Road tunnel-Tutorial 2
Previous (Road tunnel-Tutorial 1) | Back to Start | Next (Parametric Runs-Tutorial)
TABLE OF CONTENT
Road tunnel tutorial 2- Fire in the tunnel
In fire simulations, the interest is focused on the possibility to prevent smoke from going backwards and reaching the vehicles locked-in by the fire. In these calculations, it is essential to calculate the temperature distribution along the tunnels, since it can generate major stack effects and excessive temperatures may destroy jet fans, required for the fire ventilation. The heat transfer between air and tunnel wall is modeled with convective and radiative heat transfer and 1D discretization along the tunnel and into the tunnel wall. 3D flow phenomena in the proximity of the fire are not explicitly modeled.
The time variation of the fire intensity is highly dependent on the type of fire. The model allows the user to describe arbitrary fire growth curves.
In the next study a 100 MW fire will be added to our sample river-passing tunnel.
The tutorial starts from the installed sample case:
ida\samples\Road\sample1_Standing.idm.
We will continue to work on this here, eventually ending up with the installed sample case
ida\samples\Road\sample1_Fire.idm.
Adding a fire to a tunnel section
- Open the section model and press ‘Edit segments’ to add a fire.
- Press Add and select Fire. Enter data according to Figure 47.
Figure 47. Edit segments dialog with fire selected
The basic fire evacuation strategy in a longitudinally ventilated tunnel is often to keep the tunnel free of smoke upstream of the fire by using jet fans, which blow the smoke in the direction of the traffic. Cars downstream of the fire are assumed to be able to drive out of the tunnel ahead of the smoke front. The fire position has in the example been selected to provide a worst case in terms of thermal stack effect. If a fire is placed even closer to the entry portal of the downhill part, it would provide a stronger stack effect, but in such a case passengers should even without fans be able to evacuate “under” the smoke layer.
The 1D fire model is capable of describing the overall effects of the fire to an accuracy which is acceptable for typical fire security studies in road tunnels. However, the effects of the fire in the near field are not described in detail. The Length parameter should reflect the size of the near field around the fire.
Heat from fire (or heat release rate) is the basic parameter, describing the size of the fire, including both radiated and convected heat. It may be given as a function of time by connecting it to an input table.
See Section Creating input tables and binding variables
If no table is specified, the fire will grow very quickly, by default in 10 s from the start of the simulation
A tunnel fire will result in a so called throttling effect, i.e. the fire plume will act as an obstacle to flow over the fire region. This effect is difficult to estimate in a 1D model and a parameter Fire pressure drop is required as input to account for this pressure drop. It has a reasonable default value but in some situations, where a conservative estimate is unavailable, more detailed studies must be performed to estimate this effect.
Another fundamental shortcoming of the 1D model occurs when the air velocity is exactly zero. This will always happen – momentarily - when the flow reverses direction and the situation is therefore unavoidable. At the point of flow reversal, the fire power must be limited, disregarding what has been specified by the user, in order to avoid a close to infinite air temperature. This limitation is done in several ways.
Firstly, the fire power is always limited to avoid physically impossible fire temperatures. The fire power is gradually reduced with respect to temperature above 1500°C to 90% reduction at 2000°C. Secondly, the fire power is by default limited to what is stoichiometrically possible with respect to available oxygen. This limiting mechanism can be turned off by unselecting Air flow limited fire power.
As a further help to facilitate the zero crossing, another power limitation factor is available: Zero air flow power reduction. By default, it reduces the fire power by 90% at low air-flows (0 = no reduction).
In rare cases, a stable solution will be found with air flow very close to zero and a reduced fire power. The 1D fire model is not valid for flow so close to zero and any such solution should be discarded. Therefore, a warning is issued if a simulation is terminated with reduced fire power. In the real world, such fire scenarios have been observed, where an oxygen-deprived fire oscillates around zero velocity for a long time. However, this phenomenon is extremely complex, with local 3D effects, including oxygen deprived combustion, interacting with global stack effects in the surrounding tunnel system and any 1D model will be insufficient to describe them.
Further fire related parameters, that govern smoke production etc. are specified in the Global tab of the system form. We will return to these in the next section.
