8/2/2019 SP_report_2011_66

1/60

SP

TechnicalR

esearchInstituteofSweden

Elec

Ra

rical

l Ochot

urre

rena,

ts an

ichael

brea dia

Frsth (

kdonosti

SP),Ma

n volc tool

ttias Elf

agesfor fi

berg (

Fire Tech

P Report 2

ases

OI)

nology

11:66

8/2/2019 SP_report_2011_66

2/60

Electrical currents and breakdownvoltages as a diagnostic tool for fires

Ral Ochoterena, Michael Frsth (SP),Mattias Elfsberg (FOI)

8/2/2019 SP_report_2011_66

3/60

3

Abstract

Electrical currents and breakdown voltages as a

diagnostic tool for fires

Two types of electrical measurements have been investigated in order to performdiagnostics of the fire dynamics in the ISO 5660 cone calorimeter. The rationale of thestudy is to take advantage of the pilot ignition electrodes that are already in place and usethese to collect additional information such as emission of pyrolysis gases and time toignition.

The first part of the project was a refinement of the method for measuring the so calledion current, which has already been investigated in a pilot study. It was found thatthorough shielding and grounding gives an excellent signal to noise ratio. An expressionfor the correlation between measured current and conductivity was also developed and

validated experimentally.

The second part of the project consisted of measuring the breakdown voltage, that is thevoltage when dielectric failure occurs. It was found that this method was more sensitiveto the fire dynamics before ignition, such as pyrolysis, but that the response to ignitionwas more ambiguous for the breakdown voltage than for the ion current.

Key words: electrical discharge, fire, ignition, cone calorimeter, Paschen's law

Skord: elektriskt verslag, brand, antndning, konkalorimeter, Paschens lag

SP Sveriges Tekniska Forskningsinstitut

SP Technical Research Institute of Sweden

SP Report 2011:66ISBN 978-91-86622-97-8ISSN 0284-5172Bors 2011

8/2/2019 SP_report_2011_66

4/60

4

Contents

Abstract 3Contents 4Preface 6Notations 7Sammanfattning 9Summary 101 Introduction 112 Theory 122.1 Current in electrode gaps without electric breakdown 132.2 Electric breakdown 153 Experimental methods 173.1 Electrical measurements 173.1.1 DC measurements of ion currents without electric breakdown 183.1.2 Measurements of breakdown voltage 203.2 Fire sources 233.2.1 Propane burner 233.2.2 Cone calorimeter 243.2.2.1 Fuels used in cone calorimeter 253.2.3 Smouldering combustion 274 Results and discussions 284.1 DC measurements of ion currents without electric breakdown 284.1.1 Improved signal and signal to noise ratio 294.1.2 Measurements on pyrolysis gases 334.1.3 Ion current vs. conductivity 334.2 Measurement results of breakdown voltage 354.2.1 Measurements in a 1 kW propane flame 354.2.2 Measurements in the cone calorimeter 364.2.2.1 Particle board 364.2.2.1.1 U

breakdownvs. HRR 36

4.2.2.1.2 Ubreakdown vs. SPR 394.2.2.2 PUR-foam 414.2.2.2.1 Ubreakdown vs. HRR 424.2.2.2.2 Ubreakdown vs. SPR 454.2.2.3 Black PMMA 474.2.2.3.1 Ubreakdown vs. HRR 474.2.2.3.2 Ubreakdown vs. SPR 504.2.3 Attempt to measure smouldering fires 535 Conclusions 546

Future Work 54

References 56

8/2/2019 SP_report_2011_66

5/60

5

8/2/2019 SP_report_2011_66

6/60

6

Preface

The Swedish Board for Fire Research (Brandforsk) sponsored this project with referencenumber 605-091 which is gratefully acknowledged. Brandforsk is owned by the Swedish

government, assurance companies, local authorities and industry and has as mission toinitiate, finance and follow-up different types of fire research.

Acknowledgment is given to the staff at SP who has contributed to this project. Specialthanks to Brith Mnsson, Sven-Ove Vendel, Allan Bergman, Sixten Bergman, andAnders Bergman. Anders Larsson at FOI is gratefully acknowledge for sharing hisknowledge on high voltage physics.

8/2/2019 SP_report_2011_66

7/60

7

Notations

Abbreviation Quantity Unit Explanation/comment

A prefactor in expression for [m-1Pa-1]B exponential factor in [Vm-1Pa-1]

expression forD prefactor in expression for [Am-2K-2]

thermionic emissiond distance between electrodes [m]E electric field [Vm-1]HAB Height Above Burner [m]HRR Heat Release Rate [W]I current [A]Ie,0 initial (no secondary [A]

ionization) electron currentat cathode

Ie total electrode current [A]at cathode

Iion,0 initial (no secondary [A]ionization) ion current at [A]cathode

Iion, total ion current at cathode [A]J current density [Am-2]kB Boltzmanns constant [JK

-1] kB = 1.38110-23 JK-1

ne electron density [m-3]

ppressure [Pa]PUR polyurethane

PMMA poly(methyl methacrylate),R resistance []S surface area of electrode [m2]SPR Smoke Production Rate [m2s-1]tignition time to ignition [s]T temperature [K]U voltage [V]Ubreakdown voltage required to overcome [V]

the dielectric strength

Townsends coefficient [m-1]for ionization

effective secondary [ ]ionization coefficient

degree of ionization [ ] work function for thermionic [J]

emission from metal surface resistivity [m] conductivity [Sm-1] or [-1m-1]

8/2/2019 SP_report_2011_66

8/60

8

8/2/2019 SP_report_2011_66

9/60

9

Sammanfattning

Denna rapport r en fortsttningen p en frstudie dr mjligheterna studerades fr att

anvnda jonstrmsmtningar som diagnostisk metod inom brandteknik, frmst ISO 5660konkalorimetern. De positiva resultaten frn frstudien ledde till dettafortsttningsprojekt.

Projektet bestod av tv delar:

frfinad metod att mta jonstrmmen mtningar av verslagspnningen

Den frsta delen gick ut p att mta strmmen mellan elektroderna utan nrvaro av gnista.Genom att skrma och jorda den utrustning som anvndes i frstudien stadkoms en storfrbttring i signal/brus frhllandet. Trots detta var det fortfarande enbart mjligt attmta sjlva antndningen. Detektion av pyrolysgaserna innan antndning var inte mjlig

pga. alltfr lga signalniver, ven med relativt hg plagd spnning (~1000 VDC). Ettuttryck fr frhllandet mellan den uppmtta strmmen och konduktiviteten ielektrodgapet togs fram och validerades experimentellt. Kunskap om konduktiviteten rviktig om man vill g vidare och gra uppskattningar om gasens tillstnd ssomtemperatur, elektrontthet och liknande.

Det andra delen av projektet initierades av ofrmgan att mta p pyrolysgaser med hjlpav strmmtningar utan elektriskt verslag. Tv elektriska kretsar designades ochtillverkades: En fr att skapa en vlkontrollerad gnista och en fr att mtaverslagsspnningen hos gnistan. Det visade sig att verslagsspnningen svarade vl p

frndringar i gasen ovanfr provkroppen ven fre antndning. Dessutom gav sjlvaantndningen ytterligare en pverkan p urladdningsspnningen, dock inte lika tydlig sompverkan p strmmen i den frsta delen av projektet.

Ett logiskt nsta steg r att ven beakta fasfrskjutningen mellan strm och spnningvilket rimligtvis ger en tydligare signal om frhllandet i elektrodgapet.

8/2/2019 SP_report_2011_66

10/60

10

Summary

This report is a follow up to a pilot study where the possibilities of using currentmeasurement for fire diagnostics, primarily in the ISO 5660 cone calorimeter, was

investigated. The positive results from the pilot study led to this project which consistedof two parts:

a refined method to measure the ion current measurement of the breakdown voltage

In the first part of the project the current between the electrodes was measured without aspark. This means that the electrodes could not be used as a spark igniter at the sametime. By thoroughly shielding and grounding the equipment from the pilot study a majorimprovement was obtained in the signal to noise ratio. Despite this is was still not

possible to measure pyrolysis gases since the signal was too weak, even with a relativelyhigh applied voltage (~1000 VDC). An expression for the relationship between themeasured current and the conductivity in the electrode gap was developed and validatedexperimentally. Knowledge about the conductivity is important in estimations of gas

properties such as temperature, electron density, etc.

