Spalling of concrete exposed to fire
Robert Jansson, Lars Boström
Spalling of concrete exposed to fire
Robert Jansson, Lars Boström
SP Technical Research Institute of Sweden
Box 857, SE-501 15 BORÅS, SWEDEN
Telephone: +46 10 516 50 00, Telefax: +46 33 13 55 02 E-mail: firstname.lastname@example.org, Internet: www.sp.se
Fire Technology SP Report 2008:52 ISBN 978-91-85829-68-2
Spalling of concrete exposed to fire
The addition of a small amount of polypropylene fibres in concrete prevents or reduces the amount of fire spalling. This effect has been investigated empirically using some fire tests and some related material testing. The main part of the study included fire testing of large concrete blocks, 1200 × 1700 × 300 mm, together with small scale fire tests on 500 × 600 × 300 mm blocks. The large fire tests were performed using the standard fire curve exposure and the RWS curve. These tests clearly show the spalling reducing effect of adding polypropylene (PP) fibres to concrete. Another finding was that no systematic difference could be seen between the effect of fibres with diameters of 18 µm and 32 µm.
To address the lack of standardized test methods for fire spalling determination a new method is recommended in Appendix E of this report.
Numerous effects of a PP-fibre addition were shown in the present research project. PP-fibres in concrete have the following effects, they:
reduce or prevent fire spalling
modify the capillary saturation close to the surface of concrete limit the internal destruction at moderate heating, 5 °C per minute. modify the drying behaviour at high temperature
introduce a plateau in the free thermal strain curve.
Key words: concrete, fire, spalling, PP-fiber
SP Sveriges Tekniska Forskningsinstitut SP Technical Research Institute of Sweden SP Report 2008:52
ISBN 978-91-85829-68-2 Borås 2008
Det är väl känt att vissa typer av betong spjälkar sönder vid brand, dvs att betongens yta succesivt sprängs bort. Speciellt känslig är tät betong och betong med hög fukthalt. Detta leder till att betong som används i tunnlar är speciellt utsatt då den typen av betong av beständighetsskäl bör vara extra tät. Dessutom kan tunnlar innehålla stora mängder fukt från t ex. kondens vilket leder till att ytan på
konstruktionen får ett högt fuktinnehåll. Ett antal uppmärksammade bränder i tunnlar har visat att brandspjälkning är ett reellt problem. Om inte brandspjälkningsrisken beaktas vid dimensioneringen kan det leda till extra långa stilleståndstider och höga reparationskostnader efter en brand.
Det mest tillförlitliga sättet att undersöka en betongsammansättnings spjälknings-känslighet när den utsätts för brand är att brandprova. Det finns en handfull matematiska modeller under utveckling med mål att förutsäga brandspjälknings-benägenheten hos betong. För närvarande fungerar ingen av dessa tillfredställande. En bidragande orsak till detta är att förståelsen för fenomenet eller fenomenen som leder till brandspjälkning ej är fullständig. Ett annat problem i sammanhanget är att det saknas materialdata till modellerna. Renodlade materialdata vid hög temperaturer är svåra att få fram. Det beror på att de i många fall inte är enbart
temperatur-beroende utan t.ex. även kan beror på hur snabbt provkroppen värmts upp samt om den varit belastad under uppvärmningen.
Expertutlåtanden är ett annat sätt att bedöma konstruktioners spjälkningsbenägenhet. Om ett expertutlåtande inte innehåller tydliga resonemang och referenser till
provningar som ligger till grund för utlåtandet är det av ringa värde. Denna rapport tillsammans med Boström och Jansson (2008) samt Häggström m. fl.(2007) är exempel på några mer omfattande provningsserier som kan användas som
bedömningsunderlag vid framtagning av expertutlåtanden. Kravnivån när det gäller brandspjälkning bör ej sättas till noll. Erfarenheter från provningar och
brandutredningar visar att det mycket ofta kan ske små avflagningar från ytan på betong som inte innebär någon risk för konstruktionen.
