ISO TR 5660-3:2003 pdf free download.Reaction-to-fire tests — Heat release, smoke production and mass loss rate — Part 3: Guidance on measurement.
Introduction
ISO 5660-I. describing a test method for rate of heal release from building products by means of a cone calorimeter, was pubtished in 1993 (first edition) follemng approximately ten years of development within ISOITC 92. Fire safety, Subcommittee SCI, Fire inifiation and growth.
The cone calorimeter is a tire test instrument in which horizontal specimens are exposed to controlled levels of radent heating by means of a truncated cone-shaped heater. Continuous spark ignition is provided and the time to Ignition is recorded for specimens wiulch Ignite. The rate of heat release from the burning specimen Is delermmed from measurements of the amount of oxygen consumed from the air flowing through the apparatus, which has been demonstrated to equate to heat release. The mass of the specimen is also measured throu’iout the burning period. The specimens are usually tested under wel ventilated conditions.
Results are expressed In terms of peak and average rates of heat release as well as total heat released and the effective net heat of combustion. iSO 5660-1 limits the specimen type to essentially flat. Several other groups are now utlltzing the cone calorimeter and a number of new parameters m addition 10 those defined In ISO 5660-1 and ISO 5660-2(°1 have been defined and used Some of these, induding smoke measurement. require that measurements be made from the beginning of the test rather than at the onset of ignition which Is commonly used as the starling point for heat release measurement.
The cone caloiTneter is also designed to allow measurement of smoke and gases such as CO and CO2. Smoke measurement is the subject of ISO 5660-2°1. Further work is under way to define a quality control tool (or measuring burning rates of building products, This will be published as ISO 175541221 and Is based on mess loss measurements using the thermal decomposition model of the cone calorimeter. A similar system wtilch measures the temperature of combustion products generated by this apparatus and has been standardized by ISO(C 61 as ISO 13927tI. The cone calorimeter fire model has also been used in a con’oslvlly International Standard developed by ISOITC 61 as ISO ll9O7.411. Th effect of the evolved gases on the resistance of a printed circuit board target is used to assess corrosivity.
During development of the cone calorimeter It became apparent that there was considerable interest in the use of the instrument for products other than building products. Several standards have been developed by various national and international groups based on ISO 5660-1 and ISO 5660-2,
This part of ISO 5660 provides recommendations for the testing of products in the cone calorimeter and gives guidance on the presentation of the results. Supplementary guidance Is given In doc1iments referred to In References (1) and (2).
I Scope
This part ci ISO 5660 examines the limitations of the cone calorimeter as claTently used for building products and onwnentis ways ii ehiCh some of these may be overcome for other tes ci products for other application areas. It compdes information from a large body of experience with regard to the use 0i the matrument Into a set of guidelmes WhICh will help to standardize the use of the cone calorimeter in this wider scope.
Pavlcular guidance is given on aspects of specimen preparation arid on the behaviour, such as melting. spaling and intumescaig, of specimens exposed to radiant heat. The relevance of specimen thickness and the use of substrate, and methods of fixing to substrate, are also discussed, Advice is given on approaches to testing a variety of non-slandard products. Recommendations are made on techniques of calibration of the apparatus, selection of appropriate heat flux levels and Ignition protocols.
In addition to the guidance given to operators, the document mattes recommendations on presentation of the lest results,
2 Normative references
The following referenced documents are indispensable for the application of this document. For dated references, only the edition cited applies, For undated references, the latest edition of the referenced document (including any amendments) applies.
ISO 5660-12002, Reacitoo-to-flre tests — Heat ,eJease smoke production and mass loss rate — Part 1: Heat ,elsasa tate (cone calorimeter method)
3 Capability and limitations of the cone calorimeter
Rate of heat release is one of the fixidamental properties of fire and should eWnost always be taken into account in any assessment of fire hazard Heat release snlflcantly affects fire ow1h. Considerable progress has been made in methods of using rate of heat release and ignition time results from the cone calorimeter to predict full scale fire characteristics such as this to flashover In a small room lined with the tested product and exposed to a high energy fire source such as that used in ISO 97O511
The design of the Instrument also provides for measurement of smoke (both gravimetrically and optically) and other gaseous products of pyrolysis or combustion. The instrument may thus be applied to the assessment of real fire hazards such as smoke and toxic and corrosive gas emission in addition to heat release, particularly when the results are expressed in terms of fundamental physically-based rather than ad hoc parameters.
When functioning purely as a rate of heat release apparatus the parameter which Is measured In the plume from the specimen is the concentration of oxygen. Temperature measurements are made, but these are not used to measure the heal output from the specimen In the manner of a conventional calorimeter, This is a crucial point in understanding heat release by oxygen CoflsLaTiptiofl calonmetry. The theory of oxygen consumption calorimetry is discussed m more detail in Clause 11,
The instrument limited to bench scale specimens and it uses a simple Ike model which provides continuous free ventilation and removal of the products of partial and complete combustion. Specimen behaviour during the expenment such as shflnking and swelling can be tolerated if this happens within smal margins, but if the specimen Intumesces so that it touches the igniter or the cone, or if it exhibits spelling, this behaviour will invalidate the resulls generated.
