(will be inserted by the editor)
Effects of the coupling of insulating and conductive materials to limit the temperature at chimney-roof penetration
First Author · Second Author
Received: date / Accepted: date
Abstract Recently, many roof fires have occured in Europe due to the presence of a chimney. Since also certified chimneys were involved in the fires, chimneys producers have attempted to propose some solutions However, these latter may sometimes not be effective, especially if the chimney operating conditions are more severe than those in the chimney certification procedure. Consequently, given the relevance of the problem, the scientific community was asked to identify the causes of the fires, and many studies presented in the literature have shown some points to be taken into account when designing and testing a chimney. Since heat transfer at chimney-roof penetration depends on many variables, recently, it was proposed to limit the temperature at this point by means of a device. The device should limit the temperature in very thick and insulated roofs, and in any chimney operating condition, that is, in normal chimney operating conditions and during soot fire events. Such a device is called CEIL and its features were identified in a numerical study presented in the literature. Differently from the device in the market, the CEIL device is made of conductive elements and insulating materials. The function of the insulating layer is the limitation of the heat flux from the chimney to the roof, whereas the function of the conductive elements is the facilitation of the heat transfer from the device to the surrounding. In other words, the conductive ele- ments act like fins that dissipate heat to the ambient. Given that no experimental test was performed, this paper presents the results of an experimental campaign performed to assess the efficacy of the CEIL device and to investigat aspects not taken into account in the preliminary numerical investigation. The experimental campaign has taken into account the most critical chimney operating conditions identified in the literature and it has consisted into three phases. Firstly, it has been investigated whether a device of fixed dimension can effectively be installed in any
F. Author first address Tel.: +123-45-678910 Fax: +123-45-678910
E-mail: fauthor@example.com S. Author
second address
Authors: M. Neri, P. Leppanen, M. Alanen, D. Luscietti, S. Bani & M. Pilotelli
roof. In the second phase it has been investigated whether a conductive element installed in the clearance can actually limit the roof temperature in very critical operating conditions represented by high exhaust gas temperature in the chimney and small air infiltration in the materials at the chimney-roof penetration. Finally, the effectiveness of a device 200 mm thick in height and 100 mm thick in width has been assessed in different operating conditions. Results show that the coupling of an appropriate number of conductive elements and insulating layers can limit the roof temperature effectively. For example, the presence of a conductive element reduces the roof temperature from 140◦C to 70◦C with a device 5 cm thick when the exhaust gas temperature is 500◦C. A device made of an insulating layer 100 mm width and two conductive elements keeps the roof temperature at 61◦C when the exhaust gas temperature is maintained at 650◦C until the achievement of the steady condition and it is risen at 970◦C for 30 minutes.
Keyword
Chimney ; Fire safety ; Roof fire ; Soot fire ; Wooden structures ; Chimney-roof penetration ; Device
Introduction
In recent years many roof fires have occurred in Europe due to the presence of a
1
chimney and also chimneys certified according to the EN 1859 [1] standard were
2
involved in the fires [2–6].
3
The aim of the certification procedure described in [1] is the determination of
4
the safety distance between chimney and flammable materials and the chimney
5
temperature class, that is, the maximum temperature of the exhaust gas temper-
6
ature in the chimney. The certification procedure consists of two tests, the heat
7
stress test (HST) and the thermal shock test (TST). The HST test reproduces
8
the normal use of chimneys. The TST test reproduces the condition during soot-
9
fire events, a less frequent but more critical condition in which soot deposited on
10
chimneys inner surface burns rising the exhaust gas temperature up to 1100◦C for
11
few minutes [7–9]. In both the tests, the chimney to be tested must be installed in
12
the structure shown in Figure 1 a) made of two walls at right angle and two roofs
13
positioned at different heights. The lower roof is 232 mm thick and its thermal
14
resistance is 5.90 m2K/W, and the upper roof is 132 mm thick and its thermal
15
resistance is 3.04 m2K/W. The tests consist in feeding the chimney to be certified
16
with exhaust gas at a predetermined temperature until the achievement of the test
17
final condition, and in measuring the temperature of the test structure in several
18
positions. For the HST test, the exhaust gas temperature ranges between 100◦C
19
and 700◦C according to the chimney temperature class, and the final test condi-
20
tion is reached when the variation in temperature measured on the test structure
21
is less than 2◦C in 30 minutes. As regards the TST test, exhaust gas at 1000◦C
22
is fed in the chimney for 30 minutes. Chimneys are certified if the temperature
23
measured on the test structure is lower than 85◦C and 100◦C in the HST test and
24
in the TST test respectively. The result of the certification procedure is a label
25
reporting the chimney temperature class and the safety distance from flammable
26
Fig. 1:Test structure prescribed by the EN 1859 standard [1] a), and test structure used in the experiments presented in this paper b). Position of the device in the clearance and variables analyzed in the paper c). Dimensions are in centimeters.
materials. According to the standard [1], installing a chimney respecting the safety
27
distance avoids the overheating of nearby structures A certified chimney can be
28
installed in any roof, as long as the distance from flammable materials and the
29
class temperature are observed.
30
The number of chimney fires recorded in different European areas and in differ-
31
ent periods are reported in Table 1. In Finland, the 73% of metal chimney fires oc-
32
curred at roof penetration, while the 22% occurred at a wall penetration [4,14,15].
33
The 30% of the total number of fires was due to defective insulation, the 29% to
34
Table 1:Number of chimney fires recorded in different European areas.
