Вы находитесь на странице: 1из 13

CISM - Effect of heat on concrete Thermal properties & analysis

International Centre for Mechanical Sciences (CISM)

Course on
Effect of Heat on Concrete
9-13 June 2003, Udine, Italy

Thermal properties & analysis Anderberg 2

Yngve Anderberg
Fire Safety Design, Malmö, Sweden

Abstract. This lecture describes the thermal properties and thermal response of concrete,
reinforcing steel and stainless steel. Boundary conditions are described and the effect on
thermal response of varying emissivity, convection and adiabatic conditions at the
boundaries is studied. The effect of scatter in thermal conductivity, moisture content and
voids on the thermal distribution in structural concrete members are analysed. All
computer simulations are performed with the software SUPER-TEMPCALC.

1. Introduction

The prediction of temperature distribution of the fire-exposed structural member is one


important step in the design process. The reliability of computed temperature is dependent on
accurate and sufficient knowledge about the thermal properties of building materials used.
Properties normally used in computer programs are thermal conductivity and volumetric
specific heat (cp⋅ρ). These properties are relatively easy to determine for steel but for concrete
it is far more difficult especially as concerns the thermal conductivity. The thermal
conductivity as function of high temperatures of concrete is very complicated to measure and
the scatter from different tests is considerable. However the specific heat and density
determination is relatively accurate and there is not so much scatter between different
measurements.

The most reliable way to obtain the thermal conductivity of concrete is a combination of well-
defined temperature measurement at different depths of a fire exposed concrete member and
computer simulations based on an assumed conductivity curve.

The influence of the scatter (uncertainties), in the parameters given below, on the thermal
distribution will be illustrated for

Anderberg, Y 1
CISM - Effect of heat on concrete Thermal properties & analysis

• varying thermal conductivity


• varying thermal capacity
• varying boundary conditions on exposed as well as unexposed side of a concrete
member
• varying moisture content

2. Concrete

The thermal conductivity and volumetric specific heat as function of high temperature are
necessary input in temperature predictions but these properties can also be used in the
definition of thermal diffusivity D often used in literature as defined below and can be
described as the rate at which the temperature changes take place

λ
D= eq (1)
cpρ
where
λ is thermal conductivity (W/mK)
cp is specific heat J/kgK
ρ is density (kg/m3)
cp⋅ρ is volumetric specific heat J/m3K

There exist measurement problems to


obtain reliable values on thermal
diffusivity and the scatter between
different sources are considerable. As
an example of this scatter the thermal
diffusivity of normal weight and
lightweight concretes as function of
temperature is shown in Fig 1
(RILEM, Schneider (1985). The
thermal diffusivity is decreasing about
50% at 600 oC and continues to do that
up to 800 oC.

Fig 1 Scatter of thermal diffusivity for concrete and


lightweight concrete as function of temperature

2.1 Thermal conductivity

Thermal conductivity as function of high temperature is very difficult to measure due to the
influence of many parameters as moisture content and its continuous movement, the type and
amount of aggregate etc. Furthermore measurement methods and devices are not sufficiently
accurate to be fully trusted. Steady state tests are normally applied neglecting the dynamic
temperature and moisture condition, which consequently create a measuring problem.
Unfortunately in most countries testing standards do not exist. These problem means that you

Anderberg, Y 2
CISM - Effect of heat on concrete Thermal properties & analysis

cannot take measured values as correct for granted. Published values on thermal conductivity
have a wide scatter and seems to be a little too high in general when applied in computer
programs.

Due to the uncertainty to measure thermal conductivity the only way to judge about the
correctness is to simulate fire tests and compare measured temperature values all over the
cross-section with computed values based on measured thermal conductivity and by a fixed
volumetric specific heat curve. If there is a discrepancy (normal situation) the values are to be
modified by trial and error until a good agreement is attained.

The thermal conductivity of siliceous


concrete presented in Anderberg (1978) 2
Upper limit
Lower limit
and in the CIB-bulletin and used in Swedish

Swedish fire engineering design is


Thermal conductivity [W/m°C]

based on this computer simulation 1.5

procedure and shown in Fig 2 as the


dotted curve. It is however also stated
1
that the composition of concrete and the
types of aggregates used have an
influence on thermal conductivity. 0.5

Where relevant documentation is


provided it may be appropriate to adjust
the material properties. The curve 0
0 200 400 600 800 1000 1200
Temperature °C
presented is valid for siliceous concrete
and is on the safe side when used for Fig 2 Thermal conductivity of siliceous aggregate
calcareous concrete. concrete

The conductivity curve is valid for the first heating and when cooling down the value attained
at maximum temperature remain during cooling.

