International Journal of Heat and Mass Transfer 55 (2012) 5343–5349

Contents lists available at SciVerse ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier .com/locate / i jhmt

Vacuum insulation properties of phenolic foam

Jongmin Kim, Jae-Hyug Lee, Tae-Ho Song ⇑School of Mechanical, Aerospace and Systems Engineering, Korea Advanced Institute of Science and Technology, Guseong-dong 373-1, Yuseong-gu, Daejeon, Republic of Korea

a r t i c l e i n f o

Article history:Received 13 December 2011Received in revised form 20 March 2012Accepted 16 May 2012Available online 8 June 2012

Keywords:Phenolic foamVacuum insulation panelGuarded hot plateThermal conductivity

0017-9310/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.ijheatmasstransfer.2012.05

⇑ Corresponding author. Tel.: +82 42 350 3032; faxE-mail address: thsong@kaist.ac.kr (T.-H. Song).

a b s t r a c t

Characteristic properties of phenolic foam as the interstitial material of a vacuum insulation panel areinvestigated experimentally. For the measurement of effective thermal conductivity, a vacuum guardedhot plate (VGHP) apparatus is used and the conductivity is measured at various vacuum levels. Radiativeproperties are found using a Fourier transform infrared spectroscopy (FT-IR) device. Solid conductivity isestimated using the porosity of the foam. Effective thermal conductivity at high level of vacuum is mea-sured to be 5 mW/m K which is sum of solid conductivity (2.56 mW/m K) and radiative conductivity(2.44 mW/m K) with 5% of measurement uncertainty. The pore size of the foam is estimated to be260 lm using rarefied gas conduction theory. This ensures insulation performance of phenolic foam upto about 10�3 atm. Other practical characteristics of phenolic foam as the VIP core material are alsodiscussed.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Today, the greatest share of energy is consumed in the buildingsector [1]. Especially, energy for space heating/cooling takes themajor portion of building energy consumption [2]. This energy isfinally dissipated to the environment. Therefore, thermal insula-tion technology is very important for saving energy and reducingCO2 emission. A vacuum insulation panel (VIP) is the most promis-ing insulator with very low thermal conductivity. It is composed ofan envelope to maintain vacuum and a core to endure the externalatmospheric pressure. Generally, porous materials such as PUfoam, glass wool and fumed silica are used as the core. VIPs gener-ally have thermal conductivity of 3–8 mW/m K [3] thanks to theevacuated inner space. It is much lower than those of most insula-tion materials (above 30 mW/m K) and aerogel (17 mW/m K [4]).Despite of the good insulation performance, VIPs are not widelyused yet because of the high production cost and its infant stageof development. Despite of worse insulation performance, conven-tional insulation materials are still competitive in various applica-tions mainly due to their low prices and long life. To increase theuse of VIPs, it is thus imperative to develop economical materialsand manufacturing processes. Phenolic foam is considered to bea possible candidate in this sense, although not many researchesare found in the literature and it motivated this study.

Apart from the insulation performance and cost, propertiesregarding flame, smoke and toxicity (FST) are also important prop-erties of insulators. In this regard, phenolic foam has great FST

ll rights reserved..051

: +82 42 350 3210.

properties. It has high ignition temperature (595 �C) and emits lit-tle smoke and toxic gases during combustion [5]. For this reason,phenolic foam is very useful for insulation of buildings and trans-port vehicles. However, mechanical strength is worse than otherones thus researches about phenolic foam have been usually fo-cused on enhancing mechanical properties. Fiber reinforcement isknown to be an effective way to increase mechanical strength offoam [6]. Rangari et al. [7] increased strength and modulus of phe-nolic foam by 60 and 80%, respectively using a sonochemical tech-nique. Meanwhile, Deng and Xu [8] reported thermal expansioncoefficient of phenolic foam in the range of 77–293 K using thedilatometer, and Tseng and Kuo [9] derived radiative conductivityof phenolic foam from transmittance measured by an FT-IR device.However, no reports regarding the thermal performance in vacuumwere available to the authors.

The objective of this study is to analyze the heat transfer of phe-nolic foam and examine the possibility as the core of VIPs. For thisobjective, structure conduction is modeled first. Radiative propertyof phenolic foam is measured using an FT-IR and effective thermalconductivity at different vacuum levels is measured by a vacuumguarded hot plate (VGHP) apparatus. The overall performance isevaluated finally.

