RECOMMENDATION OF TESTS FOR ASSESSING FLAME SPREAD OF MATERIALS IN HONG KONG

Transcription

1 RECOMMENDATION OF TESTS FOR ASSESSING FLAME SPREAD OF MATERIALS IN HONG KONG by Wan Ki CHOW and Cheuk Wai LEUNG Review paper UDC: : BIBLID: , 11 (2007), 2, pp Per for mance-based de sign for pas sive build ing fire safety pro vi sions is ac - cepted by the au thor ity in Hong Kong since This is also known as the fire en gi neer ing ap proach, though the per for mance-based fire code is not yet avail able. To cope with the use of new build ing ma te ri als, ap pro pri - ate flame spread tests on ma te ri als and com po nents should be spec i fied. Af - ter re view ing four stan dard tests in the lit er a ture, i. e. ASTM E a, BS476: Part 7: 1997, ASTM E84-99/NFPA 255, and ISO 9705: 1993(E), it ap pears that ISO 9705: 1993(E) is suit able for as sess ing the flame spread of ma te ri als. Key words: fire, flame spread, standard tests, assessment, materials Introduction Pas sive build ing de sign for fire safety is spec i fied by pre scrip tive codes is sued by the Build ings De part ment (BD) [1-3] of the lo cal gov ern ment. Apart from spec i fy ing the fire re sis tance pe riod of wall, flame spread over ma te ri als was not clearly de scribed. Note that flame spread over lin ing and fin ishes ma te ri als is one of the key el e ments [4] in pro vid ing fire safety. Up ward flame spread along a wall lin ing and hor i zon tal flame de - vel op ment be neath a com bus ti ble ceil ing would give flashover rap idly. Pro vi sion of fire ser vice in stal la tions in Hong Kong is con trolled by the Fire Ser - vices De part ment (FSD) [5]. BS 476: Part 7 [6] on sur face spread of flame test was spec i - fied for some ap pli ca tion such as fin ishes ma te ri als in their fire ser vice in stal la tions code [5]. But this test is ba si cally for as sess ing build ing ma te ri als, not the en tire build ing el e - ment. It is not fea si ble to as sess mod ern sand wich pan els [7]. Suit able tests on flame spread should be spec i fied for as sess ing mod ern ma te ri - als and com po nents, es pe cially in ap ply ing fire en gi neer ing ap proach [8, 9]. This is sim i lar to en gi neer ing per for mance-based fire codes (EPBFC) over seas [10]. At the mo - ment, EPBFC is not yet avail able but the gov ern ment would con sider the per for - mance-based de signs of fire safety pro vi sions through this fire en gi neer ing ap proach [8, 9]. Heat re lease rate of ma te ri als [11] is the most im por tant pa ram e ter in fire haz ard as sess ment while im ple ment ing per for mance-based de sign. Fire prop er ties of ma te ri als DOI: /TSCI C 53

2 THERMAL SCIENCE: Vol. 11 (2007), No. 2, pp can be as sessed by study ing the re la tion ship be tween flame spread and heat re lease rate. Note that fire prop er ties can be in ter preted [12] as in put ma te rial data for fire mod els. Those val ues might not be unique and de pend on the burn ing con di tions. Ap pro pri ate tests with math e mat i cal mod els have to be used for pre dict ing the real-scale fire. In fact, flame spread tests should be spec i fied in the karaoke establishments bills [13] de vel oped af ter a big karaoke ar son fire. Karaokes are typ i cally par ti tioned into many boxes with long corridors by timber partition materials [14]. Timber materials are still more com monly used [15] than gyp sum plasterboards even now as they are cheaper and eas ier to in stall. Those tim ber prod ucts can be ig nited eas ily un der a flashover fire even with re tar dants [16]. Up ward and hor i zon tal flame spread over the lin ings and wall might give a heat re lease rate high enough to flashover. Items placed ad ja cent to the par ti tion wall might be ig nited. Fur ther, the par ti tion wall would fall, lose the compartmentation ef - fect and block the way. The par ti tion walls are re quired to have a fire re sis tance pe riod of 1 hour in new codes [1-3]. As pointed out [17], lin ing ma te ri als that can be eas ily ig nited in a fire but can not sus tain flame spread af ter re mov ing the fire source would im ply some de - gree of fire safety. Flame spread over the par ti tion ma te ri als should be watched to give compartmentation ef fect. This point will be stud ied in this pa per. Pre vi ously, four stan dard tests on spread of flame com monly used were pre lim i - nary re viewed [15]. The aim was to see whether rec om men da tions can be made to the Au - thor ity on spec i fy ing the flame spread be hav iour of ma te ri als and com po nents. Those four stan dard tests are: ASTM E a Standard Test Method for Determining Material Ignition and Flame Spread Properties [18], also referred to as the Lateral Ignition and Flame Spread Test or LIFT, BS476: Part 7: 1997 Method of test to determine the classification of the surface spread of flame of products [6], ASTM E84-99 or ANSI/NFPA Standard test method for Surface Burning Characteristics of Building Materials [19, 20], and ISO 9705: 1993(E) Fire Tests Full-scale room test for surface products [21]. Comparison of the four tests A com par i son of the four tests was sum ma rized be fore [15]: Both ASTM E1321 and BS476: Part 7 are bench-scale tests for assessing specimens of smaller size relative to the actual construction. It is difficult to assess full-scale effects such as structural performance of a material or construction component in real fires. Thermostructural failures and falling of non-structural members [22] in actual fire conditions might affect the flame spread. Horizontal flame spread is tested with materials mounted in vertical position for ASTM E1321 and BS476: Part 7. This may not be good for materials mounted horizontally such as carpets. Even for testing walls, the spread of flame is along the vertical direction [23]. Further, materials are tested in a ceiling position for ASTM E84/NFPA