Vid brandprovning provas betongen på en ugn. Grundläggande för all sorts brandprovning är att man försöker använda en provuppställning som liknar den riktiga konstruktionen så mycket som möjligt. När det gäller tunneltvärsnitt är detta ofta svårt, dels p.g.a. storleken på elementen samt dels p.g.a. de olika yttre
belastningarna som kan finnas i en tunnel. Vissa kompromisser måste därför göras. Appendix E i denna rapport innehåller ett förlag på metodik för brandprovning av spjälkningsbenägenheten hos betong. I förslaget ingår en metod för att belasta konstruktionen under brandprovet.
Målet med projektet var att studera risken för brandspjälkning hos en typisk betong som används i tunnelkonstruktioner, samt att om risken visade sig vara stor även att föreslå ett sätt att minska denna risk d.v.s. att tillsätta polypropylenfibrer (PP) i rätt mängd och dimension. En annan faktor som undersöktes var inverkan av olika brandbelastning på resultatet. Ofta ställs det krav på att tunnlar skall dimensioneras för extremt kraftiga bränder som t.ex. RWS brandkurvan. Brandkurvor är
med RWS kurvan är väldigt dyrt jämfört med tester enligt standardbrandkurvan (ISO 834-1 alt. SS-EN 1363-1). Det är känt sedan tidigare att betongprovkroppar
uppvärmda i elektriska ugnar, d.v.s. utan den snabba uppvärmningshastigheten som sker initialt vid en riktig övertänd brand leder till annorlunda resultat jämför med provkroppar utsatta för en traditionell brandkurva. Däremot är det inte känt hur stor skillnaden i brandspjälkningsbenägenheten det är mellan olika brandkurvor. Genom att inkludera tunna PP-fibrer i betongen kan man reducera eller i många fall helt undvika spjälkning vid brand. PP-fibrer finns att få i olika dimensioner. I denna studie har effekten av fibrer med två olika tjocklekar undersökts, 18 µm och 32 µm båda med längden 12 mm. Generellt kan sägas, baserat på de genomförda försöken, att fiberdiametern inte ger någon signifikant skillnad i resultat med avseende på brandspjälkning. Detta trots att en ”hörsägen” i branschen säger att hälften så mycket 18 µm fiber ger samma effekt som 32 µm varianten. Med ”hörsägen” menas att ingen har kunna presentera empiriska bevis för de påstådda förhållandet. Resultat från brandtester
Totalt brandprovades 20 olika betongblandningar. Parametrar som varierades i projektet var:
vatten-cement-tal (vct): 0,4 och 0,45 max stenstorlek: 16 och 32 mm mängd PP- fibrer: 0, 1 och 1,5 kg/m3
diameter på PP- fibern: 18 och 32 µm
Under brandprovningarna utsattes provkropparna för en trycklast på 10 % av tryckhållfastheten. Tre olika typer av test genomfördes:
RWS brandkurvan (stora provkroppar, 1200 × 1700 × 300 mm) standardbrandkurvan (stora provkroppar, 1200 × 1700 × 300 mm) standardbrandkurvan (små provkroppar, 500 × 600 × 300 mm)
I figur 1 visas en jämförelse mellan spjälkningsdjup uppmätta på stora plattor utsatta för en standardbrand i 60 minuter. Värt att notera är att betongen utan tillsats av PP- fibrer, betong A, spjälkade betydligt mycket mer än de övriga betongerna.