In Eurocode 2 (2002) and Eurocode 4 (2003) a compromise has been made so that the thermal
conductivity can be chosen between an upper and a lower limit as shown in Fig 2. This is of
course an undesirable solution when concrete must have the same properties in reinforced
concrete as in composite concrete - steel structures

The upper limit of thermal conductivity λc of normal weight concrete may be determined from:
2
λc = 2 - 0,2451 (T /100) + 0,0107 (T/100) W/m K for 20°C ≤ T ≤ 1200°C eq (2)

where T is the concrete temperature.

The lower limit of thermal conductivity λc of normal weight concrete may be determined from:
2
λc = 1,36 - 0,136 (T/100) + 0,0057 (T/100) W/m K for 20°C ≤ T ≤ 1200°C eq (3)

Anderberg, Y 3
CISM - Effect of heat on concrete Thermal properties & analysis

2.2 Specific heat

The uncertainty of measured


specific heat values are much less
compared with thermal
conductivity values even if a wide
scatter from one test to another
may sometimes exist. The
influence of water and
vaporisation on specific heat is
illustrated for different concretes
in Fig 3 with a peak at 100 oC in
the specific heat curve. The
impact of latent heat due to
chemical reactions as Portland
cement dehydration of CAOH at
about 500 °C and of CaCo3 at
about 550 °C is not really studied Fig 3 Specific heat capacity for dry concretes by different
thoroughly. authors

A try of calculations of this impact upon the specific heat has been done by Harmathy ( ) but is
never validated.

The vaporisation is considered in the design curve for specific heat in Eurocode 2 (2002). This
is illustrated by a peak between 100 and 200 °C for three moisture contents 0, 1,5 and 3% of
weight in the curve for specific heat in Fig 4a . However the Portland cement dehydration and
other latent heat impacts is not considered due lack of real studies and to uncertain accuracy of
measurements and the very small influence on temperatures when computed.
cp (θ ) [kJ/kg°K] cv [kJ/m3 °K]
2,2 5000
2 u = 3%
1,8
4000
1,6
u = 1,5%
1,4
3000
1,2
1
2000
0,8
u = 0%
0,6
0,4 1000

0,2
0 0
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
θ [°C] θ [°C]
Fig 4a Specific heat, cp(θ) as function of Fig 4b Volumetric specific heat, cv(θ) as func-
temperature at 3 different moisture contents, u, of tion of temperature at a moisture
0, 1,5 and 3 % by weight for siliceous concrete content of 3% by weight and a density
of 2300 kg/m3 for siliceous concrete

Anderberg, Y 4
CISM - Effect of heat on concrete Thermal properties & analysis

In order to assess the volumetric specific heat the specific heat must be multiplied with the
density as function of temperature as illustrated above.

The volumetric specific curve in accordance to Eurocode 2 (2002) is shown in Fig 4b

2.3 Density

The density of concrete is changing


with temperature and is needed in
order to assess the volumetric specific
heat, which is used in the computer
programs. Due to the loss of first free
water up to about 200 °C and then
chemically bound water the density is
decreasing by increasing temperature.
In literature this has been studied by
(Anderberg-Thelandersson) (1976),
Schneider and Bazant and the relative
decrease of density up to 1000 °C has Fig 5 Water loss of quartzite concrete as function of
been between 11-13 %. In Fig 5 this is temperature (Anderberg & Thelandersson 1976)
illustrated for quartzite concrete.

These findings of water loss are reflected in the definition of density in the Eurocode 2 and
presented in Table 1 below:

Table 1 Variation of density with temperature influenced by water loss


ρ  = ρ20°C for 20°C ≤ T ≤ 115°C
ρ =  ρ20°C (1 - 0,02(θ - 115)/85) for 115°C < T ≤ 200°C
ρ =  ρ20°C (0,98 - 0,03(θ - 200)/200) for 200°C < T ≤ 400°C
ρ =  ρ20°C (0,95 - 0,07(θ - 400)/800) for 400°C < T ≤ 1200°C

3 Steel

3.1 Thermal conductivity

The thermal conductivity of reinforcing steel has a very high value 54 W/mK at room
temperature and is then decreasing to 27,3 W/mK at 800 °C which is illustrated in Fig 6. The
thermal conductivity of stainless steel is very different and has a low value 14.6 W/mK at room
temperature and increases linearly to 30 W/mK at 1200 °C in accordance with Eurocode 3
(2002) Annex C. As can be seen in Fig 6 the curves are crossing each other at 1000 °C.

