Modeling Jet Fans, Part 3: Car Park Simulation

In this post, we demonstrate using FDS to model jet fans that extract smoke from a car park. The jet fans are modeled using two approaches: (1) an HVAC duct with a downstream shroud and, (2) a velocity patch with a short shroud. Two ventilation designs are presented: a first design that did not perform satisfactorily and an improved design developed by James Allen at Fläkt Woods Group (james.allen@flaktwoods.com).

You can find Parts 1 and 2 of model jet fans here (Part 1: Background and Convergence Study and Part 2: Validation with Experimental Data). 

Application to Car Park

The purpose of these posts is to demonstrate the feasibility of using FDS for jet fan simulations such as a car park. Based on what we have learned in Part 2, we will use a relatively coarse mesh with the jet fans modeled using HVAC ducts with downstream shrouds or velocity patches with short shrouds.

First Design

Figure 1 shows the basic car park dimensions and the first smoke control design that used five fans.  This is not a real design, it was developed purely with the intent of illustrating an approach. The total volume of the car park is 5400 m3. In the first design, there are two 4 x 3 m exhaust openings and one 4 x 3 m supply opening. The five fans (0.25 x 0.25 x  2.5 m) are arranged to direct the flow towards the exhaust openings. For 6 air changes per hour, the flow needs to be 9 m3/s or about 11 kg/s. We will present results for a case with a fan velocity of 18 m/s – the speed corresponding to the Novenco jet fan tested by Giesen et al. (2011). This case gives about a 30 kg/s replacement air flow (about 16 air changes per hour).

Figure 1: Car park model. Five fans in a 60x30x3 m garage.
Figure 1: Car park model. Five fans in a 60x30x3 m garage. Click for a larger, clearer image.

The mesh size was 125 mm surrounding the fans and 250 mm away from the fans. Simple models of beams, columns, and cars were included.

Figure 2: Detail of mesh around HVAC jet fan duct.
Figure 2: Detail of mesh around HVAC jet fan duct. Click for a larger, clearer image.

For this example, we used a very sooty fire that represents burning of Polyurethane GM27. The soot yield is 0.198 which results in a large amount of smoke. The fire has a surface area of 6 m2 with a peak Heat Release Rate per Unit Area of 500 kW/m2, giving a peak Heat Release Rate (HRR) of 3000 kW. The 3000 kW value was chosen based on the paper by Jones et al. (2007). A simple ramp up to peak HRR and ramp down was used. This time history was selected to speed the calculation.

Figure 3: Specified HRR of fire.
Figure 3: Specified HRR of fire.

Results for First Design

We will first look at the initial flow patterns before the fire starts. Figure 4 shows the corresponding flow vectors for the fast fan (18 m/s) case. In both cases, the primary flow is towards the exhaust exits, but there is some reverse flow in the lanes between the fans. The contours are shown just below the fans which are centered at Z=2.75 m (the ceiling is at 3 m and the lower edge of the beams is at Z=2.5 m). Finding a fan configuration with no reverse flow is not simple and requires design calculations.

Figure 4: Flow vectors at Z=2.5 m at 200 seconds.
Figure 4: Flow vectors at Z=2.5 m at 200 seconds. Click for a larger, clearer image.

The fire starts at 200 seconds, reaches a peak at 300 seconds, stays at the peak to 500 seconds, then ends at 600 seconds. At 250 seconds, when the Polyurethane GM27 fire has only reached half its peak value, the smoke already extends throughout much of the car park, as shown below. Figure 5 does show that the flow due to the jet fans tends to push the smoke to the exit. However, by 500 seconds, smoke completely fills the car park for both cases.

Figure 5: Smoke at 250 seconds.
Figure 5: Smoke at 250 seconds. Click for a larger, clearer image.

At 500 s, the time of peak fire, smoke completely fills the car park.

Figure 6: Smoke at 500 seconds. Click for a larger, clearer image.
Figure 6: Smoke at 500 seconds. Click for a larger, clearer image.

After the fire stops, the jet fan flow begins to clear the air. At 1600 seconds, the fast fan case shows nearly complete clearing of the smoke.

