There is a growing recognition that safety in the event of fires in enclosed spaces such as tunnels and car parks is an issue that cannot be ignored. On 24 March 1999, a fire in the Mont Blanc tunnel, which links the Department of Haute-Savoie (France) with the province of Aosta (Italy), caused the deaths of 39 people. As a direct consequence of this tragedy, the administrative services of the Department of Haute-Savoie required the city of Annecy to carry out checks on the safety of the town hall car park.
The work was carried out by Scetauroute Tunneling and Underground Works Division with help from Arcofluid, under the control of the Annecy City Road Division.
The aim of the exercise was to understand the behaviour of heat and smoke within the car park so that modifications could be undertaken to ensure that any additional safety installations would be able to confine the fire while evacuation was carried out and while the emergency services were active in the area.
The car park is composed of a helical slab, which forms the six parking levels, surrounding a central atrium; the total depth is about 20m. Ventilation is both natural and mechanical. Natural ventilation is provided by the central atrium and by two shafts per level; there are also two extraction shafts on each level, equipped with fans. Exit from the car park is by staircases and elevators.
The PHOENICS CFD simulations were set up to take account of the whole of the geometry, including ventilation systems. The most dangerous fire location was assumed, at the lowest level of the car park, and an average heat release rate of 5MW was used.
The development of the smoke and heat distribution was established by the simulations, and the influence of various parameters was investigated. Smoke propagation was observed through the central atrium and also up the helical slab, through the parking levels. The atrium allows rapid transmission of smoke and heat to the upper levels; this is not serious, as the contact with fresh air causes significant cooling and dilution which keeps conditions safe for evacuation.
As a result of the flow to the atrium, the hot gases only propagate slowly through the parking levels; again, evacuation is not compromised by this. There is, though, a problem in the atrium, where the stairwells can become surrounded by hot fumes. It was realised that the most important safety requirement was to ensure the isolation of the stairwells from the atrium, while retaining access from the parking levels for evacuation; pressurisation of the stairwells is also important to their role as emergency exits.
The understanding gained from the PHOENICS simulations enabled safety proposals to be put forward. The changes required to the design were comparatively minor and can provide fire safety without great expense and without compromising the architecture of the building.
The Annecy Town Hall car park is composed of a helical slab, which forms the six parking levels, surrounding a central atrium. The radius of the atrium is 8.5m, as is the width of the slab; at the centre of the slab the slope is 2.5%, and the total depth is about 20m. Ventilation is both natural and mechanical. Natural ventilation is provided by the central atrium and by two shafts per level; there are also two extraction shafts on each level, equipped with fans. Exit from the car park is by staircases and elevators.
The PHOENICS simulations were set up to take account of the whole of the geometry, including ventilation systems. The most dangerous fire location was assumed, at the lowest level of the car park, and an average heat release rate of 5MW was used.
A cylindrical grid was used for the computations, with VR objects used to represent the structural components: parking levels, ventilation ducts, staircases, vehicles. The computational grid was selected after grid refinement studies, to obtain well-converged stable solutions and sufficient accuracy. The grid lines were more finely spaced near the walls and the ventilation ducts where the gradients are the steepest. The chosen grid was composed of 430000 cells.
The simulation was transient and had to take account of the key
- buoyancy phenomena generated by the fire;
- heat and smoke release at the fire.
A k-e turbulence model was used in the simulations. In view of the importance of buoyancy the k-e RNG variation (a standard PHOENICS built-in option) was adopted.
The calculations were carried out without a model of combustion.
Fire was thus represented as a volumetric source of heat,
characterised by a heat release rate as a function of time.
In all the simulations the average heat release rate of the fire
was approximately 5 MW. This power corresponds to the burning of
a large private car, or 2 or 3 vehicles of small size. The
evolution of the thermal heat release rate was as follows:
- a linear increase of 10 seconds;
- a constant maximum power of heat (5 MW) during 20 minutes.
The fire source takes into account a density of thermal emission of about 1 MW/m3.
The walls were considered as isotherms (fixed temperature of 15°C). Near the walls the standard wall functions were applied for heat and momentum fluxes. The specific materials of the walls and obstacles were taken into account: steel for the vehicles and concrete for all the other structures.
Fans were placed at the top of each ventilation duct. Each extraction fan was represented by a constant volumetric flow rate (8.8 m3/s) with a turbulence intensity of 5%.
In order not to force the flow in the interior volume of the car park by the choice of a pressure condition on the upper part of the atrium, the selected technique was to represent part of the external atmosphere. Here a wind speed of 2m/s was imposed on one side, at a temperature of 15°C and with a turbulence intensity of 5%; on the opposite surface a pressure condition was imposed (1 atmosphere).Click here to see animated smoke-movement pictures.