Anderberg, Y 5
CISM - Effect of heat on concrete Thermal properties & analysis

The thermal conductivity of


stainless steel is very different
from reinforcing and prestressing 50
steel as shown in Fig 6. The
thermal conductivity for stainless

Thermal conductivity [W/m°C]


40
steel is much lower than b)
reinforcing steel and is constantly
30
increasing by temperature when
reinforcing steel is decreasing a)
from a value more than 3 times 20

higher. This means that below


950 °C stainless steel is conduc- 10

ting heat slower than reinforcing


steel. 0
0 200 400 600 800 1000 1200
Temperature °C

Fig 6 Thermal conductivity of a) stainless steel and b)


reinforcing steel as a function of temperature

3.2 Specific heat and density

The specific heat of reinforcing steel


1200
cps is increasing from about 425
J/kgK at room temperature up to 600
1000
J/kgK at 1200 °C but it has a peak at
735 °C which ends at 5000 J/kgK 800
Specific heat [J/kg K]

which is illustrated in Fig 7 b)

600 a)
The specific heat of stainless steel as
presented in Eurocode 3 (2002) 400
Annex C is somewhat lower than for
reinforcing and prestressing steel and 200
the peak at 735 °C for normal steel
doesn’t exist for stainless steel as 0
shown in Fig 7 This means that 0 200 400 600
Temperature ºC
800 1000 1200

stainless steel per weight needs


Fig 7 Specific heat of a) stainless steel and b) reinforcing
somewhat less energy to be heated
steel as a function of the temperature
up than normal steel.

Anderberg, Y 6
CISM - Effect of heat on concrete Thermal properties & analysis

3.3 Thermal elongation

The thermal elongation ∆L/L of 0.025


reinforcing steel is increasing relative
linearly with temperature except the
plateau at 750-850 °C and reaches the 0.02

value of 1,7 % at 1200 °C.

Related thermal elongation


0.015
a)
The thermal elongation of austenitic b)
stainless steel is somewhat higher (up 0.01
to 35% at 800 °C) than for reinforcing
steel (cf Fig 8 ). The thermal elongation
0.005
at 800 °C for concrete is about 0.010
and for stainless steel 0.015. This must
be considered for the bond between 0
0 200 400 600 800 1000 1200
concrete and stainless steel due to the Temperature ºC

fact that the difference in thermal Fig 8 Thermal elongation of a) stainless steel (Eurocode
elongation will create thermal stresses. 3 (2003) and b) reinforcing steel (Eurocode 2
(2002) as a function of temperature

4. Thermal analysis

When undertaking heat transfer calculations it is essential that relevant thermal properties of
materials and heat transfer coefficients at the boundaries are defined for the entire temperature
interval. The following information is therefore essential when calculating nodal temperatures
as function of time during a specified fire scenario:

• boundary conditions
• fire exposure
• thermal properties of materials

A tailor made program (SUPER-TEMPCALC (1991)) which is one part in a software package
called TCD (Tempcalc-Design) for advanced thermal analysis has been used to illustrate how
variation in different thermal parameters influence the thermal response of concrete structures.
The program is a fire-adapted two-dimensional finite element program developed by Fire
Safety Design for use preferably on PC. The program is widely used in the field of passive fire
protection, and as part of structural analysis, in buildings and on offshore platforms. It is
accepted for North Sea applications by a number of countries and organisations.

The program solves the two-dimensional, transient, heat transfer differential Equation (eq. 4)
incorporating thermal properties, which vary with temperature.

∂  ∂T ∂  ∂T ∂T
 λx + λy +Q = ρ c eq. (4)
∂ x  ∂ x  ∂ y  ∂ y  ∂t
where
Q internal heat generation

Anderberg, Y 7
CISM - Effect of heat on concrete Thermal properties & analysis

The program allows the use of rectangular or triangular finite elements, in cylindrical or
rectangular co-ordinates. Heat transferred by convection and radiation at the boundaries can be
modelled as a function of time. Structures comprising several materials can be analysed and the
heat absorbed by any existing void in the structure is also taken into account. Boards falling
down and spalling of concrete can be considered and the change in geometry is automatically
updated.

A materials properties database is integrated with the main program. Also integrated with the
program are pre- and post-processors which allow fast and user-friendly input/output
procedures. The pre-processor generates the finite element division and retrieves the relevant
information from the database for use in the calculation. Finally, the post-processor presents
the results graphically on screen or on plotter in a variety of forms, including time-temperature
curves, isotherms, and temperature gradients. The software has also been verified and validated
against a large number of tests.