4 CalibratIon of the cone calorimeter
4.1 General
Regular and accurate calibration of several measuring devices is essentat in order for valid results to be obtained from the cone calorimeter. The blowing calibration procedures are outlined in ISO 5660-12002, Clause 10 (respectively 10.1 to 10.3):
— preliminary calibration;
— operating calibration:
— less frequent calibrations.
Table 1 gIves details of the m*r calibration requirements together with recommended intervals.
Calibration procedures are to some extent controlled by the apparatus and the comments below may not apply to all makes of cone calorimeter.
Some guidelines are given on actual operating experiences with these calibrations and follow the clause headings given en ISO 5660-1. In addition there are some additional comments on low orifice calibration factors and the cause thereof. The clause numbers ii parentheses refer to clauses given in ISO 5660-1:2002.
4.2 Heat flux meter calibration (see 6.12 and 10.3.1 of ISO 5660-1:2002)
Early work on cross calibration of the heat flux meters showed some differences in the calibration constant as given by the supplying company. Ills therefore very wise to attempt some cross calibration with another cone caiorimeter user.
Greet care should be taken of the heel flux meter which is in regular use and care should be taken to use this
always with water coding. It should be checked regularly agatnst a primary mater as set out in Annex E of
ISO 5660-1:2002 to ensure Its continued correct waning. Ills false economy to omit this regular check.
ISOITS 14934 gives guidance on heat flux meter calibration.
4.3 Heater calibration (see 10.12 and 10,2.5 of ISO 5660-1:2002)
The setting of the required heat flux is set out in the manuals of the different instruments Once a steady slate value has been obtained (emal fluctuations of the order of ± 0,1 ‘C may occur) this value should be noted for future reference and act as an early warning of some change. In parhcuiar, users should ensure that the control thermocouples which should be situated behind and touching the heater helix (i.e. the face remote from the specinen) do not penetrate the heater helix and experience the temperature of the flame rather than that of the heater winding.
4.5 Determining orifice plate calibration factor
4.5.1 CalIbration using methane (see 10.2.4 of ISO 5660-1:2002)
Ills recommended thai the calibration cons*sbng of burning methane be carried out when the healer has been set at the required heal flux. This allows the differential pressure transducer (DPT) to warm up. The fan is shut down and the DPT re-adjusted to zero. The fan is then set to the required air flow and then the burmng of methane is carried out.
ISO 5660-1 requIres that at the start of each day, one heat release calibration corresponding to a heat release flow rate of 5 kW of the supplied methane be carried out. An critic. constant between 0.040 and 0.046 should be obtained with 999% or 99,5 % methane at a flow rate of 8 lflThn referenced to standard temperature and pressure, or 0,1 gte (6 glmin). Daily calibration factors should agree within approximately 1 %.
It should be noted that the heat release calibration using methane does not constitute an absolute calibration of the instrument, but rather that 4 verIfies the orifice plate constant, which appears In the calculations [see Equation (5) in 12.1. Equation (7) in 12.3.2 and Equation (9) in 12.4 of ISO 5660-1:2002], it is not a direct measurement of heat release,
Black polyrnethy1me4hacr4ate (PMMA) (with a thickness of 6mm or greater) can also be used within ascii laboratory to check repeatability of the cone calorimeter performance,
When zeroing the differential pressure transducer (DPT). ensure that the duct fan and any decoupled’ extractor system are switched off. Air should be prevented from flowing over the open end of the stack and across the orifice plate. if necessary, a plastic bag or equivalent should be used to block the open end of the slack.
It is important to keep records of the values of X (oxygen analyser reading, mole fraction of oxygen). T (absolute temperature of gas at the orifice meter) and p (orifice meter pressure differential) which lead to good calibration factors which should also be noted every time the calibration is carried out In this way any discrepancy is knmedaatety identified and early correction can be carned out.
4.5.2 Calibration using liquids
II should b. noted that Mien calibrating using lku.ds, Which usually have low flash points, Ills essential that calibrations be performed on a cold system (the cone heater is not powered). The liquid should be held in a stable vessel, and the vessel should be stable under the cone before ignition of the liquid, The burning liquid should not be disturbed until it is all bunied.
In addition to burning methane for calibration, users have used a variety of materials such as alcohols. The heats of combustion of ethanol and piopan-2-ol ate 26,8 kJIg arid 30.2 kJIg, respectively. It is desirable to use propen-2-ol with a purity 99,5%.
4.6 Weighing device calibration (see 10.1.3, 10.1.4 and 10.2.3 of ISO 5660.1:2002)
The weighing device is simple and generally problem free, Foaow the manufac*ure?s instructions on recommendations for talvig, calibrating and selling the damping of the weighing device.
