other PEM FC technologies, conventional m-CHP with various fuels and systems where the electricity and heat are produced separately by conventional grid/source or renewable source.
The LCA results obtained showed a very high contribution of the bioethanol production and delivery for all indicators. The sensitivity analyses on bioethanol showed that the use of second generation bioethanol is an interesting way to reduce the overall impacts and that depending on the distance from bioethanol production place to the use place, the less impacting bioethanol can correspond to various dilution. The sensitivity analyses on the system design showed that the younger the system, the better the efficiency, the lower the environmental impacts, and that increase the fuel cell area enable to decrease the impacts while doubling the fuel processor membrane area does not. Based on the results of the detailed LCA, key leads for environmental performance optimization were identified: all the elements improving the efficiency in absolute and over time can improve the environmental performance, the type of bioethanol to use should be selected considering the feedstock used, the origin, the transportation, the infrastructure could be decreased trying to reduce the amount of Pt, Pd and Ru use (while keeping the same efficiency), or reducing the weight of the big metallic pieces.
See FluidCELL_Publishable_final-report-18102018_v01.pdf
Project Context and Objectives:
According to the Future of Rural Energy in Europe (FREE) initiative , rural areas represent 90% of all territory in the EU-27 and over 50% of the population (around 46% of population in the world ). They generate 43% of Europe’s gross value and support 55% of all employment. From these, there are at least 30 million homes and businesses which will probably never have access to the natural gas grid and instead largely rely on high carbon intensive energy sources. Paradoxically, rural communities have a higher carbon footprint per person than they need to and often higher than their urban compatriots. This is, among others, because of the fuel sources, as well as the inefficient conventional technologies used.
Major concerns on anthropogenic CO2 emissions and related greenhouse effect have pushed several governments to support greenhouse gas emission reduction policies. EU, for example, set a very high target for reduction of greenhouse gas emissions by 40% compared to 1990 levels within 2030 (at least 80% by 2050), together with an increase by 27% of energy efficiency and renewables share in the energy consumption . Stationary fuel cells offer a clean and efficient source of electricity in systems ranging from 1 kW up to 1 MW or more . With an appropriate fuel processing technology, fuel cells are able to tap into established or accessible sources of fuels such as natural gas but also various other fuels including biofuels and bio-gases.
Distributed power generation via Micro Combined Heat and Power (m-CHP) systems (cogeneration unit with a maximum capacity below 50 kWe ), has been proven to over-come disadvantages of centralized generation since it can give savings in terms of Primary Energy consumption and energy cost (residential applications are about 30% of the total consumption) and CO2 emission reduction. The main advantage is that m-CHP systems are able to recover and use the heat that in centralized systems is often lost. In a distributed generation scenario, fuel cells systems could lead to particularly high efficiencies (electrical efficiency up to 60%, first law efficiency in cogeneration of more than 90%), thereby attaining considerable primary energy saving whilst avoiding transmission losses. However, wide exploitation of these systems is still hindered by high costs and low reliability due to the complexity of the system.
FluidCELL aims at developing a high efficient PEM fuel cell micro Combined Heat and Power cogeneration system (net energy efficiency > 40% and overall efficiency > 90%) for decentralized off-grid applications. The new m-CHP will be based on a novel bio-ethanol fluidised bed catalytic membrane reformer and the most advance technology at the fuel cell level. Taking advantage of, a) bio-ethanol, representing a non-toxic, high energy density, easy handling and commercially available worldwide as a renewable energy supply, b) the fuel cell, that combines high efficiency, low emissions, and low noise, and finally c) Catalytic Membrane Reactor (CMR) achieving, high hydrogen conversion rates, lower working temperatures and smaller physical footprint, FluidCELL targets giving an answer to the large number of off-grid decentralized energy consumers that actually depend on expensive and high polluting sources such as LPGs, bottle gas, heating oil or solid fuels.
This new high efficient PEM fuel cell mCHP system is based on:
- The design, construction and testing of an advanced catalytic membrane reactor for pure hydrogen production (3.5 Nm3/h) from bio-ethanol reforming. The reactor components (catalyst, membranes, heat management etc.) were deeply investigated and optimized. The novel reactor is more efficient than the state-of-the-art technology due to an optimal design aimed at circumventing mass and heat transfer resistances.
- The design and optimization of the subcomponents (BoP) for the integration of the membrane reformer to the fuel cell stack.
The main idea of FluidCELL was to apply the concept of process intensification in the production of hydrogen. FluidCELL develops a novel more efficient and cheaper membrane reactor by intensifying the process of hydrogen production through the integration of reforming and purification in one single unit. While traditional reformers include several steps for producing H2 with enough quality to feed the fuel cell stack (see Figure 1), the new concept addressed in FluidCELL reduces this process to one step (Catalytic Membrane Reactor, see Figure 2). In addition, there is a reduction of the other components in the reformer (heat exchangers) and in the BoP (auxiliary elements) when integrating the membrane reformer to the stack and building up the m-CHP system. In addition, there is a reduction of the reforming temperature.
Figure 1. Layout of PEM m-CHP unit using traditional reforming (TR) for fuel processing.
Figure 2. Layout of PEM m-CHP unit using membrane reformer (MR + sweep gas) for fuel processing.
This general objective is directly related to the development of a novel catalytic membrane reactor (CMR) for hydrogen production with:
- Improved performance (high conversion at low temperature for the autothermal reforming reaction).
- Enhanced efficiency (electrical efficiency of > 40 % compared to conventional 34 %).
- Lifetime ambition (>40,000 hours) under CHP system working conditions.
- Extremely reduced CO2 emissions compared to conventional fossil fuels.
- Good recyclability of its individual components and safety aspects for its integration in domestic CHP systems.
The technical objectives on component level needed to achieve these goals with the bio-ethanol Catalytic Membrane Reformer based m-CHP system are the following:
- Application of advanced, active and selective, catalysts under moderate (< 500ºC) conditions and reduced cost.
- Application of new hydrogen permeable membrane materials with improved separation properties, long durability, and with reduced cost, to be used under reaction conditions.
- To assess the large-scale production of the membrane development.
- Understand the fundamental physico-chemical mechanisms and the relationship between structure/property/performance and manufacturing process in membranes and catalysts, in order to achieve radical improvements in membrane reactors.
- To design, model and build up novel more efficient (e.g. reducing the number of steps) bioethanol catalytic membrane reactor configurations based on the new membranes and catalysts for small-scale pure hydrogen production (3.5 Nm3/h of hydrogen).
- To validate the new membrane reactor configurations, and design a semi-industrial Reforming prototype for pure hydrogen production (3.5 Nm3/h).
- To improve the cost efficiency of membrane reactors by increasing their performance, decreasing the raw materials consumption and the associated energy losses.
Other technical objectives are related to the integration and validation of the bioethanol reformer into the PEM fuel cell CHP system and the proof of concept of the new m-CHP:
See FluidCELL_Publishable_final-report-18102018_v01.pdf
Project Results:
4.1.3. Main S&T results/foregrounds
4.1.3.1. Industrial specification of Fuel Cell CHP-System
One of the first steps in FluidCELL was the assessment of the requirements for the introduction of the novel PEM fuel cell m-CHP system, including the innovative multi-fuel processor, in the market. Therefore, state-of-the-art reactors, PEMFC, CHP systems were identified and in-deep assessment of the components and process parameters was carried out.
Nowadays the environmental policies, focused on carbon reduction and a rational use of primary energy, are leading to a development of various μ-CHP technologies in order to improve efficiency in energy conversion to useful heat and electric power compared with the separate generation.
Different technologies are available for μ-CHP such as Fuel Cell (FC), Internal Combustion Engine (ICE), Stirling engine, micro Gas Turbine (μGT), Organic Rankine Cycle (ORC), Thermo-Photovoltaic (PV-T). Table 1 summarizes the main characteristics of each system.
Table 1. m-CHP system comparison.
Fuel cells based micro-CHP systems are more interesting and studied despite the higher costs than other technologies. Indeed, fuel cells, which convert fuel directly into electricity without combustion, appear to offer very low emissions and high efficiency; furthermore, there are no moving parts which results in very low noise levels.
Table 2 summarizes the available information on the state-of-the-art and the available system (the information is based on internet search and direct contact with vendors at industry fairs). It can be noted that the costs of these systems are higher than normally predicted and can be over 14,000 €/kWe depending on the technology used. Nevertheless, many systems are available by now and several fuel cells based micro-CHP systems have been installed. In Europe both German Callux project and European Enefield project have been installed hundred 1 kWe systems, while in Japan thanks to Ene-Farm project about 50,000 1kWe systems have been installed so far.
The most available systems presented in the previous table have an electrical production of around 1 kWe, lower than the one being developed within the FluidCELL project that is a 5 kWe. Nevertheless, the comparison of these systems is a good guideline for defining the context in which the FluidCELL is being developed and understands the possibilities of commercial success.
As far as the electrical efficiency is concerned, the strength of fuel cells over other system is evident: some FC based systems could achieve values over 40% with also a great total efficiency around 90%.
Some companies have already a commercial product, while others have systems in field trial. Some others plan to have a commercial product from next years. Thus, the FC technology can be considered commercially available by now.
Table 2. Fuel Cell based CHP system comparison.
The systems shown in the previous table are all fed with natural gas or biogas. Few companies (Helbio, Tropical S.A. Prototech and Zsw-Bw) have products suitable to be fed with bio-ethanol, based on PEM Fuel Cell (LT and HT) with size from 1 to 20 kWe. However, no information about systems’ costs or electrical efficiency was available.