4.1 Boundary conditions

The nature of the ambient conditions is essential to the determination of the temperature field
of the member. Depending on the geometry, the view factors and ambient conditions, various
types of boundary conditions (fire exposed, unexposed and adiabatic) may be considered.

A fire exposed boundary is characterised by heat transferred from the fire gases to the exposed
structure through radiation and convection (see eq. 5). At high temperatures the radiation
dominates. The radiation is expressed by the resulting emissivity factor εr which takes into
account the emissivity of the fire source, εf, and the absorptivity of the heated surface, εm,, as
expressed in Eurocode 1 (2002). The convection is calculated from the temperature difference
between the structure and ambient gases, depending on the gas velocity.

Emissivity and convection factors suggested in Eurocode 1 (2002), are shown in Table 1.

Table 1 Resulting emissivity and convection factor for exposed and unexposed surfaces
Emissivity/Convection εr hc
[-] [W/m2K]
Exposed surface (concrete, steel) 0.56, 0.8 25
Unexposed surface 0.8 9

q n = hc (T g − Tb ) + ε r σ (T g4 − Tb4 ) eq. (5)


where

σ Stefan-Boltzmann constant [5.67x10-8 W/m2 K4]


Tg absolute temperature of radiation source [K]
Tb boundary temperature [K]
εr resulting emissivity factor of the radiation source and the heated surface [-]
qn heat flow at the boundary [W/m2]
hc convection heat transfer coefficient [W/m2K]

Anderberg, Y 8
CISM - Effect of heat on concrete Thermal properties & analysis

Surfaces of a structure that are not exposed to fire will imply a certain potential for cooling of
the exposed structural material. Heat entering the structure along an exposed boundary may, to
a certain extent, leave the profile in unexposed areas. The parameters of emissivity and
convection as proposed in Eurocode 1 (2002) are given in Table 1.

A boundary where no heat is said to pass (qn = 0) is often referred to as an adiabatic boundary.
These are for example symmetry lines. Structures with extreme extension in two of the three
directions are often considered to have a one-dimensional heat flow and consequently adiabatic
boundaries will be adopted in the calculation.

4.2 Effect of varying boundary conditions

The effect of varying convection factor on the unexposed surface for fire-exposed concrete
members is very important to know because in furnace testing this factor may vary in different
laboratories and also depends on the air conditioning in the testing room. Consequently it is
almost impossible to know what convection factor is relevant for the test and the surface
temperatures measured are hard to couple to a certain convection factor. This influence can be
studied theoretically for varying convection factor (0, 9 and 25) by SUPER-TEMPCALC and
the result is illustrated in Fig 9. The difference in surface temperature after 2 hours standard
fire exposure is 45 °C when based on convection factor 0 and 9 and 60 °C on convection factor
9 and 25, which is considerable. A similar simulation has been performed for the emissivity
factor on the unexposed side and the result is shown in Fig 10. This factor, ε, is varying
between 0.56 and 1.0 and has a comparatively small influence, i.e maximum 20 °C

400

350
250

300

200
Temperature [ °C ]

250
Temperature [ °C ]

200
150

150

100
100 eps =0.56
eps =0.8
hc=0 eps =1.0
50
hc=9 50
hc=25
0
0 20 40 60 80 100 120 0 10 20 30 40 50 60
Time [minutes] Time [minutes]

Fig 9 Temperature development on unexposed Fig 10 Temperature development on unexposed


side (=50 mm) with varying convection side (=50 mm) with varying emissivity on
on unexposed side exposed side

Anderberg, Y 9
CISM - Effect of heat on concrete Thermal properties & analysis

The influence of varying emissivity on the 1200


fire exposed side is as high as 50 °C after
20 min but is then decreasing with 1000

temperature which can be observed in Fig


11. The influence of a higher surface 800

Temperature [ °C ]
temperature due to a higher emissivity will
600
also have an effect on the temperature of
the reinforcement, The effect will be much 400 eps=0.56

more important on a fire exposed steel eps=0.8


eps=1.0

sheet used for composite beams. 200

0
0 20 40 60 80 100 120
Time [minutes]

Fig 11 Temperature development on exposed side


(=0 mm) with varying emissivity on
exposed side

4.3 Effect of varying thermal conductivity and thermal capacity

The effect of using the upper or the lower 450


limit of thermal conductivity in accordance
with Fig 2 is illustrated in Fig 12. The 400

effect at a depth of 50 mm (normal position 350

of prestressing wires) on a concrete slab is 300


about 35 °C after 40 min of ISO 834
Temperature [ °C ]

exposure. This temperature difference after 250

40-120 min means a lower strength of 200

prestressing wires of about 7% corre- 150


Upper (50 mm)
Lower (50 mm)
sponding to the upper curve.
100

50

0
0 20 40 60 80 100 120
Time [minutes]