4.7 AdditIonal comments on the orifice calibration factor
Some variation of the orifice plate calibration factor (also known as the methane calibration factor) may be observed for various reasons. However, any large (5 %) changes in values are ir cative of malfunction .ithe system. In the rnaority of cases, the problem Is caused by leaks into the sampling lines, In which case the recorded factor will be higher than usual. Other items that can cause problems are
— blockages in the gas sampling line.
connections between the orifice plate and the differential pressure transducer,
— leaks at the methane supply line,
faulty defterential pressure transducer, or
— faulty methane how meter.
The most common cause foe- this is some leakage in the system However, In one instance a user encountered low values of the order indicated above. After extensive testing Indicating that no leeks were present In the system and everything else was functioning correctly, the refrigeration system was suspected and it was flushed with propanone, As the effluent was heavily discoloured. it was thought that the instability was caused by a severely clogged refrigeration column. Subsequent is-calibration gave a satislaclory calibration factor. A low calibration factor may also be the result of inactive CO2 removal agent If CO2 le not removed from the gas stream entering the oxygen analyser, the heat release determined using the standard equations will be higher than expected, hence the calibration factor will be lower.
4.8 Calibration of smoke measurement system
Calibration using filters assumes that the system used to calibrate the fitter is superior to the optical system in the cone calorimeter. The photodiodes used In the cone calorimeter specify a high degree of linearity. The optical density quoted for a commercialy supplied filter is usuaky the average over a range of wavelengths and the value at the frequency of the monoctvomalic laser used In the cone may not be this average value. Therefore, the use of the filter is better confined for deity routine checkng of the proper functioning of the system rather than as a primary calibration,
The user may therefore calibrate by diecking zero and 100% values and utilizing the linearity of the photodlode.
If filters calibrated at the correct wavelength are used, the following routine may be used. The smoke measurement system shoeid be checked weekly using neutral density glass filters of 0,3 nominal optical density. This procedure assumes that the smoke system is the conventional split beam laser descnbed in
ISO 566O.2l.
Place the filter in the beam between the duct and the detector. Collect data for a period of 60 s The measured calibration extinction coefficient, k, is obtained from the equation:
= lre(10t1)id
where d is the duct diameter.
The correct k vakee, is given by the equation:
• 2,303D’ld
where D’ is the optical density of the calibration filter.
A correction factor.!, is calculated from these two k values and is used to correct all subsequent measured k values
The filter used should be of the doped type because coated filters can give nsa to interference effects with laser light and can deteriorate with time. The filter should have a reliable catibration covenng the wavelength of the measurement
4.9 Precautions in relation to wat.rICO2 removal
Where carbon monoxide and carbon dioxide are analysed in the gas stream il Is Important to select the correct drying agenl Some drng agents (e.g. silica gel) lead to tailing of the carbon dioxide peaks due to absorption on the drying agent. Anhydrous calcium sulphate has been found to be the most reliable drying agent and is recommended when carbon dioxide le to be analysed.
4.10 RoutIne maintenance
It should be noted that all safety precautions regarding potentially toxic or carcinogenic dusts should be carefully observed when cleaning the ductwork and traps. Particular precautions should be taken when dealing with fluorinated or other compounds with high toxic potencles. The study of fluorine-containing compounds needs to be conducted with care as the generation of hydrogen fluoride can result in chemical attack on the glass beads of the refrigeration colLm’ln as well as on human tissue due to the highly corrosive nature of this compound.
The equipment will always coilect a certain amount of soot. Some will inevitably be deposited on the inside of the ductwork. This should be removed regularly by brushing end vacuum cleaning.
Performance will not normaly be affected unless the gas sampling ring holes are blocked. The gas sampling probe and the associated tubing. which connect to the oxygen analyser, require periodic cleaning. One indication of blocking is the need to adjust the waste regulator repeatedly to maintain the proper flow to the oxygen analyser. Cleaning of the interconnecting tubes consists of disassembling the various sections of king and blowing them tt’wough with compressed aIr 10,70 MPa (100 psi)J
Never direct high pressure air into the analyser and remember to vent the dirt and soot to a safe place. The pump situated In the gas sampling train should be cleaned or serviced following the manufadure?s instructions.
5 Test specimen preparation and presentation
5.1 General
Specimen preparation Is described in Clause 8 of ISO 5660-1:2002 and the advice given therein is generay recommended Products used at thicknesses between 6 mm and 50mm should be tested at the finished thickness. For products thicker than 50mm the advice In 8.1.4 and 8.1.5 011505660-1:2002 recommends that the product is cut down to 50mm from the unexposed face. Care should be taken when reducing the total thickness to ensure that the resulting product is representative of the onginal specimen. Products used at thicknesses thinner than 6 mm should be tested at the finished thickness as in end-use or fixed to a typical substrate. Guidance on the selection and use of suitable substrates Is given in ISO1TR 146971281. Systems using air gaps should be studied carefully since this can influence the results end special protocols should be determined Flgire 1 shows that the same piece of material can give very different results when tested:
5.2 Specimen trays and edge retainer frame
The 25 mm metal trays and assoaated retainer frames used for testing should be constructed from stainless steel as specified ii ISO 5660-1. Use of incorrect steel may mean that there is a large mass range for the trays that would men have differing thermal Inertia and lead to different results being obtained.