The main components of these systems are: fuel, fuel processor, fuel cells, inverter, BOP, controller. The fuel processor must be suitable for supplying pure H2 or syngas of adequate quality for feeding the stacks involved. However, there are few commercial fuel processors having the size of interest for the project. In addition, they are very expensive due to also the use of precious metal (palladium, platinum...). The BoP includes all the remaining components, such as the demineraliser, pumps, valves, etc. For reasons connected to reproducibility, cost and quality, it is essential to use commercial components and not to develop them or require suppliers to develop them ad hoc. Reference systems in the market use fuel cells with polymer membranes (LT-PEM and HT-PEM) as well as ceramic ones (SOFC).
From the collected data it is possible to think that reducing the system cost achieving cost bellow 5,000 €/kWe in mass production (target of this project), could make this promising technology compatible with the current CHP market. Furthermore, the net electric efficiency target higher than 40% using bioethanol and an overall efficiency higher than 90% make this system more attractive than other for an energy, economic and environmental saving.
Economic feasibility analysis
An economic evaluation has been carried out to quantify the real economic feasibility of the FluidCELL system, compared also with commercially available CHP systems. Economic indexes (i.e. Pay-Back Time (PBT), Net Present Value (NPV) and Internal Rate of Return (IRR)) of different cases have been calculated, or solutions to maximizes that indexes have been investigated. an investment results interesting if NPV>0, IRR is highest, and PBT is lowest. Variables considered, assumption and methodology used for the analysis are discussed in the FluidCELL public report: Industrial and market introduction requirements for bio-ethanol fuelled Fuel Cell CHP systems .
It is well known that the energy market (both natural gas and electricity) can be different from country to country, depending on the primary energy used to produce these commodities, which is strictly related to the context and resources of the location. To better understand the impact of micro cogeneration, it is a good practice to evaluate the specific situation of the individual country in which the interventions are performed.
On the other side, cogeneration systems have maximum PBT when all electrical and heat energy produced is used. Furthermore, as efficiency is normally higher when working at full capacity rather than with reduced loads, CHP full power is always considered both in winter and in summer. If the heat and electricity produced are used completely when working at full capacity, it must be considered that installations cover the energy demand of different dwellings, in which a back-up boiler is surely required. The results of a 5 kWe size FluidCELL system which supplies energy for several dwellings are summarized in Table 3 and Figure 4. The analysis consists to find the minimum dwellings number which make interesting the investment, minimizing PBT
Figure 4. FluidCELL system PBT across Europe.
Table 3. FluidCELL system economic indexes and saving per year.
The same indexes have been calculated also for the 1 kWe ICE, 20kWe ICE and a μGT14. The last two have very good indexes, but buildings with several tens dwellings (over hundred in some cases) are necessary. The technical features that the FluidCELL system must have, to be industrially attractive, are summarizes in Table 4.
Table 4. FluidCELL CHP target.
Production costs [€/ kWe] 5,000
Electrical efficiency [%] 40
Total efficiency [%] 90
Lifetime (at least) [year] 10
Availability [hours/year] 7,500
Maintenance per year 1
starting time hours 3
Regulation [%] 25 - 100
CO content < 20 ppm
Catalyst lifetime 5 years
Membrane lifetime 5 years
Main conclusion from the analysis are detailed hereafter:
- FluidCELL system cannot supply only one average dwelling even reducing its size (it’s hold good for all countries).
- It becomes competitive if it can supply several dwellings, also when compared with different systems (as well as when compared with other fuel cell systems),
- The system is not competitive in the market if the natural gas and electricity costs are low, but it becomes very suitable if the electricity to natural gas price ratio is highest.
- Around 5 years PBT is however still a long time if the estimated lifetime is 10 years. In order to start large sales volumes, it is necessary to drop below 3 years. This means a reduction in specific price.
- Until production costs cannot be reduced, it is indispensable to make use of government grants.
- Good lifetime and good availability are essential.
- The recycling of precious materials and the use of standard components seems to be a way to reduce production costs.
4.1.3.2. Novel catalytic materials
Literature surveys of ethanol steam reforming catalysts revealed that rare-earth oxides based catalysts are less prone to deactivation with respect to other common supports (i.e. Al2O3) employed for the above reaction . In particular, bimetallic formulations (Pt-Ni/CeO2) displayed carbon formation rates among the lowest found in the recent literature , which has been selected as the first-generation catalysts. However, despite the very interesting catalytic performances of the above formulation, the results of fluidization tests performed at TU/e revealed very weak mechanical resistance. Therefore, to render fluidizable the selected catalyst, silica gel was chosen as mechanical support. Such material is characterized by a very high specific surface, which is expected to enhance active species dispersion and, consequently, the specific activity of the final catalyst. The second-generation catalysts were prepared by wet impregnation. Before preparation, SiO2 gel powder (Sigma-Aldrich) was calcined in air at 600 °C for 3 h (dT/dt=10 °C/min) in a muffle furnace. The impregnation procedure started with the salt precursor dissolution in bi-distilled water at ambient temperature: Ce(NO3)3∙6H2O, Ni(NO3)2·6H2O and PtCl4, supplied by Strem Chemicals, were selected as CeO2, Ni and Pt precursors, respectively. Once dispersed SiO2 in the aqueous solution of CeO2 precursors, the impregnation was carried out for 1 hour at 80 °C and then drying overnight at 120 °C and calcination, carried out at the same conditions of the support, occurred. Two following impregnations were performed for the active species and Ni was deposited earlier than Pt on ceria/silica support. The whole procedure (impregnation, drying and calcination) was repeated until reaching the desired amounts of CeO2 and active species. The final loads are equal to 20%wt for CeO2 with respect to silica, 3wt% for the noble metal and to 10wt% for the non-noble metal with respect to ceria. The values for metal loads have been optimized in the previous studies . To develop the proper formulation for the three generation catalysts, dedicated tests in the fixed bed reactor were carried out at 500 °C, H2O/C2H5OH ratio of 4, O2/C2H5OH ratio of 0.5 and at a WHSV (weight hourly space velocity, referred to ethanol flow-rate) of 4.1 h-1, in order to optimize CeO2/SiO2 ratio. All the bimetallic catalysts prepared at different CeO2 content exhibited excellent stability: during 100 h-reaction, the samples maintained a stable catalytic activity without deactivation and the average hydrogen yield was 41%. However, for higher times, the faster drop in C2H5OH conversion was recorded over the PtNi25Ce sample, which deactivated after almost 115 h. In the other two cases, 100% of ethanol conversion was measured for almost the same hours but the growth of C2H5OH concentration was more pronounced over the PtNi40Ce with respect to the catalyst containing 30% of CeO2. The spent catalysts were characterized by thermogravimetric and XRD analysis to investigate the factors resulting in their different stability performance. Figure 5 displays the dependence of carbon formation rate from the Ni crystallite sizes: the lowest dimension was recorded for the PtNi30Ce catalyst which showed the minimum carbon selectivity. Also, other authors observed that coke deposition is strongly dependent from the nickel crystallites, proving that the largest Ni crystallites caused the greater carbon deposition. In fact, low Ni surface area retained high residence time of coke precursors, thus increasing the probability of more stable carbonaceous deposits formation. Moreover, to verify the effectiveness of the developed catalysts for oxidative reforming of ethanol, their stability performances were compared, in terms of carbon formation rates, with the data found in the recent literature .Other studies were rarely focused on stability tests higher than 100 hours. Moreover, for WHSV higher than that applied in the present work, total conversion was only observed for C2H5OH concentrations lower than 10% and/or r.a. values double with respect to stoichiometric feeding conditions. At comparable operative conditions (temperature, space velocity and water/ethanol ratio), the catalysts developed in the present study assured CFR values three order of magnitude lower than that found in other works.
Figure 5. Carbon formation rate as a function of Ni average crystallite sizes for the different catalysts. Figure 6. Results of stability tests on CeO2-SiO2 based catalysts at 450°C, H2O/C2H5OH=3 and WHSV=4.1 h-1.
Once optimized the CeO2-SiO2 ratio, a series of bi-metallic and tri-metallic catalysts (Pt-Ni, Rh-Pt-Ni, Rh-Ni, Pt-Ni-K, Pt-Ni-Cs) were tested under more stressful operative conditions (450 °C, H2O/C2H5OH ratio of 3 and in the absence of water); WHSV was fixed to 4.1 h-1. The results of stability tests are shown in Figure 6 in terms of ethanol conversion (X). It was found that Pt-Ni and Rh-Pt-Ni catalysts maintained initial catalytic performances for at least 240 min without apparent deactivation. Conversely, similar C2H5OH profiles were recorded over the Rh-Pt-Ni and Pt-Ni-Cs samples, with an activity reduction from 100 to 97% in the time range of 0-1900 min. The K-containing catalyst displayed the worst performances among the prepared catalysts with a very fast conversion decrease to 93%. The Pt-Ni conversion profile was higher than that recorded over the other samples, with an X value of 95% after 5,000 min. It is also interesting to note that the bimetallic catalyst containing Pt also assures an initial hydrogen yield better than the other samples (26.5%) and very close to the value predicted by thermodynamic equilibrium. The unsatisfactory behaviour of the Pt-Ni-K catalyst can be ascribed to its low surface area and a subsequent worse active species dispersion, which reduced ethanol conversion in reforming reactions. Therefore, a negative effect of support basification, Rh addition as well as Pt substitution by Rh on the long-term stability and H2 yield of Pt-Ni/CeO2-SiO2 catalyst was observed. Based on the above discussion, the Pt-Ni catalyst was selected as the most interesting sample in terms of catalytic performances and employed for further tests.