Fig 12 Effect of thermal conductivity on temperature


development 50 mm inside a concrete slab

4.4 Effect of moisture content

Moisture in the concrete will delay the temperature rise in the specimen and this phenomenon
is studied for different moisture contents by computer simulations. However moisture
movement is not considered in the simulation because all water is assumed to be vaporised at
the original position with a constant moisture level through the specimen. It is however
possible to assume a moisture gradient by a trial and error simulation procedure where the
gradient will be changed until a good agreement is obtained between measurements and
calculations. Such a simulationprovides a more realistic temperature gradient around 100 °C.

Anderberg, Y 10
CISM - Effect of heat on concrete Thermal properties & analysis

This has been performed for lightweight concrete where the moisture movement is more
pronounced and a very good agreement with measurements has been obtained. When the
temperature has passed 100 °C it will coincide at about 200-300 °C with the curve
corresponding to no moisture gradient.

Temperatures around 100 °C has no 350


importance for fire design but can be u=0%
u=1.5%

of interest in special application or 300 u=3.0%

only theoretically. A square concrete 250

column (0.3x0.3 m) exposed to ISO

Temperature [ °C ]
200
834 on all four sides during 180 min is
studied for three moisture contents (0, 150

1.5 and 2.0 % by weight) and the result 100


for the centre point of the column is
shown in Fig 13. The effect is evident 50

and the water plateau at 100 °C for 1.5 0


% compared with 3 % occurs after 100 0 20 40 60 80 100
Time [minutes]
120 140 160 180

and 110 min respectively. The diffe- Fig 13 Temperature development in centre point of a
rence in temperature is maximum 80 square cross section (300x300mm) exposed on
°C after 140 min when comparing the four sides for different moisture contents
curve of 0 and 3% moisture content.

4.5 Effect of voids

The heat flux of an enclosure is transferred through radiation and convection. At high
temperatures the radiation dominates. The non-linear system involving convection and
radiation exchange between surfaces and/or surfaces and enclosed air is solved at each time
step. Radiation exchange between surfaces and enclosed air usually applies to smoke-filled
enclosures only (second term in eq. 6). A combination of both is theoretically possible for
partly smoke-filled enclosures, but in most cases radiation is only considered between surfaces
(term 3 in eq. 6)

N
q n,c = hc (T g − Tb ) + ε r σ (T g4 − Tb4 ) + ∑ Eb, j T j4 eq. (6)
j =1

with

q n ,c = heat flow at enclosure boundary [W/m2]


E = surface radiation matrix
T = boundary temperature vector [K]
N = number of surfaces

The surface radiation matrix is defined as

Anderberg, Y 11
CISM - Effect of heat on concrete Thermal properties & analysis

−1
 Ik, j 1− ε j 
Ek , j = 
 εj
− Fk , j
εj


(Fk , j − I k , j )σ eq. (7)
 

with

F = view factor matrix


I = unit matrix [1 if k=j; 0 if k≠j]
εj = emissivity of surface j (set to 0.9 for concrete surface in example)
k = 1..N
j = 1..N

The influence of taking into account radiation and convection in a void is studied for a hollow
slab with a geometrical size in accordance with Fig 14. ISO 834 fire exposure is used for 2
hours on the bottom surface. The predicted temperature after 2 hours especially at the bottom
of the void is considerably higher about 625 °C with an adiabatic border around the void and
only 330 °C than heat is transferred into the void which is shown in Fig 14 and 15. At the top
of the void the temperature is about 100 °C warmer for the “correct” cavity model in Fig 15.
The temperature of prestressing steel normally positioned close to the bottom of the void is
very much influenced by using the “correct” model.

Fig 14a Cavity model without convection and Fig 14b Cavity model accounting for convection
radiation and radiation: Temperatures after 2
hours

Anderberg, Y 12
CISM - Effect of heat on concrete Thermal properties & analysis

700 120

110
600
100
500
90
Temperature [ °C ]

Temperature [ °C ]
400 80

70
300
60
200
cavity model
50
no cavity model cavity model
100 40 no cavity model

30
0
0 20 40 60 80 100 120
20
Time [minutes] 0 20 40 60 80 100 120
Time [minutes]

Fig 15a Temperature development in point at the Fig 15b Temperature development in point at
bottom of cavity the top of cavity

Anderberg, Y 13

Вам также может понравиться