the performance in fire of concrete filled shs columns protected by intumescent paint

EDWARDS - The Performance in Fire of Concrete Filled SHS Columns Protected by Intumescent Paint - 1 - 19/03/13

1 INTRODUCTION Filling a Structural Hollow Section (SHS) column with concrete can markedly increase its load carrying capacity. The characteristic performance of such a structural member is well understood. Over the last two decades, various ambient temperature design manuals have been produced based on both UK (British Steel 1992) and European (CIDECT 1995) Ultimate Limit State (ULS) Design codes.

It is also known that a concrete infill will improve the thermal characteristics of a composite column and reduce the amount of external fire protection necessary when such a column is fully utilised as a load carrying member. However, at present, only limited deeply conservative design advice is available on how to utilise this (see Table 1). The need to quantify this effect became more urgent after the introduction of a new UK ULS Fire Design code (British Standards 1990) for steel structural members based on a variable limiting steel temperature obtained by reference to a reduced fire design load.

Accordingly, in 1995, British Steel Tubes and Pipes, in collaboration with the intumescent paint manufacturer Nullifire, initiated a series of indicative furnace tests (Edwards & Wainman , in prep) on a range of unloaded SHS sections.

2 PRESENT SPECIFICATION METHODOLOGY In the UK, the required thickness of external fire protection to a steel member is specified by reference to the Section Factor of the SHS shell, where;- Exposed Perimeter Section Factor (Hp/A) = ————————

c.s.a. of the SHS

A Circular Section is a uniform pure shape, so when fully exposed to a fire, the Section Factor will be based on the full perimeter and cross-sectional area, i.e;-

.D 1 Hp/A = ————————— = —————

/4.D2 - /4.(D - 2.ts)

2 ts.(1 - ts/D)

An unprotected Hot Finished SHS has rounded corners, but ignoring the corner effects will still produce a closely accurate assessment of its Section Factor, so;-

1 Hp/A = ———————

ts.[1 - 2ts/(B + D)]

The Performance in Fire of Concrete Filled SHS Columns Protected by Intumescent Paint

M. Edwards Technical Marketing, British Steel Tubes and Pipes, UK

ABSTRACT: Filling a Structural Hollow Section column with concrete will increase its load carrying capacity. The concrete infill will also improve the thermal characteristics of the composite column and reduce the amount of external fire protection necessary. However, only limited design advice is presently available on how to quantify this effect. British Steel in collaboration with the intumescent paint manufacturer Nullifire have undertaken a series of investigative fire tests on a range of unloaded SHS specimens. The test data has produced a specific assessment method to predict the performance of an intumescent paint in fire. Generalised expressions have also been developed to assess the Effective Section Factors of protected filled SHS in fire that can be used with any type of external protection. Additional fire tests on fully loaded columns have also confirmed that the same ULS limiting temperatures can be used with both filled and unfilled SHS columns.


For a square section this simplifies to;- 1

Hp/A = ————

ts.(1 - ts/B)

The Section Factor is intended for use with

tables giving specified thicknesses of both generic and proprietary inert protection systems. A table for assessing the permissible reduction in external protection thickness due to the presence of a concrete core is given in BS 5950 Part 8, viz;-

Table 1. Protection modification factor (C)

Hp/A(e) C Hp/A(f)

50-75 1.00 50-75

100 0.92 92.0

125 0.88 110.0

150 0.81 121.5

175 0.75 131.8

200 0.69 138.0

260 -300 0.55 143 -175

i.e, if the specified thickness of protection by reference to the EMPTY SHS shell = ti , then the reduced thickness required = tir = C.ti This is equivalent to reducing the Section Factor by the same proportion (also see Table 1). In practice, it has been observed during real fire tests (Edwards 1998) that filled sections appear to have substantially lower Section Factors.