The third-generation catalyst was also tested in a fluidized bed reactor, properly designed to perform catalytic tests at operative conditions close to that of the prototype. A very long stability test (400 h) was performed over the Pt-Ni/CeO2-SiO2 catalyst at 500°C, H2O/C2H5OH mol ratio of 4, O2/C2H5OH mol ratio of 0.5 and 4.1 h-1 under fluidized bed conditions. Ethanol was completely converted for more than 150 h; for longer TOS, only a partial deactivation was observed which resulted in ethanol conversion as well as hydrogen yield lowering from 100% to 80% and 43.5 to 36%, respectively. In fact, after almost 300 h of TOS, the system reached a stationary condition with no more activity decay. The test was performed for further 100 h and the system was found to hold the plateau condition. The results shown in Figure 7 suggest that carbon formation does not follow a constant trend during the test; conversely, it is supposed that coke deposited during test progressively deactivate the most reactive catalytic sites, involved in both reforming and by-products formation pathways. As a result, catalyst activity remains unchanged with no further increase in catalyst deactivation. To verify the above assumption, a further test in the fluidized bed reactor was performed over the Pt-Ni/CeO2-SiO2 catalyst: at regular intervals, the test was stopped and a sample was withdrawn and characterized by TGA to study the variation of CFR with TOS. However, to accelerate the achievement of the stationary condition, WHSV was triplicate: the results of stability test are shown in Figure 8. At WHSV=12.3 h-1, the system reached a stationary condition after almost 150 h; CFR was quite high at the beginning of the test and displayed a decreasing trend with TOS. In particular, the carbon formation rates after 175 and 200 h were almost the same, proving that the net rate of carbon formation (i.e. the difference between carbon formation and carbon gasification rate) was equal to zero. Also, other authors studied the evolution of coke during reforming reaction in fluidized bed reactors over Ni-based catalysts . However, a markedly different behaviour was observed during operation: almost steady performances were followed by a severe drop in ethanol conversion and finally a slow decrease in ethanol conversion until total deactivation.
Figure 7. Stability test over the Pt-Ni/CeO2-SiO2 catalyst in the FBR, 500°C, H2O/C2H5OH=4, O2/C2H5OH=0.5 WHSV=4.1 h-1.
Figure 8. Trend of C2H5OH conversion and CFR over Pt-Ni/CeO2-SiO2 in the FBR, 500°C, H2O/C2H5OH=4, O2/C2H5OH=0.5 WHSV=12.3 h-1.
Once selected the final formulation for the pilot plant test, two different silica gel materials were employed for the preparation of TU/e and HYGEAR catalysts. In fact, for TU/e tests, a particle size distribution between 150 and 250 µm was required. Conversely, for the tests in the pilot reactor, lower particle sizes (90-115 µm) was necessary. At that hand, two batch of supports, having a pore size of 60 Å, were provided by Sigma-Aldrich. However, concerning the materials for the pilot reactor, UNISA supplier was only able to provide a product with a particle size distribution wider than that required (53-115 µm). Therefore, it was necessary to preliminary sieve the silica gel material for HYGEAR catalysts. The total amount of material prepared was 1.4 dm3 of CeO2/SiO2 and 0.14 dm3 of Pt-Ni/CeO2-SiO2 for TUE tests and 7.4 dm3 of filler (CeO2-SiO2) as well as 2 dm3 of Pt-Ni/CeO2-SiO2 catalyst for the prototype tests. Figure 9 reports some steps of catalyst preparation.
Figure 9. Samples after drying post CeO2 (a), Ni (b) and Pt (c) impregnation.
4.1.3.3. Membranes development
The Department of Energy of the United States of America (DOE) has defined the following targets for membranes for hydrogen separation by 2015: a purity of 99.99% for a H2 flux of 100 Nm3/m2·h at 6.90 bar of differential pressure, a life time higher than 5 years and a cost lower than 1,000 US$/m2 . Among the membranes for H2 separation, Pd based membrane shows the highest permeability and exclusive selectivity for H2 due to the unique permeation mechanism. In order to achieve DOE’s target, very thin Pd membranes (less than 5 µm) are required. Thin Pd based membranes are generally supported on porous substrates including stainless steel and ceramic materials of planar or tubular configuration , . However, pure Pd membranes are often damaged by hydrogen embrittlement due to the α-β phase transition of palladium hydride, which occurs at below the critical temperature (293 °C) and pressure (2 MPa). In comparison to Pd, Pd-Ag alloy membranes are stronger against hydrogen embrittlement (PdH α-β transition at low temperature) and have higher H2 permeability up to 70% (Pd77Ag23) . Pd membranes are very sensitive to sulphur poisoning leading to the decrease of membrane performance , .
There are several technologies reported in the literature for the deposition of Pd and Pd-alloy dense layers on supports, such as physical vapor deposition (PVD, including magnetron sputtering, thermal evaporation, or pulsed lased evaporation), chemical vapor deposition (CVD), electroless plating (ELP), and electroplating (EP)23, . The strength and weaknesses of each of these technologies are also reported on these references.
Membrane activities in the frame of FluidCELL aimed at developing suitable membranes for the prototype (i.e. high permeation and selectivity, long term stability). Both PVD and ELP technologies as well a combination of both technologies were applied for the development of thin film (<6 µm) and/or ultra-thin film (<2 µm) Pd-based membranes. In addition, pore filled membranes were also studied.
Pd-Ag films are not dense when directly deposited on porous supports by PVD-MS. The H2/N2 ideal perm-selectivity selectivities are very low and not suitable for H2 purification. Ultrathin Pd-Ag membrane layers were also prepared by a combination of PVD-MS and ELP techniques, showing a very high H2 permeance (8 x 106 mol m2 s1 Pa1) at 400 °C and 1 at with H2/N2 ideal perm-selectivity around 500. Detailed information and characterization of these membranes can be found in the work of Fernandez et al.
The dependence of the Pd-Ag membrane thickness as a function of the plating time is shown in Figure 10-a 30. The orange and green points refer to the membranes prepared in references [ ] and [ ] respectively. The red dashed line represents a 1.0 µm/h metallic layer growth rate. In the zoom-in graph presented at the right bottom part, the correlation between the membrane thickness and the plating time is presented showing an almost linear dependence. Membrane thicknesses obtained with longer plating times (see orange and green dots) seems to follow this correlation. There may be a small decrease at longer plating times indicating that the membrane growth becomes somewhat slower probably due to the decrease in the concentration of the reactants along the time or to a decrease in the specific surface area for the autocatalytic Pd deposition.
a)
b)
Figure 10. a) Dependence of the time of plating with the thickness of Pd-Ag membranes prepared by ELP; b) H2 permeance and H2/N2 ideal perm-selectivity as a function of the membrane thickness at 400 ºC and 1 bar30.
The H2 permeance and H2/N2 ideal perm-selectivity as a function of the membrane thickness at 400 ºC and 1 bar transmembrane pressure difference is shown in Figure 10-b 30. The H2 permeance of the 0.46 µm thick membrane is extremely high (1.53 x 10-5 mol m-2 s-1 Pa-1) but shows a relatively low H2/N2 ideal perm-selectivity (close to 50). When the thickness increases from 0.46 to 1.17 or 1.29 µm, the H2 permeance reduces to 0.9 x 10-5 mol m-2 s-1 Pa-1. On the other hand, 1.17 and 1.29 µm thick membranes showed similar hydrogen permeances. However, the H2/N2 ideal perm-selectivity was almost double for the thicker membrane. This may indicate that after around 1 hour of plating time the entire surface of the support was covered by a Pd-Ag layer. In these membranes, defects in the porous support are significant which can explain the difference in selectivity.
The ultra-thin membranes developed in the present work show one of the highest H2 permeances reported in the literature for supported membranes (see Table 3 in reference [31]. Despite the high permeation and long-term stability of these Pd-Ag supported membranes (Figure 11), the selectivity of the membranes is not high enough for the membrane reactor of the FluidCELL.
Figure 11. H2 and N2 long-term permeances (mol m-2 s-1 Pa-1) and H2/N2 ideal perm-selectivity of the 1.29 µm thick membrane at 400 °C and 1 bar pressure difference.
Membranes for the prototype
Thin film (≈5 µm) Pd-Ag membranes developed by direct simultaneous Pd and Ag electroless plating (ELP) were selected for the prototype to ensure membrane stability under fluidization regime and ideal H2/N2 perm-selectivy above 8,000. Selective layers were deposited onto asymmetric alumina 50 cm long porous supports with thicker wall (10/4 mm o.d./i.d. provided by Rauschert Kloster Veilsdorf) than in previous developments to improve the mechanical properties compared to the 10/7 mm o.d./i.d. supports (a higher torque could be applied in the connections when using the thicker support). In addition, to reduce the number of sealings in the reactor prototype, it was decided to scale up the length of the membranes from 15-20 cm (original plan) to around 40 cm (Figure 12). Overall, 40 membranes (37+ 3 spare parts) were manufactured.
Cost analysis
A cost analysis of the semi-industrial production of Pd-based membranes considering thin films (< 6 µm) onto ceramic and metallic supports developed by ELP and PVD and ultra-thin film onto ceramic supports was performed. Depending on the production method and type of support applied (ceramic or metallic), a cost per square meter of membrane ranging from 5,000 - 10,800 euro/m2 was calculated. Higher costs are related to the use of the metallic supports instead of ceramic supports. Recycling was not considered in this analysis. When the recycling of both the Pd-Ag waste and the membranes (selective layer and supports, mainly for the metallic supports) is considered in the cost analysis, we could estimate that the cost of the membranes could be between 2,000 - 5,000 euro/m2.
4.1.3.4. Lab-scale bio-ethanol catalytic membrane reformer
The experimental tests have been carried out at the TU/e using the permeation setup under reactive conditions (Figure 13).
Figure 13. The lab facility at the TU/e.