The second part of the project was initiated from the inability to measure pyrolysis gasesfrom current measurements without electric breakdown. Two circuits were designed andconstructed: One for producing well defined high voltage pulses and one for measuringthe breakdown voltage. It was found that the breakdown voltage responded clearly tochanges in the gas composition above the tested sample even before ignition. Whenignition occurred an additional change in the breakdown voltage could be observed,although not as distinct as the current pulses measured in the first part of the project.

A logical next step would be to also measure the phase difference between current andvoltage. This is expected to give a signal which more clearly characterizes the status ofthe gas/plasma in the electrode gap.

8/2/2019 SP_report_2011_66

11/60

11

1 IntroductionThis project is a follow up to a pilot study regarding the use of ion current measurementsas a tool for ignition detection in the ISO 5660 cone calorimeter [1, 2]. In the pilot study

it was found that ignition could easily be detected by applying a DC voltage of 200 Vover the pilot ignition electrodes in the cone calorimeter [3] and measuring the ion currentin the ~3 mm air gap between the electrodes. It was found that the ion current wasvanishingly small before ignition and that ignition could easily be detected since one orseveral current pulses occurred when the tested sample ignited. The pulse height wastypically on the order 1-10 A and its length was on the order of some 10 ms. Onedrawback of this method was that it was not sensitive enough to detect for example theonset of pyrolysis. Another drawback was the fact that the electrodes of the conecalorimeter become assigned for the ion current measurements, using a DC voltage of200 V, and therefore they cannot be used to create the pilot ignition spark that is

prescribed in the cone calorimeter standard [3].

Current measurements as a method to monitor flame behaviour is not a new concept inthe combustion sciences. Ionic flame monitoring is the measurement of ion currents dueto an applied voltage between two electrodes in a flame. This is commonly used as asafety mechanism in burners [4, 5]. The function is to close the gas supply to the burner ifthe ion current disappears, that is, if the flame is extinguished. The objective is to avoidthe risk that a malfunctioning burner might fill up a space with a combustible or explosivegas mixture. More advanced versions of these so called flame rods have been presentedwhere the ion current is characterized by its DC amplitude, AC amplitude, and flickeringfrequency. This gives more detailed information concerning the status of the flame and ithas been proposed that these three parameters combined can give an early warning that a

problem is developing in the combustor [6]. From a fire safety perspective conductivity

of flames is also important in various other fields such as for example when assessing riskfor electrical breakdown between power lines and earth during forest fires [7].

In recent years ion current sensors in internal combustion engines have gainedconsiderable interest [8-10]. Measurement of the ion current over the gap of the spark

plug is a cost effective alternative to more expensive pressure sensors used for on boardengine diagnostics. In a recent study the relationship between ion current andtemperature was explored [11].

Conductivity of flames [12] and hot air [13] has been studied for over 100 years and is anarea of on-going research. It is easy to understand the complexity of the subject given thefact that the chemistry of combustion, not including ions, is still far from well-known for

most fuels and combustion conditions. Including the ion chemistry makes the feat evenmore difficult. Using electric fields to control the combustion has been proposed byseveral authors for different applications such as gas turbine control [14, 15] and formetallurgical processes [16] for example. In a recent study [17] laser diagnostics wereused to do fundamental research on the effect of electric fields on premixed methane-airflames. Direct numerical simulations [18] and experimental measurements [19] have been

performed to study the ability of electric fields to stabilize flames. An exponentialrelation between applied DC voltage and the change in burning velocity of premixedmethane/air flames has been reported [20] while another study indicated a rather linearrelationship between AC voltage and velocity in a propane flame [21]. The effect ofelectric field on soot was studied in reference [22] and one conclusion was that themajority of soot particles were positively charged. Microwaves were used to enhance

flame stability in a methane-air stagnation flame in reference [23]. Finally flameflickering induced by magnetic fields was observed in reference [24].

8/2/2019 SP_report_2011_66

12/60

12

One rationale for exploring the possible use of the electrodes in the cone calorimeter forflame diagnostics is to obtain an objective and well defined method for detecting ignition[25, 26]. In the current standard procedure for the cone calorimeter [3] an operatorvisually determines when ignition occur. This will by necessity be a subjective measure.Especially for flame retarded materials the flame can be indistinct and unstable [27] andwhen smoke is obscuring the test object it can be very difficult to objectively determinewhen ignition occurs. According to the standard for the smoke chamber test method [28]it is required that the inspection window, used be the operator to observe the test, isclosed when a certain smoke density is reached. This obviously makes it impossible todetect ignition visually after this point. Thus the detection of ignition can be a weak linkin the study of the fire properties of a material. By introducing an automatic and objectiveignition detection system more accurate information could be obtained. It is suggestedthat measurement of ion current or dielectric breakdown voltage could be the input signalfor such a system.

In this report two electrical phenomena were explored:

The ion current that flows between the electrodes under moderate voltages. Thisis a refinement of the previous pilot study [1, 2].

The voltage that is required to overcome the dielectric strength of the mediumbetween the electrodes. In other words, the voltage required to create an electricbreakdown.

Section 2 of this report contains the basic physical theory for the explored phenomena.The experimental materials and methods are described in Section 3 while the results are

presented and discussed in Section 4. The report ends with conclusions in Section 5 and adiscussion on suggested future work in Section 6.

Since this project was a direct continuation of the pilot study presented in reference [1]some parts in the present report are overlapping with the previous report.

2 TheoryAn electric force is exerted on electrons and ions in an electric field. Due to these forcesthere will be a flux of charged particles, creating a current. If two metal plates separated

by air are connected to a voltage difference on the order of 10 V no visible effect willoccur [29]. However, with a very sensitive ampere meter a current on the order of 10-15 A

would be detected. The source of this current is electrons and ions created by naturalradioactivity and cosmic rays. If a flame zone passes through the electrode gap the currentwill increase considerably. Charged species have been studied in a methane-oxygen flame[30, 31]. The most important of these species are electrons, CHO+, H3O

+, C2H3O+,

CH5O+, O2

-, OH-, O-, CHO2-, CHO3

-, and CO3- [32]. Due to these electrons and ions, the

current increases and for the electrode gap in the cone calorimeter typical currents on theorder of 10-6 A have been observed with an applied voltage of 200 V [1, 2]. Section 2.1

presents the most important parameters affecting this current.

If the applied voltage is further increased the current between the electrodes will rise oncea certain voltage is reached, Ubreakdown, and a discharge will be seen. This happens whenthe electrons gain sufficient energy, due to the electric field, between collisions with other

species. At this point, when the kinetic energy of the electrons reaches the atomicionization potential of the involved elements, each electron will knock out one additionalelectron upon collision. Immediately after the collision there will therefore be two slow

8/2/2019 SP_report_2011_66

13/60

13

electrons that again will accelerate in the electric field and then knock out two moreelectrons, and so on. In other words there will be an electron avalanche and a self-sustained electric discharge will remain as long as the high voltage is applied. The basictheory forUbreakdown is given in Section 2.2.

2.1 Current in electrode gaps without electricbreakdown

It has previously been shown [1, 2] that when the applied voltage is below Ubreakdown thecurrent between the cone calorimeter electrodes follows Ohms law:

(1)For a homogeneous electric field in an areaA with electrode distance dthe resistanceR is

(2)where

is the conductivity in the gas between the electrodes [Sm-1] (or [-1m-1]).The current is therefore

(3)The conductivity is due to charged particles, that is electrons, positive ions, and negativeions. Since electrons are much lighter than ions they are most easily accelerated by theelectric field. Therefore it is the electron concentration the determines the conductivity, aslong as the electron density is not much lower than ion concentrations. Negative ionsaffect the conductivity negatively since they are electron depleting [9].

Experiments have shown that for air [29]:

9.610 (4)where

ne is the concentration of electrons [m-3], and

p is the pressure [Pa].

Combining Equations (3) and (4) yields:

9.6 10 (5)

8/2/2019 SP_report_2011_66

14/60

14

This indicates that for a given air pressure it is the electron density that mostly influencesthe current. Simulations have shown that for a flat laminar lean methane-oxygen flame[32] the molar fractions of electrons, the degree of ionization , is on the order of 10-9.This information can be used in expression (5) by using the ideal gas law:

(6)which gives the electron density ne as:

(7)Expression (5) transforms into:

9.610 (8)Making the very bold assumption that = 10-9 is valid in the flame zone between theelectrodes in the cone calorimeter and that the flame temperature is 1300 K this can beevaluated numerically (see Section 4.1.3 on the geometry of the electrodes):

9.610 10 200 1.2 10 1 .65101.3810 1300 3 10 22 A (9)This is more or less in the same order of magnitude as the results of the measurements inthe cone colorimeter, where the current was in the range 1 10 A [1, 2]. The current inexpression (9) is not more than an indication since the degree of ionization may varysignificantly between different flames [32, 33] and the contact area between flame andelectrode may be smaller than the full area of the electrode [9, 34].