Place Period Number of chimney fires Reference
Brescia (Italy) 2008 308 [3]
England 2015−2016 4193 [10]
Wales 2015−2016 432 [10]
Northern Ireland 2016-2016 1225 [10]
Poland 2010-20015 47414 [11, 12]
Norway 2017 1139 [13]
Finland 2012 126 [4]
defective minimum safety distance, the 20% was caused by overheated chimney,
35
and 10% to other causes such as rust away, and joint of the chimney. As regards
36
fires occurred in Brescia district (Italy), in a fire brigade report [3] it was pointed
37
out that the 53% was due to an incorrect chimney installation, the 18% to a weak
38
or totally absent maintenance of the chimney, and the 17% occurred during the
39
construction phase. In particular, the incorrect installation mode is represented by
40
one or more of the following conditions:
41
– Chimney temperature class lower than the temperature of exhaust gas from
42
the heating system;
43
– Presence of flammable materials in the vicinity of the chimney at a distance
44
lower than that declared in the certification label;
45
– Chimney not tested for soot fire condition and connected to a heating system
46
that burns solid fuel;
47
– Chimney installed in a not correct way with consequent hot external surface
48
(the external surface temperature is higher than that measured during certifi-
49
cation tests).
50
The high number of roof fires as resulted in some documents about the concern on
51
fire safety risk due to the CE1 marking of chimneys products [2]. Consequently, a
52
study has begun to investigate the roof fires problem and to propose solutions.
53
According to some studies in the literature [14, 16–22] described in detail in
54
the following paragraph, real chimney operating conditions may be exceptionally
55
severe and lead to the overhating of the nearby structures. It was also shown that
56
these conditions can be more severe than those prescribed by the EN 1859 [22]
57
even if the acceptance criteria proposed in [22] is universally recognized. For exam-
58
ple, the roof in which a chimney is installed can be very thick and very insulating.
59
Nowadays, roofs can be made of standard materials but also materials based on
60
biomass such as cellulose [23,24], wood fiber [25], hemp [26], and straw [27], whose
61
self ignition temperature may be very low. For example, wood self ignition temper-
62
ature was defined ranging between 250◦C and 300◦C for short time exposure [28],
63
66◦C in [29, 30], and 100◦C [31]. However, in [32, 33] spontaneous and piloted igni-
64
tion temperatures of convectively heated wood are defined ranging between 492◦C
65
and 452◦C. Paper self ignition temperature ranges between 204◦C and 370◦C [34].
66
Some of the more traditional insulating materials, such as PIR (Polyisocyanurate),
67
polystyrene, may undergo to softening processes with a consequent lost of proper-
68
ties. As regards polystyrene, its self ignition temperature is about 488◦C [35] and
69
1 CE marking ensures that products are in compliance with legislation in force in the Euro- pean Community.
its softening temperature is 100◦C [36]. Therefore, guidance given by the label
70
reported on certified chimneys is often ignored or not understood. Moreover, the
71
certification procedure does not allow to test chimneys in the most critical oper-
72
ating condition. For this, it has was felt the necessity of a device able to limit the
73
roof temperature also in very critical chimney operating conditions. These latter
74
are represented by exhuast gas at high temperature, very thick and very insulat-
75
ing roofs, and soot fire events after a certain period of functioning of the heating
76
system.
77
Heat transfer at chimney roof penetration: state of art
78
Chimneys must be installed by qualified experts; however, a certain degree of free-
79
dom in the installation phase is possible. For example, since in the label reported
80
on certified chimneys it is not specified how to seal the space between chimney
81
and roof (the clearance), the installation mode in real buildings may be differ-
82
ent from the one in the certification procedure, and this aspect was investigated
83
experimentally in [18, 22]. Three roofs and four clearance sealing modes were con-
84
sidered: one roof was representative of the most insulating roof prescribed by the
85
standard [1], and the other two were representative of roofs in energy-saving build-
86
ings characterized by greater thickness and thermal resistance. The clearance was
87
left open, sealed with metal sheets, sealed with insulating panels, and filled with
88
insulating material respectively. Since in real installations chimneys are usually
89
installed completely surrounded by a roof and not in the vicinity of two walls as
90
in HST and TST tests, the influence of the chimney position in the roof was inves-
91
tigated. In [22] the chimney was installed in a corner test structure, while in [18]
92
the chimney was installed completely surrounded by a roof (axi-simmetric test
93
structure), in order to limit the heat transfer from the chimney to the ambient. It
Fig. 2:Representation of the heat transfer at the chimney-roof penetration: a) with insulating material in the clearance, b) with insulating material and conductive elements (CEIL device) in the clearance.
was found that the temperature at the chimney-roof penetration can be very high
94
if the clearance is filled with insulating material, or if the chimney is positioned
95
at the center of the roof. If a chimney is tested with the clearance open, but in
96
real installations it is installed by filling the clearance with insulating material, an
97
overheating may occur. To investigate the roof fires problem extensively, 2D and
98
3D numerical models were defined [17]. By comparing the temperature measured
99
experimentally according to [1] and those estimated numerically, it was found that
100
the temperature measured at the end of the HST test may be lower than the actual
101
steady temperature: the difference is mainly affected by the roof characteristics
102
and the clearance sealing mode [19]. To account for this, it was proposed to esti-
103
mate the steady temperature from the temperature-time curves measured in the
104
tests by means of theHeating Curve Model presented in [17].
105
The studies allowed an understand of the heat transfer at the chimney-roof
106
penetration and it emerged that it depends on:
107
– The exhaust gas temperature. The higher the exhaust gas temperature, the
108
higher the roof temperature.
109
– The roof thickness. Higher temperature are measured in thicker roofs.
110
– The roof layers position (wooden and insulating layers). For a roof of given thick-
111
ness and thermal resistance, a higher temperature occurs when the insulating
112
layer is placed at the top.