ISO 5660-1 requfres that an edge retainer frame be used when testing in the honzontal orientation. For composite products edge coverage Is mportanl to be consistent with the end use apphcation and reduce any edge burning effects. However, some industry sectors may find that the use of the edge retainer frame is unnecessary for theE appcaIlons. Babcauskasi4i has studied this extensively. This may be the case where low heat release rates from low thermal inertia materials are being masked because of the high thermal capacity of comparatively bulky specimen hardware. The CBUF studyl5) examined furniture composites without using any edge frames. The CBUF test recommendations were accepted by the ASTM committee E5 in 1996 when It replaced the 1993 protocol In ASTM E1474l31)
If an edge retainer frame is used, the surface area will be less than the 0.01 m2. The exposed area should be cleedy stated In the reports
6 Selection of heat flux
ISO 5660-1 does not prescrte the level of irradiance to be used In testing. It is the responsility of the end user or the product committees to ensure that the recommended heat flux levels are selected appropriate to end use. This will require carefri consideration of the available research data on the application. In general. two approaches are available:
tesleig at a heat flux deemed to be that of the design fwe.
— measuring the burning properties at a heat flux at which the material readily lgnes.
Most measurements will be requ Wed low post ignition test conditions end the user should first decide if that is the case, as most properties measured wil be very different depending on whether igriltion has at has not occurred. When the Intention is to assess post ignition properties of the product, it Is important 10 lest at a heat flux which causes ignition to occur withm. at most, about 10 mm because at lower heat fluxes specanens may show irreproducible ignition behaviour, In circumstances where materials show variable performance because they have been tasted close to their critical i9niuon fluxes, it ts suggestible to consider testing at least 5 kW& above this irraclance to obtain more reliable data concerning the intrinsic performance of the materiaL
It is suggestible to select levels from 25 kWTh2. 35 kWIm2. 50 kW?m2 and 75 kWAn. These levels are not special but are useful levels within the operating range of the cone calorimeter, (10 kWIm2 to 100 kW/m), and they are suggested in order that users produce data at a limited number of heat flux levels rather than at a wide number of arbitrary heal flux levels within the operating range.
Users wil have the following objectives for resuls from the cone calorimeter
a) to generate fundamental bulk material properties;
b) to generate input data Into fire engineering models;
C) to predict performances in larger scale tests;
d) to develop specifications which require the use of the cone caloflmeter.
Users with objoctivos a) are likely to need to map the properties of materials over a range of heat flux levels. For exploratory testing Ills suggested that Irradlances of 35 kW b. used. Further information can be found for materials that resist ignition at 35 kWIm2 by retesting at 50 kW/m2. and for those which readily ignite at 35 kW/m2 by retesting at 25 kW/m2. To obtain bulk material properties it Is Important to test thick specimens and to analyse data during the period when the material is behaving as a thermally thick speamen. In practice, this means testing specimens that are at least 6 mm thick but preferably 10 mm thick and analysing data obtained over a period of a few minutes starling about I mm after ignition.
Users with objectives b) will generally be guided by the requirements of the models to be used
Users with objectives c) will be making their decisions from recent research results, as this a developing area Predictions can b€ based on statistical correlations or the use of mathematical models. In both cases, It is important to match the level of heat received by the specimen in the cone calorimeter to that to be received by the majority of the specimen In the larger scale assessment.
Examples are as follows:
— If a material supplier is interested in assessing the possible opplication of a matenal for sheathing on an electric cable, indication of its performance ‘in an IEC 60332-3-101) test could be gained from testing It at 50 kW/m2, Testing at 75 kWIm2 woutd be required In order to assess the likelihood of the same material surviving the UI 9io1l lest required for US plenum cables,
— Many upholstered furniture composites will ignite and bum in a repealable mariner when tested at a heat Ikix of 25 kW?m2 (defined in NFPA 264A1341) but the mote egnitlon.reslstant composites may not, and the higher heat flux of 35kW/rn2 (specified in ASTM E1474I311) may be necessary.
Users with objectives d) have similar requirements to those with objectives C). This group should note that the simplest types of specification would require the selection of a design’ Wradiance to which materials will be exposed. There will generally be materials that will ignite under the Wradiance and those that will not. It need not necessarily be a requirement of the specification that the material showing ‘good reaction (I.e. little reaction) to the radiation be tested at a higher Wradiation to wve good repeatability provided that the specification is written so as to foresee any possible variability with sud, materials. Generally comparisons between materials for ranking or passifail-based test specifications would need to be made on common irradiance tests but it Is important to specify the relevant heat flux values since the relative rankings of materials can change at different heat flux values (see Reference (6]).