Other parameters also affect the current, for instance the availability of electrons. If theelectron emission from the negative electrode is dominated by thermionic emission thecurrent density J on the surface is given by [9, 29]

(10)

Where

C is a constant [Am-2K-2] is the work function of the metal, that is the energy required to leave the metal

surface [J]

The work function in its turn depends on the external electric field [35]. Other parametersaffecting the current is gas flow [36] and gas composition [29].

8/2/2019 SP_report_2011_66

15/60

15

2.2 Electric breakdownIf electron losses, due to for example recombination and attachment to walls, are ignoredthe current to the anode will equal the current of emitted electrons from the cathode,I0.

This is valid as long as the voltage over the electrode gap is low enough that no ionizationdue to collisions between accelerated electrons and molecules occur. If the voltageincreases further ionization will subsequently start. This is characterized by Townsendscoefficient for ionization, [m-1].

(11)where

A is a constant [m-1Pa-1]

B is a constant [Vm-1Pa-1]E is the electric field [Vm-1]

is the number of ionization events caused by one electron per unit length [29]. Due tothe ionizations the current at the anode becomes:

(12)Obviously the total current will be the same at the cathode. The current at the cathodeconsists of the initial electron currentI0 and an ion current which is

1 (13)In other words each electron in the initial electron current generates 1 ions in theelectrode gap. For sufficiently high electric fields the positive ions hitting the cathode willknock out electrodes. The number of so called secondary electrons that each ion hittingthe cathode knocks out is denoted .

The total electron currentIe from the cathode therefore becomes

, (14)WhereIion,means that it is the total ion current, that is

, 1 (15)In expression (15) the ion current is calculated based on the total electron currentIe andnot based on the initial electron current I0. This reason for this is obvious; whensecondary emission is taking placeI0 should be replaced byIe in both expressions (12)and (13). The total electron current leaving the cathode becomes

8/2/2019 SP_report_2011_66

16/60

16

1 1 (16)Finally the total current at the anode becomes, taking into account the emitted electroncurrent from the cathode and the ionization in the electrode gap:

1 1 (17)A transition from a non self-sustained current to a self-sustained current (that is anelectric breakdown) occurs when the denominator becomes zero:

1 1 (18)

that is

1 1 (19)Substituting (11) into (19), and usingE=U/d gives

1 1

(20)

that is

ln ln 1 1 ln

(21)

Expression (21) is known as Paschens law [37]. Whereas reasonably well definedexperimental data of the gas phase propertiesA andB exist in the literature, information

on is very scattered since this is a quite complex parameter depending on, among manyfactors, the state of the cathode surface. Often values of 10-1 10-2 are assumed [29].See Table 1for a list of tabulated data from the literature.

8/2/2019 SP_report_2011_66

17/60

17

Table 1 Coefficients in different gases for Townsends coefficient for ionization (11) and for

Paschens law (21).

Gas A [m-1Pa-1][29]

B [Vm-1Pa-1][29]

[ ]

N2 9 257 >1.310-6 [29]O2 7 206 10

-7 - 4.510-2 [29]Air 11 274 810-6 - 1.510-4

10-2[29][38]

H2 4 99 10-6 2.410-3 [29]

H2O 10 218CO2 15 220

As an example, the breakdown voltage for a 3 mm electrode gap (such as in the conecalorimeter) at atmospheric pressure (101 kPa) would become

274 101 10 3 1 0ln 11ln 110 1 ln101 10 3 1 0 13 kV

(22)

The can be compared with typical values for the dielectric strength of air which is3.2 kVmm-1 at atmospheric pressure [29].

Strictly speaking the breakdown voltage in expression (21) is rather dependent on themolecule concentration than on pressure. This means that if the pressure is constant andtemperature increases the breakdown voltage will decrease.

Furthermore, in a flame environment the gas properties obviously differs from theproperties of air. For example electrons and ions much more abundant in flame zonesthan in air. In one study [22] it was found that the flame reduced the dielectric fieldstrength to one seventh that of air. In other words the breakdown voltage is expected todrop when ignition occurs.

The theory for breakdown voltage described here is valid for moderate products ofpressure and gap distance,pd< 300 Pam [29]. Since the experiments in this study hasbeen performed at atmospheric pressure and with a gap distance of 3 mm this is really atthe limit of the applicability of the theory. For high products ofpdthe breakdown is betterdescribed by the faster processes of spark discharges (streamers) [29, 39]. However, the

theory above is only used for a qualitative interpretation of the experimental results so thephysics of streamers will not be described.

3 Experimental methods3.1 Electrical measurementsThe majority of tests were performed with the type of electrode assembly originally usedin the cone calorimeter. Figure 1 shows such an electrode, supplied by Fire TestingTechnology Limited, East Grinstead, UK.

8/2/2019 SP_report_2011_66

18/60

Fi

3

T

to

wio

b

or

c

Fi

gure 1. O

.1.1

e basic sch

the circuit

as more thorn current w

ndwidth) th

der to incre

rrent to pas

gure 2. S

riginal equip

C meas

reakdow

matics of th

sed in a pre

oughly grous measured

at measured

se the signa

through th

chematics for

ent electrode

rements

n

e circuit for

ious study

nded, see Fiusing an osc

the voltage

l to noise rat

1 M resis

the DC-measu

18

assembly for

of ion cu

the DC mea

1, 40]. The

gure 3, andilloscope (T

ver a 100 k

io, the shun

tance of the

rements circu

the cone calor

rents wi

surements, s

ain differe

ealed, see Fektronix TD

shunt resi

resistance

oscilloscope

it.

meter.

hout ele

ee Figure 2,

ce is that th

igure 3 andS 1002B, 60

stance. For s

as removed

.

tric

were simila

e new circui

igure 4. ThMHz

ome tests, i

, forcing the

t

e

8/2/2019 SP_report_2011_66

19/60

Fi

Fi

gure 3. C

n

gure 4. T

areful ground

oise.

he shunt resis

ing and shield

ance and con

19

ing of the lead

ections were

s was necessa

enclosed in a s

y for reductio

teel box.

n of electric

8/2/2019 SP_report_2011_66

20/60

In

th

1

el

Fi

3

T

gi

o

o

T

di

se

tr

se

an

T

re

ce

tr

m

th

fi

te

T

it

re

T

th

wc

v

addition to

is purpose t

00 VDC. A

ectrodes, se

gure 5.

.1.2

e circuit cr

ven frequen

ce the poten

the mediu

e circuit ba

rect current

condary win

nsformer is

condary an

d secondar

e power tra

ceives a squ

ases. The fr

nsistor clos

agnetic fiel

e circuit, the

ld, which w

rminals.

e circuit w

of 10 V and

petition rate

e rising tim

e capacitor (

ith higher capacitance to

ltage to the

reducing the

e applied v

lso, in order

Figure 5.

odified electr

easure

ates a fast i

y. The incr

tial between

contained i

sically consi

source that i

ding is dire

of the step-

the primar

windings o

nsistor is co

are pulse fro

quency and

es the circui

commence

coil faces a

as stored in

s supplied

0.55 A, res

of 50 Hz.

e of the puls

C3) which i

pacitance lea faster risi

electrodes is

noise, actio

ltage was, f

to increase t

odes with enh

ents of b

crease in th

ase in the el

the electro

n the measu

sts of a trans

s switched o

tly connect

p type with

windings cl

X1 are 9.9

figured to a

m the contr

duration of

, the primar

to rise with

rapid chang

it, is transfo

ith direct c

ectively. T

e could be c

s connected

ads to a slog time. A s

shown in F

20

ns were also

or some test

he ion curre

nced surface

reakdow

e electrical

ectrical pot

es is high e

ing volume

former with

n and off by

d to the ele

an iron core

lose to 100:1

mH and 44

ct as a fast s

l unit and o

this pulse c

coil of the

out reachin

e in the curr

med into a l

rrent with a

e high volt

ontrolled by

in parallel t

er rising tihematic of

igure 6.

taken to inc

s, increased

nt, tests wer

area.