113
– The distance between chimney and roof. The greater the distance between chim-
114
ney and roof, the lower the roof temperature.
115
– The chimney thickness. A thicker thickness entails a lower roof temperature.
116
– The chimney thermal resistance. A greater thermal resistance entails a lower
117
roof temperature.
118
– The method of sealing the clearance. The clearance sealing mode strongly af-
119
fects the roof temperature. In particular, the highest roof temperature is related
120
to a clearance filled with insulating material, whereas the lowest temperature
121
is related to a open clearance.
122
Other studies tried to highlight important aspects of the heat transfer at the
123
chimney-roof penetration. The increase in temperature at the chimney-roof pen-
124
etration due to the smoldering combustion of some binder materials was inves-
125
tigated in [20, 21]. It was found that it occurs only the first time that a certain
126
temperature is achieved, and the consequent increase in the roof temperature can
127
be 250◦C. The performance of the heating generators connected to chimneys was
128
investigated in [37–39]: the exhaust gas temperature in the chimney was analyzed
129
to evaluate the efficiency of combustion.
130
By comparing the conditions in real installations and those in the certifica-
131
tion procedure, Lepp¨anen et al. [40] proposed modifications to the certification
132
procedure, and the main points are:
133
– The position of the chimney in the test structure. It was proposed to install
134
the chimney at the center of the roof so as to limit the dissipation of heat
135
towards the ambient and, consequently, to reproduce the worst condition.
136
– The exhaust gas measurement point. It was suggested to measure the exhaust
137
gas temperature in the vicinity of the chimney-roof penetration (Tgas1 and
138
Tgas2 in Figure 1 b) instead of in the vicinity of the heat generator (Tgas∗ in
139
Figure 1 a).
140
– The maximum temperature measured in the tests. It was suggested to install
141
the thermocouples in the clearance vertically, because temperature may vary
142
significantly along the vertical direction, especially if the clearance is filled.
143
– The initial condition of the TST test. Given that a soot fire may occur after a
144
certain period of functioning of the heating system connected to the chimney,
145
it was suggested to perform the TST test immediately after the HST test.
146
– The materials used in the test structure As air infiltration through the mate-
147
rials has a strong influence on the roof temperature [17, 40], it was suggested
148
to reduce them as much as possible by using stiff materials.
149
A set of safety measures were proposed with the aim to reduce the risk of
150
fires due to the presence of a chimney, as described in the following. Given the
151
high number of variables, the problem was analyzed statistically to obtain a se-
152
ries of tables for assessing the maximum temperature at the chimney-roof pen-
153
etration in normal chimney operating conditions [41]. Other solutions consist in
154
high-performing chimneys [42], and devices to be installed at the chimney-roof
155
penetration [43–47]. The majority of these devices has an additional insulating
156
layer, or systems that exploit air flows that reduce the roof temperature. A device
157
to be installed in any roof that can prevent the roof overheating was numerically
158
investigated in [48]. The device must be installed as shown in Figure 1 c). The
159
result was an innovative device called CEIL (Conductive Elements and Insulating
160
Layer). Differently from the devices in the market that try to reduce the heat
161
flux from the chimney towards the roof by mean of insulating material only, the
162
CEIL device reduces the heat flux towards the roof by increasing the heat flux
163
dissipated in the ambient. The operating principle of the CEIL device is shown
164
in Figure 2. If only insulating material is installed between chimney and roof the
165
majority of the heat flux from the chimney flows by conduction towards the roof
166
(Figure 2 a). If also conductive elements are installed in the insulating layer, part
167
of the heat flux is intercepted and conveyed towards the ambient where it can
168
be dissipated by convection and radiation (Figure 2 b). The conductive element
169
consists of a vertical surface partially immersed in the insulating layer and a wing
170
facing the ambient. The variation in temperature due the shape and the number
171
of conductive elements was investigated numerically in [48] The study focused on
172
the variation in temperature obtained with different device configurations because
173
materials self ignition temperature depends on several variables such as physical
174
and chemical properties, and specimen size [49,50]. Positioning the wing at the top
175
reduces the roof temperature because in this position the convective heat transfer
176
coefficient is greater: due to buoyancy, at the top of the roof hot air is immediately
177
replaced by colder air, while at the bottom this replacement is more difficult. It
178
was shown that the higher the number of conductive elements, the lower the roof
179
temperature. Therefore, it was proposed to install the conductive elements both
180
at the top and at the bottom of the device. To make the CEIL device suitable
181
also for energy-saving buildings [51–53], it was suggested to install the upper and
182
the lower conductive elements spacing of some centimeters in order to avoid the
183
formation of thermal bridges between indoor and external ambient.
184
Table 2: Tests details. According to Figure 1 c),Scis the thickness of the chimney,Tchis the exhaust gas temperature measured at the chimney-roof penetration,Tenvis the ambient temperature,SDis the horizontal thickness of the device, andSvis the vertical thickness of the device.
Test Conductive Test Tch Tenv Sc SD SV Test
name element condition [◦C] [◦C] [◦C] [mm] [mm] configuration
P0-WC no HS 605 19 50 50 200 Figure 3 a)
P1-WC no HS 500 13.9 100 40 240 Figure 3 b)
P1-D1 D1 HS 500 14.6 100 40 240 Figure 3 b)
P2-WC no HS-SF 500 – 1000 28.7 100 100 200 Figure 3 c)
P2-D2 D2 HS-SF 500 – 980 27 100 100 200 Figure 3 c)
P3-D2 D2 HS-SF 650 – 970 26.5 100 100 200 Figure 3 c)
P4-D2 D2 SF 900 30 100 100 200 Figure 3 c)
Table 3:Main differences between the conditions in the certification procedure proposed in the EN 1859 standard [1] and those in the tests presented in this paper.