This experimental set-up can perform reactive permeation experiments with a maximum of five membranes. There are three thermocouples, and two pressure sensors distributed over the reactor height. A vacuum pump can be switched on to decrease the partial pressure of the permeate side, as low as 10 mbar. The system does further include feed connections to pure components as O2, H2, N2, CO, CO2, CH4 and pressurized air. The components can be fed through a mass flow controller, which depending on the type of mass flow controller can have a maximum flow of 2 L/min (CO) up to 22 L/min (N2). The retentate can then be analysed by FT-IR. The FT-IR can measure concentrations (after calibration) for water and ethanol, which was not possible at a lab scale. The retentate then passes through a cooler where the water and ethanol are condensed and separated in a flash column. The retentate is connected to a Sick® analyser, which can measure concentrations of the dry retentate. A similar analyser is also connected to the permeate side, which can measure hydrogen concentrations in the range of 0-100% and COx concentrations on a ppm accuracy. The membranes were prepared and delivered by TECNALIA. However, some difficulties were encountered when placing the graphite ferrules. All the membranes used for the experiments had an external diameter in the same range (10.38 mm). The graphite ferrules ordered for these experiments had an average diameter of 10.4 mm. Theoretically, it should be possible to seal the membranes with these graphite ferrules. However, the graphite ferrules started to break when they were stretched the to the desired size. This was solved by gently warming up the graphite ferrules with a heating gun, no visual ruptures could be observed prior to placing the graphite ferrules. For the new tests in future and for the prototype reactor it is suggested to use ferrules of the right size. TUE has already paid the company for the tools to prepare ferrules with different sizes, which can now be ordered also by the partners of FluidCELL.
The membranes connectors were placed by closing with a torque wrench at 11 Nm. Closing the connectors at the same torque (up to14 Nm) as used in earlier experiment has resulted in breaking of the ceramic support. Helium permeation tests were performed after closing the membranes, and the results can be seen in Table 5.
Table 5. Membrane dimension and He leakage test.
The membrane permeation tests were performed at 450 and 500 °C, in excess of hydrogen (to maintain overall reactor concentrations more or less equal to the inlet concentrations), and at different feed pressures. The lab scale tests were performed at hydrogen fractions in the range of 40-100%. There was observed that the slope of the permeation tests does not differ that much as a function of concentrations. The hydrogen fraction ranges for the pilot scale tests were set to be in the range of 25-100%, to assess the permeation at lower concentrations. The membranes were tested individually and all together during the permeation tests. The membranes performed relatively better individually, in comparison to operation with more than one membrane.
The membrane assisted fluidized bed model merged the concept of reaction kinetics with bed hydrodynamics and gas separation. The complete description of the model is presented in our paper C. Ruocco et al .
a)
b)
c)
Figure 14. Nitrogen free retentate compositions (%vol) for multi-membrane reactive experiments at a) 450, b) 500 and, c) 550 °C. By varying feed pressure on x-axis (bar(a)).
The dry retentate composition was continuously analyzed by a Sick ® analyzer. Furthermore, during these experiments it was also possible to measure steam and ethanol compositions due to the presence of an FT-IR analyzer. Due to the rather high nitrogen dilution, the retentate composition will be compared by using nitrogen and steam free compositions.
The optimum numbers of CSTR for the emulsion and bubble phase were found to be 6 and 18 respectively. From Figure 14 we can observe the same as from the performed lab scale experiments, namely higher methane compositions at low temperatures (which indicates low SMR conversions) and low methane compositions at high temperatures (which indicates high SMR conversions). Further, it can be observed that the model predicts the composition accordingly to the experimental results, even if the nitrogen is excluded from the composition (which should make existing deviations even larger). The large CO deviation at 450 °C is not observed, which was the case during the lab scale experiments. Further, it can be observed (just as in the lab-scale results) that the hydrogen fraction increases significantly at higher temperatures (Figure 15) due to the higher CH4 conversion. However, during these experiments no larger deviations are observed when operating at 550 °C and 4 bar(a). The permeate pump connected to the experimental set-up at this scale performs better than the one utilized during single-tube membrane experiments, and is thus able to pump higher gas flow rates. This is beneficial to maintain a low, and constant, permeate pressure which is required in the model as the permeate pressure is there given as a constant value.
a)
b)
c)
Figure 15. Hydrogen permeation, model vs. experimental results at a) 450 °C, b) 500 °C, and c) 550 °C by varying feed pressure on the x-axis.
The HRF during these standard experiments are tabulated in Table 6. From the table we can clearly observe an increasing trend of the HRF with increasing temperature. The HRF is defined as the maximum pure hydrogen obtained relatively from the ethanol present after combustion. This is mainly due to the higher methane conversion at higher temperatures, which enables more hydrogen to permeate through the membrane.
Table 6. HRF vs. T (feed pressure: 4 bar(a)).
By then analyzing the retentate composition from FT-IR at multiple experiments, it can be concluded that no ethanol is present in the retentate stream, as expected from the kinetics study performed and discussed in C. Ruocco et al35.
The permeate concentration was frequently checked during experiments on the CO, CO2, and H2 concentrations. The permeate CO and CO2 concentrations resulted to be in the range of 100 – 300 ppm, whereas the hydrogen concentration was measured constantly at values of >98%, the remaining ±2% is expected to be nitrogen throughout all the experiments.
All the experiments of the study have been plotted in one single parity plot, in which every component is highlighted separately in Figure 16.
Figure 16. Parity plot for all measured and predicted value of the retentate stream.
Components CH4 and CO result to have multiple predicted points with an error larger than 10%, this, however is explained due to the relative high read-off error for these components when utilizing the SickTM-analyzer. The analyzer measures concentrations round off up to one decimal, which gives high deviations for components having low concentrations. From the parity plot above there can be concluded that the proposed reaction kinetics, proposed reaction route and corresponding hydrodynamics are suitable for the description of the system.
4.1.3.5. Design and manufacturing of novel bio-ethanol catalytic membrane reformer
WP6 dealt with the design of the pilot scale membrane reactor for ATR of ethanol. The reactor is intended for simultaneous production and purification of hydrogen in one single step. It uses palladium-based membrane technology developed in WP4. The ATR reaction is sustained in a fluidized bed of particles of catalyst mixed with inert filler. Table 7 summarizes the reactor design specifications.
Table 7. Specifications and definitions on the pilot reactor.
Parameter units Value
Nominal hydrogen output Nm3/h 3.2
Minimum partialization (in terms of EtOH processed) % 40
Sweep gas (nominal) Nm3/h 1.3
Sweep gas (maximum) Nm3/h 1.8
Operating temperature °C 500
Operating pressure barg 11
Permeate pressure mbarg 200
Hydrogen Recovery Factor % 76.8
Hydrogen recovery factor (HRF) is defined as the ratio between the flow of hydrogen permeated over the maximum production capacity given the feed of oxygen needed for temperature control. The calculation of HRF is shown in equation (1).
HRF=F_(H_2,perm)/(6F_(C_2 H_5 OH,feed)-2F_(O_2,feed) ) (1)
A schematic view of the reactor flow configuration is shown in Figure 17. The feed enters the reactor through the distributor in the bottom and flows vertically through the fluidized bed. The first section before reaching the membranes is intended for oxidation and pre-reforming. The membrane reactor section takes place across the height level where the gas is in contact with the membranes. The oxidation supplies the required heat for the endothermic reforming reaction. When hydrogen reaches the membrane, this is permeated across the membrane. Sweep gas flows on the permeate side of the membrane counter-current to the reforming stream. This gas is used for reducing the partial pressure of hydrogen on the permeate side.
Figure 17. Schematic view of the ATR-MR used in the model.
Table 8. Stream compositions at inlet/outlet of reactor.
Pipeline Fuel & steam Air Sweep Permeate Retentate
Comp. (%molar)
EtOH 21.7
H2 71.2 4.7
H2O 78.3 100 28.8 33.7
CO 0.8
CO2 29.7
CH4 3.9
N2 79 27.2
O2 21
Flow (Nm3/h) 3.71 1.61 1.29 4.44 4.68
The characteristics of the membrane tubes used in the pilot reactor are listed in Table 9.
Table 9. Membrane characteristics.
Support Ceramic porous
Outer diameter 10 mm
Inner diameter 4 mm
Effective length 400 mm
Sealing Swagelok union with graphite ferrule
Number of tubes 37
The design of the reactor is shown in Figure 18. The reactor holds 12 thermocouples at different radial and axial positions. An external heat source (electrical) is used to start up the system. The design of the manifold allows the distribution of sweep gas along the permeate side of the membranes and collection of the permeated hydrogen in a single outlet. The design gives the possibility of removing the inner tubes to evaluate the tightness of each membrane. If during operation the impurities were too high, it could be that a tube was leaking. By opening the top flange, it is possible to find out which tube is leaking. This is done without unmounting the membrane manifold.
Figure 18. Reactor design.
Figure 19 shows the Process Flow Diagram (PFD) of the assembled system including the membrane reactor and auxiliaries. The system, as shown in Figure 20 is skid mounted for ease of transport to the testing facility of ICI in Italy.
Figure 19. Process Flow Diagram of the membrane reactor prototype.
Figure 20. Assembled membrane reactor test setup.
The pilot scale autothermal membrane reformer of FluidCELL was successfully tested and validated in WP6. The reactor was tested at HyGear (Arnhem, The Netherlands), before it was shipped to ICI Caldaie (Verona, Italy) for the integration of the complete FluidCELL system in WP8. Initially, permeation tests were performed under non-reactive conditions, and later, test under ATR conditions. During the ATR tests, the temperature in the reactor was controlled fully by the ATR reaction, without the use of the electric heaters, which are used only for heating during start up, neither was any diluent inert gas used. During the ATR tests, several operating conditions were tested, varying parameters such as operating temperature, feed flow, water to ethanol ratio, and sweep flow. The main objective of the test campaign was to characterize the performance of the membrane reactor. The ATR tests were performed using the fully automated mode. During the tests, the controls of the reformer were fine-tuned for optimal response.