3 Chow, W. K., Leung, C. W.: Recommendation of Tests for Assessing Flame... ASTM 1321, BS476: Part 7 and ASTM E84/NFPA 255 allow testing one product by itself, or attached to an adjacent product, at one time. Flame spread from the wall to the ceiling or from the ceiling to the wall cannot be studied. In actual fires, fuel droplets would fall from the burning fuel, say the ceiling, onto the wall or floor. This, together with the radiative heat flux from the hot gas layer, would give a faster fire spreading rate [24]. A wall surface exposed to an advancing heat and flame front is simulated in BS476: Part 7. The fire is fairly well developed. It only provides a measurement of the rate of development of the flame across the materials, which are not the first items ignited and no thermal feedback from ignition is considered [25]. The lengthy preheat time of ASTM E1321, say from 7 to 10 minutes, of the specimen may allow surface pyrolysis which will then have time to change from the actual conditions after the pilot flame is inserted. Such changes can lead to poor ignition of the sample and altering its burning behaviour. This can be evidenced by the propensity for oscillating flames rather than sustained flaming, after ignition [26]. As a result, values measured in assessing flame spread depend on the consistency of flame propagation. Erratic flame fronts may lead to uncertain sets of data and affect the repeatability of the test. It was recommended [26] to eliminate the preheat period and to run the tests using the International Maritime Organisation (IMO) Res. A653(16) [27] or ASTM E1317 [28] surface flammability test protocol, which operates at a higher heat flux level without preheating the specimen, and then use the ASTM E1321 procedures for further analysis. For ISO 9705, wall and ceiling materials can be tested in their normal mounted conditions. Flame spread from one wall or falling burning objects to adjacent walls and ceiling can be observed. From the experiments reported on investigating the burning behaviours of materials under different configurations [29], the results on testing the walls and ceiling separately and putting them together are different. Typical results on the heat release rate curves of plywood and fire-retarded plywood were reported [29]. The heat release rates of testing materials with fire retardant for wall only, or for ceiling only, would be much lower than the materials without fire retardant. However, for the arrangement with both wall and ceiling, the heat release rates for the materials with fire retardant were only slightly less than those without fire retardant. In other words, the effect of putting on fire retardant to reduce the heat release rate is not significant. Updated cone results [7] illustrated that burning materials with and without fire retardants under high heat fluxes are similar. Such results can only be illustrated by modifying ISO 9705 extended to postflashover fire, not stopped at flashover. The test re sults and the clas si fi ca tion sys tem are: ASTM E1321 There is no established pass/fail criterion for the ignition and flame spread test results generated from ASTM E1321. The material parameters obtained can be used in mathematical models for fire growth and prediction of performance of materials [30, 31]. BS476: Part 7 55

4 THERMAL SCIENCE: Vol. 11 (2007), No. 2, pp Materials are classified into Classes 1 to 4 depending on the flame front distances at 1.5 and 10 min. of the test. ASTM E84/NFPA 255 The travelling distances of the flame front (at every 0.6 m or time intervals not more than 30 s) are measured and plotted against time to obtain the flame spread time-distance curve. The total area A T under the curve is used to determine the Flame Spread Index, FSI, which can be compared with the benchmark materials to provide a relative ranking. ISO 9705 There is no classification or product rating scheme defined officially in the standard. Proposed systems overseas are discussed in the later section. Both ASTM E1321 and BS476: Part 7 are bench-scale ex per i ments. ASTM E84/NFPA 255 is a rel a tively large test, while ISO 9705 is con sid ered as a full-scale burn ing test. These four stan dard tests were de vel oped based on dif fer ent sce nar ios. Ma - te ri als were there fore only as sessed un der the con di tions des ig nated. Prac ti cal rank ing or clas si fi ca tion sys tems were de duced from the test ing re sults. At tempts were made to de - rive cor re la tions among the mea sured data ap peared in the lit er a ture [15]. How ever, cor - re la tion ex pres sions de rived are not con vinc ing due to the dif fi cul ties in search ing for tested data on the same ma te ri als un der the four tests. Recommendation of using ISO 9705 More re al is tic fire per for mance data than tests for in di vid ual com po nents can be ob tained [32]. It ap pears that ISO 9705 is a suit able test ing method for as sess ing flame spread for Hong Kong [33] with jus ti fi ca tions as fol lows: Measurement of heat release rate and smoke production The answer to the question How big is a fire? is on estimating the heat release rate from burning the building materials. This was identified to be the most important parameter in hazard assessments. On the other hand, most of the people killed in a fire were due to smoke. Therefore, the smoke production rate is also important. These two important parameters can be measured by ISO Not only that, the temperature, thermal radiation levels, concentration of gases, and production of toxic gases can all be measured accurately. Possibility of flashover Time to flashover can be measured by this test [33]. Note that the room size for the ISO 9705 test is very close to the minimum size of a building. A lining fire might develop faster in a larger room. Scale of test and orientation Flame spread depends on the conditions of the surrounding fluid. Air cavity and substrates for some products may become the heat sink. With ISO 9705, it is possible to test the lining and surface finish products that cannot be tested properly in smaller scale tests. Different orientations, such as wall and ceiling, of the specimen as well as its real configuration of installation can be included. 56