A more formal description of the fire model is available in Fire model of the Tunnel theory.
- Close the Edit segments dialog by pressing OK.
Specifying global data
- Click on the Global tab of the system form. Inputs that affect the whole system are shown (Figure 48)
Figure 48. The Global tab of the system form
The Vmax parameter governs the general speed limit in all tunnels. It is overridden by any local speed limit prescribed in the Edit segments dialog of a tunnel segment.
TWall0 is the starting temperature of the wall at the beginning of the simulation.
NO2 /NOx is the fraction of NOx which is regarded to turn into NO2. In the Default wall field, a wall construction is selected.
See Section for the way new wall construction objects are defined.
Local selection of wall construction can be done in the Edit segments dialog.
The parameters under ‘Wall spatial discretization’ affect the way the wall is divided into suitable computational layers; tauWall together with thermal material properties set the depth of the radial domain so that relative temperature changes at that depth caused by any variation at the wall surface are negligible within the time interval tauWall. tauWall should thus normally be approximately equal to the simulation time range. The parameter nWallNod sets the number of nodes into the wall and the grid expansion factor GridFac is used to refine the grid close to the wall surface, where the biggest temperature gradients appear.
For a more thorough discussion of wall temperature computation see Section Wall in the Tunnel theory.
Further details of the fire model may be specified in the Fire model tab (Figure 48b) of the Global form.
Figure 48b. The Fire model tab of the global form
See also Sections Fire model and Extinction coefficient of the Tunnel theory.
H, K, kic and Ys relates the amount and optical density of produced smoke to the heat release rate. In situations where the amount of smoke produced is relevant to the design task, these parameters will normally be provided by fire experts.
The fire2air parameter determines the amount of fire heat that acts to directly heat the tunnel air, i.e. (1 - fire2air) will radiate from the flame to the surrounding tunnel wall. This seemingly critical empirical parameter is often assumed to be around 70%. Fortunately, it turns out that results in the far field are not as sensitive to this input as one could expect, since radiated heat will heat the wall which then, in turn, will heat the air by convection.
Accurate computation of the air temperature, and thereby density, downstream of the fire is naturally critical to predict the stack effect. Care must therefore be taken to resolve the exponentially decaying temperature. For this purpose, an automatic refinement of the grid around the fire is provided. (Another possibility to refine the grid ‘manually’ is provided by the dxMax parameter in the section form.)
The parameters dxFire0, dxFireFactor and dxFireExtent control the automatic grid refinement around a fire. dxFire0 gives the cell size next to the fire cell, where the gradient is steepest. dxFireFactor determines how much larger the next cell is and dxFireExtent specifies the total extent in all directions that should get a refined grid.
The grid refinement algorithm will refine the grid in all directions around a fire, i.e. it has no notion of the direction fire gases will take or the distance it will take for gases to cool. When a branch or plenum component is encountered in a possible air path, the refinement will continue in all connecting branches.
It is always good practice to successively refine the grid until there is no change in key computational results. However, many fire studies will not be critical with respect to processing power, in which case one can use a (safely) fine grid.
- Set tauWall to 1 hour, to reflect our choice of simulation time and duration of fire scenario.
- Open the Vehicles data form (Figure 34). Select Congestion model.
With the congestion model, traffic will fill the tunnel and form a queue leading up to the fire site.
In our sample case, for simplicity, the fire starts at the start of the simulation, and since the tunnel initially is empty from traffic, there will be no cars downstream of the fire. They all stop at the fire.
In a more realistic case, the fire start should be delayed until the tunnel has been filled with traffic. This would capture the ventilating effect of the traffic driving out of the tunnel ahead of the smoke. However, for a conservative design case, this possible extra ventilation should probably be disregarded.
- Open the existing Path report object and select the following additional variables by pressing the Add/Remove variables button: Radiative heat x-fer coeff, Convective heat x-fer coeff, Air temperature, Wall temperature, Extinction coefficient.
- Make a simulation pressing the ‘Run’ button,
. - Open the section form and examine the resulting air velocity in the tunnel. Compare with the presented critical velocity. Are we OK? Open the form for More details in the Traffic section of the form. Does the traffic configuration after the simulation look as would be expected?