Spjälkningen i provkropp A och F var ganska jämt utspritt på ytan emedan det i betong I, M, och N skedde mer koncentrerat i en punkt. Betong F som innehöll 1 kg/m3 PP fibrer med diametern 18 µm var den typen av betong med
Spjälkningsdjup efter 60 minuters standardbrand (EN 1363-01), stora provkroppar (1200 x 1700 x 300 mm) 0 20 40 60 80 100 120 140 160
A, Agg16 noPP E, Agg16 1kg18PP F, Agg25 1kg18PP I, Agg16 1.5kg18PP M, Agg16 1kg32PP N, CAgg25 1kg32PP S pj ä lk ni ng s dj up [ m m ] Medel Max
Figur 1 Spjälkningsdjup uppmätt efter standardbrandexponering i en timma. (Agg16 = maximal stenstorlek 16 mm, 1 kg18PP = 1 kg/m3 PP-fibrer med diametern 18 µm)
Skillnaden i avspjälkning mellan betong med och utan PP- fibrer kunde också ses när betongen utsattes för den betydligt tuffare RWS brandkurvan, se figur 2. Under detta test sprack provkropparna på längden vilket gjorde att provet avbröts efter 30 minuter. Därför kan ej spjälkningsdjupen i sig jämföras mellan standardbrand och RWS brand. Det som dock kan jämföras är typen av spjälkning. Under RWS branden spjälkade alla betong typer, d.v.s. även med olika mängd och typ av PP-fibrer, medan vid provet med standardbrand spjälkade endast betong F innehållande 1 kg/m3
PP-fibrer med diametern 18 µm (om man bortser från punktnedfallen hos betong I, M och N).
Spjälkningsdjup efter 30 minuters RWS brand, stora provkroppar (1200 x 1700 x 300 mm) 0 20 40 60 80 100 120 140 A, Agg16 noPP B, Agg25 noPP E, Agg16 1kg18PP F, Agg25 1kg18PP I, Agg16 1.5kg18PP J, Agg25 1.5kg18PP M, Agg16 1kg32PP N, Agg25 1kg32PP S p jä lk ni ng s d jup [ m m ] Medel Max
Figur 2 Spjälkningsdjup uppmätt efter RWS brand i en 30 minuter. (Agg16 = maximal stenstorlek 16 mm, 1 kg18PP = 1 kg/m3 PP med diametern 18 µm)
Testprogrammet innehöll även 40 brandprover på mindre provkroppar, 500 x 600 x 300 mm. Testerna utfördes som dubbelprov d.v.s. två prov gjordes på varje
blandning. Som figur 3 visar repeterades trenden som visats vid proverna i större skala, d.v.s. all betong utan tillsats av PP fibrer spjälkade. När man tittar på inverkan
av vct på spjälkningsbenägenheten kan ingen klar effekt visas i denna provserie (recept A och B hade vct=0,40, C och D hade vct=0,45). Det var också en viss spjälkning i en av provkropparna vilken innehöll 1 kg/m3 PP med diametern 18 µm. När man jämför resultat från de små provkropparna med resultat från den stora provkropparna utsatta för standardbrand ser man att spjälkningsdjupet är lägre. Det beror troligtvis på mycket större randeffekter hos de små provkropparna under den senare delen av brandprovet. Dessa randeffekter är troligtvis mycket små under det tidiga skedet av branden då spjälkningen sätter igång. Ett tecken på detta kan ses när man jämför tiden till första spjälkningen mellan de två småskaliga försöken på betong A och det på den lite större provkroppen av samma sort. Vid de två
småskaliga försöken började spjälkningen efter 7,8 respektive 8,5 minuter vilket skall jämföras med provet på den stora ugnen vid vilket spjälkningen i provkropp A började efter 8 minuter. För att summera kan den småskaliga provuppsättningen (miniugnen) användas för att undersöka om en betong är spjälkningsbenägen eller ej men för att undersöka hur mycket av ett tvärsnitt som försvinner på en viss tid bör större provkroppar användas, ett förslag på testmetod finns i appendix E .
Spjälkningsdjup efter 1 timmas standardbrand, små provkroppar (500 x 600 x 300 mm) 0 10 20 30 40 50 60 70 80 90 100 A2 A5 B1 B2 C1 C2 D1 D2 E2 E4 F2 F5 G1 G2 H1 H2 S p jä lk nings d ju p [ m m ] [ m m ] Medel Max Provkropp I1-T2 ingen spjälkning
Figure 3 Spjälkningsdjup uppmätt på små provkroppar efter
standard-brandexponering i en timma. Betong A till D är utan fibrer, betong H är med 1 kg/m3
fibrer med diameter 18 μm.