3 MODELLING THE BEHAVIOUR OF PROTECTED CONCRETE FILLED SHS SECTIONS

Figure 1 shows the observed temperature time profile of a 150 sq x 6.3 concrete filled SHS column heated to the ISO curve while externally protected by a layer of Nullifire S605 intumescent paint with an initial thickness of 1819 microns. Figure 2 shows the measured temperature developments within the concrete along the diagonal. It can be seen that once intumescence has occurred, the temperature time development of the SHS is relatively linear (as is to be expected for a protected section with a low effective Section Factor). Moreover, the individual layers of concrete all appear to have a similar rate of temperature increase once the free water in each layer has been boiled off. This suggests that the temperature profiles of the column interiors can be simply modelled in terms of an underlying uniform heating rate without explicit reference to the external fire regime or the thermal behaviour of the protection layer.

Figure 1 Temperature development in the SHS shell

of a protected filled SHS

Figure 2 Temperature development within the concrete core of a protected filled SHS

3.1 Protected Circular Section Cores

For a circular core, the underlying heating rate of the concrete can be quantified as;-

(l - i) Qc = c.cc.Ac. ————

(l - wd)

T with;- Twd = ————————— [1 + cc. (l - i)/(pw.Lh)]

where;-

l = Limiting temperature of the SHS

i = Initial temperature of the SHS

c = density of the concrete

(taken as 2300 kg/m3)

cc = specific heat of the concrete (see below)

Ac = csa of concrete core

l = Time to reach limiting temperature

wd = Water Dwell Time (minutes)

L = Latent heat of water (2150 kJ/kg)

pw = unit water content of the concrete

by weight (taken as 0.04)

The temperature gradient through the core can

now be estimated as a one-dimensional heat flow by

0 20 40 60 80 100 120 140 160 180 200 220 240

0

100

200

300

400

500

600

700

800

900

Test Time (minutes)

Concre

te T

em

pera

ture

(oC

) UC6 - 10mm deep

UC7 - 20 mm deep

UC8 - 30 mm deep

UC9 - 50 mm deep

UC5 - centre point

Test 1 : 150 sq x 6.3 SHS Column

Upper Level A : DIAGONAL CONCRETE Temperatures

Time to Failure : 102 min

0 20 40 60 80 100 120 140 160 180 200 220 240

0

100

200

300

400

500

600

700

800

900

Test Time (minutes)

Ste

el Tem

pera

ture

(oC

)

S1

S2

S3

S4

S5

Test 1 : 150 sq x 6.3 SHS Column

Upper Level A : STEEL Temperatures

Time to Failure : 102 min


dividing the core up into N annuli of equal area. Each annulus will then, in turn, absorb a constant proportion (1/Nth) of the heat while allowing the remainder to be conducted through it towards the centre. So for the nth annulus;-

n = Qc.(- n/N).(1/2k).[1- (N- n)/(N+1-n)]

0.5

while n/N (1)

n-1

and n = l - n )/2 - x) 0

where;-n = Temperature gradient across the

nth

annulus x = Temperature gradient across the

xth

annulus l = Limiting temperature of the SHS n = Temperature of the n

th annulus

= proportion of the concrete area actually being heated

k = conductivity of the concrete (see below)

The temperature profile through the circular

core can be simply obtained on a spread sheet by summing the contents of N cells using trial values of . This can be done using an iterative macro, with conditional checks. Once these conditions have been satisfied, the average temperature of the core (av) and a resulting core utilisation factor () can be calculated as;-

ccc (av - i) = ——.———

scs (l - i)

3.2 Protected Square Section Cores

For a square section, the underlying heating rate of the concrete can be based on heat flow through one side face of the section and quantified as ;-

(l - i) Qc = c.cc.Ac/4. ————

(l - wd)

The temperature gradient through the core can now be initially estimated as a one-dimensional heat flow by dividing the core up from the side face to the centre line into N vertical strips of equal area. Each strip will then, in turn, absorb a constant proportion (1/N

th) of the horizontal heat

flow from that face, while allowing the remainder to be conducted through it towards the centre. However, there will also be an identical vertical heat flow from the horizontal faces. The only way

to satisfy both conditions simultaneously is to subdivide a quadrant of the core into an N x N grid and place the resulting temperature gradients along the grid diagonal. So, at position (n,n) on the core diagonal;- n,n) = Qc.( - n/N).(1/2k) while n/N (2)