Table 2 gives some indication of the ranges of Wradiances developed ii some typical fires or fire test environments. Please refer to the actual reference to find details of the position of the heat flux measurement.
For those load cell systems that carl stabilize within I s to 2s after positioning the specimen, either of the following ignition protocols will ensure that the correct data are collected at ignition
The cone healer should be provded with a removable radiation shield to protect the specimen from the radiance prior to the start of the test. The shield should be made of non-combustible material, with a total thickness not exceedmg 12 mm. The shield shall be either of the following:
a) water-cooled and coated w4th a durable maft black finish hanng a surface emissivity of = 0.95 ± 0.05; or
b) not water-cooled, which may be either metal with a reflective top surface or cerantic in order to mirWmze radiation transfer
The shield should be equipped with a handle or other suitable means for quick insertion and removal. The cone healer base plate should be equipped v.itii a mechanism for introducing the shield Into position.
8 Guidance on the testing of non-standard products
8.1 General
ISO 5660-1 . in general, applicable to essentially flat homogeneous products, whidi do not Iritumesce, swell. bend or collapse appreciably when exposed to radiant heat. This is because the space between the specimen surface and either the cone heater or the igniter is limited, and therefore physical as well as thermal damage can be caused to any expanding specimens Any change in specimen surface position with take the exposed specimen surface into a different heat flux zone. The change in exposure condition during the test would be maximized when the specimen swells to the point of almost touching the heater coils. ISO 5660-1 also contains a method to test intumescent products.
8.2 Non-planar products
Products that are non planar. i.e. corrugated. or composites such as small electric cables, can physically be housed In the specimen holder. They do. however, show two non-standard characteristics that need to be addressed by the user. The first is that dferent parts of the specimen surface can experience different heat flux levels at any time. This can b calculated from specimen dimensions, the Information given in the thermal maps given in flgure 2. and simple trigonometry. These variations are small for specimens with surface height variations less than 10mm. The second is that planar specimens have surface areas that are the product of the exposed edge lengths, but the exposure area of non-planar products should be carefully estimated and used when calculating the test results.
If the lack of planwity does not involve some parts of the specimen because of air spaces and irregular thermal contact with the substrate, th. main source of problems for these spec Wnens will derive from radiative transfer aspects. The radiative problems are caused by different parts of the specimen receiving different radiative fkjx.s, which may arise from three main areas:
a) the variation of irradiance with distance from the come due to the varnng heights of the specimen surface;
b) the vanatlon of irradiance with the angle presented by the vanous specimen surface oneritation (configuration factor);
C) the post-Ignition feedback of radiation (and In extreme cases, convective heat) to the burning surface as a result of the specimens own combustion.
The separation of these contributions is generally very difficult arid it ie important to realize that data generated from such tests are comparative and not absolute.
Some experimental data are available in the literature l141151, which show how the flux varies with the distance fran, the base of the cone heater, and also radially out from the centre.
enhance fire performance. These faikires had been partially overcome by use of a raised grid as descnbed by bkkolai7I, but it is consIdered now that they are better overcome by testing at ranges greater than 25 mm (e.g 60 mm). In the latter case, the swelling specrien would then avoid physical contact with the spark igniter or the heater coils and experience a less severe thermal overexposure after sweaing. The calculations described 83 can be used to calculate the thermal regimes the specimen surface would pass through.
ISOTC 61 is now recommending that a spacing of 60 mm between the specimen surface and the cone calorimeter baseplate be used for testing intumescen* specimens.
It should b noted that although the heat flux can be adjusted to be the same as it was at 25 mm. the change m geometry mght affect the burning behaviour of the sample,
8.4.3 SpecImens that retreat from the heat source
Some specimens colapse on initial exposure to the radiant heat source. In the case of some thermoplastic cellular polymers, this can be a reduction of almost the full specimen thidmess, so there has been considerable difference in opinion on the level of reduction in potential heat fkix received at the surface after collaps. of the specimen. The variation of heat flux with distance is different from those discussed eartler because the specimen is housed in a tray with edge frames and reflective aluminium foil trays which reflect more heat into the collapsed specimen than a free-standing surface would expenence Lukasl18l has made a study of thés situation and found that the heat flux at the bottom of a 50 mm aluminium specimen tray is only 1 % lower than the 25 kWm2 measured at what would b. the specimen surface. When the edge frame was used in similar experiments, the heat flux reduction was 8%. This work implies that reduction in heat transfer to the specimen is not a significant problem These empirical resultsl’61 show smaller differences than would be expected from 8.3.
9 ComposItes and layered products
9.1 General
Laminated products may requwe some form of edge sealing to ensure that the burning process used ii the cone calorimeter reflects that found in practice. This may be done by the use of edge frames, which reduce edge flaming and specimen bowing, or by sealing the edges of laminates or composites using impermeable cements. Care should be taken when using results from products where:
a) the end use orientation differs from the test orientation;
b) one of the layers acts as a fire-resisting barrier under the conditions of test;
C) the end use application has different boundary conditions to that used in the test:
d) the pyiolysls products are vented in a nOn-representative way.