2. Thermal conductivity of phenolic foam

A phenolic foam sample is dark green and made by Asia horti-cultural product co., Ltd. It has density of 25 kg/m3, size of30 � 30 � 1 (cm3) and porosity of 0.98. Fig. 1 shows the microstructure of the sample. It has an open-cell, non-uniform structure.Heat transfer in phenolic foam occurs in these modes: conduction

Nomenclature

a speed of soundA area (m2)c speed of light in vacuumcv specific heat at constant volumee emissive powerE extinction coefficienth Planck constantH height of a specimen (m)k thermal conductivity (W/m K)kB Boltzmann constantl mean free path of a gas moleculeL thickness (lm)P gas pressure (Pa)q heat transfer rate (W)T temperature (K)x thickness of a strut

Greek symbolse porosity/ pore size (m)

q density (kg/m3)k wavelengthr Stefan–Boltzmann constants transmittance

Subscriptsb blackbodycr criticaleff effectivef fluidg gaseousr radiativeR Rosselands solidst struttot totalw wall

5344 J. Kim et al. / International Journal of Heat and Mass Transfer 55 (2012) 5343–5349

through the structure of the foam, conduction through gas in voidsand radiation. Thus effective thermal conductivity keff of phenolicfoam can be expressed as

keff ¼ ks þ kr þ kg ; ð1Þ

where ks, kr and kg are solid, radiative and gas conductivities,respectively.

2.1. Solid conductivity

Solid conductivity ks of the foam can be estimated by properlymodeling a cell of the foam. Placido et al. [10] suggested a dodeca-hedral cell model and derived ks using the strut diameter and thecell diameter. This method needs geometric factors of the foamstructure such as thicknesses of the wall, strut and so on. If thefoam had a uniform structure, it could be safely applied. Phenolic

Fig. 1. SEM micrograph of the phenolic foam.

foam, however, has non-uniform micro structures as shown inFig. 1. Therefore, it is not easy to express ks using this theory. Analternative approach to estimate ks is using the porosity of thefoam. To develop the equations, let’s assume that the foam has aunit cubic cell (see Fig. 2 (a)). If the unit cell is composed of strutsof length 2 and thickness x, ks of the unit cell can be expressed as[11]

ks ¼ kstx2 þ kf ð1� x2Þ þ 2xð1� xÞkstkf

kstð1� xÞ þ kf x; ð2Þ

where kst and kf are thermal conductivities of the strut and the fluidin the void, respectively and the relative thickness x of the strut is afunction of the porosity e as

x ¼ 12þ cos

13

cos�1ð2e� 1Þ þ 43p

� �: ð3Þ

If the cubic cell is composed of walls as shown in Fig. 2(b), ks can beexpressed as [12]

ks ¼kf e

23 þ kwð1� e2

3Þkf ðe

23 � eÞ þ kwð1� e2

3 þ eÞkw; ð4Þ

where kw is thermal conductivity of the wall.Schuetz and Glicksman [13] suggested simpler approximations

for both strut and wall-cubic cell models. In Fig. 2(a) and (b), one-third of the strut and two-third of the wall are oriented to the heatflow direction. Therefore, ks’s of the strut (a) and wall-cubic (b)cells can be intuitively approximated as

ks ¼ kf þ13ð1� eÞkst ; ð5Þ

and

ks ¼ kf þ23ð1� eÞkw; ð6Þ

respectively. When the voids of phenolic foam are filled with air, kf

in Eqs. (2)–(6) is the thermal conductivity of air. However, since theobjective of this study is to find thermal conductivity in vacuum,the effect of air is temporarily ignored here and treated later. Inthese equations, kst and kw are both 0.179 W/m K which is the ther-mal conductivity of phenolic resin. As the results, solid conductivityby Eq. (2) and its approximation by Eq. (5) of the strut-cubic cell

Fig. 2. Unit cubic cells consist of struts (a) and walls (b).

Fig. 3. Measured transmittance (a) and extinction coefficient (b) for differentsample thicknesses.

J. Kim et al. / International Journal of Heat and Mass Transfer 55 (2012) 5343–5349 5345

model are calculated to be 1.26 and 1.19 mW/m K, respectively. Incase of the wall-cubic cell model, solid conductivity by Eq. (4) andits approximation by Eq. (6) are calculated to be 2.42 and2.39 mW/m K, respectively. More discussions about them are madelater in Section 4.1.