Temperature and pressure control accuracy were widely satisfied. With tests closer to nominal conditions a temperature distribution with lower than +/- 4 °C were observed, which would be practically inconceivable in conventional fixed bed ATR.
The results from the test campaign were overall satisfactory. Hydrogen recovery factors (HRF) of 26 % and separation factors (SF) of 59 % were achieved with the first tests. In terms of hydrogen production factor (H2Prod), levels of 44 % were achieved. Hydrogen productivity was in all cases higher than the theoretical hydrogen production of a traditional ATR at the same feed and operating conditions, reaching up to 3.4 times higher with the membrane reactor.
With regards to hydrogen permeation flux, the tests at pilot scaled confirmed the experience obtained at lab scale. Such result means that the hydrogen output is lower than that predicted at the beginning of the project with the design of the reactor in mid-2015. Studies on the phenomena behind this reduction in permeance were very recently submitted for publication to a special issue of Chemical Engineering & Processing on Process Intensification, by project partners TUe and Politecnico di Milano. Based on previous experience at lab scale, it is expected that hydrogen permeation values would further increase by means of a more effective membrane activation.
During the tests at HyGear much knowledge was obtained on the performance of such fuel processor. The reactor of FluidCELL is worldwide first in its kind, capable of producing pure hydrogen from bio-ethanol by means of autothermal reforming in a fluidized bed membrane reactor.
At the end of the testing campaign, the fuel processor was packaged for transport taking appropriate measures to safeguard the fragile membranes, and shipped to ICI Caldaie.
4.1.3.6. Fuel Cell Stack
A state of the art about PEMFC for stationary µCHP application has been proposed by POLIMI showing technologies for low and higher temperature technologies, showing operation features or degradation issues. The SoA report about the µCHP PEMFC systems allowed to define the status and get available data about the commercial fuel cells implemented on systems, showing few existing information or product directly adapted for the FuildCELL needs, like available data in specific conditions, thus confirming relevance to select and test fuel cell components.
Beginning of the project, experimental activities have been conducted in order to prepare the test of selected fuel cell components. Single cell equipment has been adapted for operation with polluted fuel. MEAs and components have been selected as first reference for testing. Small size MEAs have been provided for single cell tests at UPORTO, with start-up, conditioning and testing conditions, so as to enable comparing single cell and stack information. MEAs have also been made for tests in CEA PEMFC short stack of 8 cells. Sensitivity studies were conducted on the single cell and on a short stack (8 cells) made with reformate type reference MEAs (Figure 21).
Wide range of conditions has been defined with the consortium with regard to reformer developed and to system application specifications. Single cells tests have shown, after some adaptation of the testing device, similar performance as same MEAs in stack. All Electrochemical measurements gave information about the impact of conditions mainly related to fuel and air feeding. Simulation of single cell behaviour was conducted to help defining best suited components to be tested for selection. Results showed the influence of the fuel composition, like performance losses due to presence of some CO in the hydrogen produced. On the stack, dead end operation has been tested to check stack voltage compared to fuel circulating mode.
Figure 21. Single cell and 8 cells stack (including the S++ device with 480 segments for the measurement of local current densities) installed on the test stations at UPORTO and at CEA.
During the middle of the project, a deep investigation has been conducted in close collaboration between CEA and POLIMI, in the frame work of the PhD of S. Foresti, dedicated to the analyses of fuel cell data including in-situ measurements of the distribution of local current density focusing on the impact of fuel feeding conditions with mixtures of hydrogen, inert gases like nitrogen, methane, carbon dioxide and small amounts of carbon monoxide, all these components being possibly produced by the reforming process or the fuel feeding strategy like recirculation or dead-end. These results have been published .
Figure 22. Protocol applied to get data for CO/H2-inerts/H2 maps (left side) for different sets of operating conditions for the stack temperature and the fuel and oxidant pressures, flow rates and relative humidity. Example of a map showing the impact of fuel composition on the voltage (right side).
Before the end of the project, additional work has been conducted with the analysis of experimental data, from single cell and stack testing, and modelling on fuel cell components performance.
The electrochemical characterization was completed for the benchmark Membrane Electrodes assemblies (MEAs) provided by CEA. Polarization curves and potential-time history for poisoning tests were used to analyse CO tolerance. The impact of applying internal or external air bleeding to reduce anode pollution when fuel is containing CO was also explored in single cell tests.
Figure 23. Potential-time history at 0.5 A cm-2 with internal air bleeding (left-hand side) and direct air bleeding (right-hand side).
Additionally, a transient multi-dimensional non-isothermal multiphase model for simulating PEM fuel cells operating on pure hydrogen and reformate fuel was developed and validated against experimentally obtained results.
All information needed for the integration of the stack into the ICI system have been detailed and provided to the work package in charge of developing and operating the system.
Main final actions concerned the preparation and delivery of the fuel cell prototype (Figure 24). Electrodes and Membrane Electrode Assemblies were manufactured at CEA with lab scale pilot lines. Test of an 8-cell stack was conducted in several conditions for the validation of the MEAs before making the prototype. End-plates of the prototype stack was specifically designed and ordered.
Figure 24. Fuel cell prototype 120-cell stack tested at CEA and delivered to ICI (left side). Comparison of the polarisation curves in nominal conditions get on CEA test bench and measured at ICI with different conditions for the temperature and gases humidity (right side).
Stack prototype has been assembled, conditioned on a test bench and delivered to ICI in May 2017. ICI could implement the stack on a system but not with the reformer. Few data were obtained operating the stack at ICI but only under pure hydrogen,
Few data could be obtained at ICI, particularly no operation was conducted on the system including the reformer. A polarization curve could be obtained under pure hydrogen with conditions slightly different compared to the tests conducted at CEA before delivery, with lower temperature and probably different humidity for gases; thus, the performance was lower (estimated -7% on the stack power at nominal current point).
As a conclusion about the fuel cell activities in the project FluidCELL, investigations conducted at single cell and short stack levels on test benches allowed to improve the knowledge about the operation of PEMFCs under reformate fuel containing various amounts of contaminants in different sets of operating conditions which is giving valuable information for this technology. The very detailed studies with specific mixtures and in-situ measurements coupled with modelling allowed to further understand and improve the system management with the specific design considered. In addition, the selected core components were manufactured, validated and assembled in a 120-cell stack prototype delivered to ICI for being implemented in the system.
4.1.3.7. Integration and Proof of Concept of the CHP system
The optimization of the system was carried out in two steps. The first step consisted in the definition of the layout to be adopted for the prototype. Two configurations of the membrane reactor were considered in order to enhance the hydrogen permeation across the membranes: the vacuum pump layout and the sweep gas layout. The main output of the simulations are (i) maximize the heat recovery from the exhausts (ii) reduce the water excess to the reformer are fundamental to achieve the target efficiency (>40% electric); (iii) the efficiency of the vacuum pump layout is more sensitive to the reactor pressure than the sweep gas layout; The vacuum layout requires less membranes but has a lower electric efficiency due to the parasitic consumption of the vacuum pump, therefore the sweep gas layout was preferred for the prototype. When defining the layout not only the Pd-membranes area and the number of PEM cells for the stack were considered, for example the arrangements of flows in the heat recovery to avoid extremely high temperatures, which requires expensive heat exchangers, as well as no additional ethanol to the retentate combustor to avoid a liquid/gas burner.
The second step was the detailed simulation of the membrane reactor and fuel cell stack to achieve the best performances of each component, taking into account also off-design operation.
About the membrane reactor a 1D dynamic model was developed by POLIMI including the bubble-emulsion phases, the reaction kinetics of ethanol autothermal reforming, and the gas permeation across the Pd-layer and its support (which resulted one of the most limiting factor in the operation of the sweep gas layout). The model was validated against the experimental data at lab-scale and used to simulate the partial load operation of this unit. An example of the parametric analysis performed with the membrane reactor model is depicted in Figure 25: the gas diffusion through the thick support is a strong limiting factor in hydrogen permeation.
Considering the selectivity issues, a modelling for the optimization of the PEM operating conditions as function of the hydrogen purity has been carried out. In principle, Pd-based membranes can produce high-purity hydrogen (target= 99.99%, i.e. selectivity=10,000). However, the selectivity may decrease and low-purity hydrogen (e.g. 99.9% with 100ppm CO) could be obtained. A methanator is therefore added along the hydrogen cooling line to prevent CO-poisoning of the FC anodes.
Figure 25. Comparison of permeated hydrogen between the ideal (dotted line) and the phenomenological (bars) models for different configurations of the membrane reactor (vacuum/sweep gas layouts).
The quality of hydrogen affects the cell voltage and the overall system performance; therefore, FC control strategies were investigated. Build-up of inert and poisoning species in the hydrogen recirculation loop can be limited by venting a fraction of the anodic off-gas, whose amount can be optimized. These circumstances were investigated by simulating the FCs subsystem that includes the stack, the air and hydrogen blowers and the air humidifier. An experimental campaign on an 8-cells stack (cell area 220 cm2) was performed to characterize the FC operation with reformate gas. The impact of fuel composition, containing up to 20% of inert gases (N2, CO2, CH4) and CO up to 40 ppm, and operative conditions (pressure and humidity) was analysed on the overall stack performance as well as on the current density distribution along the cell surface. Experimental data constituted a valuable source for the validation of the model developed by POLIMI. Main results from the model show that (i) voltage losses can be minimized by adjusting the amount of hydrogen vent; (ii) the most critical regions of the PEM fuel cells are the cathode inlet, due to water deficiency and the anode outlet due to inert gases and CO concentration while hydrogen is consumed.