There will also be a secondary temperature gradient occuring between the the core diagonal and core centre line. This can be closely approximated by assuming that;-

n,N) = n,n) .( 1 + (N - n)/n ) (3)

1

The temperature profile along the diagonal of a square core can now be simply obtained on a spread sheet by using equation (2) to sum the contents of N

. Again this can be done using an iterative macro, with a conditional check. Temperature off-sets within the core quadrant can then be assigned to the individual elements of the N x N grid using equation (3). The average temperature of the core (av) and the resulting utilisation factor (

3.3 Unprotected Filled Sections

Within limits, the temperature of an unprotected section after a given ISO heating time can be obtained using the published formula;- T = 0.54 .s - 50)/(Hp/A)e

0.6

In fact, a more accurate estimate can be made

using the modification;- T = 0.135 .s

2/(Hp/A)e ]

0.604 (4)

This formula can be incorporated into the

procedures given above for protected sections by continually replacing l with s and recalculating as the iteration progresses. It is then possible to assess both the final steel section temperature and effective section factor after a fixed heating time (15, 30 or 45 minutes).

3.4 Concrete Thermal parameters

The specific heat and conductivity of concrete vary markedly with temperature. According to the base document of Eurocode 4.0 part 1.2 ;- c = 0.90 + 0.08.(c/120) - 0.004.(c/120)

2 kJ/kg/

oC

k = 2.0 - 0.24.(c/120) + 0.012.(c/120)

2 W/m/

oC


Comparison between assessed and measured temperature profiles showed that only average values need be used for these parameters during the iteration processes described above and that these could be quantified using the temperature at the 1/2 area depth on the centre line.

4 INDICATIVE TEST SPECIMENS A total of 34 indicative specimens were heated to the ISO temperature-time curve using a private gas fired test furnace owned by the Nullifire Company at Coventry. These specimens can be divided into three groups;- a) Specimen Group 1 - Empty/Protected Sections Test group 1 comprised seven empty SHS specimens coated with Nullifire S605 intumescent paint. These specimens were used to initialy assess the performance of the paint. Table 2 gives details of these specimens including the time taken for each to reach the average temperature of 550oC used to quantify the paint performance. An initial regression was performed on these specimens using the standard formula for inert external protection, viz;-

ti0.77

T = C. ————

(Hp/A)e

0.77

It was found that this formula had to be modified to include a size effect, i.e;-

B + C.(1 + s/D).ti0.77

T = A + ————————— (5) (Hp/A)e

0.77

where;- s = section size

This additional parameter probably reflects the actual pattern of intumescence that occurs on an SHS, namely a corner/diameter effect that becomes less pronounced as section size increases.

Values for the coefficients and resulting correlation are given as regression group r1 in Table 4.

b) Specimen Group 2 - Filled/Protected Sections

Test group 2 comprised twenty three filled SHS specimens, again coated with Nullifire S605 intumescent paint. These specimens were used to quantify the performance of the paint on filled sections and compare preformance with the empty sections. Table 3 gives details of these specimens, again including the time taken for each to reach the average temperature of 550

oC .

An initial regression was performed using equation (5), then a second, more conservative regression performed that excluded the three best specimens. The coeffients and correlations obtained are given as regression groups r2 and r3 in Table 4.

c) Specimen Group 3 - Filled/Unprotected Sections

Table 4. Regression Coefficients and Correlations

Case No. A B C D r2

r1 - unfilled 7 14.0 219 2.49 401 0.984

r2 - filled 23 2.5 795 1.66 287 0.964

r3 - filled 20 1.5 792 1.60 264 0.985

r4 - all spec. 30 10.6 559 1.61 243 0.984

Table 2. Test Group 1 (7 Specimens)

Spc Size ti Hp/A Time to 550

oC (min)

mcrons m-1

test reg 1 reg 3

nk1 114.3 x 3.6 471 286.8 21.0 21.3 14.9

nl2 114.3 x 6.3 2111 168.0 38.0 40.7 32.8

na1 100 x 4.0 481 260.4 21.0 22.0 15.9

nc1 100 x 10.0 2230 111.1 53.0 51.1 44.6

nd4 200 x 6.3 2730 163.9 54.0 50.9 41.5

ne2 200 x 10.0 617 105.3 35.5 34.6 34.3

nf3 200 x 16.0 2810 67.9 86.5 88.1 81.4

Table 3. Test Group 2 (23 Specimens)

Spec.

