1 June 2007

Fuel cells go pro

A new approach to uninterruptible power could mean curtains for traditional battery power

Fast Forward

Replacing the battery in uninterruptible power supplies could mean more reliability and less cost.
Chemical process facilities see hope in proton emitting membrane fuel cells for process control, emergency shutdown, and fire detection.
Some companies now provide small businesses with off-the-shelf products for IT systems' continuous power.

By Nick Harbud and Tom Sperrey

The Swedish nuclear power plant at Forsmark experienced a short circuit at an outdoor switching station, causing workers to spend a frantic 45 minutes conducting an emergency shutdown. The preliminary report from that 25 July 2006 incident said the cause was the malfunction of vital safety-related equipment in which two out of four emergency generators failed to start. The most serious aspect of the incident, in the view of Statens Kärnkraftinspektion (SKI), the Swedish nuclear power inspectorate, was safety systems that should have been independent were not sufficiently separate. In February, SKI initiated investigations to prosecute the management of Forsmark for not bringing the reactor to a cool shutdown fast enough.

As the Forsmark incident shows, diesel generators are not a particularly reliable source of emergency or standby power. This so-called emergency power supply has a number of failure modes in common with the mains power, such as the electrical distribution and switchgear. Consequently, we generally calculate uninterruptible power supply (UPS) battery requirements as a standalone reserve on the basis of sufficient duration to allow safe shutdown of the plant as one instance. The capacity is not based on the length of time required to start the diesel generator. Therefore, when assessing fuel cell reliability, it might be wiser to use the batteries as a benchmark. An alternative to the battery would provide more reliability and duration and reduce costs and space requirements. An emerging technology that meets all these criteria is the proton emitting membrane (PEM) fuel cell.

PEM fuel cells over the years have matured into a commercially viable product. (See related article on page 38.) Within a hydrocarbon processing facility, the hydrogen fuel generates onsite, and costs are minimal. Its storage infrastructure already exists, and the industry is completely familiar with its use. There is a possibility one day most process plants will operate comfortably with the reassurance of the fuel cell UPS units, and we will have banished the battery.

Progress obstacles

In some cases, it is important to know certain non-technical obstacles to progress exist. One of the most commonly used fire system standards, NFPA 72, mandates technology we can use for standby power supplies, which inhibits innovation within its scope. However, we can still gain benefits from integrating the other systems, making the search for an alternative UPS solution worthwhile.

Desirable features in a solution could include minimum 24 hours duration for all systems requiring UPS, space requirements within buildings not to exceed that required for traditional battery rooms, fast recharge and renewal of backup storage following depletion, and reduced lifetime costs. Can we find a solution that will meet these criteria? We believe within the context of a hydrocarbon processing facility, replacement of batteries with modern fuel cells provides an opportunity to reduce costs while improving the quality of uninterruptible power supplies.

Large chemical process facilities, such as oil refineries or petrochemical complexes, include electronic systems providing discrete functions-process control, emergency shutdown, fire detection, telecommunications, closed circuit television (CCTV), and public address. Tradition used to dictate having separate systems to perform these functions. However it is possible to uniquely combine or integrate these systems with technological developments. Most distributed process control systems (DCS) can accept a video stream from a CCTV system and display a camera image on any of the operator workstations, eliminating the need for separate CCTV monitors. Similarly, software controls in the DCS can replace hardware camera control units. Plenty of products combine the fire detection and public address systems to issue prerecorded messages during fire detection. It is quite common for a fire detection system to automatically direct a CCTV camera in the direction of an active fire alarm. CCTV image recognition software can detect some types of fire.

In addition to cost savings, production facility staff can enjoy operational and maintenance benefits from these technologies. Yet the power supply is a major and unlikely obstacle in their path. The required duration of the UPS batteries differs between applications. Process control and shutdown systems usually require between 30 minutes and two hours battery backup to shut down the plant; communications, CCTV, and security systems can require four to eight hours; electrical switchgear two to eight hours; and fire systems more than 24 hours.

A battery backup requirement for any combined system will be the longest duration of any of its constituents. So using a DCS to display CCTV images, we can increase the battery backup duration from two to eight hours, quadrupling the number of batteries required for the DCS.
The cost of extra batteries alone will generally exceed any savings from combining systems without even considering additional space requirements, even though modern Nicad and sealed batteries have done away with acid-resistant floor coatings, drains, and hydrogen detectors associated with a traditional battery room. The control buildings and remote instrument buildings on a hydrocarbon processing facility must be blast-resistant. Plus, space in these buildings is expensive and at a premium.