Now, as can be seen in the last column of the Data for segments section, the air velocity upstream of the fire is around 2 m/s while the estimated critical velocity is about 3.36 m/s. We need to add more fan power.
Critical velocity is an estimate of the upstream air velocity that is required to avoid so called back-layering of smoke, i.e. that smoke propagates against the prevalent air flow direction.
The traffic density is about 110 pcu/km per line upstream of the fire (click More details button in the bottom right) and close to zero downstream, so the traffic seems OK.
- Open the Edit segments dialog in the section model form. Select (click on) the fan row at 300 m under Points. Press the Add button and select Fan(s). A duplicate fan group is created at 300 m. Edit the Position of the new fan group to 450 m. Create additional identical new fan groups at 1050 m in the same way.
- Make a simulation pressing the ‘Run’ button,
. - Open the section form and verify that the current fan power is sufficient to maintain a velocity above the critical. The air velocity upstream of the fire should be greater than critical velocity (We would actually be OK with just two additional fan groups.)
In the next couple of section we will look a bit closer at the obtained solution.
Viewing and printing path reports
- Double click on the path report with name PATH-REPORT at the top of the system form under heading ‘Results:’ to open the path report form Figure 3 .14.
- Click on the 3D Tab and select e.g. wall temperature by double clicking to view the wall temperature as a 3D coloured surface as a function of position along path and time (Figure 4 .19).
Figure 49. 3D coloured surface plot of wall surface temperature
- Examine also the other recorded variables.
The surface plot can be rotated (click, hold and move cursor within plot) and zoomed in/out (right click, hold and move cursor within plot).
Results can also be shown as graphs along cuts parallel to the x-, y- or z-axis of the 3D surface plot. Press the x-, y- or z-key, click with cursor within shaded cut surface, hold and drag until desired time or position is reached.
See also the separate on-line help instructions for the 3D window by pressing the Help button.
Printing path reports and other 3D diagrams can be done by selecting Print from the File menu. This will write a tiff file and open the default Windows’ viewer. From here one can print to paper. By default, the generated tiff file is magnified three times with respect to the screen resolution. If one holds the Shift key while selecting Print, one is allowed to enter any magnification factor.
Any IDA form (including 2D diagrams) can, in fact, be printed from the File menu.
Note that the wall temperature is significantly higher in the fire segment. Some experiments will reveal that this maximum temperature is quite sensitive to both the empirical fire2air parameter (Global tab of system form) and to the chosen size of the fire segment (Length parameter in Fire description in Edit Segments dialog).
The computed wall temperature in the near field around the fire is in fact not a reliable result in the 1D model and may even exceed physically possible temperatures (higher than any reasonable flame temperature). Fortunately, the computed air velocity in the tunnel is rather insensitive to changes in these parameters. (The established measure of critical velocity, on the other hand, is sensitive, which is a problem with this design approach. However, discussing this is beyond our scope here.)
In the next two sections we will learn how to locate and examine individual variables in the model in order to understand more. The reader who is less interested in the underlying equation model may want to conclude the tutorial here.
Open also the other sample models that are installed with IDA Tunnel to see examples of slightly more complex systems.
Viewing 2D fields
Let us look at the computed temperature field into the wall surrounding the tunnel at the end of the simulation.
- Open the section component by double clicking on it. To get access to the computed temperature field into the wall click on the ‘Outline’ tab. Display variables computed in segments by clicking on ‘seg’ and place the cursor on TGroundr. Right click and select ‘Open with Plot3D ’ to present a 3D surface plot of the temperature field into the tunnel wall.
Figure 50. 3D coloured surface plot of wall temperature
The temperature field is computed relative Twall0, which here is the initial wall temperature. The x-axis runs along the tunnel section from left to right. Note that the scale of the y-axis is in grid points (1 to 5), not in distance into wall, which makes the spatial variation incorrect since the grid size increases into the wall.
As mentioned above, results can be shown as graphs along cuts parallel to the x- or y- axis of the 3D surface plot. Press the x- or y-key, click with cursor within shaded cut surface, hold and drag until desired position is reached.
This concludes the tutorial part of the Getting Started Guide.