5 Chow, W. K., Leung, C. W.: Recommendation of Tests for Assessing Flame... Review on classification systems based on ISO 9705 There are rec om men da tions on clas si fy ing and rat ing prod ucts based on ISO A rank ing sys tem can be de rived by eval u at ing some of the crit i cal per for mance as - pects. As dis cussed in the lit er a ture [34], the fol low ing can be used to pre sume the haz - ards of ma te ri als such as sand wich pan els: peak heat release rate, total heat release, time to flashover (during 100 kw burner exposure in the first 10 minutes or during the subsequent period of 300 kw burner output), and amount of smoke evolution. The time to flashover was con sid ered as the cri te rion for clas si fi ca tion [e. g. 34, 35]. The abil ity of a ma te rial in sus tain ing flame spread and its con tri bu tion to fire growth can be as sessed. Four lev els of room fire per for mance, A, B, C, and D were sug gested [36] for the reg u la tion con trol of ma te ri als: A: for fire-isolated passageways (exits), no flashover after 10 minutes, B: for assembly areas and corridors providing access to exits, flashover after 6 minutes, C: for general areas, flashover after 4 minutes, and D: not permitted, flashover in less than 4 minutes. Good ef forts were made in the Nordic coun tries on eval u at ing a pro posed sys - tem [37]. A five-scale clas si fi ca tion sys tem was pro posed [38, 39]. Heat re lease rate, smoke pro duc tion and the time to flashover are con sid ered with a graph i cal pre sen ta tion of the sys tem [39]. In the clas si fi ca tion sys tem, both the peak and av er age val ues of the heat re lease rate are con sid ered. A limit is put on long last ing fires which give off a sig nif - i cant amount of to tal heat. Cred its are given to those prod ucts that burn out quickly even though the peak heat re lease is high. To eval u ate the sys tem, eleven prod ucts were tested and clas si fied un der the pro posed sys tem. The re sults were com pared with the reg u la - tions used in Eng land, France, Ger many, It aly, and Den mark un der their EUREFIC programme [39, 40]. It was found that there was no gen eral agree ment be tween the clas si - fi ca tion sys tems, ex cept for class A and prod ucts like plas ter board and ply wood. A com - par i son of the pro posed EUREFIC sys tem by SP with BS 476: Part 7 is sum ma rized in tab. 1. This clas si fi ca tion sys tem is open for a more de tailed as sess ment. Hav ing five classes in a new clas si fi ca tion sys tem might not be prac ti cal, but can give flex i bil ity [39]. It can dif fer en ti ate var i ous prod ucts and can there fore be re lated to var i ous ex ist ing na - tional sys tems. There is an other clas si fi ca tion sys tem based on ISO 9705 in the USA. The High Speed Craft code (HSC) was im ple mented on Jan u ary 1, 1996, as part of the Safety of Life at Sea (SOLAS) on the con struc tion of high-speed crafts by com bus ti ble ma te ri als. In that code, bulk head lin ings, com part ment lin ings, and ceil ing ma te ri als are re quired to be tested us ing ISO Ma te ri als are clas si fied into ei ther fire re strict ing or non-fire re - strict ing. A fire re strict ing ma te rial is de fined as a ma te rial hav ing low flame spread char - ac ter is tics, lim ited rate of heat re lease, and low smoke pro duc tion [41]. It should meet the ac cep tance cri te ria as pub lished in the res o lu tion MSC.40(64) [42] of the IMO [26]: 57

6 THERMAL SCIENCE: Vol. 11 (2007), No. 2, pp average heat release rate over the entire testing time shall not exceed 100 kw, maximum 30 s average heat release rate shall not exceed 500 kw, average smoke production rate shall not exceed 1.4 m/s, maximum 60 s average smoke production rate shall not exceed 8.3 m/s, no flame spread to the area below 0.5 m from the floor at a distance greater than 1.2 m from the corner, and no flaming droplets or debris may reach the floor, except in the area within 1.2 m from the corner. Table 1. Comparison of the EUREFIC system for ISO 9705 and BS 476: Part 7 [Hovde 1991, Bluhme 1991] BS 476: Part Pro posed EUREFIC clas si fi ca tion sys tem A B C D E Prod uct Remarks A1: Painted gypsum paper plasterboard A2: Melamine faced high density non-combustible board A3: Plastic faced steel sheet on mineral wool A4: FR particle board D1: Textile wallcovering on gypsum paper plasterboard D2: FR particle board type B1 D3: PVC wallcarpet on gypsum paper plasterboard Class 2 (BS 476: Part 7) if flashing and transitory flaming are taken into account E1: Ordinary plywood E2: Plastic faced steel sheet on polyurethane foam Class 3 (BS 476: Part 7) if flashing and transitory flaming are taken into account 4 U1: Combustible faced mineral wool Invalid U U2: FR extruded polystyrene foam 58

7 Chow, W. K., Leung, C. W.: Recommendation of Tests for Assessing Flame... How ever, some short com ings with the ISO 9705 stan dard were iden ti fied while car ry ing out the tests [26]. There are no clear spec i fi ca tions on the ex haust duct vol u met - ric flow rate or range of flow rates, ex cept the ex am i na tion of the ef fect of the duct flow rate (at 300 kw only) in the cal i bra tion pro ce dure. The ex haust vol u met ric flow is be - lieved to af fect the mea sure ments of heat re lease rates, es pe cially when the flow rates and heat re lease val ues are low. The heat re lease rate might give un real spikes when the duct vol u met ric flow rate is sud denly in creased. Rapid in crease in the duct flow rate is found when there is a sud den in crease in the smoke pro duc tion rate. The IMO fail ure cri te rion con cern ing the flame spread to the area be low 0.5 m is an other con cern. Tests on nine com pos ites rep re sent ing a range of fire re strict ing and non-fire re strict ing ma te ri als were car ried out with ISO The flame spread fail ure cri te rion was found to be not so rep re sen ta tive since the up per lim its of the smoke and heat re lease rate are usu ally ex ceeded be fore the flames spread to the 0.5 m level when flashover oc curs. On the other hand, the flames are usu ally con fined to the wall and ceil - ing ar eas in the im me di ate vi cin ity of the burner flame if flashover does not oc cur dur ing the 20 min. test ing pe riod. Thin wall pa per type cov er ings might be an ex cep tion as flame would spread rap idly with a small amount of heat and smoke re leased. The flame spread cri te rion was sug gested to be re-ex am ined. There are also some res er va tions on the fall ing drop lets or de bris cri te rion. In test ing paper backed tex tile wallcovering, wall pa per would be sep a rated from the sub - strate and fall to the ground. Ma te ri als show ing this be hav iour would be re garded as non-fire re strict ing. How ever, the quan tity of fall ing de bris would be very small and so flam ing would cease in a few sec onds. This would not pro duce any sig nif i cant prob lem in such ap pli ca tion. A room par tially filled with wall-cov er ing ma te ri als was sug gested to be good enough to as sess the fire-re strict ing na ture of that ma te rial. It was ob served that only the panel sec tions ad ja cent to the burner, at the top of the side walls and on the ceil ing were burnt in most cases. Re main ders of the ma te ri als did not con trib ute sig nif i cantly to the fire. Such par tially lined room would re duce the test ing cost and the amount of ma te ri als to be pro vided by man u fac tur ers. Other alternatives An agree ment was made in June 1994 on the har mo ni za tion of test pro ce dures and clas si fi ca tion sys tem for sur face lin ing ma te ri als used in build ings by the mem ber coun tries in the Eu ro pean Com mu nity. Re ac tion to fire for con struc tion prod ucts will be clas si fied with a Euroclass sys tem into six dif fer ent classes, A to F [43]. A new in ter me di - ate-scale test method de vel oped and ap proved by the Eu ro pean Com mit tee for Stan dard - iza tion in 2001 the Sin gle Burn ing Item (SBI) test was spec i fied as the main test pro ce - dure for flame spread. The test was also es tab lished as a Brit ish Stan dard BS EN [44] in In the new 2000 edi tion of the UK Build ing Reg u la tions [45], the SBI test was in cluded for as sess ing flame spread over in ter nal lin ings. Build ing prod ucts should ful fill the re quire ments of the SBI test and the Euroclasses, un less ap pro pri ate test and 59