Figure 7: Smoke at 1600 seconds.
Figure 7: Smoke at 1600 seconds. Click for a larger, clearer image.

Heating of the air due to the fire has a large impact on the evacuation of smoke. Obviously, the heated air rises and then spreads when it contacts the ceiling. But in addition, the heating of the air causes it to expand with two consequences: (1) less fresh air flows into the car park, and (2) although air velocity at the jet fan exits remains the same, the reduced density of the air means that the fans have less thrust to move the air.

This effect can be seen in the plots of the entry air mass flux, Figure 8. Here we show both a slow and fast fan case. In both cases, when the fire starts, the fresh air flow into the model is reduced significantly. As the hot air is exhausted the fresh air flow again increases.

Figure 8: Mass flow rate of supply air into car park. Image on left is for jet fan velocity of 4.5 m/s and image on right for jet fan velocity of 18 m/s.
Figure 8: Mass flow rate of supply air into car park. Image on left is for jet fan velocity of 4.5 m/s and image on right for jet fan velocity of 18 m/s. Click for a larger, clearer image.

Summary of First Design Results

The first design was  not developed with any design calculations, it was just a guess that used the same jet fan parameters as used for the convergence study. The results show that the first design was not adequate:

  • The design has locations with reverse air flow.
  • After the fire starts, the high velocity of the hot air and smoke was able to overcome the air velocity due to the jet fans.
  • For this first design, the smoke spreads throughout the car park.
  • The change in air density due to heating reduced the efficiency of the fans (volume flow the same, mass flow and thrust reduced) and significantly reduced the amount of fresh air flowing into the car park.

Because of the poor performance of the first design, a re-design was required.

Second Design

An improved ventilation design was developed by James Allen at Fläkt Woods Group (james.allen@flaktwoods.com) and is shown in Figure 9.

Figure 9: Improved design of car park ventilation (James Wood)
Figure 9: Improved design of car park ventilation (James Wood) Click for a larger, clearer image.

Changes to the original design include:

  • Use of more powerful fans with a minimum thrust of 65 N. The recommended fan had a thrust of 73 N with a diameter of 400 MM and a flow rate of 2.77 m3/s.
  • Use 9 fans.
  • Rearrange the fans to enhance flow from the entry and a line of fans that moves air uniformly toward the exits.
  • Increase the area of the open supply from 4 x 3 m to 6 x 3 m, to reduce inlet flow to less than 2 m/s.

Although not a design change, because of the higher flow, the HRR time history was shortened, so that the ramp to full HRR occured from 60 to 120 s. The fire then remained constant for 300 s to ensure that a steady state condition was met with respect to smoke. It then ramped down from 420 to 480 s.

Figure 10: Specified HRR of fire for new design calculation.
Figure 10: Specified HRR of fire for new design calculation.

Validation Run for New Fan

Because the new design uses different fans, it was necessary to verify that the calculation gives acceptable results. The new fans have a diameter of 0.4 m. Using the 0.125 m mesh size, the representation of the fan is a 0.375 x 0.375 m square. This gives a different area than the circular fan. The thrust of a jet fan is given by:

    \[T_{g}=Q_{d} V_{d} \rho\]

where:  T_{g} is the gross thrust, Q_{d} is the volume flow of air at discharge, V_{d} is the discharge velocity, and \rho is the air density.

To maintain the same theoretical thrust with a different area, the equivalent discharge velocity is 20.7 m/s and the flow rate is 2.9 m3/s. The results for a 0.375 x 0.375 jet fan with these equivalent conditions are shown below. In both cases the total jet fan length was 2.5 m. The velocity patch model used a 0.5 m short shroud and the HVAC model included a 0.375 m length downstream shroud. For both models, the centerline velocity decay and flow entrainment are less than the experimental correlations. Ideally, the mesh would be refined, but to keep reasonable run times, the assumed mesh dimensions will be used. Although the jet fan model with a velocity patch performs slightly better, it was decided to focus on the HVAC model because of its reliability. Using the HVAC model will provide conservative results with respect to the car park simulation, since it will be a lower flow entrainment corresponds to a longer time to clear the smoke.

Figure 11: Centerline velocity and flow entrainment for the 73 N thrust fan. Click for a larger, clearer image.
Figure 11: Centerline velocity and flow entrainment for the 73 N thrust fan. Click for a larger, clearer image.