2.2. Radiative conductivity

Radiative heat transfer rate through an optically thick mediumlike phenolic foam can be estimated by the diffusion approxima-tion using radiative conductivity kr as [14]

qr ¼ �krdTdz¼ �16rT3

3ER

dTdz; ð7Þ

where r is the Stefan–Boltzmann constant, T is mean temperature,and ER is the Rosseland mean extinction coefficient. The Rosselandmean extinction coefficient is expressed as

1ER¼ C1C2

4rT5

Z 1

0

1Ek� 1k6 exp

C2

kT

� �� 1

� ��1

dk: ð8Þ

Here, C1 = 2phc2 and C2 = hc/kB, where h is the Planck constant, c isthe speed of light in vacuum and kB is the Boltzmann constant. Thespectral extinction coefficient Ek can be expressed from the spectraltransmittance sk as [14]

Ek ¼ � lnðskÞ=L; ð9Þ

where L is the thickness of the medium. Therefore, the Rosselandmean extinction coefficient ER can be derived by measuring sk. Mea-surement of sk is made by an FT-IR device (IFS 66/s from Brukercorp.). The wavelength range of the FT-IR is 2.5–20 lm and themeasurement temperature is 298 K. Transmittances of three pheno-lic foam samples with different L’s are measured and Ek’s are ob-tained using Eq. (9) as shown in Fig. 3. Using Ek’s, the Rosselandmean extinction coefficients ER’s can be calculated by Eq. (8). Theaverage ER and kr of the samples at 298 K are calculated to be3300 m�1 and 2.44 mW/m K, respectively.

2.3. Gas conductivity

Gas conductivity kg in a continuum range is given as [15]:

kg ¼13qcv la; ð10Þ

where q is the density, cv is the specific heat at constant volume, l isthe mean free path of a gas molecule and a is the speed of sound.Density q is proportional to the gas pressure but l is inversely pro-

portional thus kg is independent of the gas pressure in a continuumlimit. However, the gas conduction mechanism is changed when thegas pressure is sufficiently low so that the distance of conduction issame or smaller than the mean free path of a gas molecule. Smolu-chowski established a gas conduction theory for such rarefied gasregion [15] and based on the theory, Kwon et al. suggested an equa-tion of kg at room temperature as [16]

5346 J. Kim et al. / International Journal of Heat and Mass Transfer 55 (2012) 5343–5349

kg ¼kg0

1þ 0:032P/

; ð11Þ

where kg0 is thermal conductivity of gas at a continuum range, P isthe gas pressure (Pa) and / is the conduction distance (m), that is,the pore size of porous foam. Fig. 4 shows the variation of kg accord-ing to Eq. (11). Note that across a critical pressure (Pcr to be definedlater), kg decreases to zero as P decreases.

3. Measurement of effective thermal conductivity

3.1. Measurement apparatus

Effective thermal conductivity keff of phenolic foam is measuredat different vacuum levels. The guarded hot plate method (GHP) isapplied because it is the most precise among various measurementmethods [17]. Since the gas pressure needs to be controlled duringmeasurements, a vacuum GHP (VGHP) is manufactured (Fig. 5). Itis composed of conductivity-measuring components, i.e., theGHP, and vacuum control components. The former is installed inthe vacuum chamber which has a cylindrical shape with 510 mmof diameter and 375 mm of height. The vacuum level in the cham-ber can be controlled from atmospheric pressure to 10�4 Pa by athrottle valve and a diffusion pump. A pressure pad can exertexternal force to the specimen when needed. In the GHP part, aheater block is placed at the center. An electric heater is equippedin the heater block and controlled by a power supply. It is made ofpure copper and has size of 150 � 150 � 50 mm3. The guard-ringsurrounds the heater block across a narrow gap (2 mm) and it ismade of pure copper, too. Below the heater block/guard-ring, thehot plate is located. The cold plate is placed on the heaterblock/guard-ring and a specimen is sandwiched between the coldplate and the heater block/guard-ring. Both of cold and hot platesare made of pure aluminum and have a size of 300 � 300 �50 mm3. Temperatures of the guard-ring and the hot plate aremaintained constant using a hot-bath circulator, while the coldplate is connected to a separate cold-bath circulator. Temperaturesof each part are measured by K-type thermocouples. Also, 5-pair K-type thermopiles are attached between the heater block, guard-ring and hot plate to magnify the temperature difference betweenthem (Fig. 5(b)). They are all calibrated following the calibration