8 THERMAL SCIENCE: Vol. 11 (2007), No. 2, pp clas si fi ca tion meth ods are yet avail able where the ex ist ing na tional test, BS 476: Part 7 can be used un til such a time that a gen er ally ac cepted guid ance is pub lished by the Of fice of the Dep uty Prime Min is ter. Since it has just been de vel oped and as sessed re cently, the ISO 9705 Room/Cor ner test was taken as a ref er ence sce nario [46] and to be used to spec - ify the lev els for the clas si fi ca tion. The Lat eral Flame Spread (LFS) is con cerned [47]. In the test, the re quire ment of up ward flame spread ex pressed as a value of length dur ing a spec i fied time (F s ) should be ful filled. In ad di tion to the SBI test, all ma te ri als should be tested by the Non-com bus ti bil ity test (EN ISO 1182) [48], the Gross Cal o rific Value test (EN ISO 1716) [49] as well as the Ig nita bil ity test (EN ISO ) [50]. Based on the re sults, ma te ri als are clas si fied into six groups from A to F [51], with class A fur ther di - vided into A1 and A2. An other set of tests and clas si fi ca tion sys tem was con sid ered [47] to be used for in te rior lin ings in or der to meet the per for mance-based build ing reg u la tions and in ter na - tional de mands for har mo ni za tion in Ja pan. The test with a cone cal o rim e ter [52] was pro posed to be used to gether with the ISO 9705 room/cor ner test by tak ing it as a ref er - ence sce nario. Three pro pos als were pro vided [47]. A re duced-scale model box (RMB) test [53] was pro posed for as sess ing fire spread. It is a 1/3 scale model of the room/cor ner test used to de ter mine the be hav iour of quasi non-com bus ti ble ma te ri als [47]. At tempts have been made to de rive the cor re la tion equa tions be tween the re sults of the RMB test and the ISO 9705 full-scale test [54]. Re sults [47] mea sured by the ISO 9705 test and the RMB test were roughly com - pared [15]: RMB: no flashover, ISO 9705: no flashover, RMB: peak heat release rate <100 kw, ISO 9705: time to flashover >10 min., RMB: peak heat release rate >200 kw, and ISO 9705: time to flashover <5 min. Fur ther stud ies on ad di tional pa ram e ters are re quired. Comparison with ASTM E84/NFPA 255 ASTM E84/NFPA 255 tunnel test is widely ac cepted and the FSI rat ing sys - tem is readily un der stood in the USA. How ever, it was not de signed as a stand-alone test to de scribe the com bus ti bil ity of a ma te rial [55], but rather, to eval u ate the FSI of in te rior fin ishes and to com pare the per for mance of dif fer ent ma te ri als with the bench mark ma te - ri als with out nec es sar ily be in dic a tive of the be hav iour of the ma te rial in its end-use ap - pli ca tion. At the same time, de sign ers use a wide range of ma te ri als for in te rior fin ishes and dec o ra tion in the mod ern build ing de signs now a days. In some cases, par tic u larly those com pos ite and dec o ra tive ma te ri als with un usual ge om e tries or con fig u ra tions, the tun nel test is in ap pro pri ate [56]. It was found that al though some wallcoverings may achieve a low flame-spread rat ing in the ASTM E84 test con di tions, the ma te ri als may ac tu ally spread flame readily 60

