Click here to enlarge image

A French company and its research centre have begun a long-term trial of the operation of a stationary fuel cell-based cogeneration system installed in a seasonal district heating application and fuelled by natural gas. The first performance measurements are promising, reports Alexandre Lima

In 2003 Veolia Environnement and Dalkia France launched a programme of tests in real conditions on stationary fuel cell-based power cogeneration systems. So far, this programme has led to the installation of four such systems, the most recent and largest of which is in Paris. This project, which we have called CELLIA, is led in association with Paris’ urban planning authority, OPAC, and EDF R&D and began in early 2003 with initial talks between the partners. In 2003, we compiled a specifications document for the construction of a system with an electrical power of up to 250 kW, the site having been chosen beforehand, in inner Paris. At that point we had not yet focused on a particular technology, or on a minimal power figure (the range chosen was between 50 and 250 kW of electricity). Nevertheless we wanted the system to run on natural gas because of the easy availability of this fuel in the French capital.

Finally, since the aim was to conduct tests in real working conditions, the demonstration project needed to be long-term (at least six years) and in line with the French cogeneration rhythm: i.e. the system needed to work for five months of the year (from 1 November to 1 March of the following year). This would allow us to compare the system with direct rival technologies such as gas turbines or gas engines, even if the useful electrical power produced is not really on the same scale.

Midway through 2004, we closed the call for tender (after consultation with over 20 constructors in Europe and around the world) and it became apparent that few constructors were capable, at that time, of proposing a credible offer to get the system up and running by 2006. We realized that in 2004 there was a difference between the many proposals made by the constructors and the reality of the offer, which was extremely limited.

We finally chose the German constructor CFC Solutions, which develops systems based on molten carbonate technology. Its system, the HOTModule, works on natural gas and produces a gross electrical power of 245 kW for a net power approaching 215 kW. The useful heat energy is around 200 kW. The system is very large: the machine is a parallelepiped (a geometric solid whose six faces are all parallelograms) of 8 metres long, 3.5 metres high and 3 metres wide, and weighs 26 tonnes.

At the same time, we compiled, assisted by our partners, requests for subsidies on three levels: on the European level via the European Commission’s LIFE environmental programme, on the national level via the French Environment and Energy Management Agency (ADEME), and on the regional level via the Greater Paris Regional Council (CRIDF).

In total, if we add together the cost of the civil engineering work, the preliminary studies, the purchase of the fuel cell, its construction, the system running costs and six years of studies and tests by our research centre, the total budget of the operation stands at €6.5 million.

The principal objectives of this demonstration are:

  • knowledge of the offer, mainly on the European level and in 2004, of supplying medium-power stationary fuel cell systems
  • evaluation of the technical performances of a stationary fuel cell system running on natural gas, especially knowledge of the electricity and heat outputs and the useful production
  • evaluation of the environmental impact of this system, locally and globally: in particular, evaluation of gases released into the atmosphere, water consumption, etc.
  • evaluation of running costs: maintenance, spare parts, consumables
  • knowledge and possible development of the specific skills required to operate the unit within the technical teams in charge of the system
  • recording and dealing with possible specific risks linked to operating a fuel cell
  • evaluation of how the main sub-systems fare over time, since the demonstration project is spread over a relatively long period, unique in Europe to our knowledge, and intermittently (operating for five months of the year only). It is particularly interesting to analyse how the system reacts to these seasonal shutdowns.

Finally, there is also a pedagogical aspect to the project, particularly with regard to the construction of premises capable of receiving numerous visits, and with on-site multimedia resources, but also through communication in specialized conferences or via more generalist media.

The system installed in Paris

The system operating in Paris is based on molten carbonate technology, which has two distinctive features.

  • The fuel cell stack operates at a temperature of 650ºC, which provides the operator with useful heat at a sufficiently high temperature to prepare hot water and heating; compared, for example, with PEMFC or even PAFC technologies, the system can be operated independently from a backup boiler, even when the client’s heat requirements are low.

  • The fuel cell stack uses a molten carbonate-based electrolyte (CO32-). During the oxidation reaction between the hydrogen on the electrodes, the ionic transfers between the electrodes cause the cathode carbonate ions to move towards the anode. This movement must be offset by adding CO2 at the cathode. Thereby, the CO2 produced at the anode during the oxidation of the hydrogen is introduced to the cathode. This point, despite being invisible to the end user, adds to the complexity of the energy production process compared to the other fuel cell technologies.
Click here to enlarge image

Apart from these specific features of molten carbonate fuel cell (MCFC) technology, this system has all the usual components that make up a fuel cell system: a reformer (referred to here as the ‘media supply’) destined to produce hydrogen from natural gas, the fuel cell stack coupled with a catalytic after-burner, and an inverter to convert the continuous electrical current into alternating current compatible with the electricity network.

