Fire Technology

, Volume 53, Issue 3, pp 1101–1121 | Cite as

Real-Time Forecasting of Building Fire Growth and Smoke Transport via Ensemble Kalman Filter

Article
First Online:
Received:
Accepted:

Abstract

Forecasting building fire growth and smoke dispersion is a challenging task but can provide early warnings to first responders and building occupants and thus significantly benefit active building fire protection. Although existent computer simulation models may provide acceptable estimations of smoke temperature and quantity, most simulations are still not able to achieve real-time forecast of building fire due to high computational requirements, and/or simulation accuracy subject to users’ inputs. This paper investigates one of the possibilities of using ensemble Kalman filter (EnKF), a statistical method utilizing the real-time sensor data from thermocouple trees in each room, to estimate the spread of an accidental building fire and further forecast smoke dispersion in real time. A general approach to forecasting building fire and smoke is outlined and demonstrated by a 1:5 scaled compartment fire experiment using a 1.0 kW to 2.8 kW propane burner as fire source. The results indicate that the EnKF method is able to forecast smoke transport in a multi-room building fire using 40 ensemble members and provide noticeable accuracy and lead time. Unlike other methods that directly use measurement data as model inputs, the developed model is able to statistically update model parameters to maintain the forecasting accuracy in real time. The results obtained from the model can be potentially applied to assist mechanical smoke removal, emergency evacuation and firefighting.

Keywords

Data assimilation Ensemble Kalman filter Zone model Real-time forecast Sensor integration Smoke transport 

Nomenclature

cp

Specific heat of air in constant pressure, kJ/kgK

Cs

Smoke layer height factor

cv

Specific heat of air in constant volume, kJ/kgK

H

Height, m

Hd

Smoke layer height, m

K

Kalman gain

\( \dot{m} \)

Mass flow rate kg/s

P

Pressure, kPa

Pf

Forecasted error covariance

q

Number of ensemble member

\( \dot{q} \)

Energy flux, kW

T

Temperature, °C

t

Time, second

V

Volume, m3

v

Velocity, m/s

w

Width of opening, m

x

Model states

y

Measurement

γ

Ratio of cp to cv

θ

Localization factor

ρ

Density, kg/m3

ϕ

Control parameters

Subscript

amb

Ambient

l

Lower zone

u

Upper zone

vent

Ventilation

z

Height, m

Superscript

a

Analyzed

f

Forecasted

Abbreviations

BAS

Building automation system

EnKF

Ensemble Kalman filter

CFD

Computational fluid dynamics

PIV

Particle image velocimetry

FDS

Fire dynamic simulator

RMSE

Root mean squared error

HRR

Heat release rate

This is a preview of subscription content, log in to check access.

Notes

Acknowledgement

The authors acknowledge the financial supports from the Discovery Grants of the Natural Sciences and Engineering Research Council of Canada (NSERC) (No. 402848-2012) (2012–2017) and the Concordia University Research Chair (CURC) New Scholar for the Building Fire Safety, Smoke, Airflow and Thermal Management program (2014–2019). The authors also thank Dr. Wael Saleh and Guanchao (Jeremy) Zhao for the technical supports of conducting the fire and PIV experiments.