Fig. 4. Gas conductivity according

methods of ASTM standards [18,19]. When temperatures of theguard-ring and the hot plate are maintained the same as that ofthe heater block, heat from the heater block cannot flow eitherto downward or lateral directions. Since temperature of the coldplate is lower than that of the heater block, heat from the heaterblock flows only upward to the cold plate across the specimen.Then thermal conductivity of the specimen can be calculated as

kmeas ¼qheater � HAeff � DT

; ð12Þ

where qheater is the heat transfer rate from the heater block, H andAeff are the thickness and the effective area of the specimen and DTis the temperature difference between the upper and the bottomsurfaces of the specimen.

3.2. Specimen and measurement condition

Since the phenolic foam sample is brittle, it is surrounded by asupporting structure to withstand the weight of the cold plate(13 kg) during measurement as shown in Fig. 6. The supportingstructure is composed of two cover plates (300 � 300 � 1.5 mm3

each) and four cylindrical pillars which have 2 mm of diameterand 10 mm of height. Cover plates and pillars are made of polycar-bonate. The phenolic foam sample has a slightly smaller surfacearea than the cover plate but has the same thickness as the heightof pillar as shown in Fig. 6. Heat conduction through the support-ing structure is only a few percent compared with that of specimenbecause thermal conductivity of polycarbonate (0.2 W/m K) andsurface area of pillars are sufficiently small. Thus its effect is neg-ligible. However, thickness of cover plates (1.5 mm � 2) affectsthe accuracy of the measured thermal conductivity. Therefore, itis corrected as

keff ¼ kmeasH

H þ 2 � tcv; ð13Þ

where keff is the effective thermal conductivity of the foam withoutthe support structure, kmeas is the measured thermal conductivityincluding the supporting structure and tcv is the thickness of a coverplate.

Before measuring the thermal conductivity, vacuum pressure inthe chamber has to reach the desired level. It takes a few minutes

to gas pressure and pore size.

Fig. 5. (a) Schematic diagram of the measurement apparatus and (b) detail of the GHP components (x: thermocouple, o: thermopile).

Fig. 6. Phenolic foam and a supporting structure for the measurement.

J. Kim et al. / International Journal of Heat and Mass Transfer 55 (2012) 5343–5349 5347

to reach a low level of vacuum (about 10�1 Pa) from atmosphericpressure, but several hours are needed to reach the desired high le-vel of vacuum (about 10�3 Pa). After the vacuum level is settleddown, the apparatus needs to be operated for a thermal steady-state. It takes a long time because the thermal conductivity ofthe specimen is very low. When temperatures of apparatus andsample are maintained within 0.1 K of fluctuation for a long time(usually 5 h), it is determined to be at a steady-state. The meantemperature of the specimen is maintained at 298 K during themeasurement with the heater block at 308 K and the cold plateat 288 K. In summary, 9–12 h are usually needed to measure at acertain vacuum level.

Fig. 7. Measured effective thermal conductivity (error bars: ±5%).

4. Results and discussion

Fig. 7 shows measured keff of the phenolic foam at differentpressure levels. It shows a similar shape as Fig. 4. Since ks and kr

are independent of the pressure, the change of keff means thechange of kg. It shows keff = 30 mW/m K at the atmospheric pres-sure. Around 100 Pa, keff drops sharply since it is the transition re-

5348 J. Kim et al. / International Journal of Heat and Mass Transfer 55 (2012) 5343–5349

gion between a continuum and a rarefied gas regions. Below 1 Pa,kg is virtually zero thus keff (5 mW/m K) in this region representsks + kr. This shows that VIPs using the phenolic foam must be main-tained below 10�3 atm for proper performance.