9 Chow, W. K., Leung, C. W.: Recommendation of Tests for Assessing Flame... and cause fires to grow quickly in the larger-scale and more re al is tic room/cor ner test [57]. In the 1989 At lanta high-rise of fice build ing fire, an elec tri cal arc ig nited mul ti ple lay ers of wallcoverings and fire rap idly spread in the exit cor ri dor. Al though the listed rat ings of the in di vid ual wallcoverings were ac cept able for the oc cu pancy, lo ca tions, and exit cor ri dors, the com bi na tion of ma te ri als per formed much dif fer ently in fire con di tions [58]. The U. S. Build ing codes place some re stric tions on in te rior fin ishes in build ings by lim it ing the flame spread rat ing of a ma te rial per mit ted for a given type of con struc tion and oc cu pancy. Tex tile wallcoverings in stalled in an unsprinklered oc cu pancy are re - quired by Model build ing codes [59-61] and the Life Safety Code (LSC) [62] to be tested in a room/cor ner fire test [32]. Com plete as sem bly with all lay ers and sub strates are re - quired by the LSC to be tested. How ever, the ASTM E84 test method is not de signed to test wall as sem blies. Some mod i fi ca tions of ex tend ing the skirt on the re mov able tun nel lid were re quired [32] to en sure an ad e quate seal in the liq uid trough and pre vent the es - cap ing of heat and smoke, so that the tun nel test can be used for in te rior fin ish ma te ri als on their in tended sub strate and wall as sem bly. Con sid er ing the test ing po si tions of ma te ri als, the Na tional Stan dard of Can ada for build ing ma te ri als also em ploys a sim i lar tun nel test [63, 64], in which the build ing ma te ri als can be mounted on the ceil ing or floor de pend ing on their phys i cal (whether it can sup port its own weight) and burn ing (melt ing or drip ping) char ac ter is tics. How ever, for ASTM E84, sam ples can only be tested in the ceil ing-mounted po si tion [65]. As re ported, car ry ing out an ASTM E84/NFPA 255 tun nel test is less ex pen - sive than the ISO 9705 room cor ner fire test in the USA [32]. But if a new rig is to be de - vel oped, there might not be much dif fer ence in the ini tial in stal la tion cost, staff train ing cost, op er a tion and main te nance cost. Bear ing in mind that there might be nu mer ous new ma te ri als to be de vel oped in China. De vel op ing such a test ing fa cil ity should be a good in vest ment. ASTM E1321 together with the cone calorimeter As dis cussed, ISO 9705 is rel a tively ex pen sive to run and only one sce nario can be mod eled at a time. Large quan ti ties of test ing ma te ri als are re quired. Re sults of the cone cal o rim e ter have been widely proved to be cor re lated with the ISO 9705 re sults. For de vel op ing new ma te ri als, it can be used to gether with the LIFT ap pa ra tus (ASTM E1321) to pre dict the full-scale flame spread of ma te ri als. An em pir i cal re la tion ship was de vel oped [66] for pre dict ing the time to flashover in the full-scale room/cor ner tests for the sur face lin ing ma te ri als. That was fur - ther mod i fied [67] to give some new cor re la tions based on the heat re lease rate mea sure - ments in the cone cal o rim e ter. The time to flashover was ex pressed [68] as a func tion of the ma te rial pa ram e ters from the cone cal o rim e ter and LIFT. Stud ies have also been car ried out by the USA [26, 69] to as sess the abil ity of the small-scale test re sults to pre dict the full-scale fire per for mance of com part ment lin ings in room/cor ner con fig u ra tions. A pro posal was sub mit ted to IMO by Fin land on qual i fy - 61

10 THERMAL SCIENCE: Vol. 11 (2007), No. 2, pp ing the fire-re strict ing ma te ri als on the ba sis of the cone cal o rim e ter data, in which it showed that a max i mum 60 s slid ing av er age heat re lease rate of 60 kw/m 2 or less in the cone cal o rim e ter at a heat flux of 50 kw/m 2 would be equiv a lent to meet the ISO room test cri te ria for heat re lease rate [26]. This pro posal was as sessed and found to be fea si ble to de velop ac cu rate ac cep tance cri te ria from the ISO 5660 cone cal o rim e ter test. Their over all eval u a tion sug gested that us ing the cone cal o rim e ter re sults, LIFT test re sults and math e mat i cal mod el ing re sults can pre dict the ISO 9705 re sults; and the re al is tic sce nar - ios in clud ing dif fer ent ig ni tion sources, ceil ing heights, room sizes, and ven ti la tion rates. It was pro posed [70] to use the cone cal o rim e ter and LIFT re sults to sim u late the be hav iours of ma te ri als in the ISO 9705 test us ing the SPREAD al go rithm [71, 72]. Mod - i fi ca tions were made to change the side wall al go rithm to a cor ner one. Re sults il lus trated that the heat re lease rate curves from the model are sim i lar to those mea sured from the ex - per i ments. This type of model has been proved to have po ten tial for prac ti cal ap pli ca tion. Conclusions As re ported ear lier [15, 33] on com par ing with three other tests, ISO 9705 is rec - om mended for test ing the flame spread of ma te ri als as ex plained in this pa per. In con trast to over seas, the lo cal gov ern ment can draft codes on flame spread with out wor ry ing about dis turb ing the lo cal test ing fa cil i ties. Tak ing ISO 9705 as the first choice might give good data. On rank ing ma te ri als from ISO 9705, there are clas si fi ca tion sys tems pro posed in other coun tries. How ever, only the IMO ac cep tance cri te ria [42] are for mally adopted in the HSC, though some de ci sive con di tions were not agreed on [26]. Other sys tems are still open for fur ther anal y sis and im prove ment. The lo cal gov ern ment may take those over seas ex pe ri ences as ref er ences. Fur ther, in-depth stud ies on a wider range of prod ucts are rec om mended. The re sults should be com pared with BS 476: Part 7 [6] and the other gen er ally rec og nized rank ings [e. g. 34, 38-40] in or der to de velop the most suit able clas - si fi ca tion sys tem for the lo cal con struc tion in dus try. There are al ways ar gu ments say ing that ISO 9705 is ex pen sive to be car ried out. How ever, in com par ing with the lo cal cost of build ings, a res i den tial build ing at mid-lev - els would still cost HK$ 5,000 (US$ 600) per square foot. Peo ple can not ar gue on the cost for such a test. Fur ther, higher ed u ca tion in sti tutes would serve the in dus try by test ing these ma te ri als and com po nents at rea son able costs to help pro vid ing fire safety. How - ever, smoke emis sions might bring about en vi ron men tal prob lems. De vel op ing such a full-scale burn ing fa cil ity in a re mote area is strongly rec om mended. But that does not im ply that in stall ing a smoke treat ment plant is un nec es sary. Other tests with SBI [45], RMB [47], and com bin ing the LIFT test [18] with the cone cal o rim e ter re sults [52] might also be con sid ered in per for mance-based de sign. Those tests might be used as al ter na tives in later stages of de vel op ment. 62