Figure 26. Integrated m-CHP system.
FluidCELL reactor unit arrived in ICI end September 2017. All the other parts of the system (i.e. Fuel cell, inverter, recirculating pumps, humidifier, controller, and BoP in general) were already set and ready for the connection with the pipes and wires from the reactor.
The system required different easy connections to be done (e.g. mechanical connection of piping, electrical, pneumatic, gases) correlated with a user manual for its properly start up and operation procedure.
As soon as the final integration with the rest of the system began, some unexpected actions appeared essential for the correct installation and operation of the reactor. The main issues to be solved are reported hereafter:
Reactor prototype electrical connections;
High leak in the membrane reactor after the ATR test carried out at HyGear;
Unexpected low H2 permeance in the prototype compare to lab scale;
When test for the membrane activation restart in manual mode, new problems arise and the system was out of control;
Optimization of the reactor prototype working conditions;
The cooling flowrate was too low;
The ethanol flow rate was too low.
Different values of Ethanol flow rate have been tested (higher than 12 Nl/min) but the system was not able to reach stable conditions delivering enough amount of H2 with the appropriate purity, and keep them for a time long enough to perform the analysis (Figure 27). After these tests it was decided to stop the activity because the smell of alcohol gave indication that there was a release of ethanol somewhere in the system, which may lead to unsafe operation. During the last month of activities some extra efforts have been done in changing the hardware (H2 recirculating system, humidifier installation, electrical connection) and the software of the stack module in order to perform some test with pure H2 from bottles instead from fuel processor. The tests done in ICI were highly influenced by the problems arising with system changing conditions (like new H2 humidifier that caused flooding in the stack) and they affected the final results. Since FC stack tests with H2 from bottles were out of the scope of the project and the same tests were already performed by CEA on the stack, this activity was suspended. The data analysis was carried out by POLIMI with the support of the models in order to understand the phenomena at the origin of the poor performance of the system.
Figure 27. Feed, retentate and permeate flow rates in one of the tests.
The technology of the fluidized bed membrane reactor is very promising for future investigations and studies and this project has been useful to understand where these technologies can be improved. In particular, a deeply analysis of the fluidized bed behaviour could help knowing the flexibility of the system and how it responds to the unexpected operating conditions. Improving the membrane strength would results in better long-term reliability and more flexible operations. The slow gas diffusion through the thick porous support limits the effectiveness of the sweep gas layout. Also, the software and the logic stands behind of it should be more flexible and manageable for having a better control on the reaction parameters. More precision on the reactor data collection could help in knowing exactly which conditions the catalyst is subject to. All the work done for developing and building the BoP and the control system has been very useful to improve knowledge and experience about CHP system and membrane reactor. The membrane reactor prototype will be opened to assess the status of the membranes and catalyst.
4.1.3.8. LCA and safety analysis
LCA analysis
The aim was to analyse the environmental profile of the FluidCELL system and identify key leads of improvement, based on the LCA methodology. The functional unit is defined as “Provide 2 German or Italian typical dwellings with useful heat and electricity over one year, in an off-grid situation”.
The LCA performed for this project covered the entire life cycle of the considered systems from the extraction and processing of all raw materials through the end-of-life of all product components (i.e. from “cradle to grave”). In this LCA, the IMPACT 2002+ LCIA method was considered and endpoint (damage-oriented) indicators were presented.
Environmental profile of the FluidCELL system
The LCA showed that the most important contributor to the impacts is the bioethanol production and delivery for the five environmental indicators considered (Figure 28), which is mainly related to bioethanol production part (delivery corresponding to 1 to 20% of this life cycle stage impacts depending on the indicator considered). The bioethanol production impacts are driven by the rye grain cultivation: fertilizers and pesticides production, on field emissions due to fertilizers application, agricultural machinery diesel consumption, energy and water for irrigation and land occupied by the crop. When working on improving the environmental performances of the system, the first action to consider consists of increasing the efficiency of the entire system at the maximum, in order to decrease the amount of bioethanol consumed for the same quantity of energy produced.
The remaining impacts are related to the equipment production, maintenance and end-of-life but the maintenance is the most important. The impact of the equipment is mainly due to the rare metals used: platinum and ruthenium used in the fuel cell electrode, platinum used in the catalyst of the fuel processor and palladium used for the fuel processor membrane. The other important contributors to the equipment impacts are the heavy metallic pieces (large pieces from the BoP as e.g. housing).
Key leads for the optimisation of the environmental performances of the FluidCELL system
Based on the results of the LCA and the different sensitivity analyses performed, some levers can be identified to lower the environmental impacts:
All actions of design that can increase the global efficiency should be explored, since bioethanol production remains the main contributor to the global impacts of the system (increasing the size of equipment should be explored depending on the equipment, maintaining the efficiency over the lifespan of equipment).
Encourage the use of low impacting bioethanol, by selecting a crop used as feedstock that has a high yield, low agrochemical inputs, with no or low irrigation and with no deforestation, or by selecting second generation bioethanol, or by exploring the possibility of bioethanol production on site made from biomass wastes available on site.
Decrease the impacts related to bioethanol delivery by finding a close bioethanol production place to reduce the delivery distance and ensuring to choose the right bioethanol dilution depending on the delivery distance from bioethanol production plant to the off-grid dwellings.
For the additional heat and electricity needed to fulfil the dwellings needs, having a heat pump is an interesting option but exploring wood pellet stove could be interesting too.
Decrease the impacts of infrastructures (m-CHP and heat pump production, maintenance and end-of-life): by ensuring recycling of metals at the end-of-life, by reducing the mass of these elements but always maintaining the efficiency of the system, and by trying to lighten the heavy metallic pieces and ensure that the micro-CHP is easily dismountable at the end-of-life to be able to recycle as much as possible the different pieces.
Figure 28. Climate change, resources consumption, impacts on ecosystem quality and human health and freshwater withdrawal related to the production of heat and electricity to fulfil the needs of 2 Italian or German dwellings over one year. Negative values correspond to an environmental credit and are due to the recycling of some materials in the end-of-life (EOL) stage. Indeed, the recycling provides secondary material on the market that will enable to avoid primary material use.
Safety Assessment
Based on the design of the membrane reactor as well as from the integration of the reformer into the fuel cell CHP-system a safety assessment was performed. Public deliverable D9.5 “Safety assessment of the m-CHP” summarizes the results of the assessment. The deliverable describes the safety standards and directives employed for the assessment and certification of the system. The most relevant results from the HAZOP are outlined, also describing the scoring used for the risk assessment, and the methodology used is described. Moreover, an analysis of the potential failure modes and effects is given for the CHP integration.
The fuel processor of FluidCELL was developed by HyGear. The technology here used incorporates a few chemical processes at elevated temperatures and pressures, and with both flammable and toxic gasses. Therefore, conventional standards, directives and procedures had to be applied to ensure safe design of the system. HyGear has vast experience in these type of processes, therefore internal procedures were already in line with the required standards. The fuel processor was CE certified.
For the fuel processor, the main directives applied were the Machinery Directive (2006/42/EC), the Pressure Equipment Directive (2014/68/EU), the EN 60204-1 Safety of machinery, and the EN 60079-10-1 Explosive atmospheres. A conformity assessment according to module G of PED directive was carried out for the pressurized assembly. This assessment is verified by an external notified body in order to certify that appropriate safety measures had been implemented. Moreover, an inspection for commissioning (Keuring voor Ingebruikneming) was performed by the NoBo.
An analysis of Potential Failure Mode and Effects has been carried out. To each item or function a value for Severity, Probability and Detection has been associated and reported in the table below. Based on the average values of Severity, Probability and Detection, it was evaluated that the probability of failure of the stack part was low and even if something happened it would not have a high severity. A risk analysis was done, and adequate mitigation measures were taken where needed. The safety assessment was essential to bring forward a system with safe and robust operation.
See FluidCELL_Publishable_final-report-18102018_v01.pdf
Potential Impact:
4.1.4. Potential impact
European countries are implementing environmental policies to cope with the EU 20 % emission reduction target set for 2020 (from 1990 levels) . Recently, m-CHP systems received growing interest and experienced further development, thanks to their superior performances in energy conversion. The target of this project is to develop a 5 kW fuel cell m-CHP system based on bioethanol.
Before considering the above-mentioned technology, it is useful to summarize the technologies supporting m-CHP systems. The following technologies can be integrated into a m-CHP system:
➢ ICE (Internal Combustion Engine)
➢ Stirling Engine
➢ μTG (micro Gas Turbine)
➢ ORC (Organic Rankine Cycle)
➢ PV-T (Thermo-Photovoltaic)
Table 10 summarizes the main features of these technologies.
Table 10. m-CHP systems comparison.