n voltage

otential bet

ntial leads i

ough to bre

i.e. the gap

its primary

a power tra

trodes, see

and a relati

(X1). The i

H, respecti

witch, closi

ening the ci

n be control

transformer

saturation.

ent flowing

arge electric

potential an

ge pulse wa

means of m

the switchi

e and a cohe electroni

rease the io

rom 200 V

made with

een the ele

to an electri

ak the diele

between ele

inding con

sistor (T2);

igure 6 and

n of turns b

ductance o

vely.

g the circui

ircuit when t

led on dema

is energised

hen the tr

hrough it an

al potential

current flo

s produced

odifying the

g transistor.

ponent withcs used to cr

current. Fo

C to

larger

ctrodes at a

ical dischar

tric capacit

ctrodes.

nected to a

meanwhile

Table 2. Th

etween the

f the primar

when it

he pulse

nd. When th

and its

nsistor open

d the magne

between its

wing throug

ith a

capacitance

. A compon

lowereate the hig

r

e

its

e

e

s

tic

h

of

nt

8/2/2019 SP_report_2011_66

21/60

21

Figure 6. Schematic of the electronics used to create the high voltage to the electrodes. T2 is an

IRF740, D3 and D4 are a 1N4005, C3 has a value of 8.2 nF. The inductance of the

primary and secondary windings of X1 are 9.9 mH and 44 mH, respectively.

Table 2 Component list for the electronics used to create the high voltage pulse to the electrodes.

See Figure 6.

Symbol Value/TypeR1 & R2 250 R3 & R8 10 kR4 1-10 kR5 & R6 100 R7 10 C1 4700 FC2 1 FC3 8.2 nFT1 BC337NPN

T2 IRF740D1 L-7104GDD2 L-7104YDD3 & D4 1N4005X1 B 0221119027

The potential across the electrodes was measured using a resistive voltage divider andcommercial Tektronix P6015A. The resistive voltage divider consisted of two resistors,one of 1G and other of 1M which were connected in series across the electrodes. Thelatter resistor and the oscilloscope (1 M) were connected in parallel. A schematic of thevoltage divider is shown in Figure 7.

VDC C1

R1

R2

R3

4 3 2 1

5 6 7 8

LM555

C2 R4

R5

D1

T1

R6

D2

R7

R8

T2

D3

D4 C3

X1

SparkGap

8/2/2019 SP_report_2011_66

22/60

Fi

T

fu

w

v

pr

si

me

A

c

pr

Fi

gure 7. S

v

e reason to

rthermore th

ith the high

ltage, the v

obe. One ex

gnals with f

easurement.ample pyrol

ccurate abso

mmercial p

esented in t

gure 8 T

v

p

a

b

chematics of t

ltage divider

build a simp

e plastic det

eat fluxes n

ltage divide

planation co

irly fast rise

This is of mysis and ign

lute values

obes, for fa

is report are

he cone heate

ltage coil, b)

air of the equi

e 1000:1 volt

reduces the vo

le voltage di

ails of the c

ear the cone

r measures

uld be that t

times but th

inor importition affects

f the breakd

t processes

obtained us

and burning

eads to the vo

pment, not us

22

ge divider. W

ltage a factor

ivider is that

mmercial T

calorimeter

higher volt

he inductan

at the Tektr

nce in this sthe breakdo

own voltag

such as thos

ing the low

sample with d

ltage measure

d here.

hen connecte

2000.

it is a cost e

ektronix pro

. When mea

age compare

e of the volt

onix is bette

tudy since tn voltage i

are difficul

studied in t

cost voltage

ifferent leads

ment circuit, c

c

to a 1 M os

ffective sol

be were not

uring the br

d with the T

age divider i

designed f

e goal is toa measura

to measure

his report.

divider.

ndicated. a) l

) leads to the

cilloscope, the

tion and

compatible

eakdown

ektronix

is high for

r dynamic

see if forle way.

, even with

ll results

ads to the hig

riginal electr

h

de

8/2/2019 SP_report_2011_66

23/60

Fi

A

Fi

b

o

th

T

re

el

v

3

3

In

pr

I

w

e

gure 9. H

n example o

gure 9 wher

observed th

curs betwee

e gases in th

e average o

corded and

ectrodes wa

ltage agains

.2

.2.1

the first par

opane burne

C 60695-11

ith 65030

uivalence r

igh voltage p

f a typical p

e the potenti

at the poten

n them. Her

e measurin

f sixty-four

tored by the

measured

t time. This

ire sou

ropane

t of the mea

r was used,

-2:2003 stan

l/min prop

tio of 1.6, r

lse from the c

lse, generat

al between t

tial between

eafter the po

volume.

ulses, each

acquisition

igitally for

data was filt

ces

urner

surements o

as shown in

dard [41].

ne and 10

sulting in a

23

ircuit shown i

ed by the ab

he electrode

the electrod

tential falls

of them sa

system. The

ach set of d

ered using a

breakdown

Figure 10.

ell ventilat

.5 l/min air

blue flame.

Figure 6.

ove describe

s is plotted a

es rises until

s a consequ

pled at a rat

reafter, the

ata, thus obt

low-pass fil

voltage a w

he burner c

d combusti

fed to the b

d circuit, is

s a function

an electrica

ence of the i

e of 100 MS

aximum vo

ining the br

ter.

ll controlle

mply with t

n condition

rner, corres

shown in

of time. It c

l discharge

onisation of

/s, was

ltage betwe

eakdown

d 1 kW

he

s were used

onding to a

n

n

n

8/2/2019 SP_report_2011_66

24/60

Fi

3

T

illis

m

T

S

p

to

H

st

gure 10 T

.2.2

e cone calo

ustrated insubjected to

aterial. The

e cone calo

R (Smoke

ssible to sa

measure fo

CN for exa

udy.

he propane b

one calo

rimeter [3]

igure 11. Tirradiation

irradiation l

rimeter can

roduction

ple the exh

example un

ple. No ext

rner used in t

rimeter

as used for

is is a testrom an elec

vels used in

e used to m

ate), and M

aust gases t

burned hydr

rnal gas ch

24

he measurem

most of the

here a 0.01trically heat

this study

easure time

R (Mass L

an FTIR sp

ocarbons an

racterisatio

nts of breakd

easuremen

m2 specimed conical s

ere 25 and

to ignition,

ss Rate) of

ectrometer

d toxic gase

measureme

wn voltage.

ts, schemati

, horizontalliral above t

0 kW/m2.

RR (Heat

he tested ob

r other exte

such as N

nts were per

cally

ly positionee tested

elease Rate

ject. It is als

nal analyse

, HCl, and

formed in th

,

),

o

s

is

8/2/2019 SP_report_2011_66

25/60

Fi

N

T

sh

el

w

3.

T

T

gure 11 S

ormal opera

is is achiev

own in Figu

ectrodes sim

ithout electri

2.2.1

ree fuel typ

Polyuwhere

Partic100

Blackarea o

e fuels are

chematic pict

ion of the c

ed by a spar

re 1. For the

ply acted as

ical breakdo

uels used i

e were used

rethane foa

100 mm x 1

le board wit

m x 100 m

PMMA, po

f the specim

hown in Fi

re of the cone

ne calorime

igniter act

measureme

the pilot. Fo

n, no pilot

n cone cal

in the tests

with a den

00 mm and

a density o

and the thi

ly(methyl m

ens where 1

ure 12 to Fi

25

calorimeter.

ter requires

ally consisti

nts of break

r the measu

was used.

rimeter

ith the con

ity of 211

the thicknes

f 68050 k

ckness was

ethacrylate),

00 mm x 10

gure 14.

pilot ignitio

ing of the sa

own voltag

ement of io

calorimete

kg/m3. The

s about 35

/m3. The ar

12 mm.

with a dens

mm and th

of the pyro

e spark ga

in Section

current in

. These wer

area of the s

m.

a of the spe

ity of 1180

e thickness

lysis gases.

as that

4.2.2 the tes

Section 4.1.1

e:

pecimens

cimens whe

50 kg/m3. T

as 10 mm.