Standard procedure [1] Modified procedure
Position of the Near two walls At the center
chimney in the at right of the roof
test structure angle
Initial condition of the From the Immediately after
soot fire test ambient temperature the HS test
Tchmeasurement point Near the heat generator In the vicinity of the roof
Thermocouples Horizontal Vertical
position
Roof thermal R=5.9 m2K/W R=10.1 m2K/W (roofR0)
resistance R=10.53 m2K/W (roofR1)
R=11.52 m2K/W (roofR2) Final test After 8 hours or Achievement of stationary condition T increase≤2◦C/ 30 minutes temperatures orHeating
curve model[17]
Aim of the study
185
The shape of the CEIL device was identified numerically in [48] because an ex-
186
perimental approach would had been too honerous. The aim of this paper is the
187
verification of the efficacy of the CEIL device by means of experimental tests. The
188
study is not a validation of previous results; instead, it aims to investigate aspect
189
not easy to analyze experimentally. The experimental tests were performed by
190
taking into account all the findings about chimney-roof penetration reported in
191
the literature. The study focuses on the variation in temperature due to different
192
operating conditions and device configurations. The experimental tests were per-
193
formed at the FIRE laboratory at the Tampere University (Finland), and at the
194
ANFUS laboratory of Brescia (Italy).
195
1 Methods and results
196
To assess the efficacy of the CEIL device and to improve its efficacy, an experi-
197
mental campaign characterized by three phases has been performed. Firstly, the
198
investigation evaluated the influence of thermal radiation from the chimney on an
199
unprotected roof. Indeed, the CEIL device is conceived as a universal device of
200
fixed vertical and horizontal thicknesses [48] and, consequently, it can be more or
201
less thick than the roof in which it is installed. For example, according to Figure 1
202
c), if the thickness of the device is 200 mm (SV) and it is installed in a roof 400 mm
203
thick (Hr), 200 mm of roof remain unprotected from thermal radiation. Then, in
204
the second phase, the investigation evalauted whether the presence of a conductive
205
element in the insulation between the chimney and the roof can effectively reduce
206
the temperature at the chimney-roof penetration. Finally, the performance of the
207
device in different exhaust gas temperature (Tch) was evaluated.
208
The tests are summarized in Table 2. In the tests identified with the acronym
209
WC, only insulating material has been installed in the clearance. In tests identified
210
with the acronyms D1 and D2, conductive elements have been installed in the
211
clearance. According to Figure 3 b), device D1 consists of an insulating layer
212
40 cm thick (SD) and a conductive element made of aluminum. The conductive
213
element consists of a vertical surface connected to a wing facing the ambient at
214
the top. According to Figure 3 c), device D2 is a more elaborate version of the
215
CEIL device, and it is made of a layer 10 cm thick (SD) of mineral wool and two
216
conductive elements made of copper. The conductive elements consist of upper
217
and lower parts spaced 3 cm: in this way, thermal bridges between the external
218
and the indoor ambient are reduced.
219
As prescribed in the EN 1859 [1] standard, the tests should have consisted of
220
feeding exhaust gas at a predetermined temperature in the chimney and measuring
221
the temperature at the chimney-roof penetration in different operating conditions.
222
However, to reproduce the most critical chimney operating conditions, the tests
223
were performed in a slightly different way as reported in Table 3.
224
Test structure
225
The test structure is shown in Figure 1 b). It consists of a heat generator, an
226
interchangeable roof, and two connecting flue pipes (one to connect the chimney
227
to the heat generator, and the other to convey the exhaust gas away from the test
228
structure).
229
Three roofs have been installed in the test structure. Roof R0 is shown in
230
Figure 3 a) and it is made of a lower wooden layer 17 mm thick and an upper
231
layer made of EPS (λ=0.040 W/mK) 400 mm thick. RoofR1 is shown in Figure
232
3 b) and it is made of a lower wooden layer 17 mm thick and an upper layer 240
233
mm thick made of PIR (λ=0.023 W/mK). RoofR2 is shown in Figure 3 c) and
234
it is made of a lower wooden layer 20 mm thick, and an upper layer 400 mm thick
235
made of mineral wool. The thickness of the wooden layer of roofsR1 andR2 is
236
slightly different (17 mm for roofR1 and 20 mm for roofR2) but this is irrelevant
237
for the purposes of the study. However, by comparing roofsR0, R1, R2 and those
238
prescribed in the standard [1], it emerges that the first ones are thicker and more
239
insulating.
240
Two chimneys were installed in the test structure. The chimney in Figure 3 a)
241
is made of two layers of the same thickness made of mineral wool and ceramic fiber
242
and the total thickness is 50 mm (Sc), whereas the chimney in Figures 3 b) and c)
243
is made of Rockwool 100 mm thick (Sch) and it is representative of well-insulating
244
chimneys usually used at the chimney-roof penetration. In some cases, they may
245
Fig. 3:Test configuration forP0-WCtest (a), forP1-WC, P1-D1tests (b), and forP2-WC, P2-D2, P3-D2 andP4-D2 tests (c). Dimensions are in millimeters.
be installed in contact with flammable materials. The thermal conductivity of the
246
mineral wool used in the tests isλ=0.035 W/mK [54].
247
Fig. 4: Assembling phases of the test structure forP1-D1 test. The conductive element has been positioned around the chimney a). The conductive element has been bandaged with insulating material b). The roof has been positioned around the elements c).
