presentation_vistnes.pdf

8/10/2019 presentation_vistnes.pdf

1/30

Validation of Phoenics 3.5 for Modelling Tunnel

Ventilation Systems Under Fire Conditions

Presenter: Jamie Vistnes

Stephen Grubits & Associates


2/30

Introduction

Fires within tunnels have the potential for

catastrophic consequences

Need to ensure safe evacuation of

occupants Need tunnel emergency ventilation

systems to maintain tenable conditions

Ventilation systems may be designed to acode or on a performance basis


3/30

Purpose

Increasing demand for performance-based design of

ventilation systems

CFD codes used to assess performance of systems

Accuracy of codes critical in success of design

Validation necessary for comparison between numericalprediction and reality

Reliable and well documented reference data needed

Previous study on longitudinal ventilation system (forced

flow)

Assess natural ventilation system for further validation


4/30

Memorial Tunnel Fire Ventilation

Test Program

Consisted of a comprehensive series of full-scale fire tests

Natural, longitudinal, transverse ventilation

examined Fire sizes of 10, 20, 50 and 100 MW

Conducted in an abandoned road tunnel nearCharleston, West Virginia

Tunnel equipped with instrumentation and

recording equipment at various sections Comprehensive data available on CD


5/30

Description of Memorial Tunnel

3.2 percent upgradefrom south tunnelportal to the northtunnel portal

Tunnel length of854m

Cross-sectional areaof 60m2

Fan rooms located attunnel portals whichreduce the height to4m


6/30

Description of Memorial Tunnel

3.2 percent upgradefrom south tunnelportal to the northtunnel portal

Tunnel length of854m

Cross-sectional areaof 60m2

Fan rooms located attunnel portals whichreduce the height to4m

Fan Room


7/30

854 m

Z-axis scale 1:10


8/30

Scenario Modelled

Test 501: examined natural ventilation witha 20MW fire

Objective: To measure the buoyancy

driven airflows, air temperature andstratified smoke layers when no forcedflow is provided

Ambient temperature of 7oC

Initial draught was observed from North toSouth


9/30

Phoenics Model

Phoenics Version 3.5utilising special purposeversion Flair

Geometry built using VR Editor

Cartesian coordinate system and mesh

40 time steps over 10 minute simulation period(15 seconds per time step)

Gravitational acceleration:

x = 0, y = -9.8 m/s2, z = -0.31 m/s2

Parameters from previous study: k-turbulence model

Buoyancy effect on turbulence with a coefficient of 0.1


10/30

Description of Models

Based upon elapsed time0.01m7.22 oC fixed

temperature

5040 x 37 x 4615


temperature

2540 x 37 x 4614


temperature

4031 x 23 x 3483


temperature

4068 x 62 x 6062


temperature

4040 x 37 x 4611

Heat Release RateRoughnessSolids, walls

properties

Iterations

per time

step

Grid

(X x Y x Z)

Run


11/30

Fire Modelling

Blocks of domain material with a heat flux of1 MW/m3

Radiation not modelled 70% convective component

Fire located approximately 238m north of the southern portal

Heat Release Rate - Natural Ventilation Fire Test 501

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

10.0

11.0

12.0

13.0

0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540 560 580 600

Time (seconds)

ConvectiveHRR(MW)

Natural Ventilation FireTest 501

CFD


12/30

Results Runs 1 to 5

Generally temperatures and velocities were

over-predicted near the fire in all runs

At earlier times the hot layer did not spread as

far

At 10 minutes temperatures and velocities were

over-predicted throughout (except run 2)

Run 5 gave closest results to the fire test

50 iterations, intermediate grid

Require further investigation


13/30

Comparison of Run 5 and Fire Test

Temperature Contours (oF)

22 ft 22 ft

19 ft 19 ft

15 ft 15 ft

11 ft 11 ft

7 ft 7 ft

4 ft 4 ft

214 213 211 209 208 207 307 306 305 u 304 303 302 301 202

Fan

Room

Fan

Room

Time of 1 minute


14/30

Comparison of Run 5 and Fire Test

empera ure on ours

2 ft 22 ft

9 ft 19 ft

5 ft 15 ft

7 ft 7 ft

4 ft 4 ft

214 213 211 209 208 207 307 306 305 u 304 303 302 301 202

Fan

Room

Fan

Room

Time of 10 minutes


15/30


16/30

MTFVTP Elapsed Time

Peak Heat Release Rate (HRR) considered to

be achieved upon full pan engulfment

Potentially 1 minute prior to start of Phoenics runwhere there is a rapidly growing fire within the

tunnel

Significant delay in the fire reaching its peak

HRR


17/30

Calculation of Heat Release Rate

Fire generated by pans of fuel

Fuel was constantly pumped into pans tomaintain constant level

Weigh cells beneath pan gave feedback to

controller, which determined the rate that fuelwas pumped into the pan

HRR determined by fuel consumption

Fire plume produces turbulent flows and mayproduce forces on pan

Fluctuations observed in weight measurements Feedback control system results in calculated