When considering the problems of additional battery cost and space, most control system and telecommunications engineers will stick with the traditional architecture of separate systems, each with an independent UPS. Yet this architecture, although well-proven, is far from elegant. Control buildings, remote instrument buildings, and substations could have two or three different UPS installations, each with their own quota of batteries and a dedicated battery room. On a large refinery or petrochemicals complex, this can amount to over 100 separate systems. Following a main power outage, when the UPS has the task of powering essential systems, the batteries must recharge. It can take up to eight hours to recharge a two-hour battery set. If another mains outage occurs during this period, the facility can find itself with insufficient reserve power. Despite being needed for short periods during main power failure, batteries must undergo maintenance throughout their life. Batteries have a life of five or 10 years, with replacement required after 80% of this rating, which is after four or eight years respectively. Even without the cost of occasional failures and periodic testing of battery installations, this represents a significant ongoing maintenance cost to a process production facility.

So how would such a fuel cell fit into a conventional UPS installation? Physically, fuel cells are compact. A conventional UPS consists of a rectifier/charger unit that feeds a DC bus connected to an inverter (producing clean AC power) and charging the standby batteries. Upon failure of the main power supply, the DC bus draws power from the batteries. Fuel cells generate DC power. So it stands to reason the fuel cell power output should feed the UPS DC bus. As the fuel cells will take a few seconds to start producing power, it is not possible to completely eliminate the batteries. However, we can substantially reduce their number and capacity.

A commercially available fuel cell product comprises three 10kW stacks, cooling pumps, control unit, and other ancillaries packaged in a standard data server cabinet (0.6m width x 1m depth x 2.2m height). The data server cabinet requires cooling water so building heating, ventilation, and air-conditioning systems can disperse generated heat. In temperate climates, heat from fuel cells often serves for space heating within buildings. The fuel cell's reaction produces water vapor, which needs a vent to route to an outside location. It produces nearly 2 l/h of clean condensed water for each 10 kW of fuel cell power, which can route to a normal sanitary or surface drain. As hydrogen is present, it is prudent to install suitable gas detectors in the room where the fuel cells are located. This is similar to the requirements for a conventional battery room.

The hydrogen fuel source stores separately from the fuel-cell cabinet. A modern PEM fuel cell can use ordinary industrial grade hydrogen (99.95% pure). Cylinders of hydrogen fuel come from distributors of industrial gases. The commonly used K cylinder contains sufficient hydrogen for 10 kWh of electricity and usually costs around $70 per charge. For such bottled gas installations, it is normal to have 2 x 100% cylinder racks to ensure continuity of fuel, either following a main power outage or when changing cylinders.

For the IT data center, the installation site of fuel cell UPS systems, hydrogen cylinders store just outside the building, an arrangement feasible for refinery or other process plant control buildings or substations. However it can be difficult and expensive to protect such cylinders if a process plant building is located in a blast zone. Of course, on a process site, you do not need to store fuel close to the fuel cell. It is common practice to store large inventories of flammable material, such as hydrogen, remotely and pipe it where it is required. Where multiple fuel cells install across a large site, it is simpler to distribute hydrogen in this manner from a single storage facility. A normal fixed storage system for hydrogen consists of a prefabricated rack of large, horizontal cylinders that store the gas at up to 200 Bar.

Another benefit of fuel cells for those sites already generating and using hydrogen within their chemical processes (including most refineries and petrochemical complexes) is more than likely a free UPS fuel storage infrastructure. The PEM fuel cell requires 99.95% purity hydrogen, a percentage that fits easily within the feedstock specification of say, a typical polymer plant. If a process plant does not generate a feedstock of this purity, you will need to install a small pressure-swing absorption unit.

It is nearly a given, under the circumstances of a mains power outage, the main generator of hydrogen from within the process plant will shutdown. Can a secure source of hydrogen guarantee supplying the fuel cell UPS during such events? Fortunately, sites using hydrogen as a process feedstock can anticipate such process interruptions. Consequently, such sites will have either fixed hydrogen storage or an alternate off-site source of hydrogen in order to cope with process trips of the hydrogen generating process units.

Fuel cell reliability

Battery reliability and mean time between failures (MTBF) can vary enormously, depending on the type of battery you choose, arrangement of batteries in series, or parallel, storage conditions, and charge management. You can reduce individual battery life by overcharging. Generally quoted MTBF figures for lead acid batteries vary between 20,000 and 80,000 hours, although there are examples of users obtaining in excess of 100,000 hours MTBF.

Reliable fuel cell MTBF figures can also be hard to pin down, mainly due to the scarcity of long-term use data from installations. Nevertheless, figures reveal between 20,000 and 40,000 hours MTBF for PEM fuel cells, which is comparable to that of the lead acid batteries.

However, raw MTBF figures do not give a complete comparison of battery versus fuel-cell reliability. Even if you keep them under ideal conditions and periodically trickle-charge them, batteries will constantly deteriorate. Once you discharge them, batteries can deteriorate rapidly. And repeated deep-discharge will lead to premature failure. By contrast, when not actually delivering power, a fuel cell is inert and does not appreciably decay. As the fuel cell will spend only a small proportion of its time delivering power, this will greatly extend its life when compared to an equivalent battery installation.