,

e

he

8/2/2019 SP_report_2011_66

26/60

Fi

Fi

gure 12 P

1

gure 13 P

m

olyurethane f

0 mm x 100

article board

m x 12 mm a

am used in th

m x 35 mm a

sed in the exp

d a density of

26

e experiments

nd a density o

eriments. Th

68050 kg/m

. The specime

211 kg/m3.

specimens ha

.

in the pictur

ve dimensions

e has dimensi

100 mm x 10

ns

8/2/2019 SP_report_2011_66

27/60

Fi

3

A

c

in

fo

fo

gure 14 B

m

.2.3

n attempt w

mbustion. F

cluding an u

r breakdown

am, see Fig

lack PMMA u

m x 10 mm a

moulder

s made to d

or this purp

pholstered f

voltage. Th

re 12 clad

sed in the exp

d a density of

ng comb

etect change

se EN 1021

urniture mo

e upholstere

ith blue cot

27

eriments. The

118050 kg/

ustion

s in the brea

-1 was used

k-up, cigare

d furniture

on textile.

specimens ha

3.

kdown volta

[42]. Figure

ttes, and the

ock-up con

e dimensions

ge during s

15 shows th

measureme

sisted of sta

100 mm x 100

ouldering

e setup

nt equipme

dard PUR-

t

8/2/2019 SP_report_2011_66

28/60

Fi

4

4

T

ra

d

p

fl

sel

gure 15 T

.1

e purpose o

tio could be

termine wh

rolysis gase

mes with a

ouldering fectrode gap

he test setup f

esults

C mea

lectric

f these tests

improved as

ther the ne

s. In the pre

DC-circuit b

ire. Finally twas calculat

or smoulderin

and di

uremen

reakdo

were twofol

compared t

circuit, wit

ious study i

ut no attem

he relation bed and the c

28

g combustion

cussio

ts of io

n

d. Firstly, to

o the previo

h increased

it was found

ts were ma

etween the ialculations c

according to

s

curren

investigate

s, unshield

applied volt

to be a strai

e to study p

on current aould be vali

N 1021-1.

s witho

hether the

d circuit. Se

ge, could b

htforward t

rolysis gas

d the condated by exp

ut

signal to noi

condly, to

used to det

ask to detect

s or a

ctivity in theriment.

se

ct

e

8/2/2019 SP_report_2011_66

29/60

4

Inu

Fi

Fi

to

.1.1 I

order to assshielded, ci

gure 16. I

ir

a

gure 17 sho

noise ratio

mproved

ess the newcuit is sho

n current me

radation [1].

d the shunt r

s the ion c

as improve

signal a

circuit a repn in Figure

sured at the i

he measurem

esistance was

rrent measu

considerab

29

d signal

esentative r16 [1].

gnition of a p

ent circuit wa

100 k .

red with the

ly.

o noise r

esult obtaine

rticle board s

s unshielded.

new circuit.

atio

d with the p

bjected to 50

he applied vo

It is clear th

revious,

kWm-2

ltage was 200

at the signal

8/2/2019 SP_report_2011_66

30/60

Fi

T

c

th

isra

Fi

gure 17. I

a

sts were als

rrent to pas

e voltage si

no significatio.

gure 18. I

t

r

n current me

plied voltage

o performed

through th

nal in the o

nt evidence

n current me

ere was no ex

sistance of th

sured with th

was 200V and

with the 10

1 M resis

cilloscope.

hat removin

sured with th

ternal shunt r

oscilloscope.

30

e new shielde

the shunt res

0 k shunt r

tance of the

representa

g the shunt

e new shielde

sistance, mea

circuit, see F

stance was 10

esistance re

oscilloscope

tive result is

esistance in

circuit. The

ning that all c

gure 3 and Fi

k .

oved. This

, and thereb

shown in Fi

reases the s

pplied voltag

rrent passed

gure 4. The

forces all th

increasing

igure 18. Th

ignal to nois

was 200V an

the 1 M

e

re

e

8/2/2019 SP_report_2011_66

31/60

In

re

se

o

b

is

Fi

1

[1

Fi

O

h

w

io

e

order to inc

presentative

en by comp

ly selected,

haviour of t

108. This

gure 17 wit

00 V / 101

, 40].

gure 19. I

t

ne proposal

s been to in

ith the large

n current is

ample, Figu

rease the io

result is sho

ring with th

albeit fairly

he system.

is obtained

R = 200V /

0-6

A. It has

n current me

e shunt resist

for increasin

rease the s

electrode su

elatively hi

re 19.

current the

wn in Figur

e typical res

representati

rough esti

sing Ohms

210-6

A, or

previously

sured with th

ance was 100

g the ion cu

rface size o

rfaces show

h but not si

31

applied volt

e 19. The io

ult using 20

e, results a

ate of the

law,R=U/I

approximat

een shown

e new shielde

.

rent, thereb

the electro

in Figure

gnificantly

age was incr

current cle

V, shown i

e shown her

inimum resi

, and approx

ing the resul

hat the ion

circuit. The

making th

es. Tests we

. A result is

ifferent fro

eased to 10

rly increase

n Figure 17.

e they descri

stance in th

imating the

t in Figure 1

urrent follo

pplied voltag

method mo

re therefore

shown in Fi

the current

0 V. A

d as can be

Although

ibe the gene

electrode

result in

9 withR =

s Ohms la

was 1000V a

re sensitive,

performed

gure 20. Th

in, for

al

ap

w

d

8/2/2019 SP_report_2011_66

32/60

Fi

U

elth

si

F

el

T

Fi

gure 20. I

ci

m

sing such la

ectrodes thee tested sam

gnificant thi

rthermore,

ectrodes, wh

erefore this

gure 21. S

n current me

rcuit. The ap

eaning that al

ge electrode

irradiationple, see Fig

might affe

he flames

ich could e

approach w

hadow effect

sured with th

lied voltage w

l current pass

s will introd

rom the resire 21. Since

t, for examp

ight be quen

plain the ab

as not found

n the exposed

32

e large electro

as 1000V and

ed the 1 M r

uce some pr

tively heatethe shadow

le, the HRR

ched when

sence of ion

to be intere

sample due t

des, see Figur

there was no

esistance of th

oblems. Due

d spiral willed (the non-

, which is an

hey enter th

current incr

sting for furt

the large elec

5, and the ne

xternal shunt

oscilloscope.

to the large

not reach cearkened) ar

unwelcom

e gap betwe

ase using th

her develop

trodes in Figu

w shielded

resistance,

size of the

rtain areas oea is

effect.

en the

ese electrod

ent.

re 5.

f

es.

8/2/2019 SP_report_2011_66

33/60

33

4.1.2 Measurements on pyrolysis gasesDespite several attempts it was not possible to measure a current due to pyrolysis gases.

The applied voltage was increased to 1000 VDC and the large electrodes in Figure 5 weretested but no signal above the noise level could be detected. A plausible explanation isthat the increase in electron density, or ion density, is simply not large enough to give ameasurable change in the conductivity of the pyrolysis gases.

For this reason another approach was investigated in order to measure pyrolysis gases, i.e.measurement of the breakdown voltage. As will be seen below this method can be used

both for measuring the onset of pyrolysis as well as for detecting ignition.

4.1.3 Ion current vs. conductivityWhen measurements are performed the result is a current. This is not a direct property ofthe gas between the electrodes. Rather, in order to characterize the gas in the gap theconductivity is the physical property of choice. Therefore the relation between currentand conductivity is investigated here.

The resistivity is given by

(23)Where

R is the resistance []A the cross sectional area of the electrode gap [m2], andd the separation between the electrodes [m]

and the conductivity is simply the inverse of the resistivity

(24)

The geometrical details of the electrode pair are given in the upper part of Figure 22. Thedistanced between the electrode surfaces is 3 mm and the two radii required to calculatedthe exposed electrode area in the gap are r1 = 2.4/2=1.2 mm and r2=3.3/2=1.65 mm. Thearea is, therefore

(25)

8/2/2019 SP_report_2011_66

34/60

Fi

T

In

v

T

[S

Ec

T

T

m

E

thth

c

gure 22 D

e resistance

Figure 16,

ltage was 2

e ion curre

m-1

] by mul

pression (2nductivity

erefore

is is within

easurement

pression (2

eoretical atte electrodes

rrent at elec

etails of the el

is simply c

igure 17, Fi

0 V. There

482

4

t in these fi

iplying the

) was valid=1.390 Sm

-

48

7% of the tr

uncertaintie

) will not b

mpt to explthe express

trical break

ectrode pair a

lculated usi

gure 18, an

ore

82m-1

200

ures can th

ion current [

ted by a cal1. The applie

35 10

13

ue conducti

.

e used furth

ain the ion cion is proba

own.