Results for Second Design

We first look at velocity vectors just before the fire starts. Compared to the first design, the flow velocities are more than twice as large. There are some locations of circulation, but most of the flow is towards the exits.

Figure 12: Velocity vectors at 60 s, Z=2.5 m. Click for a larger, clearer image.
Figure 12: Velocity vectors at 60 s, Z=2.5 m. Click for a larger, clearer image.

The visibility contours under full fire conditions are shown in Figure 13. This image shows that about 6.5 m from the fire, the visibility has increased to about 9 m.

Figure 13: Visibility contours at 420 s, Z=2.0 m. These contours show visibility under full fire conditions.
Figure 13: Visibility contours at 420 s, Z=2.0 m. These contours show visibility under full fire conditions. Red corresponds to greatest visibility. Click for a larger, clearer image.

The contours of temperature at 420 s and Z=2.5 m are shown in Figure 14. The distance from the fire to a location where the temperature is less than 60 C is about 2 m. Because the fans operate in relatively cool air, the efficiency of the fans does not significantly change and the replacement entrance air flow was approximately constant at 52 m3/s.

Figure 14: Temperature contours at 420 s, Z=2.0 m. These contours show temperatures under full fire conditions.
Figure 14: Temperature contours at 420 s, Z=2.0 m. These contours show temperatures under full fire conditions.Click for a larger, clearer image.

The smoke visualization at 420 s (peak fire) and 600 s (120 after end of fire) are shown in Figures 15 and 16. By 800 seconds, the car park is cleared of smoke.

Figure 15: Smoke visualization at 420 s. This shows smoke under full fire conditions.
Figure 15: Smoke visualization at 420 s. This shows smoke under full fire conditions. Click for a larger, clearer image.
Figure 16: Smoke visualization at 600 s. This shows smoke 120 s after the end of the fire.
Figure 16: Smoke visualization at 600 s. This shows smoke 120 s after the end of the fire. Smoke is beginning to clear. Click for a larger, clearer image.

Summary of Re-Design Results

The re-design provided by James Allen at Fläkt Woods Group was satisfactory, meeting both visibility and temperature criteria.

Summary

This post has demonstrated an approach to simulating fire and smoke in a car park using FDS and PyroSim. The first (naive) design was not satisfactory, but a re-design proposed based on calculations met design criteria. This highlights the requirement that design calculations be understood and implemented when developing the smoke control system for a car park fire.

It is anticipated that the results are conservative for two reasons:

  1. The selected jet fan model (HVAC duct with downstream shroud) under-estimates flow entrainment downstream of the fan.
  2. The fire used for the simulation was very sooty, so this would under-estimate visibility.

Acknowledgements

Thanks to James Allen at Fläkt Woods Group (james.allen@flaktwoods.com) for taking interest in the original calculations and for developing an improved design of the smoke control system.

All calculations were performed using the FDS and Smokeview software. PyroSim was used to create and run the FDS models.

Input Files

Download the PyroSim and FDS input files here.

References

Awbi, Hazim B., (2003). Ventilation of Buildings, Second edition, Spon Press, 2003.

Giesen, B.J.M. v.d., Penders, S.H.A. , Loomans, M.G.L.C., Rutten, P.G.S., & Hensen, J.L.M. (2011).  “Modelling and simulation of a jet fan for controlled air flow in large enclosures,” Environmental Modelling and Software, 26(2), 191-200.

Jones, J.C., Noonan, T., and Riordan, M.C. (2007). “An Examination of Vehicle Fires According to Scaling Rules,” International Journal on Engineering Performance-Based Fire Codes, Volume 9, Number 3, p.111-117.

Comments or Questions

This post was written by Daniel Swenson. For comments or questions, send email to support@thunderheadeng.com.

Update History

This post replaces a previous post that used the Dynamic Smagorinsky turbulent viscosity model, which has poor performance when simulating fire (based on discussions with NIST).

Updates to this post include:

  • 2016/06/20 – Added redesign and verification calculations for the jet fans used in the re-design.

More Tutorials Like This

See all PyroSim tutorials