Det finns ett antal olika fibertillverkare på markanden vilka kontinuerligt utvecklar sina produkter. Provningarna som redovisas i denna rapport bör därför generaliseras med försiktighet. Trots detta kan resultaten i denna rapport tillsammans med resultat från den öppna litteraturen (t.ex. Häggström m fl, 2007) fungera som grunden för en rekommendation hur man kan undvika kraftig spjälkning, men ej hur man undviker spjälkning fullständigt.
Om betong skall designa för att kunna motstå en brand motsvarande
standardbrandkurvan (SS-EN 1363-1) kan man tillsätta ett kilo PP fibrer per
kubikmeter betong för att undvika kraftig spjälkning. Både fiber med diameter 18 µm och 32 µm kan användas. Med den rekommenderade dosen minimeras spjälkningen och fortskridande spjälkning, d v s lager efter lager genom hela konstruktionen, kan undvikas.
Om en tunnel skall motstå en mer intensiv brand eller om spjälkning skall undvikas helt måste fler tester utföras för att verifiera den valda betongens funktion vid brand.
131.1 Background 13 1.2 Objectives 13 1.3 Research team 14
152.1 Concrete mixes 15 2.1.1 General 15 2.1.2 Concrete mixes 15 2.2 Manufacturing 16 2.2.1 General 16
2.2.2 Properties of fresh concrete 16
2.2.3 Concrete moulds 17
2.2.4 Instrumentation 18
2.2.5 Specimens for material testing 19
2.3 Conditioning 20
3.1 Fire exposure 21
3.2 Small slab tests 21
3.2.1 Test set-up 21
3.2.2 Temperature measurements 22
3.2.3 Load measurements 24
3.2.4 Vapour pressure measurements 24
3.2.5 Spalling measurements 25
3.3 Large slab tests 25
3.3.1 Test set-up 25
3.3.2 Temperature measurements 27
3.3.3 Load measurements 27
3.3.4 Vapour pressure measurements 28
3.3.5 Spalling measurements 29
3.4 Results from fire tests 30
4.1 Compressive strength 31
4.2 Moisture content 31
4.3 Degree of capillary saturation 32
4.3.1 Introduction 32
4.3.2 Method and preparation 32
4.3.3 Results 34
4.4 Thermal expansion 36
4.4.1 Introduction 36
4.4.2 Method 36
4.5 Stress strain 40
4.5.1 Method 40
4.5.2 Results 40
4.6 Drying tests at high temperature 41
4.6.1 Introduction 41 4.6.2 Method 42 4.6.3 Results 43 4.7 Speed of sound 48 4.7.1 Introduction 48 4.7.2 Method 48 4.7.3 Results 49 4.8 Permeability 54 4.8.1 Introduction 54 4.8.2 Method 55 4.8.3 Results 56
595.1 Introduction 59
5.2 Fire spalling theories 59
5.2.1 Introduction 59
5.2.2 “Moisture Clog” 60
5.2.3 Hydraulic pressure 60
5.2.4 Restrained thermal dilation 61
5.3 Fire spalling models 61
636.1 Spalling measurements 63 6.2 Internal pressure 65 6.3 Material testing 67
697.1 General 69
7.2 Recommendation to add polypropylene fibres to concrete to
avoid or limit fire spalling 69
7.3 Recommended test method for spalling 70
Appendix A – Drawings of test setup
Appendix B – Test results from fire tests on large slabs,
Appendix C – Test results from fire tests on large slabs
Appendix D – Test results from fire tests on small slabs
Appendix E – Test method for fire spalling determination
This project was supported by the Swedish Road Administration (registration number: AL 90B 2005:16378, FUD id: 2758) which is gratefully acknowledged. The work was expanded in the material sciences field with support from the Swedish Research Council, FORMAS, (registration number: 243-204-1992).