34

nd the measur

g Ohms la

482m-1

in most fig

2.41m-1

refore simp

] by a fact

ibration in ad voltage w

1.3 Sm-1

ity of the sa

r in this rep

urrent resultly not valid

ement of cond

w,R=U/I, y

ures in refer

V-1

ly be transfo

r 2.41.

saline solutias 13 V and

linity which

ort but is an

s. Due to effor measure

uctivity.

ielding a con

(26)

nce [1] the

(27)

rmed to con

on with knothe current 3

(28)

is acceptabl

important re

ects of the sments of vo

ductivity o

applied

ductivity

wn5 mA.

e given the

ference for

harp edgesltage and

f

8/2/2019 SP_report_2011_66

35/60

Fi

th

4

R

d

pr

b

li

4

M

st

se

b

h

B

c

Fi

nally, if a si

e cathode th

.2

esults from t

scribed in S

esented belo

rner. The la

itations of

.2.1

easurement

andardised p

tup is show

fore the mi

ight above t

urner), the b

nducted out

gure 23. B

b

f

gnificant pa

e current wil

easure

he measure

ection 3.1.2

w were con

tter was em

the measuri

easure

of the brea

ropane flam

Figure 10.

ing zone, th

he burner. It

eakdown v

side the rea

reakdown vol

urner in the fl

r the heights

t of the curr

ll depend on

ment re

ents condu

are presente

ucted usin

loyed to im

g system be

ents in a

down poten

e at two hei

It is possible

e breakdown

is also obse

ltage increa

tion borders

age for a pro

ame center, an

5 mm and 10

35

ent is due to

the cathode

sults of

ted using th

d in the foll

the cone ca

rove the un

fore it was

1 kW pr

tial done al

hts, are sho

to observe

voltage is

rved that ar

ses due to th

of the flam

ane flame. Ubd for several

mm.

thermionic

temperature

reakdo

e intermitte

wing sectio

lorimeter an

derstanding

pplied to th

opane fla

ng the sym

n in Figure

hat, at the b

igh and dec

und 180 m

e fact that th

but measur

reakdown is sho

ff-center radi

mission of

, see Eq. (10

wn volt

t spark disp

s. Measure

a well cha

of the behav

cone calori

me

etry axis, a

23. The me

ottom of the

eases as a f

HAB (Hei

e measurem

e in hot gase

ed for several

i (distance to t

lectrons fro

).

ge

ositive

ents

acterised

iour and

meter.

nd across th

asurement

flame end j

unction of

ght Above

ents are the

s.

height above

he flame cent

m

st

the

r)

8/2/2019 SP_report_2011_66

36/60

4

A

t

t

t

T

th

4.

4.

Fi

w

ig

ig

o

Fi

.2.2

fixed proto

0 s

10 s

20 s

e time to ig

e results bel

2.2.1 P

2.2.1.1

gure 24 sho

ith 50 kWm-

nition occur

nition. The

particle bo

gure 24. P

easure

ol was follo

data acqui

the sampl

protected

the radiati

irradiation

nition was o

ow.

article bo

breakdown vs.

the break2

irradiation

s at 50 s. Fi

easuremen

rd. For othe

article board.

ents in t

wed in all m

sition starte

was introd

he sample f

on shield wa

level; 25 k

bserved vis

rd

RR

own voltage

level. It is c

ure 25 show

s continued

r measurem

50 kWm-2

. tigni

36

e cone c

easurement

ced under t

om direct ir

s removed a

m-2 or 50

ally in addit

and HRR f

learly seen t

s a close-up

for approxi

ents the typi

ition=50 s.

alorimet

:

e cone heat

radiation

nd the speci

Wm-2.

ion to the dr

r a measure

hat Ubreakdownof the dyna

ately 1000

cal samplin

r

er but the ra

en was ex

op in Ubreakd

ment on a p

drops signi

ics around

s, for the ca

time was ar

diation shiel

osed to the

wn observed

article board

icantly whe

the time of

se 50 kWm-

ound 200 s.

set

in

2

8/2/2019 SP_report_2011_66

37/60

Fi

In

w

is

gure 25. P

Figure 26 a

ith different

seen when i

article board.

nd Figure 2

scales on th

gnition take

50 kWm-2

. tig

the results

time-axis.

place.

37

ition=50 s. Clo

for an irradi

ignition is no

e-up around i

tion level o

144 s and a

nition.

25 kWm-2

distinct dro

re shown,

p in Ubreakdow

n

8/2/2019 SP_report_2011_66

38/60

Fi

Fi

gure 26. P

gure 27. P

article board.

article board.

25 kWm-2

. tig

25 kWm-2

. ti

38

nition=144 s.

nition=144 s. Close-up around ignition.

8/2/2019 SP_report_2011_66

39/60

T

w

fo

m

v

br

w

4.

Fi

th

Fi

e initial Ubrhen it is 50

r this is uncl

eans lower

ltage, see E

eakdown vo

ith Figure 3

2.2.1.2

gure 28 sho

e right ordin

gure 28. P

eakdown is so

W. This is

ear since in

olecule con

q. (12) and t

ltage for lo

(PUR-foa

breakdown vs.

s the same

ate. Figure

article board.

e kV lower

asily seen b

fact the opp

centrations,

he subseque

er radiation

) or compar

PR

Ubreakdown as

9 shows a c

50 kWm-2

. tigni

39

when the irr

y comparin

site trend c

which typic

t discussio

level can al

ing Figure 4

in Figure 24

lose-up arou

ition=50 s.

adiation lev

Figure 25

uld be expe

ally would l

. The trend

so be seen b

0 with Figu

but this tim

nd tignition.

l is 25 kW

ith Figure 2

cted. Highe

ad to a low

owards a lo

comparing

e 42 (black

e compared

ompared to

7. The reaso

temperatur

r breakdow

wer

Figure 33

PMMA).

to the SPR

n

n

8/2/2019 SP_report_2011_66

40/60

Fi

Fi

gure 29. P

gure 30. P

article board.

article board.

50 kWm-2

. tig

25 kWm-2

. tig

40

ition=50 s. Clo

nition=144 s.

e-up around i nition.

8/2/2019 SP_report_2011_66

41/60

is

e

se

w

a

th

Fi

4.

T

c

se

hen studyin

completely

haust hood,

e Figure 11.

ell localized

eraging of 6

at is observe

gure 31. P

2.2.2 P

e results fo

mpared to t

conds after

g results lik

spatially ave

mixed, and

By compari

to the electr

4 samples,

d at 80 11

article board.

UR-foam

PUR-foam

e particle b

emoval of t

those show

raged since

measured in

son, the me

ode gap, alt

orrespondin

0 s, which h

25 kWm-2

. tig

are shown i

oard is that i

e radiation

41

n in Figure

all smoke fr

the smoke

surement o

ough in the

g to 1.2 s. T

as no clear c

nition=144 s. Cl

Figure 32

gnition take

shield at t =

0 it should

om the samp

ptical densit

Ubreakdown is

presented fi

his could ex

orrespondin

ose-up around

o Figure 39.

place much

20 s.

be observed

le is collect

y measurem

an in-situ m

ures there i

lain the dip

peak in the

ignition.

The major

faster, just

that the SP

d in the

ent system,

easurement

s a temporal

in Ubreakdown

SPR curve.

ifference as

one or a few

8/2/2019 SP_report_2011_66

42/60

4.

Fi

It

t

is

in

2.2.2.1

gure 32. P

is interestin

100 s altho

made with t

the noise a

breakdown vs.

UR-foam. 50

to observe

ugh the HR

he SPR, see

plitude.

RR

Wm-2

. tignition

that Ubreakdodrops. The

Figure 36.

42

21 s.

n stays at a

same phen

he major in

elatively lo

mena is obs

luence on

level in Fi

erved when

breakdown is ra

gure 32 for

a compariso

ther a decre

n

se

8/2/2019 SP_report_2011_66

43/60

Fi

In

th

gure 33. P

Figure 34 t

e drop in Ub

UR-foam. 50

e irradiatio

reakdown, espe

Wm-2

. tignition

level was

ially in Fig

43

21 s. Close-u

5 kWm-2

an

re 35.

around ignit

d tignition = 2

on.

s. This is allso reflected in

8/2/2019 SP_report_2011_66

44/60

Fi

Fi

gure 34. P

gure 35. P

UR-foam. 25

UR-foam. 25

Wm-2

. tignition

Wm-2

. tignition

44

24 s.

=24 s. Close-up around igni ion.

8/2/2019 SP_report_2011_66

45/60

4.