Ident

Size ti Hp/A (m-1

) Time to

550oC (min)

mcrns empty full test reg3

nk2 114.3

x 3.6

670 286.8 68.04 40.0 45.5

nk3 1534 -"- 61.35 61.0 61.9

nl3 114.3

x 6.3

1110 168.0 56.46 59.0 59.5

nl4 2139 -"- 54.10 80.0 76.9

na2 100

x 4.0

2580 260.4 66.00 67.5 69.9

na3 1440 -"- 66.30 66.0 56.3

nb2 100

x 6.3

1050 169.4 62.69 57.0 53.4

nb3 2130 -"- 60.53 75.0 69.2

nc2 100

x 10.0

690 111.1 56.47 56.0 52.0

nc3 1870 -"- 55.60 67.0 70.3

nd1 200

x 6.3

622 163.9 54.29 56.0 56.3

nd2 1730 -"- 43.21 111.0 93.0

nd3 2800 -"- 41.80 120.0 117.5

ne3 200

x 10.0

1810 105.3 38.69 111.0 103.0

ne4 2870 -"- 37.64 122.0 128.8

nf1 200

x 16.0

634 67.9 36.92 74.5 75.6

nf2 1800 -"- 32.82 121.0 116.5

nf4 2790 -"- 31.97 141.0 143.9

ng4 300

x 8.0

818 128.4 41.86 77.0 79.7

ng2 2660 -"- 36.05 144.0 145.0

nh1 300

x 10.0

417 103.4 41.22 63.0 66.9

nh2 1420 -"- 36.43 113.0 108.3

nh3 1910 -"- 35.24 129.0 126.1


A small group of 4 specimens were filled with concrete but left unprotected. Details are given in Table 5 which shows their assessed effective Section Factors after heating to 550

0C ; the

predicted time it should take for them to reach this temperature according to regression 3 and the actual measured time for them to do so.

Table 5. Test Group 3 (4 specimens)

Spec. Size Hp/A (m-1

) Time to 550oC

(minutes)

empty filled test predict

nl1 114.3 x 6.3 168.0 74.13 19.5 20.8

nb1 100 x 6.3 169.4 76.70 19.5 19.9

ne1 200 x 10.0 105.3 55.15 23.0 24.9

ng1 300 x 8.0 128.4 56.22 24.0 24.6

4.1 Comparative Performance

In practice, all three regressions give similar significance to each component of the formula. Figure 3 shows the estimated fire times for all 30 protected specimens calculated using the most conservative prediction formula.

Figure 3 Predicted v Measured Times to 550

oC.

5 GENERAL PERFORMANCE EQUATIONS

Modified versions of the simple assessment spread sheets and their macros were used to obtain discrete values for the effective Section Factors for the full range of Circular and Square British Steel Hollow Sections at 540

oC , the standard

limiting temperature of fully utilised columns.

Figure 4 Comparative Section Factors of Filled v Empty SHS

Figure 4 shows a comparison of the assessed effective factor obtained for protected Circular Sections. It can be seen that these values are substantially lower than those given in Table 1 according to the present recommendations.

Using the following approach, the discrete values of effective Section Factors obtained from these spread sheets were also regressed to obtain simple expressions that would directly evaluate effective Section Factors;- i) The Effective Section Factor for a section can be defined as;-

Exposed Perimeter of the SHS (Hp/A)e = ———————————— Effective csa of the filled SHS

P = ————

As + .Ac

ii) This formula can be simplified to;-

1 (Hp/A)e = —————————

(ts + tce).[1 - (ts + tce)/D] where;- tce = apparent increase in CHS wall

thickness due to the presence of the concrete

iii) The value of tce is time dependent and has been quantified to a high accuracy for filled CHS as;- either tce = 1.858.T