The uncertainty of the measurement can be calculated from Eq.(12) and the following equation [20]:

dkeff

keff¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiokeff

oAeff

dAeff

keff

� �2

þ okeff

oðDTÞdðDTÞ

keff

� �2

þ okeff

oHdHkeff

� �2

þ okeff

oqheater

dqheater

keff

� �2s

:

ð14Þ

Uncertainties of Aeff, DT and H are 0.03%, 0.4% and 0.2%, respectively,which are negligibly small but that of qheater is around 5%. Theuncertainty of qheater comes from uncertainties of voltage and cur-rent of the power supply. At a high level of vacuum, qheater is gener-ally very small (0.2–0.3 W) and it makes the relative uncertainty ofqheater to be slightly higher.

4.1. Comparison between theoretical and measured ks’s

In Section 2.2, radiative conductivity kr was calculated as2.44 mW/m K. Since ks + kr is measured as 5 ± 0.25 mW/m K, ks isaccordingly 2.56 ± 0.25 mW/m K. It agrees well with the theoreti-cal result of Eq. (4) in Section 2.1 (ks = 2.42 mW/m K) with 6% ofrelative error, however, it is much larger than that of Eq. (2)(ks = 1.26 mW/m K). This suggests that the structure of the sampleis much closer to the wall-cubic cell model than the strut-cubic cellmodel. In the micrograph of the sample (Fig. 1), however, it isapparently not composed of walls or membranes. Instead, it iscomposed of struts which have thicknesses of 15–20 lm and thestruts are connected to each other by fine skeletons. Even thoughthey are much thinner than the main struts, they clearly functionas thermal bridges and are expected to bring about additional con-duction. Therefore, solid conductivity ks can be reduced if the sub-structure between struts is eliminated somehow.

4.2. Pore size approximation by the measured kg

Pore size / of open-cell porous materials can be determined byvarious methods such as a scanning electron microscope (SEM)observation, gas adsorption, Hg penetration and so on. In addition,Lee et al. [21] suggest another method to determine / using a rar-efied gas conduction theory. We adopt their method as follows. Gasconductivity kg according to the vacuum level can be found byextracting ks + kr (5 mW/m K) from keff. Assuming that vacuum le-vel below 1 Pa is the rarefied gas region, kg is expressed by Eq. (11)and the pore size / can be derived. Actually, various pores are dis-tributed in the foam, therefore the pore size in Eq. (11) means theaverage one. Using this method with the measured data below1 Pa, the average pore size is estimated as 260 lm. It is comparablewith the SEM micrograph of the sample which shows pore sizes ofseveral hundred microns (Fig. 1).

The pore size also affects radiation in the foam. If the pore size isincreased, photons would penetrate further distance. Therefore theextinction coefficient (or the Rosseland mean extinction coeffi-cient) of the foam is inversely proportional to the pore size.Schuetz and Glicksman [13] suggested an equation for the extinc-tion coefficient of closed-cell polyurethane foam. It expresses theextinction coefficient as a function of mass fraction of the strut,porosity and the pore size and it shows an inversely proportionalrelationship between the extinction coefficient and /. Campo-ar-naiz et al. [22] measured the Rosseland mean extinction coefficientER of various foams with different /’s and reported that ER in-creases as / decreases. In this study, however, it was difficult tovalidate the relationship between ER and / due to lack of data. In-stead, when assuming a simple relationship as

ER �1/; ð15Þ

ER is estimated to be 3800 m�1. It has 16% of relative error com-pared to the measured ER in Section 2.2. However, it is still not accu-rate enough to ascertain the validity of the approximation of Eq.(15) and thus further verification needs to be made.

4.3. Discussion on the phenolic foam as the core of VIP

The most important requirement of a core material is indeedthe insulation performance, keff. At a high level of vacuum, phenolicfoam has keff � 5 mW/m K which is the sum of ks (2.44 mW/m K)and kr (2.56 mW/m K). Among them, kr can be decreased by addingopacifiers. They are used to suppress thermal radiation in porouslayers such as aerogel. Zirconium silicon oxide (ZrSiO4), silicon car-bide (SiC) and carbon black (C) are widely used as opacifiers andcarbon black is known to have the highest extinction coefficient[23]. When carbon black is added to silica aerogel powder, ER atroom temperature is increased from 3750 to 12500 m�1 [24]. IfER of phenolic foam (currently 3300 m�1) can be increased toaround 12500 m�1 as in the silica aerogel insulator, kr and keff aredecreased to around 0.64 and 3 mW/m K, respectively. Note that,however, thermal conductivities of those opacifiers are higher thanthose of general core materials thus they may affect ks. It is usuallyexpected to increase ks but the contrary effect on ks is also reported[24].