9.2 Non-homogeneous products
Products that are non-homogeneous set other limitations, Problems of specimen mounting for laminated specimens with components of different levels of combustibility can lead to significant edge effects. Specimen mounting is particularly difficult when the specimen is thermally thin and Is used #I a composite with ar In some cases, it Is Impossible to assess such composites. In other cases, for example where the specrien Is used in hollow square sections but at thicknesses greater than the thermal thickness of the material, valuable data can be produced from testing plaques of the material.
9.3 SpecImens with short test duration
Some specimens Ignite and complete burning quickly. Problems associated with load cell stabllity and other aspects of initiate,g tests can be overcome by the use of the heat shields described ii Clause 7. It is alsom necessary 10 decrease the scan time of the measurements in order to colect as much test data as possible. Scan times of 2 a should be oonsidered
10 Liquids
10.1 General
Liquids can be tested in the cone calorimeter but they introduce a number of problems thai are not apparent with solid specimens, for example effects of convection currents In the liquid, surface tension effects and sometimes vigorous boilrng of the luid.
10.2 TestIng without the radiant heater
Flammable solvent materials and materials with low flash points may be carefuly tested en the cone by this mode. A satisfactory protocol for testing in this mode is as foIlows
— the cone Calorimeter should be in the normal testing condition wh the heater power off and the heater cold;
— pour 20 g to 30g of the flammable liquid for test into a quartz dish, which should be weighed externally prior to inserting into the specimen holder of the cone;
— place the dish into the normal cone specimen holder, lined with the fibre blanket with a mininuian thIckness of 13mm;
— commence data acquisition:
— ignite the flammable liquid with a pilot flame, rather than with a spark igniter:
— allow the data acquisition to continue for at least I mm after all the liquid has been burnt.
If. for example. propan2-ol were used, the total heat released would be 30,2 kJ per gram of propane-2.ol. This can then be compared with the result from the run to check whether the orifice flow constant. C, was correct or not. Other alcohols may be used for calibration purposes The heat release rate from ethanol or propan-2-oI in the normal specimen tray gives a heat release rate of between 3.5 kW and 4 kiN.
The liquids are generally burnt In a square or circular container. These liquids bum reasonably constantly after an initial stabilizing period, although the rate of heat release versus time curve is relatively noisy due to the flames pulsing and flickering at the liquid surface, The manner in which these lIquids bum varies considerobly, for example, petroleum ether (60 C to 80 C) bums smoothly at the upper rim of the container, after the initial ignition. Other liquids, such as n-hexane. bum with relatively large ftames In both these cases, the smoke production has been shown to be affected by the position of the cone heater and can increase up to 35%. increasing in the following order:
a) 65mm above the specimen in the horizontal position:
b) in the normal testing position:
C) In the vertical position;
d) with the cone heater removed altogether, although no heat fkix is imposed.
However. liquids burning with smaller flames show no such variation.
Since some hquids are immiscible with waler, they can be burnt floating on water which ensures that they bum at steady state conditions without showing a rapid increase at the end of their burning, i.e. the conditions ensure that the material bums as a thermaly thick specimen.
A liquid depth of about 6mm Ia 12mm in a container 15mm deep ensures free access of air to the liquid burning surface whereas a deep tray might not behave in an identical way due to less ‘ being available. However, It may be necessary to have a reasonable difference between container depth arid bquld depth because of surface tension effects end also to prevent vigorous boiling which would cause the liquid to sp aver the edge of the container.
In all cases, it is essential to allow the specimen holder to return to approximately room lemperature before using It fore subsequent test to avoid any prehealing of the specimen prior to testing.
10.3 TestIng with the radiant heater
WARNING — Many liquids have law flash points and it Is very Important to ensure that volatile lIquids are NOT tested under radiant heat. Th. volatility of liquids increases with Increasing temperatur. and care is necessary to avoid hazardous test conditions. The use of a shield between the radiant heater (If used) and the test specimen is recommended.
This applies to dielectric materials such as lransfom’ier liquids, including both edicones and mineral oils, It also applies to hydraulic liqui& A study was cwled out by IECITC 10MG 8 on materials with a range of ignition points, including a mineral ad and a silicone fluid with a wscosity of 50 cSt. The conclusion of this study, which Involved other heat release equipment, was that there was no necessity for the development of a new method intended specially for electrical insulating fluids, but that ISO 5660-1 was suitable for this work.
IECtI’C 10WG a used the following lest conditions for these liquids:
— heat flux: 25 kWim2
— maximum lest duration. 20 mm;
— test vessel: quartz crucible;
— inner dIameter. 50 mm;
— heigl*l3mm;
— thickness of wall: 2,2 mm;
— volume of the specimen: 15 cm3.
The test vessel was placed on a cylindrical or square, asbestos-free, mineral fibre pad.
