PREDICTING LARGE-SCALE FIRE PERFORMANCE FROM SMALL- SCALE FIRE TEST DATA Marcelo M. Hirschler * And Marc L. Janssens ** * GBH International ** Univ. North Carolina @ Charlotte
OVERALL OUTLINE Predictive Procedures: * Models * Correlations * Pass/Fail Small-Scale Fire Tests Large-Scale Fire Tests Products Applications
Predictive Procedures Models Calculate Results Based on Theories and Equations Correlations Can Predict Relative Fire Performance and Perhaps Test Results Pass/Fail Procedures Can Assess Whether Products are Likely to Meet Certain Criteria, e.g. Flashover
Flashover Temperature/Products Growth Time Fully-developed Fire Decay
Self-Propagating Fire Products May be Unacceptable and Yet Not Lead to Flashover This Can Happen if Product Fails Certain Regulatory Pass/Fail Criteria This Can Happen if Product Leads to Self-Propagating Fire Self-Propagating Fire is Fire that Doesn t Cease Burning Without External Input
Large-Scale Fire Test Methods
Interior Wall and Ceiling Finish: ROOM/CORNER TEST STANDARDS NFPA 265 NFPA 286 ISO 9705
ROOM/CORNER TEST STANDARDS Basic Test Apparatus
ROOM/CORNER TEST STANDARDS Details & Measurements
ROOM/CORNER TEST STANDARDS NFPA 265 8 x 12 x 8 ft room with type X gypsum walls 30 x 90 in. door opening in front wall 12 x 12 in. propane gas burner, 12 in. off floor Burner at 2 in. standoff from back & side walls 40 kw first 5 min, 150 kw next 10 min Only back and side walls lined Ceiling thermocouples are present Flashover is primary endpoint
ROOM/CORNER TEST STANDARDS NFPA 286 8 x 12 x 8 ft room with type X gypsum walls 30 x 90 in. door opening in front wall 12 x 12 in. propane gas burner, 12 in. off floor Burner in contact with back and side walls 40 kw first 5 min, 160 kw next 10 min Only back and side walls lined Ceiling thermocouples are present Acceptance criteria in codes include smoke production
ROOM/CORNER TEST STANDARDS ISO 9705 2.4 x 2.4 x 3.6 m room with calcium silicate or concrete walls 0.8 x 2 m door opening in front wall 0.17 x 0.17 m propane gas burner, 0.17 m off floor Burner in contact with back and side walls 100 kw first 10 min, 300 kw next 10 min Back wall, side walls, and ceiling lined No ceiling thermocouples Criteria for Marine Fire Restricting Materials based on heat release and smoke production
ISO 9705 Test (1) Fully Lined Room Prior to Test
ISO 9705 Test (2) Start of 100 kw Exposure
ISO 9705 Test (3) Prior to Flashover
ISO 9705 Test (4) After Termination
Upholstery Items PRODUCT TEST STANDARDS ASTM E 1537/CA TB 133 ASTM E 1590/CA TB 129 CBUF Test Protocol ASTM E 1822
Key Test Measurements Heat Release: RHR & THR Smoke Release: RSR & TSR Mass Loss of Item Room Temperature Room Visibility
Furniture Test Details ASTM E 1537/NFPA 266/UL 1056 Gas Burner @ 19 kw (13 L/min) 80 s Impinges on Item from Top/Front Furniture or Mock-up Needed Everything Else Like Room-Corner Tests
Mattress Test Details ASTM E 1590/NFPA 267/UL 1895 Gas Burner @ 18 kw (12 L/min) 180 s Impinges on Item from Front Actual Mattress Needed Everything Else Like Room/Corner Tests
CBUF Test Details European Test Gas Burner @ 30 kw (19.5 L/min) 120 s Impinges on Item from Top/Front Furniture or Mattresses Everything Else Like Room-Corner Tests
Stacking Chair Test Details ASTM E 1822 Gas Burner @ 18 kw (12 L/min) 180 s Impinges on Item from Front 5 Stacked Chairs Needed Everything Else Like Room/Corner Tests
Furniture Pass/Fail Criteria Chairs: CA TB 133 & NFPA 101/IFC Mattresses: CA TB 129 & NFPA 101/IFC Stacking Chairs: NFPA 301 All Criteria Based on RHR and THR Depend on Occupancy
Cables: VERTICAL TRAY STANDARDS ASTM D 5424/ D 5537 UL 1685 & CSA FT4 w/rhr FIPEC Test Protocol IEC 60332-3 Modified
Vertical Cable Tray Test
ASTM D 5424/D5537 Cable Tray Test: Front
ASTM D 5424/D 5537 Cable Tray Test: Back
CSA FT4 Cable Tray Test: Burner and Tray
IEC 60332-3 3 or FIPEC Cable Tray Test
Fire Source Input to Tests 20 kw or 70,000 BTU/hr
Cables Pass/Fail Criteria UL 1685: RHR, flame spread, smoke CSA FT4: flame spread and smoke IEC 60332-3: flame spread FIPEC: RHR, THR, flame spread & smoke (classes under development)
Pipe Insulation Draft NFPA 274: Vertical Pipe Chase Use in Plenums Input: 20 kw (3 min), 70 kw (7 min) Criteria: RHR, THR, flame spread & smoke (standard under development)
Pipe Chase Fire Test Pipe Chase - dimensions are interior dimensions 18 80 8 6 12 48 24
Transportation Applications Same Tests & Techniques Can be Used: Aircraft Trains Buses Ships Cars
Small-Scale Scale Fire Test Methods: RHR