11 Chow, W. K., Leung, C. W.: Recommendation of Tests for Assessing Flame... References [1] ***, Code of Prac tice for the Pro vi sion of Means of Ac cess for Firefighting and Res cue Pur - pose, Build ing Au thor ity, Hong Kong Spe cial Ad min is tra tive Re gion, 2004 [2] ***, Code of Prac tice for the Pro vi sion of Means of Es cape in Case of Fire, Build ing Au thor - ity, Hong Kong, 1996 [3] ***, Code of Prac tice for Fire Re sist ing Con struc tion, Build ings De part ment, Hong Kong, 1996 [4] Bukowski, R. W., Babrauskas, V., Developing Rational, Performance-Based Fire Safety Require ments in Model Build ing Codes, Fire and Ma te ri als, 19 (1994), 2, pp [5] ***, Code of Prac tice for Min i mum Fire Ser vice In stal la tions and Equip ment and In spec tion and Test ing of In stal la tions and Equip ment, Fire Ser vices De part ment, Hong Kong, 1998 [6] ***, Fire Tests on Build ing Ma te ri als and Struc ture, Part 7. Method of Test to De ter mine the Clas si fi ca tion of the Sur face Spread of Flame of Prod ucts, BS476: Part 7: 1997, Brit ish Stan - dards In sti tu tion, Lon don, 1997 [7] Chow, W. K., Fire Haz ard As sess ment on Poly ure thane Sand wich Pan els for Tem po rary Ac - com mo da tion Units, Poly mer Test ing, 23 (2004), 8, pp [8] ***, Guide to Fire En gi neer ing Ap proach, Prac tice Note for Au tho rized Per sons and Reg is - tered Struc tural En gi neers PNAP 204, Build ings De part ment, Hong Kong, 1998 [9] Chow, W. K., Fire Safety in Green or Sus tain able Build ings: Ap pli ca tion of the Fire En gi - neer ing Ap proach in Hong Kong, Architectural Science Review, 46 (2003), 3, pp [10] ***, Ap pli ca tion of Fire Safety En gi neer ing Prin ci ples to the De sign of Build ings, BS7974: 2001, Brit ish Stan dards In sti tu tion, Lon don, 2001 [11] Babrauskas, V., Grayson, S. J., Heat Re lease in Fires, Elsevier Ap plied Sci ence, Lon don, 1992 [12] Cleary, T. G., Quintiere, J. G., A Frame work for Uti liz ing Fire Prop erty Tests, Pro ceed ings (Eds. G. Cox, B. Lang ford), 3 rd In ter na tional Sym po sium, Fire Safety Sci ence, Ed in burgh Scot land, UK, 1991, pp [13] ***, Leg is la tive Coun cil, Karaoke Es tab lish ments Bill. Press Re lease, Hong Kong Spe cial Ad min is tra tive Re gion Gov ern ment, 2000 [14] Chow, W. K., Lui Gigi, C. H., A Fire Safety Rank ing Sys tem for Karaoke Es tab lish ments in Hong Kong, Jour nal of Fire Sci ences, 19 (2001), 2, pp [15] Leung, C. W., Fire Spread ing of Ma te ri als and Its Con tri bu tion to Com part ment Fire, Ph. D. the sis, The Hong Kong Poly tech nic Uni ver sity, Hong Kong, 2006 [16] Chow, W. K., Han, S. S., Chow, C. L., Fire Be hav iour of Tim ber Prod ucts with and with out Fire Retardant, Jour nal of Ap plied Fire Sci ence, 11 ( ), 3, pp [17] Hasemi, Y., Yoshida, M., Hayashi, Y., Yabuta, T., Correlations between Bench-Scale Test and Room Cor ner Test Based on a Flame Spread Mod el ling: In ter pre ta tion of Small and In - termediate Scale Fire Testing, NISTIR 6588, 15 th Meet ing of the US-Ja pan Na tional Re - search, Panel on Fire Re search and Safety, San An to nio, Tex., USA, 2000, Vol. 1, pp [18] ***, Stan dard Test Method for De ter min ing Ma te rial Ig ni tion and Flame Spread Prop er ties, ASTM E a, Amer i can So ci ety for Test ing and Ma te ri als, West Conshohocken, Pa., USA, 1997 [19] ***, Stan dard Test Method for Sur face Burn ing Char ac ter is tics of Build ing Ma te ri als, ASTM E84-99, Amer i can So ci ety for Test ing and Ma te ri als, West Conshohocken, Pa., USA, 1999 [20] ***, Stan dard Method of Test of Sur face Burn ing Char ac ter is tics of Build ing Ma te ri als, NFPA 255: 2000, Na tional Fire Pro tec tion As so ci a tion, Quincy, Ma., USA, 2000 [21] ***, Fire Tests Full-Scale Room Test for Sur face Prod ucts, ISO 9705: 1993(E), In ter na - tional Stan dards Or ga ni za tion, Geneva, Swit zer land, 1996 [22] Babrauskas, V., Wickström, U., The Ra tio nal De vel op ment of Bench-Scale Fire Tests for Full-scale Fire Pre dic tion, Pro ceed ings, 2 nd In ter na tional Sym po sium on Fire Safety Sci ence, Hemi sphere Pub lish ing Comp., New York, USA, 1989, pp

