Research work is being carried out at the BRE to encourage the safe use of ammonia, and one of the major issues to be addressed concerns the dangers of uncontrolled discharges of the gas into the atmosphere.
Many designers and consultants believe that a safe location for an ammonia refrigeration system is in the open air on a building roof. This is based on the premise that because ammonia vapour is less dense than air, it should discharge safely skywards. The latest tests by the BRE reveal this to be an incorrect assumption.
There is little published information on the causes, frequency and rates of discharge from ammonia refrigeration plant. The rate of discharge is important as it will strongly influence both the behaviour of an ammonia leak, and the ambient concentrations that it produces.
A small leak from a seal or gland may be in the order of grammes per second. However a catastrophic release, for example caused by a fractured or severed pipe, could result in the majority of the machine charge being released in a very short time.
Depending on the cooling capacity of the machine and its design, leakage of ammonia gas may be at a rate of a 0·1-1 kg/s.
Table 1 gives details of some recently installed ammonia chillers in UK air conditioned buildings. The specific refrigerant charge varies between 0·055 kg/kW and 0·46 kg/kW of nominal cooling capacity, so chiller selection is clearly an effective way of minimising the quantity of ammonia used and the consequential hazards of accidental release.
A crucial feature of ammonia discharges is the ambient density of the released refrigerant. Ammonia vapour has a density of between 0·6 and 0·7 that of air (depending on its discharge temperature), so a cloud of released ammonia will be positively buoyant.
Liquid ammonia discharges behave differently. When ammonia is released from a part of the plant which is at high pressure, the gas can produce "flashing" clouds of boiling liquid droplets and vapour. The cooling effect of the continuously evaporating droplets reduces the temperature of the cloud so that it becomes denser than air.
Very large discharges from industrial plant, which sometimes contain tonnage quantities of ammonia, can therefore be very dangerous.
Ammonia plumes
Previous research on the behaviour of dispersing gas plumes has shown that the buoyancy of a discharging gas is not sufficient alone to determine whether the resulting plume rises or falls. Buoyancy is strongly dependent on wind speed and the gas volumetric discharge rate, as well as the gas density.
This effect can be evaluated numerically by calculating buoyancy flux (a dimensionless quantity) for a given gas discharge rate, wind speed, and gas density.
High values of buoyancy flux, above around 0·1-0·3, mean that the plume will have a positive buoyancy.
A positively buoyant discharge (for example, ammonia vapour) will rise, and a negatively buoyant discharge (for example, a heavier-than-air cold ammonia droplet and vapour cloud) will fall.
Conversely, low values for buoyancy flux, below about 0·002, mean that the plume will have neutral buoyancy and will neither rise or fall, regardless of the discharge buoyancy.
The BRE has modelled positively buoyant discharges (for example, ammonia vapour) from the roof of a 10 m high building against wind speed, to determine the specific values of buoyancy flux. The research shows that smaller release rates at higher wind speeds show negligible plume buoyancy effects, while larger release rates at lower wind speeds will show very pronounced plume buoyancy effects.
It is likely that the majority of ammonia discharges from building cooling plant would be towards the lower end of the discharge rate scale. Under typical wind conditions the release plumes would have a neutral buoyancy, with a significant risk of ammonia entering the building via ventilation inlets or other openings.
Wind tunnel tests
The flow patterns of negatively and positively buoyant ammonia discharges around buildings have been investigated using a small-scale model (representing a rectangular building with a height of 20 m) inside the BRE's dispersion wind tunnel at its Cardington test site.
It can be seen that both discharge plumes spread over the entire downwind face of the building. Clearly, both locations would be unsuitable for neutrally buoyant ammonia discharges, as would be the case with smaller discharges at high wind speeds.
In both cases there was a pronounced plume rise. In the case of the discharge at ground level, the rising plume was spread over all sides of the building, making such a location highly dangerous in the event of ammonia discharges.
Safety guidelines
Designers should always comply with the requirements of BS 4434: 19951, the Institute of Refrigeration Safety codes2 and HSE PM 813, as well as the general requirements of the Health and Safety at Work Act. However, BRE's research has shown that additional consideration needs to be given to the location of the ammonia plant and possible discharges.
The hazards of ammonia discharges can be minimised by restricting the amount of refrigerant charge in any single chiller. Careful chiller design can lead to very low ammonia-specific charges, and values of around 0·05 kg per kW nominal cooling capacity are possible with current commercially available chillers by using plate heat exchangers.
The safest location for ammonia chillers is in a "special" plantroom (again, complying with the safety requirements of BS 4434:1995 and the Institute of Refrigeration Safety codes). The external siting of ammonia chillers, either at ground level or on a building roof, would result in any accidental release or leakage of ammonia dispersing in an unpredictable manner, depending on the rate of discharge and prevailing local wind conditions.
The advantages of locating ammonia chillers in special plantrooms is that liquid spills can be contained, and external discharge rates of ammonia can be controlled. Discharge could be through a fan-assisted stack with high velocity efflux. Specific guidance on stack design is given elsewhere (for example, the ASHRAE Handbook4).
An alternative to a special plantroom for small refrigeration systems is a container or box with its own ventilation system and ammonia vapour detectors. If designed and built correctly, this should offer comparable safety to a conventional special plantroom, but with significantly lower cost and weight.
Design implications
Despite its toxicity, ammonia is an excellent refrigerant. Its use in the construction industry has shown that accidents are rare if the plant is designed, installed and maintained strictly in accordance with existing codes and standards.
BRE's research has shown that it is also necessary to consider dispersion of ammonia to minimise the risk of the refrigerant reaching building occupants or people on the ground. Further research work is needed to determine actual rates of ammonia discharge so that the degree of hazard from leaks and discharges can be more accurately assessed.
Source
Building Sustainable Design
Reference
1BS 4434:1995 Safety and environmental aspects in the design, construction and installation of refrigerating appliances and systems, BSI, 1995. 2Safety code for compression systems utilising ammonia – Part 1: Design and construction; Part 2: Commissioning, inspection and maintenance, Institute of Refrigeration, 1990. 3Guidance Note PM 81: Safe management of ammonia refrigeration systems, HSE, 1995. 4ASHRAE Handbook 1997: Fundamentals – Chapter 15: Airflow around buildings, Atlanta, ASHRAE, 1997.
Postscript
This work was carried out by the BRE with funding from the DETR Construction Directorate as part of the DETR Framework project 'New and improved cooling'.