Multi-scale Fire Modelling Method in Longitudinally Ventilated Tunnels for FDS6.1

Edmund Ang, Guillermo Rein, Joaquim Peiro, Roger Harrison, Izabella Vermesi

Contents

  1. Introduction
  2. Multi-scale Modelling
  3. Implementation of Multi-scale Modelling
  4. Cold Flow Multi-scale Modelling Results
  5. Multi-scale Fire Modelling Results
  6. Conclusion
  7. Reference

Introduction - Team

  • Work carried out in 2014
  • Led by Edmund Ang (Imperial College London), with collaborators:
    • Guillermo Rein (Imperial College London)
    • Joaquin Peiro (Imperial College London)
    • Roger Harrison (AECOM UK)
    • Izabella Vermesi (Imperial College London)
  • Special thanks to software support from Thunderhead Engineering

Introduction - Multi-scale Modelling

  • Multi-scale modelling based on Colella et al [1]
  • Previous work based on commercial CFD package and Reynolds Averaged Navier Stokes
  • Multi-scale modelling using FDS6:
    • Not attempted before
    • Enabled by the FDS HVAC feature
    • Proof of concept by Vermesi et al [2]
    • Constant flow specified at boundary
    • No coupling of jet fans to the fire behaviour

Multi-scale Modelling - What and Why

  • Increasingly denser and taller buildings
    • Scarcity of land resources
    • High rise buildings
    • Options for above ground rail or utility network limited
  • Tunnel is often the more practical solution
  • Tunnels can be used for rail, roads or utilities

Multi-scale Modelling - What and Why

  • Ventilation is needed to most tunnels
    • Life safety (fumes or CO from fire)
    • Regulating temperature (train operation)
  • Three main types of ventilation system
    • Longitudinal (jet fans)
    • Transverse (ducted supply and extract)
    • Semi-transverse (hybrid of ducted and non-ducted)

Multi-scale Modelling - What and Why

Epping2Chastwood_H

Figure 1. ECRL Rail Tunnel, Australia

Multi-scale Modelling - What and Why

  • Design of tunnel ventilation system
  • 1D / Subway Environment Simulation (SES)
    • Early stages of design
    • Fast computation
    • Provides global averaged prediction
    • Limitations on gas species or high resolution calculations

Multi-scale Modelling - What and Why

  • 3D / Full CFD simulation
    • To validate the design carried out at the earlier stages
    • Time consuming
    • Provides high resolution predictions, e.g. gas species and combustion
    • Tunnel section models often shorten to reduce computational time
  • Is there a best of both world?

Multi-scale Modelling - What and Why

  • Multi-scale modelling method
  • Using 1D for far field tunnel sections
  • Full CFD for near field, or tunnel section of interest
  • Significantly reduce the computational time
  • Off-set computational time for longer tunnel section

Multi-scale Modelling - What and Why

  • Multi-scale modelling method
  • Direct and indirect coupling methods
  • FDS6.1 is based on indirect coupling method
  • Implemented using HVAC feature
    • Acknowledge this is not the intended use of HVAC

Multi-scale Modelling - Implementation

  • Model is based on Dartford Tunnel West
  • Cold flow field measurement data available [1]
  • Dartford Tunnel West properties:
    • 1.5 km long
    • 14 jet fans pairs (JFP) with 8.3 m3/s per fan
  • Key of multi-scale model
    • Interface between 1D and full CFD sections
    • Flow need to be fully developed

Multi-scale Modelling - Implementation Diagram

Fig1

Figure 2. Representation of multi-scale model

Multi-scale Modelling - Implementation Details

  • Lp is length of portal = 50 m
  • LJF,UP and LJF,DW are length up and downstream of jet fans = 35 m and 130 m respectively
  • Lfire is length each side of fire = 170 m
  • Calibrated from running multiple models

Cold Flow Modelling Results

  • Cold flow modelling results presented separately in Tunnelling and Underground Space Technology [3]
  • Good correlation from 80 m downwind of the jet fans
  • Poorer correlation nearer to jet fans:
    • Lack of detailed information of installed jet fans
    • Difficulty to accurately model the jet fans
    • No accurate measurement of tunnel walls' surface roughness

Cold Flow Modelling Results

fig6_velocity

Figure 3. Average velocities measured in the tunnel [3]

Cold Flow Modelling Results

fig_100

fig_120

Figure 4. Multi-scale model velocity profile downwind of jet fans [3]

Cold Flow Modelling Results

fig_20

fig_40

Figure 5. Multi-scale model velocity profile downwind of jet fans [3]

Cold Flow Modelling Results

fig1_time1

Figure 6. Reduction in computational time for the multi-scale model

Multi-scale Fire Modelling Results

  • Adapted from the same cold flow multi-scale model
  • Introduction of a fire in the middle of the tunnel
  • Three fire sizes considered, 35 MW, 55 MW and 75 MW
  • Validation study conducted (Arup Tunnel case)
  • Mass flow rate is the measured variable
  • Interesting behaviour observed in the multi-scale model

Multi-scale Fire Modelling Results

  • Oscillatory mass flow observed
  • Mass flow rates in multi-scale models do not stabilise compared to the full CFD model
  • It is yet to be determined if the oscillation is numerical
  • Similar oscillation for velocity and temperature observed by Vermesi et al [2]

Multi-scale Fire Modelling Results

fig4_1

fig4_2

Figure 7. Mass flow rates (35 MW) in the tunnel. Left: Mass flow in and out. Right: Mass flow along the tunnel.

Multi-scale Fire Modelling Results

Fig4_3

Fig4_4

Figure 8. Mass flow rates (55 MW) in the tunnel. Left: Mass flow in and out. Right: Mass flow along the tunnel.

Multi-scale Fire Modelling Results

Fig4_5

Fig4_6

Figure 9. Mass flow rates (75 MW) in the tunnel. Left: Mass flow in and out. Right: Mass flow along the tunnel.

Multi-scale Fire Modelling Results

Untitled

Figure 10. Oscillating temperature and velocity in a tunnel fire [3]

Conclusion

  • Multi-scale modelling using FDS6.1 + HVAC is feasible
  • Can be used for cold flow multi-scale modelling
  • Fire modelling using the multi-scale model:
    • Oscillating mass flow
    • Mass flow rates do not stabilise compared to the full CFD model
    • Should not be used until the above two questions can be answered

Reference

[1] F. Collela, G. Rein, and J. L. Torero. “A Novel Multiscale Methodology for Simulating Tunnel Ventilation Flows During Fires.” Fire Technology 47 (January): 221–253. Jan. 2011.

[2] I. Vermesi, G. Rein, F. Colella, M. Valkvist, and G. Jomaas. “Reducing the Computational Requirements for Simulating Tunnel Fires by Combining Multiscale Modelling and Multiple Processor Calculation.” Tunnelling and Underground Space Technology. 2016.

[3] C.D.E. Ang, G. Rein, J. Peiro, and R. Harrison. “Simulating Longitudinal Ventilation Flows in Long Tunnels: Comparison of Full CFD and Multi-Scale Modelling Approaches in FDS6.” Tunnelling and Underground Space Technology 52 (February): 119–126. Feb. 2016.