Small-Scale Scale Test Methods Useful for Predictions Cone Calorimeter: ASTM E 1354 or NFPA 271 or ISO 5660 OSU Calorimeter: ASTM E 906 FM Calorimeter: ASTM E 2058 or NFPA 287 or FM Global Tests
Cone Calorimeter
OSU Calorimeter
FM Calorimeter
A Survey of Methods to Predict Performance of Wall Linings in the Room/Corner Test
INTRODUCTION Room/corner tests have been used for more than two decades to assess the fire performance of linings Several standard protocols are now available Standards are specified in codes and regulations Textile Linings (Codes) NFPA 265 All Other Interior Finish (Codes) NFPA 286 Fire Restricting Materials (IMO High Speed Craft) ISO 9705 Development of predictive methods is motivated by high cost for testing and sample size
Physical Phenomena
ROOM/CORNER FIRE GROWTH Flame Spread Modes
ROOM/CORNER FIRE GROWTH Ignition of Initially Heated Area
ROOM/CORNER FIRE GROWTH Flames Spread to Ceiling
ROOM/CORNER FIRE GROWTH Flames Spread to End of Wall
LITERATURE SURVEY 16 methods found Distinction can be made between 3 types of methods Simulation models: predict room environment and fire growth Analytical methods: predict fire growth Statistical correlations: predict particular aspect of fire growth such as the time to flashover The extent of validation varies widely
Simulation Models (1 of 2) Steckler (1983) OSU (Smith and Satija, 1983) Karlsson (1992-94) Quintiere (1993) Janssens (1995) variation of Quintiere model Wade (1996) variation of Quintiere model
Simulation Models (2 of 2) Opstad (1995) Yan (1996) HAI (1999) SwRI (1999) variation of Quintiere model WPI (1999)
Steckler Model (NBS) Described in NBSIR 83-2765 (1983) Based on conceptual framework developed by Quintiere Two-zone room environment Only considers lateral flame spread and does not address upward and wind-aided spread Accounts for oxygen vitiation effects on burning rate No validation
OSU Model Described by Smith and Satija Many revisions were published subsequently Critical review by Janssens Two-zone room environment Considers ceiling jet Upward flame spread algorithms based on data from OSU calorimeter (ASTM E 906) Lateral flame spread based on OSU data No separation between physics and numerics
Karlsson Model (Lund, Sweden) Described in detail in Karlsson s Ph.D. thesis (1992) Two-zone room environment with layer interface fixed at soffit Only considers upward and downward flame spread Upward flame spread based on RHR and ignition data from Cone Calorimeter (ASTM E 1354, ISO 5660) Downward spread based on LIFT data (ASTM E 1321) Model used to develop power law correlations between flashover time and flammability properties downward spread not relevant for lined ceiling
Karlsson Model/Inequality Described by Karlsson (1994) Based on cone data, ideally @ 50 kw/m^2 3 critical parameters: (a) Exponential Decay Factor (Lambda) (b) Equivalent Time to Ignition (Tau) (c) Peak RHR Cone RHR curve must be fitted Inequality predicts whether Self-Propagating Fire is likely to occur
Regions of Flame Front Acceleration - Bjorn Karlsson Model 8 7 6 Lambda Tau 5 4 3 lbd tau=(1+sqrta)^2 lbd tau=(1-sqrta)^2 lbd tau=a-1 2 1 0 0 2 4 6 8 10 12 14 16 a: K Pk RHR
Example of a Fire Retarded Textile Wallcovering Cone @ 50 kw/m^2 - Karlsson Analysis 300 250 200 RHR (kw/m^2) 150 100 50 0 0 20 40 60 80 100 120 140 160 180 200 Time (s) Data Fitted
Example of an Unpainted Gypsum Wall Karlsson Analysis - Cone @ 50 kw/m^2 120 100 80 RHR (kw/m^2) 60 40 20 0 0 20 40 60 80 100 120 time (s) Data Fitted
Example of Car Interior Molding Karlsson Analysis - Cone @ 40 kw/m^2 350 300 250 RHR (kw/m^2) 200 150 100 50 0 0 100 200 300 400 500 600 700 time (s) Data Fitted
Example of Vinyl Lining Karlsson Analysis - Cone @ 50 kw/m^2 90 80 70 60 RHR (kw/m^2) 50 40 30 20 10 0 0 5 10 15 20 25 30 35 40 45 time (s) Fitted Data
Quintiere Model (1 of 2) One-zone room environment T g based on modified MQH correlation Uniform T s based on 1-D heat conduction Considers all spread mechanisms Wind-aided spread based on cone calorimeter data Opposed-flow spread based on LIFT data Accounts for burnout Validation for wide range of materials and different standard room/corner test scenarios
Quintiere Model (2 of 2)
Variations of Quintiere Model Janssens (AF&PA, DC) Includes burner flame geometry and heat flux calculations Revised procedures to obtain ignition, flame spread, and heat release properties from cone calorimeter and LIFT Validated with NFPA 286 & ISO 9705 wood data Wade (BRANZ, New Zealand) Incorporates spread algorithms in two-zone room fire model Uses Janssens ignition and flame spread properties