Dr Bijan Adl-Zarrabi and Professor Ulf Wickström from SP Fire Technology have provided important input to discussions regarding the project. The fire and material testing has been carried out by Simon Fitz, who conducted the practical part of his studies at SP, Emmanuel Margerie, a French student exchange student working at SP, together with Bengt Bogren, Patrik Nilsson, Martin Rylander, Peter Lindqvist and Kent Pettersson, technical staff at SP Fire Technology. The input from all is gratefully acknowledged.
Borås, December 2008 Robert Jansson, Lars Boström
It is well known that certain qualities of concrete exhibit explosive spalling when exposed to fire. This is particularly true for dense concrete, concrete with high moisture content and structures loaded in compression. Thus concrete used in tunnels may be in danger of explosive spalling since while a dense concrete is good for the durability, the moisture content is often high and the structure is generally loaded in compression.
The objective of the proposed project was to study some well defined concrete mixtures used in tunnel constructions and investigate whether they pose a risk for fire spalling. Should there be a high risk for spalling, methods to eliminate the problem will be proposed. In particular a specific amount of addition of polypropylene fibres into the concrete will be defined. Another factor included in this study was the fire load. Often it is required that the tunnel shall be designed for a very severe fire, such as the RWS (Rijkswaterstaat) fire curve. Testing in accordance with RWS is very expensive compared to tests in accordance with the traditional ISO 834 fire curve, EN 1363-1 in the European system. When investigating the risk for spalling it is not certain that the RWS curve is necessary as the EN 1363-1 curve can provide a good prediction. If this can be confirmed the cost of testing could be dramatically reduced. The main test program was conducted using small specimens (500 × 600 × 300 mm3)
complimented by two large scale furnace tests that were also designed to compare the RWS and the standard fire curve with respect to spalling behaviour. Several concrete mixtures were examined with variation of strength, maximum aggregate size and amount of polypropylene fibre addition. During the tests the internal temperatures and vapour pressure were measured in the concrete specimens. This type of data can be useful for verification of numerical codes. A literature review was also performed to investigate the present status of modelling of fire spalling of concrete.
There is no standardized test methodology available for the determination of the risk of spalling. One aim of this project was to develop and propose such a standard methodology.
The research has been conducted simultaneously with two other projects that are in progress. One project deals with spalling of self-compacting concrete and the other project considers the development of theoretical modelling of fire spalling.
The main research work on the fire tests has been carried out by Robert Jansson and Lars Boström from SP Fire Technology. The practical work with testing has been carried out by Simon Fitz, Bengt Bogren, Patrik Nilsson, Martin Rylander, Peter Lindqvist and Kent Pettersson from SP Fire Technology. Much of the experimental work on permeability and drying was performed by Emmanuel Margerie, a French student working as a part of a student collaboration at SP. Tests were also performed by Robert Jansson at CSTB in Paris, France. The work at CSTB was conducted under the supervision of Pierre Pimienta and in collaboration with Jean Christophe Mindeguia who both contributed with experience and skills in the art of determining mechanical properties at high temperatures.
The preparation of the concrete was performed at the concrete mixing plant Färdig Betong AB in Borås, Sweden. Färdig Betong AB, included in The Thomas Concrete Group, is the largest ready-mix concrete supplier in Sweden. The concrete mixtures chosen for the study are typical Swedish infrastructure concretes that have been used in tunnels in western Sweden as well as modifications of the concretes by addition of polypropylene fibres (PP-fibres). The addition of PP-fibres has proven to be an challenging task and large differences were found in the fresh properties of the concrete as can be seen in chapter 2.2.2.
The concrete mixes used in the project are shown in Table 1, and the products used can be seen in Table 2. The main parameters that were varied were the water cement ratio, maximum aggregate size and type as well as the amount of PP-fibre addition. The names of the different mixes given in Table 1(letters A, B etc.) will be used throughout this report. Concrete U is a reference mix for comparison of data during material testing. The reason for including the reference mix is that this concrete is known to have good fire spalling resistance.