T

b

Fi

2.2.2.2

e same Ubre

t this time c

gure 36. P

breakdown vs.

akdown as in F

ompared to

UR-foam. 50

PR

igure 32 to

he SPR.

Wm-2

. tignition

45

igure 35 is

24 s.

hown in Fi ure 36 to Figure 39 below,

8/2/2019 SP_report_2011_66

46/60

Fi

Fi

gure 37. P

gure 38. P

UR-foam. 50

UR-foam. 25

Wm-2

. tignition

Wm-2

. tignition

46

24 s. Close-u

24 s.

around ignit on.

8/2/2019 SP_report_2011_66

47/60

Fi

4.

4.

Fi

ig

gure 39. P

2.2.3

2.2.3.1

nally results

nition occur

UR-foam. 25

lack PM

breakdown vs.

for black P

s at t = 51, a

Wm-2

. tignition

A

RR

MA are sh

lthough this

47

=24 s. Close-u

own. In Fig

drop is not s

p around igni

re 40 there

o distinct.

ion.

is a drop in breakdown wh

n

8/2/2019 SP_report_2011_66

48/60

Fi

Fi

gure 40. B

gure 41. B

lack PMMA.

lack PMMA.

0 kWm-2

. tignit

0 kWm-2

. tign

48

ion=51 s.

ition=51 s. Clos -up around i nition.

8/2/2019 SP_report_2011_66

49/60

In

b

Fi

Figure 42,

tween igniti

gure 42. B

here the irr

on and a dis

lack PMMA.

adiation lev

tinct drop in

5 kWm-2. tign

49

l was 25 k

Ubreakdown is

ition=151 s.

m-2

and tignivery good.

tion = 151 s, he correlati n

8/2/2019 SP_report_2011_66

50/60

Fi

4.

T

c

gure 43. B

2.2.3.2

e correlatio

se, see Figu

lack PMMA.

breakdown vs.

n between

e 47.

5 kWm-2

. tig

PR

breakdown and

50

ition=151 s. Cl

SPR is relat

se-up around

ively good,

ignition.

specially for the 25 kW

-2

8/2/2019 SP_report_2011_66

51/60

Fi

Fi

gure 44. B

gure 45. B

lack PMMA.

lack PMMA.

0 kWm-2

. tignit

0 kWm-2

. tign

51

ion=51 s.

ition=51 s. Clos -up around i nition.

8/2/2019 SP_report_2011_66

52/60

Fi

Fi

gure 46. B

gure 47. B

lack PMMA.

lack PMMA.

5 kWm-2

. tign

5 kWm-2

. tig

52

ition=151 s.

ition=151 s. Cl se-up around ignition.

8/2/2019 SP_report_2011_66

53/60

53

4.2.3 Attempt to measure smouldering firesFinally an attempt was made to measure a change in the breakdown voltage when theelectrodes were positioned above a smouldering cigarette in an upholstered furnituremock-up, see Section 3.2.3. No significant change in Ubreakdown could be measured.

There is one major difference between the measurement of smouldering fire, Figure 15,and the measurements in the propane flame and the cone calorimeter, Figure 10 andFigure 8 respectively. The temperature is very high in the flames of the propane burnerand the cone calorimeter. Even for the cone calorimeter before ignition the temperature ishigh and increasing. By contrast, the temperature above the smouldering fire is quiteconstant and not significantly higher than the ambient room temperature. This means thatthe change in Ubreakdown due to changed air density does not come into play for thesmouldering fire in the same way as it does for the propane burner and the cone

calorimeter. On the other hand the results show that increased temperature, due toremoved radiation shield, is not the supreme parameter affecting Ubreakdown. For example,in Figure 30, Ubreakdown does not start to decrease significantly at 20 s when the radiationshield is removed, but rather at ~75 s when the pyrolysis gases, as measured by the SPR,start to appear. This is in contrast to the results in Figure 29 where the decrease inUbreakdown starts immediately at the time of shield removal, but where also the SPR startsto increase at the time of shield removal. Furthermore the correlation betweentemperature (or density) and Ubreakdown is not trivial, as discussed at the end of Section4.2.2.1.1.

Although measurement of smouldering fires did not succeed in this project this does notmean that there is no hope for using the electrodes for this purpose. There are many

possible refinements possible for the method as will be discussed below.

8/2/2019 SP_report_2011_66

54/60

54

5 ConclusionsIt has been shown that the signal to noise ratio of DC current measurements can begreatly improved by careful shielding and grounding of the measurement equipment. This

enables more sensitive detection of the ignition phase. It was also shown that the currentcan straightforwardly be translated into electric conductivity which is a property that

better lends itself to a description of the status of the gas between the electrodes.

It was not possible to detect a current above the noise level for the pyrolysis phase,neither by increasing the applied voltage to 1000 VDC nor by increasing the size of theelectrodes. The explanation for this is probably that the concentration of electrons andions is much lower for the pyrolysis gases than for the flame front. In order to measure

pyrolysis gases with the electrodes another property was investigated, the breakdownvoltage, Ubreakdown.

The breakdown voltage correlated with HRR and SPR before and at ignition. Afterignition the correlation was poor. This shows that, given the simple equipment used inthis report, is possible to use the pilot ignition electrodes to detect pyrolysis gases andignition. Indeed, this is possible to do at the same time as the electrodes perform theadditional role of pilot ignition. In other words, it is possible to perform measurements ofUbreakdown without compromising with the ISO 5660 standard. In fact, according to thestandard the spark igniter should be removed after ignition. In other words the lowcorrelation after ignition is of no relevance if the standard is followed. The drop inUbreakdown is not as distinct as the current pulses that appear using the DC measurements.However, this was only a first test of measuring Ubreakdown and there is clearly a potentialfor improvement of the methodology.

6 Future WorkFurther research should aim at improving the response of the methods to weakly ionizedgases such as pyrolysis gases and especially smoke from smouldering fire. The latter hasnot yet been measurable be any of the two methods tested.

The voltage divider probe did not give the same result as the commercial voltage probe.This is probably due to the relative high inductance of the voltage divider. The inductancewill affect the probes accuracy when measuring dynamic processes. The performance ofthe voltage divider could be improved by either adding a compensating network or

changing the resistors. An improved probe should be calibrated against known highvoltage pulses and not against a commercial probe since the latter can give incorrectresults for the type of demanding measurements performed in this study (high voltage,fast processes, EMC problems).

One proposal for improving the response of the system is to study the time lag betweenthe applied voltage and resulting current. This can be done quite straightforwardly withthe method of electric breakdown as has been described in this report but an improvedvoltage probe and an appropriate current probe are required. For the DC measurements(without electric breakdown) it is clearly not possible to measure a time lag since thevoltage is constant. Therefore a signal generator producing AC voltages would berequired. When the atmosphere between the electrodes changes there will be a change inthe capacitance, and therefore also a change in the the time lag. Measuring the change intime lag (capacitance) instead of the current (conductivity) can also be described asmeasuring electric dipoles instead of electric monopoles (free charges). The change in

8/2/2019 SP_report_2011_66

55/60

55

abundance of dipoles can be expected to vary more than the change of free charges sincethe latter require ionization or emission of electrons from the cathode. Changes in dipoleconcentration can occur more easily due to non-ionized chemistry in the pyrolysis of thefuel, in the flames, or in the smoke gases.

Electric flame diagnostics is not limited to the cone calorimeter. One interesting projectwould be to conduct a similar study to the one presented here but applied to the smoke

box method [28] where under-ventilated combustion is studied.