It is also possible to increase the availability of a fuel cell UPS by installing spare stacks. Installing a spare 10 kW stack in a 50 kW UPS provides n+1 redundancy without any increase in UPS footprint. Installing an equivalent number of spare batteries for a conventional UPS would increase the size of the battery room by 20%.

Fuel cell costs

Prices for fuel cells and batteries can vary considerably. In the case of the fuel cells, this is a reflection of the technology's youth. For batteries, the more expensive types will last longer. In a typical process plant control building you might find separate UPS systems, such as a 50 kW control system UPS, with two hours battery backup and a 30 kW communications system UPS with eight hours battery backup. You could replace both these systems with a single 80 kW fuel cell UPS. An on-site bulk storage facility would supply hydrogen for the fuel cells, and as there is no requirement for battery rooms; you can make the blast-resistant control building smaller. On installation costs alone, the fuel cell will save money. However, if you consider the lifetime costs, the economic case for fuel cells becomes much clearer.

Process plants are normally designed for a minimum 25 years operation, and many operate much longer. A battery backed UPS might require a complete planned replacement of batteries every eight years or three times over a period of 25 years. The fuel cell membranes could require replacement every 10 years or twice within the same 25-year period.

About the Authors

Nick Harbud (Nick.Harbud@fluor.com) is a lead control systems engineer for Fluor Ltd., a Camberley-based U.K. subsidiary of the Fluor Corporation, an international engineering contractor to the oil, gas, petrochemical and life-science industries. Tom Sperrey (tsperrey@upssystems.uk.com) is managing director of UPS Systems plc, an integrator of power supply systems.

Fuel cells unplugged

A fuel cell is a device that generates electricity, heat, and water by reacting with fuel and air without burning. Fuel cells are not a new technology; researchers discovered their principles more than 150 years ago. However, lately much research has aimed at producing more flexible, robust, and practical types of fuel cells, especially in space vehicles and the automotive industry as an alternative to the internal combustion engine.

With the PEM fuel cell, hydrogen fuel feeds into the anode. Oxygen enters the fuel cell through the cathode. Encouraged by a catalyst, the hydrogen atom splits into a proton and an electron, which take different paths to the cathode. The proton passes through the electrolyte. The electrons create a separate current and see use before they return to the cathode, to be reunited with the hydrogen and oxygen in a molecule of water.

Most fuel cells are normally connected together to produce a fuel cell stack that can generate meaningful amounts of power at around 2448 Volts. The PEM fuel cell operates around 55°C.

It is therefore able to start rapidly from cold, typically producing electricity within seconds. Like all fuel cells, it is a generator of electricity, rather than a storage system. Its duration is therefore only limited by its fuel supply, which for this type of cell is hydrogen.

Direct current means low voltage control

By Ellen Fussell Policastro

Jim Blais is a controls engineer at Hallam-ICS, an engineering and controls integration company in Burlington, Vt., that handles gas safety monitoring, shutdown, and control system design, especially toxic gas monitoring systems for facility engineers at research and manufacturing facilities.

While alternating current (AC) uninterrupted power supplies (UPSs) might be the conventional norm in protecting a system, Blais believes in direct current (DC)-to-DC UPS units to provide greater flexibility and easy installation into user's facilities. "Incorporating DC-to-DC UPS into our UL 508A control panels simplifies the installation," he said. So he can provide "on-board UPS capability without requiring a facility provided UPS circuit, which gives users a higher gas monitoring system that's high quality, with a longer life, he said. "All we need is an emergency power circuit and we are ready to go."

Most people wouldn't think about a small compact DC UPS because it's just not the norm. But what's changed everything is a migration to low-voltage DC control. Working with hazardous production materials, such as toxic, pyrophoric, flammable, corrosive, reactive, oxidizer, and oxygen depletion inert gases, requires Blais and his team to adhere to industry best practices and comply with the Underwriters Laboratories (UL) standard for safety in industrial control panels, UL508A.

International and uniform codes require toxic gas monitoring control systems to connect to an emergency power system. Configured as a stand-alone safety control system, they are similar to a fire alarm system; they're normally static or dormant when conditions are safe, and require supervised wiring, diagnostics and periodic testing to ensure their proper detection and availability.

Low voltage DC devices wouldn't normally back up products that run on AC power, said Bruce Desmond, a product marketing manager at Phoenix Contact. They would run devices that normally run off DC power. To that extent they're meant for use inside a conventional control cabinet snapped onto a piece of DIN rail, just like a DIN-rail-mounted terminal block or power supply, or a DIN-rail-mounted relay. So din-rail mountable DC-to-DC UPSs mount the same way as other popular control devices.

Blais uses UPSs to avoid momentary power interruptions, even when connected to an emergency power system. With UPS, you can "avoid the typical delay waiting for the facilities' emergency generator to start and transfer switch operation. This avoids power cycling to our detection systems and provides uninterrupted monitoring" for users, he said.