Special attention has been paid, within the media supply, to the desulphurization stage. This organ is crucial because the slightest trace of sulphur arriving in the reformer and/or in the fuel cell stack would cause irreversible damage by blocking the reaction sites. This process consists of a bed made of a mixture of active coals capable of trapping sulphurous elements (H2S, THT, etc.). By nature of their construction, these filters must be regularly changed to ensure that they operate optimally. Given that the rate of sulphur varies from one country to another, as does the very nature of the sulphurous compounds found in the natural gas, the life cycle of active coals varies from country to country depending on the quality of the natural gas.

So, since at present we lack the necessary experience, the operator is forced to analyse, currently several times a year, the gas after it has undergone the desulphurization process. With this requirement in mind from the outset of the project, the constructor integrated the nozzles required to take samples of the gas. The results of the preventive analyses provide information about the saturation state of the active coals and allow us to anticipate any possible problem of sulphur poisoning. In real terms, two desulphurization bottles work in series and it is possible to isolate one of the two bottles to change its content, without having to stop the fuel cell.

Figure 1. Simplified working principle of the HOTModule
Click here to enlarge image

Figure 1 presents a simplified diagram of how the HOTModule works.

Energy flows

The system runs on natural gas, air and water. Aside from passing through a series of dust-removal filters, the air is not subjected to any specific treatment. The sulphur is removed from the natural gas thanks to the active coals, before going through the traditional reformer stage using steam. The public water supply is purified by a sub-system integrated within the reformer, which first softens it (using salts), then filters it and finally subjects it to a reverse osmosis.

Figure 2. Cut-away diagram of the HOTModule by CFC Solutions
Click here to enlarge image

At the output, the stack produces continuous electrical current that is transformed into alternating electrical current within an inverter, at low voltage. In the specific case of Paris, the electricity obtained after the inverter is transformed into electricity in high voltage to make it compatible with the network. Indeed, except for that used to run certain elements of the boiler room (pumps, lighting, safety mechanisms), the electricity is resold to EDF as part of a traditional cogeneration contract respecting the French regulations in force.

Therefore, this contract stipulates that the electricity is purchased at a preferential rate, solely between 1 November and 31 March of the following year. As a result, our system built in Paris is operated for only five months of the year, during the period when heating is required, in a similar way to the other CHP units operated by Veolia Environnement.

Finally, the useful heat produced is used on the unit’s flues as a flow of humid hot air at approximately 350ºC and at atmospheric pressure. So, on the roof of the media supply, we have installed a tube heat exchanger, which sends the water heated to 90ºC directly into the primary hydraulic circuit of the existing boiler room.

Thus, the fuel cell contributes to the production of heating and/or hot water for 283 rent-controlled homes that are also supplied by the main boiler room. In very cold periods, our forecasts indicate that we are capable of meeting 15%–20% of these apartments’ requirements thanks to Cellia.

Generally speaking and at the current stage of development, fuel cell systems, regardless of the technology chosen, are capable of modulating their power according to the demand, but frequent variations have an effect on their service life, especially that of the fuel cell stack. In the specific case of the fuel cell built in Paris, and given the high temperature technology, the variations are slow because the temperature ramps need to be controlled.

Therefore this tool is not the best adapted to closely follow a load curve. Thus, the host site was chosen so that the unit could run normally/on the established basis, i.e. at full power throughout the cogeneration season.

Installing a fuel cell

A word about safety
Fuel cells are often associated with hydrogen risk. However, we should remember that this system runs on natural gas. Hydrogen is indeed produced within the system, but it is just as quickly consumed within the fuel cell stack so that only a few grams of hydrogen are present in the fuel cell at any time. Therefore, the risks linked to the use of a fuel cell are fundamentally no different from the risks linked to a ‘traditional’ gas boiler.

The HotModule, prior to delivery, was the object of a system safety study by the German firm TÜV, notably on the existence of explosive atmosphere (ATEX) zones. The system is also CE labelled.

Before the arrival of the system in Paris, and during major building work, the Veolia Environnement Research Centre also launched a safety study, entrusted to an independent third party, concerning the integration of the HotModule into its new premises and close to the existing boiler room.

This safety study led to the simulation of several scenarios, mainly with regard to the possibility of the gas pipes bursting (natural gas, hydrogen, CO2, etc.).

According to these scenarios and the regulations in force, our service provider came up with recommendations that we followed. These recommendations covered safety aspects such as the type of fire doors to use, the type, positioning and number of gas detectors, and the air extraction means to install on the premises. As a result, the site does not have permanent ATEX zones. However a type 1 ATEX zone is located between the reformer and the fuel cell stack, as previously stipulated by the report provided by TÜV.