Table 1: Concrete mixes. Series w/c Gravel 0-8 mm [kg/m3] Gravel 8-16 mm [kg/m3] Gravel 16-25 mm [kg/m3] Water [kg/m3] Cement CEM I [kg/m3] Super-plasticizer [kg/m3] Fiber amount [kg/m3] Fiber φ [μm] A 0.40 852 896 - 170 426 2.98 - - B 0.40 847 143 767 168 420 2.73 - - C 0.45 871 862 - 182 405 2.43 - - D 0.45 880 134 772 171 380 2.28 - - E 0.40 852 896 - 170 426 2.98 1.0 18 F 0.40 847 143 767 168 420 2.73 1.0 18 G 0.45 871 862 - 182 405 2.43 1.0 18 H 0.45 880 134 772 171 380 2.28 1.0 18 I 0.40 852 896 - 170 426 2.98 1.5 18 J 0.40 847 143 767 168 420 2.73 1.5 18 K 0.45 871 862 - 182 405 2.43 1.5 18 L 0.45 880 134 772 171 380 2.28 1.5 18 M 0.40 852 896 - 170 426 2.98 1.0 32 N 0.40 847 143 767 168 420 2.73 1.0 32 O 0.45 871 862 - 182 405 2.43 1.0 32 P 0.45 880 134 772 171 380 2.28 1.0 32 Q 0.40 852 896 - 170 426 2.98 1.5 32 R 0.40 847 143 767 168 420 2.73 1.5 32 S 0.45 871 862 - 182 405 2.43 1.5 32 T 0.45 880 134 772 171 380 2.28 1.5 32 U 0.62 1051 881 - 192 308* 0.04 - - * CEM II
Table 2: Products used in the concrete mixes.
Cement Concrete A-T: Degerhamn CEM I 42.5 BV/SR/LA Concrete U: Skövde Bygg CEM II/A-LL 42.5R Superplasticizer EVO 26-50
Air entrainment Sika Aer-S
Polypropylene fibers φ18 μm, length 12 mm φ32 μm, length 12 mm
A petrographic analysis of the aggregate used in all concrete mixes is shown in Table 3. As seen in the table the completely dominating part of the aggregate is granite. Table 3: Petrographic analyse of the aggregate.
Size 0.063-2 mm Size 2-4 mm Size 4-8 mm Size 8-16 mm Size 16-25 mm
Granite 97% 97% 96% Quartz, Felspar 94% 94% Amphibolite 3% 2% 4% Diabase 4% 3% Mica 2%
All test specimens were moulded at SP with material delivered from the concrete mixing plant Färdig Betong AB. The different concrete mixtures were delivered by concrete lorries to pre-built moulds in the large furnace hall at SP. Two sizes of slabs were manufactured for fire tests: small slabs 600 × 500 × 300 mm3, and large slabs
1700 × 1200 × 300 mm3. Smaller specimens were also prepared for material testing.
Properties of fresh concrete
When the concrete was delivered to SP from Färdig Betong AB the fresh properties were tested on concrete taken directly from the lorry. The results of slump flow and air content measurements can be seen in Table 4. The measurements show noticeable but non-systematic variations. The variation is larger between the concrete mixtures containing PP-fibres compared to those without PP-fibres.
Table 4: Summary of properties of the fresh concrete. Slump flow [mm] Air content [%] A 85 3.9 B 110 5.4 C 170 4.2 D 120 4.0 E 190 5.9 F 135 3.0 G 45 3.8 H 35 3.6 I 192 5.2 J 54 2.8 K 55 3.4 L 50 3.8 M 102 4.0 N 89 3.0 O 120 4.9 P 63 3.4 Q 25 3.7 R 75 3.8 S 60 4.0 T 44 4.6 U 28 2.0
The concrete specimens were moulded in removable frames on a moulding floor. Figure 1 show the wooden frames for small slabs in a plan view. The large specimens were moulded in similar, large frames. After moulding the specimens were covered with plastic. Immediately after removal of the forms the specimens were put in a large water tank stored in the cellar of the furnace hall of SP Fire Technology.
