19378 01 cfd modelling vesda beijing metro a4

Xtralis

CFD Modelling of VESDA Performance to Tested Fires in Beijing Metro Station

Applications Engineering Group

November 2010

VESDA by Xtralis CFD Modelling of VESDA Performance

Doc. 19378_01 ii

This page has been left blank intentionally.


iii Doc. 19378_01

Contents 1. Background and Objectives ............................................................................................................1 2. Testing Metro Station .......................................................................................................................1 3. Testing Fires and Set-up ..................................................................................................................2 4. VESDA Systems ...............................................................................................................................3

4.1 Existed (tested) VESDA system ...............................................................................................3 4.2 Extended VESDA system .........................................................................................................4 4.3 Maximum coverage VESDA system .........................................................................................4

5. CFD Modelling ..................................................................................................................................4 6. Simulation and Calibration ..............................................................................................................5

6.1 Simulation results .....................................................................................................................5 6.2 Calibration by the tested results ....................................................................................................6

7. Prediction from Smoke Curves and Comparison ...........................................................................6 7.1 Prediction from smoke curves recorded in the tests ..................................................................6 7.2 Comparison of predictions from the different algorithms ............................................................7

8. Conclusion .......................................................................................................................................9 Appendix A. VESDA Pipe Networks & Parameters in the Tests ............................................................10 Appendix B. Recorded Smoke Curves from Tests .................................................................................11

(VLP001 – upper, VLP002 – lower) ..................................................................................................11


Doc. 19378_01 iv

This page has been left blank intentionally.


Doc. 19378_01 1

1. Background and Objectives

Series of fire tests were conducted in Bagou Station of Line #10, Beijing Metro to assess applicability of various fire detection technologies. Xtralis VESDA fire detectors were tested in a selected area of the platform. Early warning detection was achieved from the tests by VESDA’s detectors.

Since the testing area is only a portion of the platform due to limitation of access during this almost one year testing project, VESDA’s detection performances in actual protection are to be investigated and known.

Xtralis has been requested to perform this further investigation by utilising the test results and CFD modelling to predict VESDA system performance in:

1. Standard protection scheme in Beijing Metro station, which using one VESDA detector to protect half of the platform area;

2. Maximum possible VESDA protection area, limited by maximum sampling pipe length or size of fire zone allowed by local fire code. Such “max” protection capacity may be used in other size of stations or areas.

2. Testing Metro Station

The testing area, platform of Bagou Station, Line #10, has dimensions of 23.5x10x4.3m.

The platform area was separated by glass walls from the rails. Height of the glass walls is approx. 2.7m with a gap of 0.3m on top.

A perforated false ceiling presented in the platform and its height above ground floor is 3m and distance to roof is 1.2m. Two hollowing ratios exist as 70% in the centre strip area, and 50% in the outer strip areas which close to the glass walls, Figure 1.

Figure 1. Hollowing false ceiling


Doc. 19378_01 2

Under the false ceiling, there are no objects in the platform area except columns and stairs.

Inside the ceiling void, there are ventilation ducts and beams, as shown below.

Figure 2. Objects inside ceiling void

3. Testing Fires and Set-up

According to the most possible fire materials can be found in public areas of a metro station, the testing fires were chosen as cotton wicks smouldering fire, smouldering paper fire, and PU foam flaming fire.

In this modelling, two of the testing fires were simulated, cotton wick fire (TF3) and PU foam fire (TF4). Their detailed configurations can be found from a test plan document: “Proposal of testing research of selection and installation of fire detectors in Beijing Metro stations”.

Lack of details of the actual fire developments from the tests, the fires were modelled by theoretical calculation and reference from other researchers.

The cotton wick (TF3) fire was suggested by other researchers having a peak HRR of 3.2kW1. The growth rates were simulated as linear and T-square, reached its peak at 120s. Soot yield ratio was estimated as 0.035 (3.5 times higher than cellulosic materials).

Two methods were adopted to model the PU foam (TF4) fire, heat release rate (HRR) method and mass loss rate (MLR) method. These two methods use experimental data of HRR2 and MLR3 to simulate the fire source respectively.

1 W. Grosshandler, 1995, “A review of measurements and candidate signatures for early fire detection”, NISTIR 5555 2 A. Mirme et al., 1999, ‘Performance of an Optical and an Ionization Smoke Detector Compared to a Wide Range Aerosol Spectrometer’, AUBE’99


Doc. 19378_01 3

The MLR data was then converted into HRR by PU foam’s combustion heat, 20MJ/kg. The soot yield used was 0.05, as used in a modelling performed by Farouk4.

The tested curves of the HRR and MLR were shown below.

Figure 3. Profiles of HRR (left) and MLR (right) – TF4

4. VESDA Systems

There are 3 system designs assessed in the simulations.

4.1 Existed (tested) VESDA system

Two VESDA detectors were installed in the testing area, sampling under top solid ceiling (upper level) and just above the false ceiling (lower level). Both levels have identical pipe layouts as shown below.