The system is fully automated and therefore does not require the presence of personnel in the boiler room. In the case of an alarm (gas leak, sudden temperature change within the fuel cell stack, fire, unusual gas pressure readings, or similar), the Dalkia Call Centre (CTRA) is immediately and automatically alerted by telephone, which consequently alerts the technician on call who will go to the site to deal with the problem. The system has three alarm levels. The first two levels correspond to minor malfunctions that require human intervention. The third alarm level is reserved for cases that could threaten the safety of people and materials and this immediately leads to the automatic and total shutdown of the unit. As in the other cases, the alert is transmitted to the call centre.

The need to store auxiliary gases

To start-up the fuel cell, but also to ensure its safe operation, there need to be three auxiliary gas storage zones.

For nitrogen: a tank for liquid nitrogen with a capacity of 6000 litres has been built. In the event of an emergency shutdown of the fuel cell, this gas is used to purge the fuel gas pipes. There is a dual objective: to ensure that absolutely no combustible is present in the system for safety reasons, but also to protect the components, especially the electrodes in the fuel cell stack, from the corrosion that would be caused by the presence of oxygen at a high temperature on the anodic side. Therefore, premises were built outside the building to house this tank. A gauge is directly linked to the control unit and an insufficient nitrogen level activates a warning.

For hydrogen and CO2: to ensure the start-up and to control the quality of the combustible gas, a controlled mix of CO2 and hydrogen feeds the anodes almost throughout the system start-up procedure. In fact, as the temperature gradually rises in the fuel cell stack (less than 10ºC/hour, i.e. three days to rise), the system calls for the use of these start-up gases. This means that we need to store the equivalent of 10 kg of hydrogen on site. So we have set aside part of the building, in closed, ventilated and controlled premises, to install traditional pressurized bottles (200 bar) – the same goes for the nitrogen. The pressure level in the tanks is directly transmitted to the control unit and the alarm is raised if the level is too low.

Aside from the technical specificity of the unit, French regulations do not require authorization for such storage facilities. Storage of 100 kg needs to be declared to the competent authorities while authorizations are only required for storage of 1000 kg or more.

The first start-up

Once it had been built, the fuel cell was inspected by TÜV before the initial start-up. It approved the system and declared it ready. Starting up the machine, especially when getting it up and running, is a delicate operation. It consists of a slow and controlled increase in the temperature of the fuel cell stack (less than 10ºC/hour), from room temperature up to 600ºC. This rise is brought about in a controlled atmosphere within the anodes (negative poles): the electrodes are successively supplied with nitrogen then with a mixture of H2 + CO2, which explains the pressurized storage of the gas in an adapted facility. This rise stage lasts approximately three days. Once the temperature of 600ºC has been reached, the reformer, which is already running, can then supply the fuel cell stack with hydrogen created by reforming the natural gas.

The procedure of the first start-up required the permanent presence of technical personnel – at least one or two people were in the plant for each eight hour period. For the next start-up, this presence was not necessary since the increase in the temperature of the system occurred automatically.

Finally, after verifying the systems connecting the fuel cell up to the electrical network, the reformate fuel gas was introduced into the fuel cell stack, allowing the first kWh of electricity to be produced.

Monitoring performance

The workings of the system are recorded continuously on two computers located in the control unit. Some 400 parameters are recorded continuously as the fuel cell runs, but most of this data is, above all, valuable for the constructor who may need them for diagnosis or analysis purposes.

For the operator, only around fifteen or twenty of these parameters are valuable or interesting to measure the system’s performances. The others are useful for an overall understanding of the way the machine works. It should be noted that these computers are not essential to the functioning of the system. They are automats that ensure both the regulation of the parameters of the fuel cell and its safety.

To complete the experience feedback process, we have installed meters for each fluid entering and leaving the system: electricity, heat, water and natural gas. Concerning the natural gas, we have begun installing a complementary tool: an analyser of the quality of the natural gas, developed by the Veolia Environnement Research Centre, which continuously measures in real time the methane index, the LHV, the HHV, the Wobbe index as well as other information such as the density or the CO2 content. This instrument is particularly useful because, given that the quality of the natural gas is not constant, it allows us to calculate the real output of the fuel cell more precisely.

All this computerized data is centralized on a third computer. All the computers can be remotely accessed from our research centre.

The performances shown in Figure 3 concern the first period of tests conducted between March and July 2007. At the time of writing this report, the first season of cogeneration (1 November 2007–31 March 2008) is not over.

Figure 3. Main fuel cell performances between April and July 2007
Click here to enlarge image

Figure 3 shows the main performances measured during the period of tests. During these four months, we have made adjustments to the system’s settings and optimized the functioning of the peripherals in the boiler room. Thus, the fuel cell’s various separate malfunctions during the first eight weeks prevented us from attaining the system’s maximal power: the pressure of the natural gas proved to be insufficient to allow the system power to rise above 120 kW. As soon as a sufficient pressure was available (from the month of June), i.e. a minimum of 0.8 bar when entering the fuel cell, the system could reach its nominal power.