CFD-Based Models CFD grid in gas phase defines wall grid No explicit consideration of flame spread modes LIFT flame spread data are not needed Opstad (SINTEF, Norway, 1995) Extension of KAMELEON code developed at SINTEF Ignition and heat release rate based on cone data Yan (Lund, Sweden, 1996) Ignition and heat release rate based on pyrolysis submodel
US Coast Guard Models Three models were developed to simulate ISO 9705 tests on marine composites conducted at SwRI SwRI model (Janssens and Dillon) Based on Janssens version of Quintiere model Revised heat release and add smoke production calculations WPI model (Dembsey and Barnett) Based on Mitler s flame spread algorithms and CFAST HAI model (Beyler et al.) Extension of corner fire model developed by HAI for US Navy SwRI model gives the most consistent predictions
Analytical Methods Room effects are fixed single test scenario All methods require only cone data Magnusson (Lund, Sweden, 1984) Assumes exponential heat release rate curve in room test Requires ignition temperature measurements Validation: ISO 9705 and 1/3 scale room data for 13 materials Wickström and Göransson (SP, Sweden, 1992) Single cone calorimeter test needed Applied to ISO 9705, but modified for larger room Validation: ISO 9705 and large room for 11 EUREFIC materials Dietenberger (FPL, WI, 1998) measurements dynamics
Statistical Correlations Östman (Trätek, Sweden) Flashover time correlation Smoke SEA correlations Based on Swedish ISO 9705 test data for 13 materials Quintiere (University of Maryland, MD) Critical b-parameter for accelerating upward flame spread " t b = 0.01 q 1 t Originally developed for ISO 9705 Later modified for NFPA 265 and NFPA 286 ig b
Quintiere Modified b Flammability Parameter 1200 900 EPS foam tflashover (s) 600 300 Polyurethane foam 0-3.00-2.50-2.00-1.50-1.00-0.50 0.00 0.50 1.00 b NFPA 265 Protocol NFPA 286 Protocol
DISCUSSION Extensive validation shows that Quintiere s model, in original or modified form provides good predictions for a wide range of materials and different scenarios The b-parameter concept is a useful screening tool Additional modification by Dillon et al. follows, and is applied to actual data
Dillon Janssens - Hirschler Developed tools for predicting performance of materials in room/corner tests NFPA 265 & NFPA 286 Cone Calorimeter @ 50 kw/m² 36 Materials investigated Predicted likelihood of flashover (19 materials) No flashover (17 materials) Peak heat release rate Smoke production
Wall and Ceiling Linings NFPA 265 required by US codes for textile wall coverings: No Flashover NFPA 286 required by US codes for other interior finish: No Flashover and TSR < 1,000 m²
What Does the Cone Calorimeter Do? The cone calorimeter measures:! Heat release rate! Total heat released! Effective heat of combustion (all measurements done by the oxygen consumption principle)
What Does the Cone Calorimeter Do? The calorimeter also measures:! Mass loss rate! Time to ignition! Specific extinction area (i.e. smoke), and! Optionally, CO/CO 2 production
Schematic of Cone Calorimeter in Concept
Cone Calorimeter Sample Exposure:! Radiant heat fluxes from a conical heater! Exposure values range from 0 to 100 kw/m 2! Horizontal Orientation
Materials Database 6 sets of materials 36 materials. Room/corner test: NFPA 265 fi 11 materials. NFPA 286 fi 25 materials. Cone Calorimeter: all materials 50 kw/m² Horizontal 6 materials tested vertically
Occurrence of Flashover Most important factor in Room-Corner tests Time to flashover based on cone data. Flashover unlikely if t flashover > 900 s Correlation based on a wind-aided flame spread analysis by Cleary and Quintiere: b = 0.01 HRR avg 1 b > 0 indicates likelihood of flashover. t b for many materials not presented. t ig t b
Occurrence of Flashover RHR ( ) ( ( ) t t = RHR exp λ t t ig peak ig RHR represented as an exponentially decaying function of time, with l determined by matching to RHR 180
Occurrence of Flashover Fit Decay Coefficient 500 HRR (kw/m²) 400 300 200 100 0 0 60 120 180 240 300 360 420 480 Time (s)
Likelihood of Flashover b =.01 HRR 1 λ 0 180 t ig
Likelihood of Flashover 1200 Time to Flashover (s) 900 600 300 No Flashover PU Foam EPS Foam 0-3.0-2.5-2.0-1.5-1.0-0.5 0.0 0.5 1.0 NFPA 265 Protocol b NFPA 286 Protocol
Time to Flashover Materials reach flashover within 2 minutes of burner increase. No correlation developed for t flashover.