Solid conductivity ks seems to be difficult to be reduced unlessits micro structure is modified as discussed earlier. Meanwhile, be-cause of its brittleness, phenolic foam needs a supporting structureto withstand the external force by atmospheric pressure whenused as the core of a VIP. The increase of thermal conductivitydue to the supporting structure depends on its shape and material.As an example, circular pillar and cover plates made of polycarbon-ate add minimum of 0.5 mW/m K to ks. Then it is expected to becomparable to VIPs of open-cell PU foam (5–7 mW/m K [26]) butworse than glass fiber (3–5 mW/m K [25]) in terms of the insula-tion performance.

Another important requirement for a core material is the ser-vice life. The inner pressure of VIPs rises up with time by perme-ation through surfaces and heat-sealed flanges of envelope andoutgassing from the core [27]. The insulation performance is dete-riorated due to the increase of gas conduction and finally the VIPfails to function as an insulator (see Fig. 4). The critical pressurePcr is defined as the pressure at which gas conductivity is equalto kg0/2. From Eq. (11), Pcr (Pa) is expressed with / (m) as [28]

Pcr ¼0:032

/: ð16Þ

The data shown earlier ð/ � 260� 10�6 mÞ yield Pcr � 100 Pa(10�3 atm) and typical phenolic foams show / ranging between200 and 400 lm [9]. Thus, most of phenolic foams have Pcr in thevicinity of 10�3 atm. On the other hand, pore sizes of fumed silicaand silica aerogel are 30–70 nm and that of carbon aerogel isaround 100 nm. Polyurethane foam shows larger / (about 60 lm[29]) and fiber insulation materials usually have around 100 lmof / [16]. Compared with these materials, phenolic foams haveslightly large pore size and thus smaller Pcr. Therefore, phenolicfoams are disadvantageous in service-life. Despite of the large poresize, however, great FST properties and low cost of phenolic foamsare still strong advantages as the core of VIPs.

5. Conclusion

In this study, thermal conductivity of phenolic foam is investi-gated experimentally. The Rosseland mean extinction coefficient

J. Kim et al. / International Journal of Heat and Mass Transfer 55 (2012) 5343–5349 5349

is measured as 3300 m�1 using an FT-IR device. Effective thermalconductivity keff is measured at various vacuum levels using aVGHP apparatus with 5% of uncertainty. At the high level of vac-uum, the foam shows 5 mW/m K of keff which is the sum of kr

(2.44 mW/m K) and ks (2.56 mW/m K). The pore size of the foamis estimated to be 260 lm using a rarefied gas conduction equa-tion. The Rosseland mean extinction coefficient is roughly esti-mated to be 3800 m�1 by a reciprocal relation of the pore sizewith 16% of relative error to the FT-IR measurement. Phenolic foammay be used as a core of VIPs from the view point of insulation per-formance, great FST properties and low cost, but poor vacuumenveloping may result in short service life due to its large pore size.

Acknowledgments

This work was supported by the National Research Foundationof Korea (NRF) grant funded by the Korea Government (MEST) (No.2011-0027642) and the second stage of the Brain Korea 21 Projectin 2011. The authors thank Asia Horticultural Product Co. LTD. forthe provision of phenolic foam samples.

References

[1] Nature Publishing Group, Architects of a low-energy future, Nature 452 (2008)520–523.

[2] Karim Ghazi Wakili, The world energy crisis – Part 2: some more vacuum-based solutions, Vacuum 82 (2008) 679.

[3] Bjørn Petter Jelle, Traditional, state-of-art and future thermal buildinginsulation materials and solutions – properties, requirements andpossibilities, Energ. Build. 43 (2011) 2549–2563.

[4] J.M. Schultz, K.I. Jensen, Evacuated aerogel glazings, Vacuum 82 (2008) 723–729.

[5] A. Gardziella, L.A. Pilato, A. Knop, Phenolic Resins, second ed., Springer, 1999.[6] Hongbin Shen, Steven Nutt, Mechanical characterization of short fiber

reinforced phenolic foam, Compos. A 34 (2003) 899–906.[7] Vijaya K. Rangari, Tarig A. Hassan, Yuanxin Zhou, Hassan Mahfuz, Shaik Jeelani,

Barton C. Prorok, Cloisite clay-infused phenolic foam nanocomposites, J. Appl.Polym. Sci. 103 (2007) 308–314.