Figure 4. VESDA pipe layout

3 M.A. Jackson and I. Robins, 1984, “Gas Sensing for Fire Detection: Measurements of CO, CO2, H2, O2, and Smoke Density in European Standard Fire Tests’, Fire Safety Journal, Vol 22, P181-205

4 B. Farouk, et al., 2001, “Simulation of Smoke Transport and Coagulation for a Standard Test Fire”, AUBE’01


Doc. 19378_01 4

System parameters such as transport time and flow rate through each hole were modeled by a pipe network software, ASPIRE2. A brief summary of the parameters were shown in Appendix A, and were adopted into the CFD modelling. Details of the tested VESDA system can be found from an Xtralis design document, “Testing plan for VESDA system in public area of platform at Bagou Station, Bejing Metro.” Fire alarm threshold configured in the tests, 0.16%/m, was set in the simulations.

4.2 Extended VESDA system

The existed VESDA coverage is to be extended to half of the platform with identical sampling hole spacing as that in Existed option.

The hole number of each VLP is 28 and maximum pipe length increased to 80m.

4.3 Maximum coverage VESDA system

This option has maximum allowed pipe length (100m) and greater sampling hole space of 6m. Then the total hole number for each VLP is reduced to 26.

The first two options were modelled fully in the simulations, i.e. all the sampling points were modelled within the domain. While the domain size is smaller than the maximum coverage of the last VESDA option, certain number of sampling holes was sit out of the domain. Therefore, by-pass dilution was introduced.

5. CFD Modelling

Half of the platform was simulated with dimensions of 54x10x4.2m, shown below.

(Top view)

(Elevation view)

Figure 5. Simulation domain

Beam structure and ventilation ducts inside ceiling void were also simulated.

VESDA sampling points were placed in the domain following locations from the various pipe layout options, shown as green dots below.

Still environment with ambient temperature was simulated.


Doc. 19378_01 5

Figure 6. Modelled sampling points

6. Simulation and Calibration

6.1 Simulation results

It’s found that FDS couldn’t properly simulate detector response from the lower level pipe network, especially in the cotton wick fire. The reason is that in the tests performances of the lower layer detector were greatly determined by random smoke drift on the site due to uncontrollable environmental conditions. In current CFD modelling, none of existing CFD models, including FDS, can simulate such phenomenon.

Predicted upper layer VESDA Fire 1 alarm times are listed in following table. For comparison, tests results from the first testing day (19th Aug 2009) were also presented.

Table 1. Predicted & tested VESDA response times

If an average 15s of the non-significant change period was added into PU foam M-MLR method simulation, the predicted alarm time from the Existed VLP will be 58s. Therefore, both M-HRR and M-MLR methods generated quite accurate predictions, within 14% comparing to the tested data.

Errors from the cotton wick simulations are relatively higher, ranged from 30 to 44%. Based on observation from the current test and MRR curve in Figure 3, T-square growth rate is much suitable for the cotton wick fire. Therefore, this 30% accuracy is reasonable and acceptable due to great uncertainty in the fire development, environmental conditions and accuracy of the FDS model.

Existed Extended Max coverageM-HRR 50 51 60

M-MLR 43* 44 54Linear 79 104 121T-square 98 116 123

Note: * The period of no significant mass loss can be recorded, between 10 and 20s, was not included into the simulation.

Cotton wicks

Tested

58

142

Growth rate

PU foam (20kW)

TestVESDA Fire1 alarm time (s)


Doc. 19378_01 6

6.2 Calibration by the tested results

The simulated response time were then calibrated by the results from the tests. It’s performed by applying correction factors (percentages), obtained from a ratio of Tested to that from the Existed option, to the other predicted response times. Such calibrated response times can be used for comparison of performances of the various options, shown below.

Table 2. Calibrated VESDA response times

The calibrated response times from M-HRR and M-MLR methods for the PU foam fire are almost identical, and very close to the performance of the exited VLP. For the cotton wick fire, no significant increase on response time from the VESDA system that covers half of the platform.

7. Prediction from Smoke Curves and Comparison

7.1 Prediction from smoke curves recorded in the tests

VESDA’s detection performance under the half-platform protection option can be estimated from smoke profiles recorded from the fire tests. It’s known that a ratio of the sampling hole numbers from the Existed and Extended pipe layouts is 2 (28 to 14). It’s also known that smoke required to reach the detector alarm threshold (concentration) was sucked into pipe network through limited holes near the fire source. So, the extra hole in the Extended system will suck in fresh air and dilute the smoke concentration in the detector chamber, in the ratio of 2:1. Then performance of the Extended option can be predicted by doubling the alarm threshold, from the tested 0.16%/m to 0.32%/m, in the tested smoke curves.

Details of smoke profile around the threshold level can be examined by enlarging smoke curves from Appendix A, shown below.