Test procedure
248
Different chimney operating conditions have been reproduced in the tests, that
249
is, normal functioning condition, soot fire condition from ambient temperature,
250
and soot fire condition after a certain period of functioning of the heat generator
251
connected to the chimney. This latter is a condition more severe than that in the
252
certification procedure where the TST test starts from ambient temperature. The
253
tests aimed to reproduce the normal functioning of chimneys, when exhaust gas
254
temperature is constant between 500◦C and 700◦C are denoted with HS. The
255
tests aimed to reproduce the soot fire conditions characterized by an exhaust gas
256
temperature between 900◦C and 1000◦C for 30 minutes, are denoted with SF.
257
Tests consisted in maintaining the exhaust gas temperature at a predetermined
258
temperature and then rising in temperature to 1000◦C and maintained for 30
259
minutes are identified with the acronymHS-SF.
260
Temperatures have been measured by thermocouples type K and two DAQ NI
261
9213 connected to a PC. Tipe K thermocouples have an uncertainty of±2.8◦C for
262
temperatures ranging between 0◦C and 350◦C and of ±0.75% for temperatures
263
ranging between 350◦C and 1260◦C. The uncertainty of the DaQ is ±0.02◦C.
264
Fig. 5:Temperatures measured inP0-WC test. ThermocouplesT01,T02 and T03 are ex- posed to thermal radiation from the chimney and convection. ThermocouplesT1,T2,T3,T4 andT5are immersed in the insulating material and they are affected by conduction only. The temperature of the external surface of the chimney was 102◦C (thermocoupleC1in Figure 3 a).
Temperatures were recorded every 10 seconds. The exhaust gas temperature has
265
been regulated so as to ensure the predetermined temperatureTchat the chimney-
266
roof penetration: according to Figure 1 b), Tch = (Tgas1+Tgas2)/2. It was not
267
always possible to achieve steady conditions, therefore steady temperature was
268
estimated using theHeating curve model presented in [17]. From the temperature-
269
time curves measured in the tests, the steady temperatureTf can be determined
270
iteratively from equation:
271
e−lt= (Ti−Tf)/(Ts−Tf) (1) where t is the time instant, l is an unknown term to be determined, Ts is
272
the initial temperature, andTi is the temperature at the instantt. The termsTf
273
andl must be the ones that give the highest R2 coefficient. Because of test du-
274
ration, it was not possible to repeat the tests. To account for this, temperatures
275
were measured in homologous points as shown in Figure 3. The final temperature
276
calculated is the average of the temperatures measured or estimated at the homol-
277
ogous points, and the error bar has been estimated by summing the uncertainty
278
of the measurement instruments, and the difference between the maximum (Tmax)
279
and minimum (Tmin) values, that is, (Tmax−Tmin)/2.
280
1.1 Tests details
281
Influence of thermal radiation on unprotected roof
282
The influence of the radiative heat flux on an unprotected portion of roof has
283
been investigated by means ofP0-WC test. The test consisted of feeding exhaust
284
gas at 605◦C in the chimney until the achievement of the test final condition.
285
The test configuration is shown in Figure 3 a). In this test no conductive element
286
Fig. 6:Temperature-time curves measured in testP1-WC a) and testP1-D1 b). Each curve has been determined by averaging the temperature measured at the same distance from the chimney.
was installed in the clearance. The peculiarity of this test is that the insulating
287
material in the clearance does not extend as much as the roof; along the vertical
288
direction, the insulating layer in the clearance is 200 mm (Sv), and the roof is 400
289
mm thick (Hr), therefore 200 mm of roof remained exposed to thermal radiation.
290
To guarantee a sufficiently high temperature on the external chimney surface, the
291
chimney has a metallic surface insulated with a layer 50 mm thick. Thermocouples
292
have been positioned on the chimney external surface (C1), on the unprotected
293
portion of the roof (T01, T02, T03) and between the roof and the insulating layer
294
in the clearance (T1, T2, T3, T4, T5). ThermocouplesT01, T02, T03 are exposed
295
to thermal radiation from the chimney and to convection, while thermocouples
296
T1, T2, T3, T4, T5 are affected by thermal conduction only. Thermocouples were
297
positioned in these positions, in order to obtain a sort of repetition of the mea-
298
surement. If the temperature measured by the unprotected thermocouples is much
299
higher than the ambient temperature the effect of thermal radiation is strong and
300
it is not possible to design a device of fixed vertical thickness. On the contrary, if
301
the temperature measured on the unprotected thermocouples approaches the am-
302
bient temperature, the effect of the thermal radiation is limited by the contribution
303
of the convective component. Temperatures are shown in Figure 5.
304
Device with one conductive element
305
To verify whether the presence of a conductive element in the clearance can ef-
306
fectively limit the roof temperature,P1-WC andP1-D1 tests were performed. In
307
P1-WC test no conductive element was installed between the chimney and the
308
roof, while in the P1-D1 test a conductive element made of aluminum was in-
309
stalled as shown in Figure 3 b). The vertical part of the conductive element has
310
been installed in contact with the chimney, and the external wing spaced 3 cm
311
from the insulating layer in the clearance: in this way, convection occurred on both
312
sides of the wing. To perform P1-D1 test, the conductive element was removed,
313
while the other conditions not changed. Since it was shown that air infiltration
314
strongly affects the roof temperature [17], and the aim of these two tests is to
315
reproduce the worst chimney operating condition, the roof was made of PIR that,
316
given its stiffness, has limited air infiltration. To ensure a perfect contact between
317
the different layers, the test structure was assembled as shown in Figure 4. Firstly,
318
the conductive element was positioned around the chimney. Then, the conductive
319
element has been bandaged with insulating material. Finally, the roof was posi-
320
tioned around the elements. To ensure a good contact between the different layers,
321
the roof was cut into two parts placed side by side.