0.5 when di > 24.717.T

0.5

or tce = 0.1206.di.[2 - di /(17.955.T

0.5)]

when di 24.717.T

0.5

where;- T = the ISO heating time in minutes

di = the internal CHS diameter in mm iv) The apparent increase in the wall thickness of SQUARE sections has also been explicitly quantified in tabular form. Explicit formulae are still being developed to take account of the two dimensional heat flow. However, a slightly conservative set of expressions have been developed using CHS style formulae, namely;- either:- tce = 1.810.T

0.5 when bi > 20.972.T

0.5

or:- tce = 0.1193.bi.[2 - bi /(16.422.T

0.5)]

0 50 100 150 200 250 300

E m p ty CH S S e c tio n Fa c to r (1 / m )

10

20

30

40

50

60

Fill

ed

CH

S S

ect

ion

Fa

cto

r (1

/m)

2 h o u rs IS O h ea tin g

1 .5 h o u rs IS O he a tin g

1 h o u r IS O h e a tin g

Estimated using generalised Concrete Thermal

Parameters according to EC4 Part 1.2

E ffe c tiv e S ec t ion Fac tors fo r P ro tec ted Filled CH S

0 20 40 60 80 100 120 140 160

M e a su re d T ime to 5 5 0 C (min u te s )

0

20

40

60

80

100

120

140

160

Pre

dic

ted T

ime to 5

50

C (m

inu

tes)

u n fille d se c tio n s

fille d CH S

fille d S q u a re s

E xc lu d e d

S q u a re s

L P C T e s t

Co lu m n s

T = 1.4 + [791 + 1.6.(1 + d/264).ti^0.77]/(Hp/A)^0.77

r^2 = 0.985

R eg ressio n o f f illed-pro tec ted sec tion s

exc luding 3 best indicative spec imens


when bi 20.972.T0.5

where;- T = the ISO heating time in minutes

bi = the Internal square section size in mm

6 COLUMN TESTS

While the indicative series was continuing, a series of six full size protected SHS column specimens were fabricated and fire tested at the Loss Prevention Centre and used to investigate if the existing load ratio method/limiting temperature method for steel columns could be safely applied to protected composite SHS columns (Edwards, in prep)

Five test specimens were designed using SHS of markedly differing sizes but with the same wall thickness and SHS Section Factor so as to produce a clear indication of the mass cooling effect of the concrete and how the presence of load bearing concrete would modify the failure temperatures in fire. A sixth specimen using the maximum possible wall thickness (16.0 mm) was also introduced.

6.1 Fire Test Load Ratio and Limiting Temperature

The actual Load Ratio of a member in fire (and the resulting Limiting Temperature) depends both on the type of member and the relative magnitude of the Permanent and Live Loads.

Assuming that the member is being used with 100% efficiency in a room temperature design; that the Permanent Load is Equal to the Imposed Load; that the Imposed Load will reduce in fire and that Wind Load can be discounted, then;-

D + 0.8.L Fire Test Load Ratio (R) = —————

1.4.D + 1.6.L

with D = L 1.8

so, R = —— = 0.6 3.0 Table 6, below, is reproduced from BS5950

Part 8 and shows the Limiting Temperatures to be used with protected steel columns for the generalised range of possible Fire Design Load ratios ;-

Table 6. Limiting Temperatures for steel columns

Column

Slenderness

Limiting Temp. at Load Ratio of:

0.7 0.6 0.5 0.4 0.3 0.2 oC

oC

oC

oC

oC

oC

70 510 540 580 615 655 710

> 70 180 460 510 545 590 635 635

NOTE: column slenderness = effective length divided by

radius of gyration

6.2 Test Details

The six columns were fabricated and filled with a commercial structural grade concrete at the Loss Prevention Centre, Elstree. The axial capacity of each column was calculated using BS 5400 Part 5 procedures based on measured SHS material properties and the concrete 28 day cube strength and the normal material strength reduction factors. The concrete contribution () to the axial capacity was also assessed. The fire test load was taken as 60% of this capacity. Each column would then have a limiting temperature of 540

oC, if they behaved in an

identical fashion to steel columns. Table 7. LPC Column Tests - Failure Temperatures

Sp. SHS

Size

Time to

Failure /

dft ()

Contrib.