11 The theory of oxygen consumption calorimetry
11.1 G.n.ral
ThOrnlOnhll, in 1917, showed that, in the case of complete combustion, a large number of organic fuels release an approximately constant amount of heat for every unit of oxygen, which is consumed. In 1980 Hugget{201 obtained an average value of 13,1 kJ of heat released for each gram of oxygen consumed and he also showed that the value was not signhtlcantJy affected by incomplete combustion. He concluded that this value may therefore be used for practical applications and Ia accurate, with only a few exceptions, to within
±5%.
Table 3 lists some oF the values reported by Hugget(20). With the exception of three materials. ethene. ethyne and polyoxymethylene. all the calculated heats of combustion per gram of oxygen consumed lie between 12,32 kJ and 13,51 ki.
11.3 Effect of additives and fillers
It is important to realize that the heat release measured by oxygen consumption calonmetry is the heat released during burning. It does not define whether this heat Is released from the system, whether It is used in some other chemical or physical change, or if released, what form the release of the energy takes. The data from the cone test does not Inform the researcher where the heat may be In the system at any stage in the combustion process.
Heal release Is normally quoted normalized to surface area of the sample. It Is also commonly quoted per gram of sample lost. This is called the effective heat of combustion. Because some fillers can absorb heat and can also generate mass loss by dehydration as well as by combustion some concern has been shown that the oxygen depletion work erroneously reports the heat release measurements. A mathematical analysis of the thermodynamics associated with the use of high (e.g. 60 %) leadings of highly endothermec and highly hydrated fillers shows that maximum error would be 7% overestimation of the heat release (i.e. at these 60 $ leadings, the system is actually releasing 7 % less heat than the cone calorimeter indicates). The mathematical analysis Is descrted fully In Annex A. Thus the flame retardant is probably suppressing heat release even better than perceived.
12 Start and end of test
12.1 Start of test
The user should decide if it is essential to use pre-ignilion data in any test or average results that are to be used Wi the assessment. ISO 5660.1 begins measurement of heat-associated properties from the point of ignition because most materials, when present in sufficient bulk, will generate significantly more heat and smoke after Ignition than they would before. There are some examples where this is not true and son,e of these involve systems where the bulk of the material is pyrolyzed prior to Ignition. Considerable pre-ignition pyrolysis may however lead to significant generation of smoke and toxic gases that would not be accounted for if the only data used are collected after ignition occurs. In practice all commercially available cone calorimeters collect data from the point of exposure and allow the operator to define the test start point and many actually tell the operator the recorded ignition time and ask if that is to be appointed as the start of the test Thus no software or hardware changes are needed for pro-Ignition data to be Included in test reporting and test averaging.
12.2 End of test
ISO 5660-1 states that the test ceases if the sample has not generated any data or I the specimen did not ignite after the spark igniter had been sparkmg in position for a period of 30 mm or after the mole fraction of oxygen remains within ± 0,005 of that of the pro-test vakie for 10 mm.
ISO 5660.1 specIfies that if ignition occurs and the spark arm igniter is removed but the specimen then extinguishes, the spark ami igniter should be re-inserted and remain throughout the test This can be somewhat problematic with PVC, which burns for long periods and also generates electrically conductive cobweb-like structures on the spark arm. These can become substantially short to the edge frame and the loss of mass loss data can result In these cases, the spark arm igniter should be cycled in and out of the bum zone to disturb any such deposits (e.g. once every 120 a).
End of lest situations are defined as one of the following, whichever occurs Ilist:
a) 32 mm after the time of sustained flaming;
b) 30 mm have elapsed and the specimen has not ignited;
c) after the mole fraction of oxygen remams within ± 0,005 of that of the pro-test value for 10 rrNn. Observe and record physical changes to the sample such as melting, swelling and cracking.
The above timings are adequate for standard testing but researchers Interested Ni properties of materials In smouldermg phases of fires may require that these be extended until after the requ,red parameter peaks are seen. For example, those interested in carbon monoxide measurements may wish to continue the test for 5 mEn longer once peak values have been measured in the duct.
For vary accurate work on gas species the variation m atmospheric presse should be used in assessing the gas concentrations.
ISO 5660-1 requires that if the flame extinguishes after sustained flaming occurs and the spark igniter Is removed, the spark arm igniter shoiid be reinserted and the spark turned on within 5s. The spark ami igniter should then not be removed until the entre test has been completed.
13 Recommendations for presentation of data
13.1 Current situation
ISO 5660-1 gives the format for presenting data derived from heat and mass related aspects of the calorimeter. These data (all with time) inckide the heal release rate, in kWtsi2. as well as the average heat release rate for specified intervals timed from ignition.
Many test houses supply, in addition to the above data, information wIth respect to smoke emission and toxic fisne emission (commonly CO. CO2 and HCI. but others are also possible) Many of these data are given in the form of rates of production but it should be noted that these rates are calculated on a per unit rate of mass loss of specImen basis such as toxic fumes on a kg/kg (Ic. kgs l/kg 1) basis.
In ISO 56602°, the smoke production rate expressed In m2/s (smoke extinction area per second).