322
Thermocouples were positioned in the device and in the roof as shown in
323
Figure 3 b). They were positioned between the chimney and the device, that is,
324
on the internal surface of the conductive element (Line 0), on the internal surface
325
of the conductive element (Line 1), and between the device and the roof (Line
326
2). Regarding thermocouples alongLine 0 andLine 1, thermocoupleT5 is not in
327
contact with the conductive element. In the roof, thermocouples were positioned
328
at several distances from the chimney, that is, at 5 cm (Line A), at 10 cm (Line
329
B), at 15 cm (Line C), and at 20 cm (Line D). Thermocouples were immersed in
330
the roof by perforating the PIR. This procedure resulted in a small uncertainty
331
on the final position of the deepest thermocouples. The tests reproduced the HS
332
condition, and exhaust gas was maintained at 500◦C (Tch) for 7 hours. Figure
333
6 shows the temperature-time curves measured during the tests; each curve was
334
obtained by averaging the temperatures measured at the same distance from the
335
chimney. Steady temperatures estimated with Equation 1 are shown in Figure 7.
336
The temperature measured in the device in theP1-D1 test are shown in Figure 8.
337
Figure 9 shows the roof surface at the end of theP1-WC test.
338
Fig. 7:Estimated steady temperatures forP1-WC (without conductive element) andP1-D1 (with conductive element) tests.
Device with two conductive elements
339
This set of tests was conducted in a traditional way; the reduction of air infiltra-
340
tion was ignored. For this, traditional materials that do not completely prevent
341
air infiltration were used. To assess the influence of two conductive elements in
342
the clearancP2-WC test was performed with only insulating material within the
343
clearance, while inP2-D2 test two conductive elements were installed in the clear-
344
ance as shown in Figure 3 c). Also inP3-D2, P4-D2 the conductive elements were
345
installed in the clearance. As shown in Figure 3 c), thermocouples were positioned
346
Fig. 8:Temperature measured in testP1-D1. The thermocouples position is shown in Figure 3 a):Line 0is between the chimney and the device,Line 1 is on the internal surface of the conductive element, andLine 2 is on the external surface of the conductive element.
on the chimney’s external surface (Line 00), on the external conductive element
347
(Line 0A), and between the device and the roof (Line 0B). InP2-WC andP2-D2
348
tests, the exhaust gas temperature was maintained at 500◦C until the test final
349
condition and, then, it was increased up to about 1000◦C for 30 minutes. Roof
350
temperatures are compared in Figure 11. P3-D2 test was performed in the same
351
way, except the exhaust gas temperature in the first phase was maintained at
352
650◦C, that is, 150◦C higher than in the previous tests. Roof temperatures mea-
353
sured inP3-D2 are shown in Figures 11 and 12. InP4-D2 test, the exhaust gas has
354
been maintained at 900◦C for 30 minutes. Even if the temperature exhaust gas
355
temperature is 100◦C lower than the temperature in TST test prescribed by the
356
standard [1], this test has allowed to assess the effect of the device in a dynamic
357
condition. Temperatures measured in P4-D2 test are shown in Figure 13 where
358
they are compared with those measured in the previous tests at the end of theHS
359
condition.
360
2 Discussion
361
The temperatures measured in the tests were discussed to assess the efficacy of one
362
or more conductive elements installed in the clearance between chimney and roof
363
and, consequently, of the CEIL device. The focus is on the variation in temperature
364
obtained with different device configurations and in different chimney operating
365
conditions.
366
Influence of thermal radiation on unprotected roof
367
To assess the influence of thermal radiation on unprotected roof, temperatures in
368
Figure 5 are analyzed. The temperature on the chimney external surface (thermo-
369
coupleC1 in Figure 3 a) was about 102◦C. Temperatures measured by thermo-
370
Fig. 9:Roof surface at the end of testP1.
couplesT01, T02, T03, that is, those subject to thermal radiation and convection
371
from the chimney, is lower than 33◦C, which is only 14◦C higher than the ambient
372
temperature (Tenv=19◦C). Temperatures measured by the immersed thermocou-
373
ples are higher and they approach 70◦C. This highlights that thermal radiation
374
from the chimney does not significantly affect the roof temperature: convection re-
375
moves the heat from the roof, even if air does not flow through the entire clearance
376
because of the insulating layer. The difference between the temperature measured
377
by the unprotected thermocouples and those measured by the thermocouples im-
378
mersed in the insulating layer (T1, T2, T3, T4, T5) is because the latter are not
379
cooled by convection. This is in agreement with [17] that stated that leaving the
380
clearance open is the best way to reduce the roof temperature. However, the dis-
381
tance between the chimney and the roof must provide a sufficient movement of
382
air. For this, if the chimney is sufficiently insulated (about 5 cm of mineral wool)
383
and the distance between the chimney and the roof is about 50 mm, there is no
384
risk of overheating by radiation. Therefore, the idea of a universal device with a
385
defined vertical extension seems achievable.
386
Device with one conductive element
387
Figure 6 shows the temperature-time curves measured inP1-WC andP1-D1 tests.
388
The curves have been obtained by averaging the temperatures measured at the
389
same distance from the chimney, namely at 5 cm, 10 cm, 15 cm and 20 cm. It
390
can be seen that if a conductive element is installed in the clearance, the roof
391
temperature is lower and the steady condition is achieved earlier.