Peak Heat Release Rate Materials that do not reach flashover. 17 materials HRR peak approximation: Peak HRR from Cone @ 50 kw/m² Area in contact with burner Exposed sample area: A = 2 D H flame Flame height at 50% intermittency: NFPA 265 fi 150 kw fi H flame = 1.71 m NFPA 286 fi 160 kw fi H flame = 1.79 m
Peak RHR Correlation 500 Predicted Peak HRR (kw) 400 300 200 100 0 0 100 200 300 400 500 Measured Peak HRR (kw) NFPA 265 Protocol NFPA 286 Protocol
Revised RHR peak Correlation Extended burning area. Requires t ig from Cone (13 materials). Area ignites t ig seconds after burner increase. Taller flame Based on RHR of burning area Heated section ignites t ig seconds later At 2 t ig the RHR is a sum of the two areas RHR peak depends on rate of exponential decay Provides a reasonable, conservative correlation
Revised Pk RHR Correlation 500 Predicted Peak HRR (kw) 400 300 200 100 0 0 100 200 300 400 500 Measured Peak HRR (kw) NFPA 265 Protocol NFPA 286 Protocol
Total Smoke Release Based on the TSR measured in the cone: TSR = A Room TSR Cone TSR Cone based on measured cone properties: TSR Cone = THR HOC Relies on an estimate of the burning area. Area assumed to be 4 m² 1/6 of total wall surface σ
Total Smoke Release Correlation Cone Calorimeter TSR (m²/m²) 400 300 200 100 0 2 m² 4 m² 8 m² 0 200 400 600 800 1000 1200 1400 1600 Room Test TSR (m²) NFPA 265 Protocol NFPA 286 Protocol
Interior Finish Predictions Data for 36 materials obtained from 6 studies Method to estimate NFPA 265 & 286 room/corner test performance was successful Cone data at 50 kw/m². Modified Quintiere b parameter provides reasonable estimation of the likelihood of flashover. Based of exponential decay factor l. Discrepancies for low-density foams.
Successful Predictions Using Dillon et al. Model Peak RHR conservatively estimated From RHR peak measured in the cone Area of material exposed to burner flames Limited wind-aided flame propagation Conservative estimate of Total Smoke Release From TSR in the cone Assuming an area of 4 m²
Cable Tray Tests
CSA FT4 Predictions @ 20 Flux Total cables tested: 21, with 7 Failures All Failures: Cone Pk RHR > 200 kw/m^2 11 of 14 Passes: Cone Pk RHR < 200 kw/m^2 3 of 14 Passes: Cone Pk RHR: Safe Errors
FIPEC Cable Test & Cone @ 50 If Pk RHR 150 kw/m^2 & SMOGRA 10: 88% Correct Predictions on Smoke 95% Safe Predictions on Smoke 2 of 43 Cables with Unsafe Errors
Cable Tray Predictions Using cone calorimeter data CSA & UL 1685 results can be predicted by Correlations FIPEC test results can be predicted by Pass/Fail methods
Furniture Predictions Using Cone Calorimeter Data Chair & Mattress test results can be predicted by Correlations, occasionally Chair & Mattress test results can be predicted by Pass/Fail Methods, often
Cable Tray or Furniture Predictions (Correlations or Pass/Fail) Using cone calorimeter data Self-Propagating Fires can be predicted, often Furniture Flashover can be predicted, usually
Transportation Environments Aircraft predictions use OSU Calorimeter Surface Mass Transportation predictions can use the same techniques as other rooms Cars: Self-Propagating Fires are being predicted based on cone tests
Conclusions Prediction Techniques Exist and Can be Successful Work on Interior Finish, Furniture and Cables to Various Extents Flashover or Self Propagating Fires Can Be Predicted, Often There is Much More Work to be Done