Features ICE Stirling μTG ORC PV-T
Electrical Efficiency 25-30 % <20 % 25-33 % 14 % 25 %
Overall Efficiency 80-85 % 65-85 % 70-85 % 80-85 % 80-85 %
Cost €/kWe 6,000 10,000 1,500 --- ---
Availability Commercial Pre-Commercial Commercial Under research Under research
The FluidCELL system is based on a m-CHP Fuel cell unit that runs on bioethanol; the system can be broken down into its main components/subsystems; these are:
➢ Fuel
➢ Fuel Processor
➢ Fuel Cells
➢ Inverter
➢ Balance of Plant (BOP)
➢ Controller
Fuel
Bio-ethanol is gaining strength in the U.S. market. In 2016, it witnessed record production along with unprecedented demand and a growth in exports. Moreover, with the new U.S. administration, it seems that many of the regulatory obstacles impeding the further ethanol growth, may be eliminated. Ethanol is also popular in Brazil and Canada, which represent the two top export consumers for the U.S. market. Ethanol price is significantly lower than that of gasoline on an equivalent heating value. Additionally, ethanol contains low sulphur and metal elements and it is CO2 neutral since it is produced from biomass. Ethanol has a relatively high hydrogen content, it can easily be carried and stored; finally, ethanol blend contains water, which is useful for steam reformers. On the other hand, there are some compounds which could poison or damage the reformer catalyst, including carbon formation. Indeed, bio-ethanol blend includes one-, two- and three-carbon unbranched alcohols: methanol, ethanol and propanol, four- and five-carbon branched alcohols: isobutyl alcohol and the two isomers of pentanol (also known as amyl alcohol) 2-methyl 1-butanol (active amyl alcohol) and 3-methyl, 1-butanol (isoamyl alcohol). Furthermore, ethyl acetate and, the di-ether, 1,1-diethoxyethane could be present as well.
Fuel processor
The Fuel Processor provides pure H2 or suitable syngas to the stacks and it is therefore the core of a m-CHP fuel cell based-system. Under current development there are systems reforming natural gas, gasoline, diesel, and renewable fuels. Companies like Precision Combustion Inc., Innovatek, WS-Reformer-GmbH provide small sized prototypes that are of interest for the project, although they are not yet fully automated, and an operator presence is still necessary. No price information is available; nevertheless, it is known that they are very expensive. Due to the small number of fuel processors produced, a reasonable price can be in the order of thousands € per Nm3/h of hydrogen supplied. Thus, this technology is not yet commercial at all and only research laboratories systems are available.
The fuel processor must supply pure hydrogen or the best suitable syngas to feed PEM stacks. Indeed, the fuel cell behaviour is influenced by the CO content in the syngas which cannot exceed 1 % for HT-PEM and 20 ppm for LT-PEM. Thus, particular attention must be paid to design and build the fuel processor keeping in mind the aim is a commercial product in a competitive market. Each component and the final assembly should not be labour-intensive to avoid quality instability and a subsequent cost increase. Additionally, high costs are incurred due to the noble metals required as catalysts for the fuel processor, as an example, palladium costs 27 €/g . Further attention should be paid to recover the catalyst once the fuel processor has reached the end of its life. Finally, the fuel processor should tolerate a small amount of impurities such as methanol, gasoline (denaturant), butanone, propanol, methyl butanol and thiophene.
TUE and TECNALIA have been developing small scale fuel processors based on membrane reactors. A first prototype of 2 m3/h of pure hydrogen using metallic supported membranes has been constructed and is under test. The 20 m3/h version is being constructed.
Fuel Cells
A Fuel Cell based m-CHP system requires stable, reliable and cost competitive stacks. At present, this technology is making its way through a commercial product, mature, resistant and with feasible costs. Fuel Cell manufacturers are continuously trying to develop technologies to provide a more cost-effective reduction to commercialize competitive products in market. Companies such as Ballard or Serenergy foresee a cost of around 500 €/kWe for the whole stack in the short period. But costs are still high, and not at all competitive; for example, a PEM stack costs around 1,300 €/kWe for Ballard and 1,800 €/kWe for PowerCell.
The most commercially ready technologies available within fuel cells are LT-PEMs, HT-PEMs and SOFC; the last two are still a step behind the other. FluidCELL project uses the LT-PEM, which is the most mature technology available at acceptable prices, even if there are several uncertainties about its stability and duration. Using LT-PEM, poses stringent requirements on the fuel processor because this kind of cells needs pure hydrogen or, according to stacks manufacturers, CO content in syngas up to 20 ppm, even if experimental data have shown that CO values less than 5 ppm give the best electrical performance. Thus, a close restriction on syngas composition is underlined.
Power Conditioning
Power conditioning has already been largely used in solar applications, as a result, it is a well-known technology, available on a large scale and on a wide size range, and finally, it can be easily adapted to the fuel cell technology. Therefore, this component does not pose any significant problem.
Balance of plant
The term Balance of plant identifies all the remaining m-CHP system components such as pumps, blowers, heat injectors, demineralizer and analysers. All of these components must be chosen from commercially available products since they offer lower costs and higher reliability than ad-hoc built ones. Failure of one BoP component causes the whole system malfunction, especially when a fluid management component (like pumps or valves) failure occurs. Experience proves that statistically, the main system problems come just from these components. Thus, particular attention must be paid towards ordinary maintenance of each element to avoid problems in normal operation.
Controller
The development of a fully automated CHP system is the industrial final aim. Therefore, the controller plays a key role in managing the system and seems to be one of the most critical components. To analyse the correct system behaviour, several variables must be acquired and monitored.
Country assessment - most attractive countries
Some requirements for the market introduction for the m-CHP system is hereafter discussed. The aim is to find the right match both in terms of size and configuration for each country as well as establishing economic viability for the FluidCELL system. The following key variables change from one country to another:
➢ Standard and regulation
➢ Government grant
➢ Cost of electricity
➢ Cost of natural gas
➢ Dwellings average consumption of electric energy
➢ Dwellings average consumption of hot water
Investment and O&M costs are assumed constant across of the countries being studied. Standards and regulations tend to differ across countries but all of them come from European Directives. Thus, it was decided to only consider European Standards and not those of the individual countries, assuming that, in the near future, all member states will implement and comply with the various European Directives regarding micro cogeneration. In particular, reference has been made to the CHP Directive 2004/08/EC (On the promotion of cogeneration based on useful heat demand in the internal energy market) and its correlated documents (2007/74/EC and 2008/952/EC).
Moreover, government grants differ among countries. For example, the Italian grant (ESC Energy Saving Certificate) corresponds to an average remuneration of 120 €/toe (0.022 €/kWh) for electricity produced by a high energy efficiency system. In Netherlands the grant is given by the price difference between fossil and renewable power production, up to a max price of 0.15 €/kWh. Switzerland provides grants to programs that are most efficient based on the kWhe consumption avoided . Instead, Portugal government grants are fiscal benefits on Individual Income Tax for high energy class level homes, which are an indirect subsidy for high efficiency systems installed. Thus, for the analysis the Italian grant for all countries has been assumed, only for Netherland will be applied their own grant (the max price of 0.15 €/kWh applicable for CHP from biogas). This is due to lack of information on the other countries government grants and the difficulty to apply their own grants to Portugal and Switzerland. The choice of the Italian grant instead of the Dutch one is due to the lowest impact that it should have in the economic analysis. Energy costs and consumptions of sample countries across Europe, will be presented in next paragraph, while for the bioethanol cost, it has been assumed the same price for all countries because its market isn’t enough widespread like that of gas and electricity. The price assumed is the quoted price at the beginning of June (2018) of 1.462 $/gal (0.034 €/kWh).
European countries – Cross section
The European countries taken into account as sample are the participants to this project, representing different Europe areas (Italy, France, Spain, Portugal, The Netherlands, Switzerland), plus other countries like Germany, Finland, Poland, Austria, Greece, Ireland and Sweden, considered interesting for the electricity to natural gas price ratio (Table 11). This index identifies countries where systems like the one studied in this project could be much profitable due to high electricity price and low natural gas price. Some of these countries are also interesting for the dwelling average energy consumption rate (Table 12), which is very high. Therefore, households might be more sensible towards economic savings arising from m-CHP systems.
The following tables summarize data on energy cost and consumption, useful for an economic evaluation, obtained from the European database. Data includes all taxes and levies and it is referred to the first semester (S1) of 2017.
Table 11. NG and electricity price (2017) .
Country Natural Gas 2017 S1 [€/kWh] Electricity 2017 S1 [€/kWh] EE/NG Ratio Spark spread [€/kWh]
AUSTRIA 0.067 0.195 2.893 0.128
FINLAND 0.051 0.193 3.784 0.142
FRANCE 0.064 0.169 2.645 0.105
GERMANY 0.061 0.305 4.989 0.244
GREECE 0.056 0.194 3.457 0.138
IRELAND 0.063 0.231 3.647 0.167
ITALY 0.070 0.214 3.043 0.144
NETHERLANDS 0.076 0.156 2.047 0.080
POLAND 0.042 0.146 3.494 0.104
PORTUGAL 0.077 0.228 2.955 0.151
SPAIN 0.067 0.230 3.442 0.163
SWEDEN 0.121 0.194 1.597 0.072
SWITZERLAND 0.097 0.152 1.567 0.055
Table 12. Unit consumption per dwelling by end uses (2014 Report).
Country Electricity [kWh/y] Space heating [kWh/y] Water heating [kWh/y]
AUSTRIA 2,675 15,468 2,326
FINLAND 4,187 16,980 2,442
FRANCE 2,793 12,432 1,841
GERMANY 2,210 13,723 2,210
GREECE 3,024 10,816 1,047
IRELAND 2,576 15,835 3,666
ITALY 2,185 9,380 1,169
NETHERLANDS 2,752 11,688 2,460
POLAND 2,358 11,791 2,358
PORTUGAL 1,996 1,980 1,986
SPAIN 2,232 4,793 2,912
SWEDEN 3,722 12,677 1,977
SWITZERLAND 3,275 13,194 2,681
Since within the same country a lot of variability can occur in consumption of electricity and heat, the financial analysis will account for dwelling average data about energy consumption. Indeed, in Italy, the heating consumption difference between Sicily and Alpine location is remarkable. Moreover, variables such as family size, energy class, dwelling size and seasonality can vary greatly as well. As a result, the data presented is semi-quantitative and every application must be considered individually to obtain the maximum benefit in consumption reduction. In this respect, the Energy Service Companies, ESCO, can provide ad-hoc configuration for each situation.