Factor

SHS Limiting

Temperature (oC)

Nominal

Actual

min oC

oC

C1 168.3

x 6.3

116.5

(2444 0.288 540 610

C2 323.9

x 6.3

165.8

(1785 0.465 539 692

S1 150

x 6.3

101.5

(1819 0.217 540 613

S2 200

x 6.3

108.8

(1783 0.250 540 616

S3 200

x 16.0

116.5

(1940 0.117 540 550

S4 300

x 6.3

145.5

(1791 0.417 540 661

Table 8. LPC Column Tests - Actual and Predicted

Times to 550oC

Spec

No

Size ti Hp/A (m-1

) Time to 550oC

(min)

mcrn empty full test reg3

C1 168.3 x 6.3 2444 164.9 42.76 102.0 104.2

C2 323.9 x 6.3 1785 161.9 41.08 115.5 115.5

S1 150 x 6.3 1819 165.7 47.81 95.0 82.9

S2 200 x 6.3 1783 163.9 44.07 105.5 92.8

S3 200 x 16.0 1940 67.93 33.00 116.0 119.6

S4 300 x 6.3 1791 162.1 39.67 128.0 112.0

The temperature-time developments of these

columns were also measured so that estimates could be made of their effective Section Factors for a limiting temperature of 550

oC. These Factors and


the resulting predicted times for the LPC columns to attain this temperature are given in Table 8 and included in Figure.3

6 CONCLUSIONS The indicative test data has produced a specific method to predict the performance of an intumescent paint in fire, In addition, generalised expressions were also obtained for the Effective Section Factors of protected filled SHS that can be used with any type of external protection. Moreover, the full size column tests have confirmed that the same ULS limiting temperatures can be used with both filled and unfilled SHS columns. This should allow full advantage to be taken of the load carrying capacity and improved Section Factor of composite SHS columns in fire.

REFERENCES British Steel 1992.Design Manual for Concrete Filled SHS

Columns Part 1, TD 296: Part 1/3E/92, British Steel

Tubes and Pipes, Corby, Northants, 1992

The Steel Construction Institute 1994.Composite Column

Design to Eurocode 4, (based on DD ENV 1994-1-

1:1994 Part 1.1, with reference to the UK NAD), SCI

Publication No. 142, The Steel Construction Institute,

Ascot, 1994

CIDECT 1995. Design Guide for Concrete Filled Hollow

Section Columns under Static and Seismic Loading,

Verlag TUV, Rheinland, Koln, 1995

British Standards Institution 1990. BS 5950 Structural use of

Steelwork in building, Part 8. Code of Practice for fire

resistant design, British Standards Institution, London,

1990

.Edwards M. 1998. Reinstatement of Concrete Filled

Structural Hollow Section Columns following Short

Duration Fires, Phase 2 - Standard Fire tests on Full

Size Columns, British Steel Tubes and Pipes, Corby,

Northants, 1998

Edwards M. & Wainman D. in prep. British Steel Project

S2808: Fire Tests on Concrete Filled SHS Indicative

Specimens protected by Intumescent Paint, British Steel

Tubes and Pipes, Corby, Northants. (to be published)

Loss Prevention Council 1998. TE89574; Fire Resistance

test on a Loaded SHS Column (168.3 dia x 6.3) filled

with concrete and protected by a Nullifire paint system,

Loss Prevention Council, Borehamwood, UK, 1998


test on a Loaded SHS Column (323.9 dia x 6.3) filled




test on a Loaded SHS Column (150 x 150 x 6.3) filled



Loss Prevention Council 1998. TE89577; Fire Resistance test

on a Loaded SHS Column (200 x 200 x 6.3) filled with

concrete and protected by a Nullifire paint system, Loss

Prevention Council, Borehamwood, UK, 1998









Edwards M. in prep. British Steel Project S2940: Fire Tests on

Fully Loaded Concrete Filled SHS Columns protected by

Intumescent Paint, British Steel Tubes and Pipes, Corby,

Northants. (to be published)

the performance in fire of concrete filled shs columns protected by intumescent paint

Documents