HRR lagging actual HRR


18/30

Adjustments to Phoenics Model

Time of run extended by 1 minute to allow

for burning before elapsed time

Fire grown to maximum HRR during this 1

minute of extra time before the elapsedtime

Maximum HRR was averaged out to

compensate for uncertainties in measured

HRR


19/30

Adjusted HRR

Adjusted Heat Release Rate

0

1

2

3

4

5

6

7

8

9

10

0 20 40 60 80 100 120 140160 180 200 220240 260280 300 320 340360 380 400 420440 460 480 500 520 540 560 580 600 620 640 660

Time (seconds)

ConvectiveHRR

(MW)


20/30

Description of Models

Heat Release RateRoughnessSolids, walls

properties

Iterations

per time

step

Grid

(X x Y x Z)

Run

Adjusted to take intoaccount burning before

elapsed time

0.001mAdiabatic5040 x 37 x 4618

Adjusted to take into

account burning before

elapsed time

0.001m7.22 oC fixed

temperature

5040 x 37 x 4617

Adjusted to take into

account burning before

elapsed time

0.01m7.22 oC fixed

temperature

5040 x 37 x 4616


21/30


22/30

Comparison of Run 5, Run 6 and

Fire Test

Time of 1 minute

Temperature Contours ( F)

22 ft 22 ft

19 ft 19 ft

11 ft 11 ft

7 ft 7 ft

4 ft 4 ft

0 ft 0 ft

214 213 211 209 208 207 307 306 305 u 304 303 302 301 202

Fan

Room

Fan

Room


23/30

Results Runs 6 to 8

Adjustments to HRR improved results in theearlier times

At 10 minutes, run 6 gave results that were ingood agreement with the fire test data

A smaller wall roughness resulted in the hotlayer spreading further, however flow patternsand characteristics of the hot layer were differentat 10 minutes

Adiabatic walls and solids increased spread ofthe hot layer, however this came with gross

over-predictions


24/30

Comparison of Run 6, Run 7,

Run 8 and Fire Test

Veloci y Profiles (fpm)

0

4

7

11

15

19

22

26

TunnelHeight(ft)

209 208214 207 307 305 304 302 301 202

-1200

-6000

600

1200

Velocity Profile Scale (feet/minute)

0

4

7

11

15

19

22

26

FanRoom

FanRoom

Time of 10 minutes


25/30

Comparison of Run 6, Run 7,

Run 8 and Fire Test

Temperature Contours (oF)

2 ft 22 ft

9 ft 19 ft

5 ft 15 ft

1 ft 11 ft

7 ft 7 ft

4 4

0 ft 0 ft214 213 211 209 208 207 307 306 305 u 304 303 302 301 202

Fan

Room

Fan

Room

Time of 10 minutes


26/30

Conclusions

Phoenics run 6 gave results which best

represented results from the fire test

Grid size of 40 x 37 x 461 cells;

50 iterations per time step; Solids and walls set to 7.22oC fixed

temperature

Global wall roughness of 0.01m

HRR adjusted to take into account burningbefore elapsed time


27/30

Conclusions

Conditions at 10 minutes in good agreement

with the test data

Under-prediction at earlier time steps, however

not necessarily due to Phoenics CFD code as

uncertainty in HRR

Based upon observations made in the

assessments:

Defining parameters and assumptions for natural

ventilation is more critical than for forced flowproblems


28/30

Conclusions

Fire growth in initial stages has significantimpact on conditions within the tunnel

Upon reaching 10 minutes, steady stateconditions were approached

Adiabatic conditions resulted in gross-over-predictions

When using data obtained from a fire test it isimportant to understand:

The procedures used, including the fire source

Nomenclature such as elapsed time


29/30

Conclusions

Phoenics 3.5 may be used to predict the

conditions within a tunnel under fire

conditions utilising natural ventilation

Particularly once buoyant driven airflowsbecome well established

Recommend further research on validation

of naturally ventilated scenarios using data

from other fire tests


30/30

Thank You

presentation_vistnes.pdf

Documents