392
Fig. 10: Temperature measured inP2-WC andP2-D2 tests at the end of theHScondition characterized by exhaust gas at 500◦C (Tch). According to Figure 3 b), temperatures denoted withLine 00 have been measured between the chimney and the insulating layer in the clear- ance, temperatures denoted withLine 0Ahave been measured at the center of the insulating layer, and temperatures denoted withLine 0B have been measured between the insulating layer in the clearance and the roof.
Figure 7 compares the steady roof temperatures estimated with Equation 1 for
393
the cases with (P1-D1 test) and without (P1-WC test) the conductive element in
394
the clearance. The chimney used in this test is made of 100 mm of mineral wool:
395
this dimension is usually used at the chimney-roof penetration and, according to
396
some chimney producers, can be installed in contact with flammable materials.
397
The measured temperatures obtained without the conductive element are shown
398
on the left side hand of Figure 7. It can be seen that, when no conductive element
399
is installed in the clearance, the roof temperature exceeds the limit of 85◦C pre-
400
scribed by the standard [1]. Because temperatures denoted withLine Ahave been
401
measured 50 mm from the chimney’s external surface, which is greater distance
402
than that recommended by the producer, it can be stated that reducing air infil-
403
tration in the clearance and in the roof may cause roof overheating. For example,
404
the temperature of 139◦C measured alongLine Acould represent a dangerous con-
405
dition that may lead to softening and blackening (pre-charring) processes if the
406
roof is made of PIR. This is confirmed also by Figure 9 that shows the surface
407
of the roof at the end of the P1-WC test: the roof has dilated and some of the
408
roof has been affected by the blackening process. However, the high temperature
409
in the roof is probably also due to the roof characteristics - roof R1 is more in-
410
sulating than the roofs prescribed in the standard [1]. Therefore, it can be stated
411
that if real operating conditions are different from those in the certification pro-
412
cedure, installations may not be safe. Moreover, this shows that the conditions
413
reproduced in the tests presented in this paper are more critical than those in the
414
tests prescribed by the standard [1].
415
By comparing the temperatures shown on the left side hand in Figure 7 with
416
those on the right side hand, the influence of the conductive element in the clear-
417
Fig. 11: Roof temperature measured in P2-WC and P2-D2 tests at the end of the HS condition characterized by exhaust gas at 500◦C, and at the end of theSF condition where the exhaust temperature has been maintained at about 975◦C for 30 minutes.
Fig. 12:Temperature measured inP3-D2test. In theHScondition exhaust gas temperature gas has been maintained at 650◦C, in theSF condition the exhaust gas temperature has been maintained at 970◦C for 30 minutes.
ance can be assessed. It is worth noting that the reduction in calculated temper-
418
ature is the difference between the maximum temperatures measured with and
419
without the conductive element is 72◦C at 5 cm from the chimney (Line A), 43◦C
420
at 10 cm (Line B), 16◦C at 15 cm (Line C), and 2◦C at 20 cm (Line D). This is
421
in agreement with the numerical findings in [48] where the difference in tempera-
422
ture was of 60◦C in the same condition of the test, except for the position of the
423
external wing of the conductive element that in the numerical test was in contact
424
Fig. 13: Roof temperature measured inP2-D2,P3-D2 and P4-D2 tests. For P2-D2 and P3-D2tests temperatures are those measured at the end of theHScondition.P4-D2 test has lasted 30 minutes only.
with the insulating layer in the clearance. The contact between the wing and the
425
insulating layer reduces the heat exchange surface, and a higher roof temperature
426
is expected. In both the P1-D1 andP1-WC tests, the highest temperature was
427
measured at the center of the roof (thermocouples T3 andT4), except for the
428
values measured at 15 cm where the higher temperature has been measured by
429
thermocouple T5. This is probably due to a thermocouple displacement because
430
thermocouples were immersed in the roof by perforating the PIR, as explained
431
in Section 1.1. As the change temperature measured with the conductive element
432
are flatter, it emerges that the conductive element removes the heat also from the
433
inner part of the insulation within the clearance. At the contrary, if no conductive
434
element is installed in the clearance, the inner part of the insulating layer can not
435
be cooled because it is too far from the point where convection takes place, that
436
is, the upper and the lower surfaces of the roof. These results confirm that insu-
437
lating the chimney as much as possible is not the correct way for making chimney
438
installations safer unless very thick insulating layers are installed. On the other
439
hand, the presence of the conductive element keeps the roof temperature lower.
440
To understand how the device works, temperatures measured inP1-D1test and
441
shown in Figure 8 are analyzed. In the analysis only temperaturesT1,T2,T3 and
442
T4 are considered because thermocoupleT5 has not been protected by the con-
443
ductive element, and a higher temperature was expected at this point.Line 0 and
444
Line 1 are separated by a layer 3 mm thick made of aluminum, whileLine 1 and
445
Line 2 are separated by a layer 20 mm thick made of mineral wool. It can be seen
446
that the difference in the maximum temperature between the internal (Line 0) and
447
the external surface (Line 1) of the conductive element is about 7◦C. This value
448
is comparable with the difference in temperature measured between the external
449
surface of the ring (Line 1) and the roof (Line 2). In other words, the difference in
450
temperature measured through 3 mm of aluminum is comparable to the difference
451
in temperature measured through 20 mm of mineral wool. This is because the
452
conductive element intercepts the heat flux from the chimney and, given its rela-
453
tively high thermal conductivity, conveys the heat towards the ambient where it
454
can be dissipated. Therefore, the conductive element reduces the temperature im-
455
mediately near the chimney. To limit the roof temperature in a limited space, the
456
coupling of conductive and insulating elements is the right approach. Therefore,
457
the presence of a conductive element limits the roof temperature.