[8] D.Q. Deng, L. Xu, Measurement of thermal expansion coefficient of phenolicfoam at low temperatures, Cryogenics 43 (2003) 465–468.

[9] Chung-Jen Tseng, Kuang-Te Kuo, Thermal radiative properties of phenolic foaminsulation, J. Quant. Spectrosc. Radiat. Transfer 72 (2002) 349–359.

[10] E. Placido, M.C. Arduini-Schuster, J. Kuhn, Thermal properties predictive modelfor insulating foams, Infrared Phys. Technol. 46 (2005) 219–231.

[11] G.N. Dul’Nev, Heat transfer through solid disperse systems, J. Eng. Phys. 9(1965) 399–404.

[12] H.W. Russell, Principles of heat flow in porous insulators, J. Am. Ceram. Soc. 18(1935) 1–5.

[13] M.A. Schuetz, L.R. Glicksman, A basic study of heat transfer through foaminsulation, J. Cell. Plast. 20 (1984) 114–121.

[14] Michael F. Modest, Radiative Heat Transfer, second ed., Academic Press, 2003.[15] J.M. Lafferty, Foundations of Vacuum Science and Technology, John Wiley and

Sons, New York, 1998. pp. 50–51.[16] Jae-Sung Kwon, Choong Hyo Jang, Haeyong Jung, Tae-Ho Song, Effective

thermal conductivity of various filling materials for vacuum insulation panels,Int. J. Heat Mass Transfer 52 (2009) 5525–5532.

[17] R.E. Collins, C.A. Davis, C.J. Dey, S.J. Robinson, J.Z. Tang, G.M. Turner,Measurement of local heat flow in flat evacuated glazing, Int. J. Heat MassTransfer 36 (1993) 2553–2563.

[18] Standard Test Method for Calibration of Thermocouples by ComparisonTechniques, E 220-07a, ASTM, 2007.

[19] Standard Practice for Preparation and Use of an Ice-Point Bath as a ReferenceTemperature, E563-11, ASTM, 2011.

[20] J.P. Holman, Experimental Methods for Engineers, sixth ed., McGraw-Hill,1994.

[21] Ok-Joo Lee, Kun-Hong Lee, Tae Jin Yim, Sun Young Kim, Ki-Pung Yoo,Determination of mesopore size of aerogels from thermal conductivitymeasurements, J. Non-Cryst. Solids 298 (2002) 287–292.

[22] R.A. Campo-arnaiz, M.A. Rodriguez-Perez, B. Calvo, J.A. De Saja, Extinctioncoefficient of polyolefin foams, J. Polym. Sci. B. Polym. Phys. 43 (2005) 1608–1617.

[23] Gui Lu, Xiao-Dong Wang, Yuan-Yuan Duan, Xiong-Wei Li, Effects of non-idealstructures and high temperatures on the insulation properties of aerogel-based composite materials, J. Non-Cryst. Solids 357 (2011) 3822–3829.

[24] Th. Rettelback, J. Sauberlich, S. Korder, J. Fricke, Thermal conductivity of silicaaerogel powders at temperatures from 10 to 275 K, J. Non-Cryst. Solids 186(1995) 278–284.

[25] C. Stark, J. Fricke, Improved heat-transfer models for fibrous insulations, Int. J.Heat Mass Transfer 36 (1993) 617–625.

[26] A.J. Hamilton, An evaluation of the practical application and use of VACPACpanel technology, Vuoto Scienza et Tecnologia 28 (1999) 27–30.

[27] Jae-Sung Kwon, Choong Hyo Jang, Haeyong Jung, Tae-Ho Song, Vacuummaintenance in vacuum insulation panels exemplified with a staggered beamVIP, Energ. Build. 42 (2010) 590–597.

[28] J. Fricke, U. Heinemann, H.P. Ebert, Vacuum insulation panels – from researchto market, Vacuum 82 (2008) 680–690.

[29] G. Reichenauer, U. Heinemann, H.-P. Ebert, Relationship between pore size andthe gas pressure dependence of the gaseous thermal conductivity, ColloidSurface A. Physicochem. Eng. Aspects 300 (2007) 204–210.