Comparing FLUIDCELL with existing m-CHP systems
To provide an overview of the context in which the FluidCELL system is found, a comparison with the main technologies for m-CHP systems is now discussed. Table 13 provides a comparison among the systems based on electrical and thermal performances and efficiencies. Table 14 summarizes the yearly amount of electricity and thermal energy output for each system.
Table 13. Systems powers & efficiencies.
System EE Power (Size) EE Efficiency Tot. Efficiency TH Power
[kWe] [%] [%] [kWt]
ICE 1kW 1 26.3 85 1.8
ICE 20kW 20 30 85 36.7
Stirling 1 15 85 3.3
μTG 30 26 85 47.3
FC 1 40 90 1.3
FluidCELL 5 40 90 6.25
Table 14. System costs & yearly energy produced.
System Investment Cost O&M Cost Electricity Thermal energy
[€/kWe] [€/year] [kWhe/year] [kWht/year]
ICE 1kW 6,000 86 7,500 13,750
ICE 20kW 2,000 1,716 150,000 275,000
Stirling 10,000 114 7,500 24,375
μTG 1,700 3,104 225,000 354,545
FC 20,000 117 7,500 9,375
FluidCELL 10,000 587 37,500 46,875
Financial Analysis
To perform the financial analysis for the FluidCELL system, three parameters have been considered: pay-back time (PBT), net present value (NPV) and internal rate of return (IRR). The rationale for establishing economic feasibility requires NPV>0, high IRR and low PBT.
To perform this analysis the following costs have been taken into account:
➢ Natural gas and electricity for separate generation
➢ Operation and Maintenance costs of back-up boiler
➢ Cost of Natural Gas for back-up boiler
➢ Government grants
Moreover, the investment cost considered the system cost, the bio-ethanol consumed by the system and the O&M costs. For the evaluation of the economic indexes the cash flow is discounted with an inflation long term forecast rate of 2 % and a nominal real risk-free rate of 4.5 %.
A service lifetime of 10 years and an availability of 7,500 hours per year have been assumed. Moreover, it was assumed the system will operate 3,924 hours in winter season and 3576 hours in summer. Since dwelling energy consumption data is an average value, a simplification on system partial load has been introduced with two periods of regulation, winter load and summer load.
The financial analysis prioritized payback time (PBT) reduction in order to let the system become more attractive for the costumer. For doing that, it is assumed that during all seasons the system has the same (maximum) load tuned on minimum thermal requirement by the dwellings. In this way, all the electrical and thermal power is used and electricity from the grid as well as a backup boiler is also required.
Even though sharing one FluidCELL system among a number of dwellings higher than 10 may not be feasible, it provides valuable information about FluidCELL financial performances in this optimistic setting. To assure the shortest PBT, the system must run 24/7 at full workload in both summer and winter and all the energy produced must be used. Thus, several dwellings are required.
Determine the cost of the investment
Table 18 includes the costs for the FluidCELL components provided by all the partners of this project for three different case scenarios, that is, for 1, 100 and 1,000 units production per year. While for the Pd based membranes TECNALIA provided their own cost estimates, the price for the other components was calculated from the prototype with historic cost evolution from ICI purchasing department for new products while moving from prototype to large sales volume.
The starting point for the costs analysis is to divide the components in three types of product:
➢ Market-standard components based on mature technology, the cost of this commodity doesn’t change so much with the quantity
➢ Market-specific components based on mature technology, study-time have been spent by the supplier to reduce significantly the cost when quantity increases.
➢ Immature component that has required a lot of study time to develop, it is just a prototype phase.
In the tables below are presented how the quantity of production affects the costs evolution for each component’s typology described above (cost evolution based on ICI purchasing department unit).
Table 15. Market-Standard components ratio cost evolution/quantity.
Production quantity impact for a mature technology
Standard components (such as electrotech components)
Quantity Ratio Product cost /
single standard component cost
1 100.0
10 99.0
100 96.5
1,000 95.0
Table 16. Market-Specific components ratio cost evolution/quantity.
Production quantity impact for a mature technology
Specific components (sheet metal, electrical board...)
Quantity Ratio Product cost /
single specific component cost
1 100
10 70
100 50
1,000 30
Table 17. Market-“Immature” components ratio costs evolution/quantity.
Production quantity impact for an immature technology
Reference used: ICI burner designed and developed with ad hoc features
Quantity Ratio Product cost /
Prototype cost
1 1
10 0.753
100 0.452
1,000 0.070
Additionally, the cost for the Fuel Cell is based on an estimate from the 2017 report by the Department of Energy (USA) that set a cost of 50 $/kWe when mass production is considered. The most cost-relevant component appears to be the Fuel Processor subsystem.
Table 18. FluidCELL Cost per Unit.
Price per 1 unit [€] Price per 100 unit [€] Price per 1000 unit [€]
Description Company Component/
material Cost Labor
Cost Total
Cost Component/
material Cost Labor
Cost Total
Cost Component/
material Cost Labor
Cost Total
Cost
Catalyst UNISA 2,500 2,500 5,000 250 1,250 1,500 200 200 400
Pd Membrane TECNALIA - - 13,000 - - 3,400 - - 2,500
Fuel Processor HYGEAR 65,000 15,000 80,000 40,000 10,000 50,000 20,000 5,000 25,000
Fuel Cell CEA - - 35,000 - - 10,000 - - 250
Integration ICI 35,240 123,200 158,440 26,599 36,960 63,559 16,382 6,160 22,542
291,440 128,459 50,692
FluidCELL System Analysis
To decrease the payback time (PBT) the final user has to fully consume all the electrical and thermal output of the system and run the system at full load all the time. Furthermore, efficiency is higher when working at full capacity rather than with reduced workloads. Therefore, to maximize efficiency the system has to cover the energy demand of multiple dwellings, thus, it requires a back-up boiler. With this consideration and merging data from Table 11 (NG and electricity price in different European country) and Table 12 (Unit consumption per dwelling by end uses) the following Table 19 can be obtained.
Table 19. FluidCELL Financial Analysis.
Country Dwellings PBT [Years] NPV [€/Y] IRR Saving[€/Y]
AUSTRIA 14 9 2,049 1% 7,253.15
FINLAND 13 >10 - 16,224 - 7% 4,858.38
FRANCE 26 >10 -2,645 -1% 6,637.87
GERMANY 18 5 33,327 11% 11,352.16
GREECE 40 9 195 0% 7,010.16
IRELAND 14 7 12,588 5% 8,634.26
ITALY 17 8 4,492 2% 7,573.22
NETHERLANDS 19 >10 - 8,058 - 3% 5,928.54
POLAND 16 >10 - 25,245 -11% 3,676.11
PORTUGAL 19 6 16,558 6% 9,154.57
SPAIN 17 7 16,001 6% 9,081.62
SWEDEN 19 6 22,607 8% 9,947.32
SWITZERLAND 17 8 5,095 2% 7,652.33
Figure 29 visualizes how the different countries performed based on PBT. The current analysis yields a cost of 10,000 €/kWe. Germany appears to be the most attractive country according to all three financial indicators. Across the countries examined, the average Payback Time is 8 years while the life of the FluidCELL system is 10 years.
Figure 29. FluidCELL PayBack Time.
Main conclusions
➢ FluidCELL has to provide energy to several dwellings, from the analysis emerged 19 dwellings are associated on average with the system. Multiple dwellings have been introduced due to the necessity to let the system always operate at full workload, while also utilizing all of the electrical and thermal energy produced.
➢ The system cost is 10,000 €/kWhe.
➢ The lowest PBT for the FluidCELL system has been achieved by Germany which scored 5 years PBT, still, the average PBT for the FluidCELL system is 8 years.
➢ To achieve an attractive 3-year PBT, the individual component costs have to decrease.
➢ Currently, the Fuel Processor is the most critical cost component for FluidCELL, accounting for 49 % of the system total cost.
➢ Since production costs are still high, it is advisable to make use of government grants.
➢ If possible, to further reduce production costs, it is advisable to increase the use of standardized components.
➢ Further work is required in order to reduce investment cost through research in optimization and industrialization of fuel processor and system integration
See FluidCELL_Publishable_final-report-18102018_v01.pdf
List of Websites:
The address of the public Website of the Project as well as relevant contact details.
Project public website with further information of the about the project and consortium and main contacts details are detailed hereafter:
https://www.fluidcell.eu/
Project manager: Dr. José Luis Viviente
e-mail: joseluis.viviente@tecnalia.com
Technical manager : Prof. Fausto Gallucci
e-mail: F.Gallucci@tue.nl
Dissemination manager: Associate Prof. Giampaolo Manzolini
e-mail: giampaolo.manzolini@polimi.it
Exploitation manager: Dr. Leonardo Roses
e-mail: leonardo.roses@hygear.com
List of all beneficiaries with the corresponding contact name and associated coordinates
Nº Participant short name Contact name E-mail
1 TECNALIA José Luis Viviente joseluis.viviente@tecnalia.com
2 TU/e Fausto Gallucci f.gallucci@tue.nl
3 CEA Sylvie Escribano sylvie.escribano@cea.fr
4 POLIMI Giampaolo Manzolini giampaolo.manzolini@polimi.it
5 UNISA Vincenzo Palma vpalma@unisa.it
6 UPORTO Adelio Mendes mendes@fe.up.pt
7 ICI Carlo Tregambe carlo.tregambe@icicaldaie.com
8 HYG Leonardo Roses leonardo.roses@hygear.com
9 QUANTIS Violaine Maguad violaine.magaud@quantis-intl.com
See FluidCELL_Publishable_final-report-18102018_v01.pdf