458
Device with two conductive elements
459
Figure 10 shows the temperature measured in the clearance inP2-WC andP2-D2
460
tests at the end of theHScondition. It can be seen that the conductive elements
461
lower the roof temperature significantly. The difference in temperature with and
462
without the conductive element is 102◦C between the chimney and the device
463
(Line 00), 63◦C at the center of the device (Line 01) and of 35◦C in the roof
464
(Line 0B). By comparing the roof temperature measured inP1-D1 test (Figure 7)
465
andP2-D2 test (Figure 10), it can be seen that the roof maximum temperature
466
is 95◦C in the first test with one conductive element, and 28◦C in the second test
467
with two conductive elements. Then, even if the test conditions were different, it
468
can be stated that with two conductive elements the heat from the chimney is
469
mostly dissipated in the ambient. The difference in temperature is due to both the
470
number of conductive elements and because in the second case the insulating layer
471
is less thick (SV) and air infiltration has been prevented. The higher the number
472
of conductive elements, the higher the heat flux intercepted and conveyed to the
473
ambient. By reducing the vertical thickness of the device, the path for the heat
474
flux to be dissipated in the ambient is reduced. Indeed, in [18, 19, 22] it was shown
475
that if the clearance is filled with insulating material and the roof is made of only
476
a single material, the maximum temperature occurs at the center of the roof. A
477
smaller vertical thickness of the device corresponds to a lower distance between
478
the inner part of the device and the surfaces where convection takes place.
479
Figure 11 compares the roof temperature measured at the end of theHS and
480
the SF conditions in the P2-WC and P2-D2 tests. The increase in temperature
481
between theHScondition and theSF condition is 32◦C inP2-WC test, and 11◦C
482
in the P2-D2 test. By comparing the temperature measured in the two tests at
483
the end of theSF condition, it emerges that the difference is 57◦C. Therefore, the
484
conductive elements reduces the heat flux to the roof and, consequently, the roof
485
temperature. It is interesting noting that without the conductive elements the roof
486
temperature is 96◦C, only 4◦C lower than the limit temperature prescribed by the
487
standard [1], whereas the presence of the conductive elements in the clearance
488
maintains the temperature at 39◦C. Therefore, the conductive elements reduces
489
the effect of a sudden increase in the gas temperature.
490
Figure 12 shows the roof temperature measured in P3-D2 test. It can be seen
491
that, even with a higher exhaust gas temperature, the roof temperature does not
492
exceed the limits of 85◦C and 100◦C prescribed by the standard [1]. The increase
493
in the exhaust gas temperature of 320◦C in the SF condition corresponds to an
494
increase in the roof temperature of only 14◦C. By comparing these temperatures
495
with those measured inP2-D2 test and reported in Figure 11, it emerges that an
496
increase in the exhaust gas temperature of 150◦C corresponds to an increase in the
497
roof temperature of only 19◦C in theHS condition, and 22◦C in theSFcondition.
498
Since this test reproduces the most critical chimney operating conditions, it can
499
be stated that if the CEIL device is installed on a 100 mm thick (Cc) the roof
500
temperature does not achieve dangerous values.
501
Temperatures measured inP4-D2 are shown in Figure 13. It can be seen that
502
the roof temperature does not exceed the limit of 100◦C. By comparing these
503
temperatures with those measured in the other tests, it emerges that inP4-D2test,
504
despite the exhaust gas temperature at 900◦C for only 30 minutes, the temperature
505
in the device is higher than in the other two tests. This shows that the device is
506
more sensitive to fast changes in temperature. However, even if in theP4-D2 test
507
the temperature measured in the vicinity of the chimney (Line 0) is much higher
508
than the that measured in the other two tests, the roof temperature does not vary
509
significantly. By comparingP4-D2 andP3-D2 tests, the difference in temperature
510
measured between the chimney and the device (Line 00) is 64◦C higher but the
511
roof temperature (Line 0B) is only 4◦C higher. By comparing temperatures in
512
P3-D2 andP2-D2 tests, the temperature is 118◦C higher near the chimney (Line
513
00), and only 23◦C higher on the roof (Line 0B). These results show the efficacy
514
of the CEIL device.
515
3 Conclusions
516
Given the high number of roof fires occurred in Europe due to the presence of
517
a chimney, several solutions have been proposed in the literature. Some of them
518
regard modifications to the chimney certification procedure. Another solution is
519
represented by the CEIL device to be installed between chimney and roof for lim-
520
iting the temperature of this latter. This device is made of conductive elements
521
immersed in an insulating layer. It was numerically predicted that it can reduce
522
the roof temperature in very critical chimney operating conditions, whereas the
523
purpose of the present article has been the verification of these results experimen-
524
tally. The final version of the device consists of an insulating layer 100 mm thick
525
and four conductive elements, two positioned at the top and two at the bottom.
526
Since the higher the number of the conductive elements the lower the roof tem-
527
perature, the increasing of the number of the conductive elements is beneficial
528
especially when it is necessary to reduce the horizontal thickness of the insulating
529
layer due to the limited space between joists in the roof. Results have shown that
530
the device limits the roof temperature effectively. The device can be installed any
531
time there is uncertainty about the safety of a chimney installation. For example,
532
when a chimney must be installed in a roof whose thickness is not included in the
533
certification procedure. Also, it can be installed whenever there is the desire or the
534
need of safety greater than that guaranteed by the certification procedure. The
535
results presented in this paper confirm that the heat transfer from the chimney-
536
roof penetration point to the ambient must facilitated as much as possible, and
537
the conductive elements in the CEIL device do it effectively.
538