12 THERMAL SCIENCE: Vol. 11 (2007), No. 2, pp [23] Walker, A.G., Flammability Test ing in Eu rope, Fire and Ma te ri als, 4 (1980), 3, pp [24] Murrell, J. M., Barnfield, J. R., The Role and Ap pli ca tion of Stan dard Tests, Fire and Ma te ri - als, 13 (1988), 1, pp [25] Allender, P. J., Spec i fi ca tion of Com bus ti ble Ma te ri als for Pas sen ger Train De sign, Fire and Ma te ri als, 8 (1984), 3, pp [26] Janssens, M. L., Garabedian A., Gray, W., Es tab lish ment of In ter na tional Stan dards Or ga ni - za tion (ISO) 5660 Ac cep tance Cri te ria for Fire Re strict ing Ma te ri als Used on High Speed Craft. Fi nal Re port, No. CG-D-22-98, U. S. Coast Guard Re search and De vel op ment Cen ter, South west Re search In sti tute, San An to nio, Tex., USA, 1998 [27] ***, Rec om men da tion on Im proved Fire Test Pro ce dure for Sur face Flammability of Bulk - head, Ceil ing, and Sur face Fin ish Ma te ri als, IMO Res o lu tion A 653(16), In ter na tional Mar i - time Or ga ni za tion, Lon don, Re pro duced in the FTP Code, pp [28] ***, Stan dard Test Method for Flammability of Ma rine Sur face Fin ishes, ASTM E a, American Society for Testing and Materials, West Conshohocken, Ma., USA, 1997 [29] Dowling, V. P., McArthur, N. A., Webb, A. K., Leon ard, J. E., Blackmore, J. M., Large-Scale Fire Tests on Three Build ing Ma te ri als, Pro ceed ings, 3 rd In ter na tional Con fer ence on Fire Re search and En gi neer ing, So ci ety of Fire Pro tec tion En gi neers, Chi cago, Ill., USA, 1999, pp [30] Bhatnagar, S. K., Varshney, B. S., Mohanty, B., An Ap praisal of Stan dard Meth ods for De - termination of Flame Spread Be hav iour of Materials, Fire and Ma te ri als, 16 (1992), 3, pp [31] Quintiere, J., Harkeroad, M., New Con cepts for Mea sur ing Flame Spread Prop er ties, Fire Safety Sci ence and En gi neer ing, ASTM STP 882 (Ed. T. Z. Harmanthy), 1985, pp [32] Dyar, C. G., Boser, R., Test ing Low-Flame-Spread Wallcoverings on Typ i cal Con struc tion Sub strates in ASTM E84, Fire Tech nol ogy, 32 (1996), 3, pp [33] Chow, W. K., Han, S. S., Su per po si tion of Heat Re lease Rate Curves for Com bus ti bles with Bench-Scale Tests, Poly mer Test ing, 25 (2006), 1, pp [34] Babrauskas, V., Sand wich Panel Per for mance in Full-Scale and Bench-Scale Fire Tests, Fire and Ma te ri als, 21 (1997), 2, pp [35] Chow, W. K., Based Fire Codes, In ter na tional Jour nal on En gi neer ing Per for mance (Special is sue on full-scale burn ing tests), 6 (2004), 3, pp [36] Gardner, W. D., Thomson, C. R., Flame Spread Prop er ties of For est Prod ucts Com par i son and Val i da tion of Pre scribed Aus tra lian and North Amer i can Flame Spread Test Meth ods, Fire and Ma te ri als, 12 (1988), 2, pp [37] Hovde, P. J., Com par i son be tween Nordtest and ISO Fire Test Meth ods. Eval u a tion of Clas - si fi ca tion Sys tems, EUREFIC Sem i nar Pro ceed ings, Co pen ha gen, 1991, Interscience Com - mu ni ca tions, Lon don, pp [38] Sundström, B., Goranssön, U., Pos si ble Clas si fi ca tion Cri te ria and Their Im pli ca tions for Sur face Ma te ri als Tested in Full Scale Ac cord ing to ISO DP 9705 or NT Fire 025, SP Re port 1988: 19, Fire Tech nol ogy, SP, Swed ish Na tional Test ing In sti tute, Swe den, 1988 [39] Sundström, B., Clas si fi ca tion of Wall and Ceil ing Lin ings, EUREFIC Sem i nar Pro ceed ings, Co pen ha gen, 1991, Interscience Com mu ni ca tions, Lon don, pp [40] Bluhme, D. A., Cor re la tion of Test Re sults with Other Eu ro pean Test Meth ods, EUREFIC Sem i nar Pro ceed ings, Co pen ha gen, 1991, Interscience Com mu ni ca tions, Lon don, pp [41] Grenier, A. T., Janssens, M. L., Nash, L., Developing Cone Calorimeter Acceptance Criteria for Ma te ri als Used in High Speed Craft, Fire and Ma te ri als, 24 (2000), 1, pp [42] ***, Stan dard for Qual i fy ing Ma rine Ma te ri als for High Speed Craft as Fire-Re sist ing Ma te - ri als, Res o lu tion MSC.40(64), In ter na tional Mar i time Or ga ni za tion, Lon don, 1994 [43] ***, Eu ro pean Com mis sion, Com mis sion De ci sion of Sep tem ber 9, 1994, Im ple ment ing Ar - ti cle 20 of Di rec tive 89/106/EEC on Con struc tion Prod ucts, Of fi cial Jour nal of the Eu ro pean Com mu ni ties, No. 6, Brussels,