In addition to the time to ignition, the following five parameters are commonly reported as a function of time:
— specanen mass loss rate (g/s);
— heat release rate (kWm1
— smoke production rate (m2/s);
— carbon monoxide yield (kg/kg):
carbon dioxide yield (kg/kg).
The first of these, mass loss rate, is calculated using the set of five-point difference equations. i.e. the mass loss of the specimen as a function of time,
The second parameter, the heat release rate, is normalized to the surface area of the specimen and since al standard test specimens have the same surface area, heat release rate values from different materials can be directly compared.
The last two parameters are customarily normalized to the rate of mass loss, i.e. CO or CO2 produced per unit time Is divided by the mass of specimen lost In that unit of time. Since different materials lose mass at different rates, direct comparison of these parameters can be misleading.
The dimensionless units for CO and CO2 arise because what Is being reported is the mass of gas produced per unit mass of specimen volatilized. The quantity is therefore a dimensionless ratio.
This approach of normalizing several of the measured parameters to mass loss rate has some Wnporlant consequences. Whenever the mass loss rate falls to low levels or tends towards zero (e.g. the later stages of testing a flame retarded product, or during pyrolysis prior to ignition) then these normahzed parameters will become large or become indeterminate. An additional effect is to incorporate into the parameter of interest an amplification of tIre noise in the mass loss data.
It also means that It Is Imposstble to study the pre•lgnition pyrolysis behaviour of malertals, because before ignition the mass loss rate will be close to zero. or in some cases it may even be positive when the increase in mass clue to oxidation Is greater than the loss of mass due to the evolu4on of volatiles.
The cone calorimeter does record data between the start of the test and the ignition time, but these data are not normally analysed.
Users wishing to convert the mass loss rate normalized data to ‘as measured’ data or to surface area normalized data lace considerable problems in that the averaging procedures used by the venous software programs cause the 1:1 correspondence between the ‘as measured’ data and the rate of mass loss from the specimen to be lost. This means that for accuracy end convenience It is necessary to return to the raw data and to recommence the entire calculation procedure.
It is proposed to rationalize the basis on which data are presented to overcome this problem.
13.2 Additional useful data
It is proposed that an additional data set and graphs in which all the parameters are given on an ‘as measured’ basis would be valuable. These would be calculated diredity from the raw cone data.
With this as the basis, ills possible to report data from ,=0 which, because of the small rates of mass loss, is not practicable on the rate of mass loss basis. The use of the 0 orIgin Is considered to be particularly .i’iportant for smoke and toxic fume aspects.
It is recommended that two basic data sets be used as follows:
a) Sell:
I) graphs of the cumulative value as a function of time of the following parameters from s 0 to ,(end):
2) heat release, expressed in kilojoules (kJ);
3) smoke production. expressed in square metres (m2):
4) CO and CO2 production. expressed in grams (g);
5) specimen mass, expressed In grams (g);
a) Se12:
1) graphsofraledataasafunctionoftimeforthefoaowtngparamelersfromt=Oto,(end);
2) heal release rate, expressed in kilowatts (kW) (kJ!s):
3) rate of smoke production, expressed in square metres per second (m2/s):
4) rate of CO and CO2 production, expressed in grams per second (gte);
5) mass loss rate, expressed in grams per second (gIs).
Such a scheme wouki mean that all the basic data measured by the cone calorimeter would be reported and It would allow users to calculate derived data sets. e.g. on a rate of mass loss basis, or a surface area basis. eaUy If they so desired.
An example of a typical graphical report is shown in Figure 5.
13.3 Recommendations
13.3.1 Provision should be made en the software for mass, heat release, smoke. carbon monoxide and carbon dioxide data to be recorded as a function of time on an as measured bases, e. not normalized with respect to either mass loss rate or specnen surface area
13.3.2 Data should be collected, following extinction of the specimen, for a sufficient period to ensise that carbon monoxide and smoke production and any other relevant miormation during smouldenng not lost Test durations of 3D mm may be required.
13.3.3 Two sets of graphical data shouid be presented, as detaded ii 13.2 above, the first set being cumulative values as a function ci time and the second set being rates of production as a function of time.
13.3.4 II requwed an effective heat 01 combustion plot shoiid be train r(ignltlon) to i(end) Al graphs mentioned in 132 should be from i = 0 to t(end).
13.3.5 Numerical data for all the graphs should be readily accessible. Also numerical data should be supplied wiuich states allot the following:
— the value at all cumulative parameters at ,end);
— the maximum rates of production;
— the times at which these maxima occurred.
13.3.6 The effective surface area ci the specimen should be stated and software should be available to allow users to normalize the data with respect to the effective surface area and to the rate of mass loss if required
13.3.7 Software should be available to allow average values to be calculated between selected times. For example, between v —0 and t(end), v.0 and ,(lgnltion). t(mgnltion) and r(end), r(lgrultlon) and s(hgnlbon +2 mlii), s(ignition + I mm) and ,ignition +3mm). etc.