Existed Extended Max coverageM-HRR 58 59 70M-MLR 58 60 73

Cotton wicks T-square 142 168 178

Growth rate

TestVESDA Fire1 alarm time (s)

PU foam


Doc. 19378_01 7

Cotton wick fire

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

134 136 138 140 142 144 146 148 150 152 154 156 158 160Time (s)

VE

SD

A r

ead

ing

(%

/m)

Recorded existed VLP alarm time

Predicted extended VLP alarm time

(a) from the cotton wick fire test

(b) from the cotton wick fire test

Figure 7. Smoke concentrations recorded from the tests

Using this method, estimated response times from VLP covered half platform are 150s for the cotton wick fire test and 66s for the PU foam fire test.

7.2 Comparison of predictions from the different algorithms

It’s noted that predictions of VESDA’s performances from the two algorithms have good agreement, with differences in 10 to 12%.

The predictions from the smoke curves have higher accuracy while the calibration is required to achieve similar outcomes from the modellings. After applying the calibration, more conservative predictions were obtained from the CFD modellings.

Generally, CFD modelling of the PU foam fire has relatively higher accuracy with or without the calibration. This is because of the smoke movement (plume) from the PU foam fire, a flaming fire with higher heat released, having a relatively easy predicable pattern. Repaid rise of smoke level from both the test and modelling, which are shown in curves below and in Appendices, lead to a high level of agreement.

PU Foam

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80

Time (s)

VE

SD

A r

ead

ing

(%

/m)

Recorded existed VLP alarm time

Predicted extended VLP alarm time


Doc. 19378_01 8

(a) PU foam fire – MLR method

(b) PU foam fire – HRR method

(c) Cotton wick fire

Figure 8. Simulated smoke curves from the fires

Cotton wicks (T-square)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0 50 100 150 200 250 300Time (s)

VE

SD

A r

ead

ing

s (%

/m)

Existed VLP

Extended VLPMax Coverage VLP

VESDA Fire1 threshold

PU foam fire (MLR method)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

10.0

0 20 40 60 80 100 120 140 160 180

Time (s)

VE

SD

A r

ead

ing

s (

%/m

)

Existed VLP

Extended VLP

Max coverage VLP

PU foam fire (HRR method)

0.0

5.0

10.0

15.0

20.0

25.0

30.0

0 20 40 60 80 100 120 140 160 180 200

Time (s)

VE

SD

A r

ead

ing

s (%

/m)

Existed VLP

Extended VLP

Max coverage VLP


Doc. 19378_01 9

8. Conclusion

From the CFD modelling, it shows no significant delays on detection time from an extended VESDA system covering half of the platform. For a VESDA system with maximum allowed pipe length, the predicted response times to the tested fires are within 1.5 minute (the PU foam fire) and 3 minutes (the cotton wick fire).

By applying the correction factors from the tests, the CFD prediction can be improved in accuracy.

Analysing recorded VESDA’s absolute smoke levels, VESDA’s performances under other system options can be estimated with relatively high accuracy. Such testing data are also valuable for validating the CFD modelling.


Doc. 19378_01 10

Appendix A. VESDA Pipe Networks & Parameters in the Tests

No Item Lower level Upper level

1 Detector model and number. 1 x VLP-000 1 x VLP-000

P1: 54m P1: 57m 2 Sampling pipe number and length

P2: 49m P2: 53m

4 Pipe spacing 5m 5m

5 Distance to safety door 2.25m 2.25m

6 Sampling hole spacing 2-4m 2-4m

7 Sampling hole size 3mm 3mm

8 Sampling hole number 14 14

9 End-cap vent 3mm 3mm

10 Estimated transport time Max 90s Max 90s


Doc. 19378_01 11

Appendix B. Recorded Smoke Curves from Tests

(VLP001 – upper, VLP002 – lower)

(From cotton wick fire)

(From PU foam fire)


www.xtralis.com

The Americas +1 781 740 2223 Asia +852 2916 8894 Australia and New Zealand +61 3 9936 7000

Continental Europe +32 56 24 19 51 UK and the Middle East +44 1442 242 330

The contents of this document are provided on an “as is” basis. No representation or warranty (either express or implied) is made as to the completeness, accuracy or reliability of the contents of this document. The manufacturer reserves the right to change designs or specifications without obligation and without further notice. Except as otherwise provided, all warranties, express or implied, including without limitation any implied warranties of merchantability and fitness for a particular purpose are expressly excluded. This document includes registered and unregistered trademarks. All trademarks displayed are the trademarks of their respective owners. Your use of this document does not constitute or create a licence or any other right to use the name and/or trademark and/or label. This document is subject to copyright owned by Xtralis AG (“Xtralis”). You agree not to copy, communicate to the public, adapt, distribute, transfer, sell, modify or publish any contents of this document without the express prior written consent of Xtralis.

Doc. 19378_01

19378 01 cfd modelling vesda beijing metro a4

Documents