These figures are obtained taking into account the real value of the LHV of the gas. So we are able to note that the average electrical output is stable during the test period, in terms of both a partial load and the nominal power. It is between 42% and 45%. Peaks were recorded at almost 47% net, according to the quality of the natural gas.

While these performances represent a very good level, particularly compared to what is obtained by rival technologies, they remain slightly inferior to the constructor’s initial announcements. The commercial presentations of CFC Solutions generally announce an average of 46% net electrical output. However, we should put these initial results into their context: they were recorded during the adjustment and optimization period and, after the seasonal shutdown, adjustments were made by the constructor and by Dalkia. Therefore, we can hope for a distinct improvement of the outputs in the seasons ahead. But even now, we can note one of the major advantages of fuel cell systems: their high energy output, mainly in electrical power.

Click here to enlarge image

Looking at the environmental performances, they are in line with our expectations – see Table 2. The main performance of note is the absence of N2O in the exhaust gas. The absence of SOx is normal: there is a sulphur removal mechanism at the entry of the fuel cell. However this raises the question of treating these waste gases: the sulphur-saturated active coals have to be treated and recycled.


At the time of writing this report, the first season of cogeneration of the fuel cell built in Paris is not yet over. It is, therefore, too soon to establish an initial performance evaluation. Nevertheless, even at this point, we can say that the system is working well thanks to its good performances in terms of its power output. What’s more, the first analyses of the composition of the exhaust gases emitted indicate that the system is running optimally and in line with our expectations.

Since the beginning of the CHP season, we experienced a partial change of charcoals (desulphurization), suggesting that the active coals sub-system is correctly sized for one complete season, but needs to be changed at least once a year, for example between two CHP seasons. We are at the moment trying to confirm this assertion by analyzing the desulphurized gas with chromatography and other laboratory techniques.

We also led a new sampling of exhaust fumes using Fourier transform infrared techniques at several power levels and we are still analyzing the results. The goal for us is first to understand the complete behaviour of the plant and possibly analyse some dysfunctions if they occur(ed). The complete analysis of the first season should be over by the end of summer, so that we can integrate the results during the next CHP season.

We have created a complete working schedule for this demonstration, which is planned to last for six years:

  • We will monitor how the system ages over time, notably via any changes in its energetic or environmental performances, but also by monitoring certain specific characteristics such as the electric voltage of the electrochemical cells or the temperature of the fuel cell stack. Access to the many parameters that are constantly recorded will enable us to diagnose correctly any possible difficulties with the technology. The service life of the system is announced at a minimum of 20,000 working hours, perhaps even 30,000 hours. In fact, this limitation corresponds to the lifetime of the fuel cell stack. We have already planned to be able to replace a fuel cell stack in order to extend the demonstration.
  • If malfunctions do occur, especially on the fuel cell stack’s auxiliary components (control unit, water treatment, reformer, desulphurization of the natural gas, after-burner module), the difficulties will be identified and the circumstances of the problem will be noted.
  • The life cycle of certain consumable components will be monitored particularly closely, especially that of the two desulphurization units or that of the consumables of the integrated water treatment module.
  • A life cycle analysis study will be initiated on MCFC-type fuel cell systems based on the fuel cell in Paris.
  • Lastly and above all, we are going to evaluate the maintenance requirements especially in order to evaluate the running costs of such a system.

Also from now on, we feel that it is important to remove the mystery surrounding this technology and break with certain preconceived ideas concerning fuel cells. Firstly, there is a general view that a fuel cell produces water. This is only true if we consider the fuel cell stack only: the electrochemical reaction of oxidation of hydrogen does indeed produce water, but the overall system uses water: pure water is needed to reform the natural gas and the production of water in the fuel cell stack is not enough to compensate for the water consumed by the reformer. On this subject, several constructors are working on the autonomy of fuel cells in terms of water. Otherwise, the question of how much noise the system makes is often raised. Since the fuel cell stack does not contain any moving parts it does not make any sound, but the system does include a number of auxiliary parts and mechanisms: pumps, ventilators and compressors. However, all in all, a fuel cell system is quieter than a gas turbine or a gas engine.

We are now at a point where stationary fuel cell systems have left the laboratory drawing boards and are now being tested in real situations. These systems must now prove their worth and should normally progress even further. In our opinion, constructors should soon be capable of offering end users systems that are more compact (by increasing the power density), less expensive in terms of the investment required, and capable of working with fuels other than natural gas. There are already a few systems working in Europe (Germany) and in the rest of the world that use biomass fuels, and also waste products (for example landfill gas or gas produced by the fermentation of sludge at water purification plants).

Alexandre Lima is with the Veolia Environnement Energy Research Centre, Paris, France