8/2/2019 SP_report_2011_66

56/60

56

References

[1] M. Frsth, A. Larsson, On the use of ion current measurements to detect ignition in

the cone calorimeter, Brandforsk project 311-081, SP Report 2008:50, in, SP FireTechnology, Bors, 2008.[2] M. Frsth, A. Larsson, Ion current measurements as a tool for ignition detection in thecone calorimeter, Fire and Materials, 34 (2010) 421-428.[3] ISO 5660-1:2002 Reaction-to-fire tests - Heat release, smoke production and massloss rate - Part 1: Heat release rate (cone calorimeter method), in: InternationalOrganization for Standardization, 2002.[4] Honeywell, C7005A,B Gas Pilot and Flame Rod Assemblies, in, Honeywell Inc.[5] B. Chian, T.J. Nordberg, Circuit diagnostics from flame sensing AC component, in,Honeywell International Inc., USA, 2007.[6] J.M. Niziolek, W.M. Clark, J.M. Bialobrzeska, T.M. Grayson, Diagnostic ionic flamemonitor, ABB, USA, 2002.[7] K. Mphale, M. Heron, Measurement of Electrical Conductivity for a Biomass Fire,International Journal of Molecular Sciences, 9 (2008) 1416-1423.[8] P. Mehresh, D. Flowers, R.W. Dibble, Experimental and numerical investigation ofeffect of fuel on ion sensor signal to determine combustion timing in homogeneouscharge compression ignition engines, International Journal of Engine Research, 6 (2005)465-474.[9] A. Franke, Characterization of an Electrical Sensor for Combustion Diagnostics, in:Combustion Physics, Lund Institute of Technology, Lund, 2002.[10] S. Saxena, J.Y. Chen, R.W. Dibble, Increasing the signal-to-noise ratio of sparkplugion sensors through the addition of a potassium acetate fuel additive, Proceedings of theCombustion Institute, 33 (2011) 3081-3088.

[11] Z. Gao, X. Wu, C. Man, X. Meng, Z. Huang, The relationship between ion currentand temperature at the electrode gap, Applied Thermal Engineering.[12] T.M. Sugden, Elementary combustion reactions. Charged Species, in: 10thSymposium on Combustion, Combustion Institute, 1965.[13] J. Prager, G. Baldea, U. Riedel, J. Warnatz, Equilibrium Composition and ElectricalConductivity of High-Temperature Air, in: 21st International Colloquium on theDynamics of Explosions and Reactive Systems (ICDERS), Poiters, France, 2007.[14] T.H. Dimmock, W.R. Kineyko, The electrical properties of ionized flames. Part I.Study of flame ionization, in, Defense Technical Information Center (DTIC), 1961.[15] T.H. Dimmock, G.F. Miller, J.J. Nichol, W.R. Kineyko, The electrical properties ofionized flames. Part II. Electrostatic and magnetohydrodynamic deflection, in, DefenseTechnical Information Center (DTIC), 1961.

[16] B. Karlowitz, Flames augmented by electrical power, Pure & Applied Chemistry, 5(1962) 557-564.[17] F. Altendorfner, J. Kuhl, L. Zigan, A. Leipertz, Study of the influence of electricfields on flames using planar LIF and PIV techniques, Proceedings of the CombustionInstitute, 33 (2011) 3195-3201.[18] M. Belhi, P. Domingo, P. Vervisch, Direct numerical simulation of the effect of anelectric field on flame stability, Combustion and Flame, 157 (2010) 2286-2297.[19] M.K. Kim, S.H. Chung, H.H. Kim, Effect of AC electric fields on the stabilization of

premixed bunsen flames, Proceedings of the Combustion Institute, 33 (2011) 1137-1144.[20] J.D.B.J. van den Boom, A.A. Konnov, A.M.H.H. Verhasselt, V.N. Kornilov, L.P.H.de Goey, H. Nijmeijer, The effect of a DC electric field on the laminar burning velocityof premixed methane/air flames, Proceedings of the Combustion Institute, 32 (2009)

1237-1244.

8/2/2019 SP_report_2011_66

57/60

57

[21] S.H. Won, M.S. Cha, C.S. Park, S.H. Chung, Effect of electric fields on reattachmentand propagation speed of tribrachial flames in laminar coflow jets, Proceedings of theCombustion Institute, 31 (2007) 963-970.[22] Y. Wang, G.J. Nathan, Z.T. Alwahabi, K.D. King, K. Ho, Q. Yao, Effect of auniform electric field on soot in laminar premixed ethylene/air flames, Combustion andFlame, 157 (2010) 1308-1315.[23] E.S. Stockman, S.H. Zaidi, R.B. Miles, C.D. Carter, M.D. Ryan, Measurements ofcombustion properties in a microwave enhanced flame, Combustion and Flame, 156(2009) 1453-1461.[24] G. Legros, T. Gomez, M. Fessard, T. Gouache, T. Ader, P. Guibert, P. Sagaut, J.L.Torero, Magnetically induced flame flickering, Proceedings of the Combustion Institute,33 (2011) 1095-1103.[25] C. Lautenberger, A.C. Fernandez-Pello, Approximate analytical solutions for thetransient mass loss rate and piloted ignition time of a radiatively heated solid in the highheat flux limit, in: Fire Safety Science - Proceedings of the Eigth InternationalSymposium, Beijing, China, 2005, pp. 445-456.

[26] G.B. Baker, M.J. Spearpoint, C.M. Fleischmann, C.A. Wade, Selecting an ignitioncriterion methodology for use in a radiative fire spread submodel, Fire and Materials, 35(2011) 367-381.[27] M.M. Khan, J.L. de Ris, Determination of Operator Independent Ignition ForPolymeric Solids, in: Proceedings of the 9th Fire and Materials conference, Interscience,San Francisco, 2005.[28] ISO 5659-2:2006 Plastics - Smoke generation - Part 2: Determination of opticaldensity by a single-chamber test, in: International Organization for Standardization, 2006.[29] Y.P. Raizer, Gas Discharge Physics, Springer-Verlag, 1991.[30] J.M. Goodings, D.K. Bohme, C.-W. Ng, Detailed ion chemistry in methane/oxygenflames. I. Positive ions, Combustion and Flame, 36 (1979) 27-43.[31] J.M. Goodings, D.K. Bohme, C.-W. Ng, Detailed ion chemistry in methane/oxygen

flames. II. Negative ions, Combustion and Flame, 36 (1979) 45-62.[32] J. Prager, U. Riedel, J. Warnatz, Modeling ion chemistry and charged speciesdiffusion in lean methane-oxygen flames, Proceedings of the Combustion Institute, 31(2007) 1129-1137.[33] J. Lawton, F.J. Weinberg, Electrical Aspects of Combustion, Clarendon Press,London, 1969.[34] A. Franke, R. Reinman, A. Larsson, The role of the electrodes for the ionizationsensor signal, SAE 2003 Transactions - Journal of Engines, 112 (2003) 2003-2001-0714.[35] W. Schottky, ber kalte und warme Elektronenentladungen, Zeitschrift fr Physik AHadrons and Nuclei, 14 (1923) 63-106.[36] A. Franke, P. Einewall, B. Johansson, R. Reinman, A. Larsson, The effect of in-cylinder gas flow on the interpretation of the ionization sensor signal, SAE 2003

Transactions - Journal of Engines, 112 (2003) 2003-2001-1120.[37] F. Paschen, Ueber die zum Funkenbergang in Luft, Wasserstoff und Kohlensure

bei verschiedenen Drucken erforderliche Potentialdifferenz, Annalen der Physik, 273(1889) 69-96.[38] E.M. Bazelyan, Y.P. Raizer, Spark Discharge, CRC Press, Boca Raton, Fla, 1998.[39] A. Bondiou, I. Gallimberti, Theoretical modelling of the development of the positivesprak in long gaps, J. Phys. D: Appl. Phys., 27 (1994) 1252-1266.[40] A. Omrane, F. Ossler, M. Aldn , U. Gransson, G. Holmstedt, Surface temperaturemeasurement of flame spread using thermographic phosphors, in: 7th InternationalIAFSS Symposium, The International Association for Fire Safety Science, 2002, pp. 141-152.[41] IEC 60695-11-2:2003, Fire hazard testing - Part 11-2: Test flames .- 1 kW nominal

pre-mixed flame - Apparatus, confirmatory test arrangement and guidance, IEC, 2003.[42] EN 1021-1 Furniture - Assessment of the ignitability of upholstered furniture - Part1: Ignition source smouldering cigarette, in, Comit Europen de Normalisation, 2006.

8/2/2019 SP_report_2011_66

58/60

58

8/2/2019 SP_report_2011_66

59/60

59

8/2/2019 SP_report_2011_66

60/60

SP TechniOur work

Sweden's

technology

and susta

universitie

organisati

internation

cal Researcis concentrat

most exten

, research an

inable devel

and institu

ns, ranging

al groups.

Institute ofd on innovat

ive and ad

d developme

pment of in

es of techn

from start-up

Swedenion and the

anced reso

nt, we make

dustry. Rese

logy, to the

companies

evelopment

urces for te

an important

arch is carri

benefit of

developing n

of value-addi

hnical evalu

ontribution t

d out in cl

customer

w technolog

ng technolog

ation, meas

the competi

se conjuncti

base of abo

ies or new i

. Using

urement

iveness

on with

t 9000

deas to