13 Chow, W. K., Leung, C. W.: Recommendation of Tests for Assessing Flame... [44] ***, Re ac tion to Fire Tests for Build ing Prod ucts. Build ing Prod ucts Ex clud ing Floor ings Ex posed to the Ther mal At tack by a Sin gle Burn ing Item, BS EN 13823:2002, Brit ish Stan - dards In sti tu tion, Lon don, 2002 [45] ***, The Build ing Reg u la tions Ap proved Doc u ment B: Fire Safety, Of fice of the Dep - uty Prime Min is ter, London, 2000 [46] ***, Eu ro pean Com mis sion, Up date No. 9 on Prog ress Re gard ing Fire Re lated Is sues, Doc u - ment, Brussels, 1998 [47] Hedskog, B., Ryber, F., The Classification Systems for Surface Lining Materials Used in Build ings in Eu rope and Ja pan A Sum mary and Com par i son, Re port 5023, De part ment of Fire Safety En gi neer ing, Lund In sti tute of Tech nol ogy, Lund Uni ver sity, Lund, Swe - den,1998 [48] ***, Re ac tion to Fire Tests for Build ing Prod ucts. Non-Com bus ti bil ity Test, BS EN ISO1182: 2002, Brit ish Stan dards In sti tu tion, Lon don, 2002 [49] ***, Re ac tion to Fire Tests for Build ing Prod ucts. De ter mi na tion of the Heat of Com bus tion, BS EN 1716: 2002, Brit ish Stan dards In sti tu tion, Lon don, 2002 [50] ***, Re ac tion to Fire Tests. Ig nita bil ity of Build ing Prod ucts Sub jected to Di rect Im pinge - ment of Flame. Sin gle-flame Source Test, BS EN ISO :2002, Brit ish Stan dards In sti - tu tion, Lon don, 2002 [51] ***, Fire Clas si fi ca tion of Con struc tion Prod ucts and Build ing El e ments. Part 1: Clas si fi ca - tion Us ing Test Data from Re ac tion to Fire Tests, BS EN : 2002, Brit ish Stan dards In sti tu tion, Lon don, 2002 [52] ***, Fire Tests Re ac tion to Fire Part 1: Rate of Heat Re lease from Build ing Prod ucts (Cone Cal o rim e ter Method), ISO : 1993, In ter na tional Stan dards Or ga ni za tion, Geneva, Swit zer land, 1993 [53] ***, Cer tif i ca tion of Quasi-Noncombustible Ma te ri als and Fire Re tar dant Ma te ri als, No ti fi - ca tion No of the Min is try of Con struc tion, 3 rd ed., The Build ing Cen ter of Ja pan, To - kyo, 1994 [54] Yoshida, M., Hasemi, Y., Tanaike, Y., Tasaka, S., Saito, F., Fangrat, J., Com par a tive Stud ies in the Re duced-scale Model Box and the Room Cor ner Test, Pro ceed ings, 8 th In ter na tional Interflam Con fer ence, Ed in burgh, Scot land, UK, 1999, 1, pp [55] ***, Stan dard Test Method for Sur face Burn ing Char ac ter is tics of Build ing Ma te ri als (E84-91a), An nual Book of ASTM Stan dards, Sec tion 4, Part 7, Amer i can So ci ety for Test - ing Ma te ri als, Phil a del phia, Pa., USA, 1993, pp [56] Torvi, D. A., Lougheed, G. D., Car pen ter, D. W., Senez, P. L., Locke, H. A., Eval u at ing Ma - te ri als and Fire Pro tec tion Sys tems Us ing Full-scale Fire Tests, Pro ceed ings, 3 rd In ter na tional Con fer ence on Fire Re search and En gi neer ing, So ci ety of Fire Pro tec tion En gi neers, Chi - cago, Ill., USA, 1999, pp [57] Belles, D. W., Fisher, F. L., Wil liam son, R. B., How Well Does ASTM E84 Pre dict Fire Per - for mance of Tex tiles and Wallcoverings, Fire Jour nal, 84 (1988), 1, pp , 29-30, 74 [58] Isner, M. S., Five Die in High-Rise Of fice Build ing Fire, Fire Jour nal, 84 (1990), 4, pp [59] ***, Stan dard Build ing Code (SBCC), South ern Build ing Code Con gress In ter na tional, Bir - ming ham, Al., USA, 1997 [60] ***, Uni form Build ing Code, The In ter na tional Con fer ence of Build ing Of fi cials, Whittier, Cal., USA, 1997 [61] ***, BOCA Na tional Build ing Code, The Build ing Of fi cials and Code Ad min is tra tors In ter - na tional, Inc., Coun try Club Hills, Ill., USA, 1999 [62] ***, Life Safety Code Hand book, Na tional Fire Pro tec tion As so ci a tion, Quincy, Ma., USA, 1991 [63] ***, Stan dard Method of Test for Sur face Burn ing Char ac ter is tics of Build ing Ma te ri als and Assemblies, ULC Standard CAN4-S102-M83, Underwriters Laboratories of Canada, Scarborough, Ont., Can ada,

14 THERMAL SCIENCE: Vol. 11 (2007), No. 2, pp [64] ***, Stan dard Method of Test for Sur face Burn ing Char ac ter is tics of Floor ing, Floor Cov er - ing, and Miscellaneous Materials and Assemblies, ULC Stan dard CAN4-S102.2-M83, Un - derwriters Laboratories of Canada, Scarborough, Ont., Canada, 1983 [65] Mehaffey, J. R., Flammability of Build ing Ma te ri als and Fire Growth, Pro ceed ings, Build ing Sci ence In sight '87, In sti tute for Re search in Con struc tion, Na tional Re search Coun cil Can - ada, Ot tawa, NRCC 2827, 1987, pp [66] Ostman, B., Nussbaum, R., Cor re la tion be tween Small-Scale Rate of Heat Re lease and Full-Scale Room Flashover for Sur face Lin ings, Pro ceed ings (Ed. Wakamatsu et al.), 2 nd In - ternational Symposium on Fire Safety Science, International Association of Fire Safety Sci - ence (IAFSS), To kyo, Hemi sphere Pub lish ing Comp., New York, USA, 1987, pp [67] Ostman, B., Tsantaridis, L., Cor re la tion be tween Cone Cal o rim e ter Data and Time to Flashover in the Room Fire Test, Fire and Ma te ri als, 18 (1994), 4, pp [68] Karlsson, B., Mod el ing Fire Growth on Com bus ti ble Lin ing Ma te ri als in En clo sures, Re port TVB-1009, De part ment of Fire Safety En gi neer ing, Lund Uni ver sity, Lund, Swe den, 1992 [69] Beyler, C. L., Hunt, S. P., Lattimer, B. Y., Iqbal, N., Lantenberger, C., Pre dic tion of ISO 9705 Room/Cor ner Test Re sults, Vol ume I, Fi nal Re port No. CG-D-22-99, U. S. Coast Guard Re - search and De vel op ment Cen ter, Groton, Ct., USA, 1999 [70] Wright, M. T., Barnett, J. R., Dembsey, N. A., Flame Spread on Cored Com pos ite Pan els for Use in High Speed Craft, Pro ceed ings, 3 rd In ter na tional Con fer ence on Fire Re search and En - gi neer ing, So ci ety of Fire Pro tec tion En gi neers, Chi cago, Ill., USA, 1999, pp [71] Mitler, H., Steckler, K., Spread A Model of Flame Spread on Ver ti cal Sur faces, Tech ni cal Re port NISTIR 5619, De part ment of Com merce, Na tional In sti tute of Stan dards and Tech - nol ogy, Gaithersburg, Md., USA (writ ten 1993, is sued 1995) [73] Mitler, H., An Al go rithm to De scribe the Speed of a Wall Fire Un der a Ceil ing, Tech ni cal Re - port NISTIR 5547, De part ment of Com merce, Na tional In sti tute of Stan dards and Tech nol - ogy, Gaithersburg, Md., USA, 1994 Authors addresses: W. K. Chow, C. W. Leung Research Centre for Fire Engineering, Department of Building Services Engineering, Area of Strength: Fire Safety Engineering, The Hong Kong Polytechnic University Hong Kong, China Corresponding author W. K. Chow bewkchow@polyu.edu.hk Paper submitted: December 15,.2005 Paper revised: January 15, 2006 Paper accepted: February 20,