References

  1. 1.
    Cowlard A, Jahn W, Abecassis-Empis C, Rein G, Torero JL (2010) Sensor assisted fire fighting. Fire Technol 46:719–741. doi:  10.1007/s10694-008-0069-1 CrossRefGoogle Scholar
  2. 2.
    Yang D, Hu LH, Jiang YQ, Huo R, Zhao XY (2010) Comparison of FDS predictions by different combustion models with measured data for enclosure fires. Fire Saf J 45:298–313. doi:  10.1016/j.firesaf.2010.06.002 CrossRefGoogle Scholar
  3. 3.
    Koo S-H, Fraser-Mitchell J,Welch S (2010) Sensor-steered fire simulation. Fire Saf J 45(3):193–205. doi:  10.1016/j.firesaf.2010.02.003 CrossRefGoogle Scholar
  4. 4.
    Jahn W, Rein G, Torero J. (2011) Forecasting fire growth using an inverse zone modelling approach. Fire Safety J 46(3):81–88. doi:  10.1016/j.firesaf.2010.10.001 CrossRefGoogle Scholar
  5. 5.
    Jahn W, Rein G, Torero J. (2012) Forecasting fire dynamics using inverse computational fluid dynamics and tangent linearisation. Advances in Engineering Software 47(1):114–126. doi:  10.1016/j.advengsoft.2011.12.005 CrossRefGoogle Scholar
  6. 6.
    Beji, T., Verstockt, S., Van de Walle, R. and Merci, B. (2014) On the use of real-time video to forecast fire growth in enclosures. Fire Technol 50(4):1021–1040. doi:  10.1007/s10694-012-0262-0 Google Scholar
  7. 7.
    Overholt, K. J. and Ezekoye, O. (2012) Characterizing heat release rates using an inverse fire modeling technique. Fire Technol 48(4):893–909. doi:  10.1007/s10694-011-0250-9 CrossRefGoogle Scholar
  8. 8.
    Overholt, K. J. and Ezekoye, O. (2014) Quantitative testing of fire scenario hypotheses: a bayesian inference approach. Fire Technol 51(2):335–367. doi:  10.1007/s10694-013-0384-z CrossRefGoogle Scholar
  9. 9.
    M. Price and A. Trouvé (2015) A Multi-observable approach to address the Ill-posed nature of inverse fire modeling problems. Fire Technol. doi:  10.1007/s10694-015-0541-7 Google Scholar
  10. 10.
    Lin C-C, Wang L (2013) Forecasting simulations of indoor environment using data assimilation via an Ensemble Kalman Filter. Build Environ 64:169–176. doi:  10.1016/j.buildenv.2013.03.008 CrossRefGoogle Scholar
  11. 11.
    Heskestad G (1984) Engineering relations for fire plumes. Fire Saf J 7:25–32. doi:  10.1016/0379-7112(84)90005-5 CrossRefGoogle Scholar
  12. 12.
    Tanaka T (1983) A model of multiroom fire spread. Fire Sci Technol 3:105–121. doi:  10.3210/fst.3.105 CrossRefGoogle Scholar
  13. 13.
    Peacock RD, Davis S, Lee BT (1988) Experimental data set for the accuracy assessment of room fire models. National Bureau of Standards 88-3752.Google Scholar
  14. 14.
    Peacock RD, Reneke PA (2007) Verification and validation of selected fire models for nuclear power plant applications. Vol. 5. Consolidated fire growth and smoke transport model (CFAST). NUREG-1824, US Nuclear Regulatory Commission, Washington, DC.Google Scholar
  15. 15.
    Ji J, Li M, Li K, Yuan M, Sun J (2015) Ambient wind effect on combustion characteristcs in compartment with simutaneous door and window opened. Energy Build 105:217–225. doi:  10.1016/j.enbuild.2015.07.046 CrossRefGoogle Scholar
  16. 16.
    Evensen G (1994) Sequential Data Assimilation with a Non-Linear Quasi-Geostrophic Model Using Monte Carlo Methods to Forecast Error Statistics. J Geophys Res 99:10,143-10,162. doi:  10.1029/94JC00572
  17. 17.
    Lin C-C, Wang LL (2015) Forecasting smoke transport during compartment fires using a data assimilation model. J Fire Sci 33(1):3–21. doi:  10.1177/0734904114548837 CrossRefGoogle Scholar
  18. 18.
    Hamill TM, Whitaker JS (2001) Distance-dependent filtering of background error covariance estimates in an ensemble Kalman filter. Mon Weather Rev 129:2276-2790. Doi:  10.1175/1520-0493(2001)129<2776:DDFOBE>2.0.CO;2 Google Scholar
  19. 19.
    Houtekamer PL, Mitchell HL (2001) A sequential ensemble Kalman filter for atmospheric data assimilation. Mon Weather Rev 129:123–137. doi:  10.1175/1520-0493(2001)129<0123:ASEKFF>2.0.CO;2 CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media New York 2016

Authors and Affiliations

  1. 1.Department of Building, Civil and Environmental EngineeringConcordia UniversityMontrealCanada

Personalised recommendations