Figure 1: Wooden frame for moulding of small slabs.
All specimens that were to be fire tested were equipped with an internal loading system. During the fire test, DYWIDAG post tension bars were used in the test specimen. To be able to insert the post tension bars the large slabs were equipped with plastic pipes of inner diameter 46.4 mm and material thickness 1.8 mm. The small slabs were equipped with aluminium pipes. Figure 2 shows a small slab with inserted pipes.
Figure 2: Small slab (standing on side) with holes for post stressing bars.
All specimens were equipped with internal type K thermocouples. In the small slabs the temperature was measured on depth 10 and 40 mm. The large slabs had two thermocouples on each depth. The depths were then 10, 20, 40, 80 and 120 mm. Stainless steel pipes for steam pressure measurement (Jansson, 2006) were inserted in 12 of the in total 20 small slabs and in 8 of the in total 14 large slabs. The inner
diameter of the pipes was 2 mm with a steel thickness of 0.2 mm. Pipes were inserted on the depths 5, 10, 15 and 20 mm.
Specimens for material testing
During moulding, standard testing cubes, 150 mm × 150 mm2, and concrete filled
plastic pipes were manufactured for the determination of the profile of capillary saturation. The plastic pipes had the following dimensions: inner diameter 46,4 mm and length 300 mm, they were open at both ends to imitate one dimensional drying of a 300 mm thick slab, i.e., the thickness used in the fire tests.
Beams with the dimensions: 100 × 100 × 600 mm3 were also manufactured for
fracture energy tests. The 50 mm deep notch necessary for the fracture energy tests was sawed when the concrete was less then one week old.
Core specimens with different diameters were drilled from concrete slabs with the dimensions of 500 × 600 × 300 mm3. The drilling was performed when the concrete
were less then one week old. The specimens used for permeability measurements and Transient Plane Heat source measurements (Jansson, 2004) were sawed in 48 mm thick slices from a core with a diameter of 150 mm. After sawing the saw surfaces were ground smooth using a diamond grinder.
Some of the specimens were stored together with the specimens used for fire testing the rest of the specimens were stored in plastic bags, see Table 5.
Table 5: Specimens for material testing, including storage conditions.
Stored in plastic Water and air
stored (along with the slabs for
fire testing) Recipe Cores φ 35 mm Cores φ 60 mm Cores φ 104 mm
Perm. & TPS cores 150 mm (height 4
surface grinded Notched Beams 100*
In plastic pipes In plastic pipes (capillary saturation test)
Cubes (150 x 150 mm 2 ) B 27 16 6 8 8 3 3 10 A 27 16 6 8 7 3 3 10 E 27 16 6 8 8 3 3 10 F 27 16 6 8 8 3 3 10 I 27 16 6 8 8 3 3 10 J 27 16 6 8 8 3 3 10 M 27 16 6 8 8 3 3 10 N 27 16 6 8 8 3 3 10 U 13 12 6 8 4 3 3 10 sum: 229 140 54 72 67 27 27 90
All specimens used in the fire tests were stored under water and then in air according to tables 6-7. The air storage was in the laboratory with an temperature of
approximately 20 °C and an average relative humidity of 50 % during the storage time. The specimens used for material testing was stored in plastic bags to minimise the moisture gradients and cracking close to the surfaces.
Table 6: Conditioning of large slabs
Moulded and then stored
under water air storage fire test
Concrete start days start Days* date age* (days) A 23-nov 230 11-jul 23 03-aug 253 B 17-nov 236 11-jul 23 03-aug 259 E 28-nov 225 11-jul 23 03-aug 248 F 06-dec 217 11-jul 23 03-aug 240 I 14-dec 209 11-jul 23 03-aug 232 J 09-jan 183 11-jul 23 03-aug 206 M 15-jan 177 11-jul 23 03-aug 200
N 17-jan 175 11-jul 23 03-aug 198
* two tests were performed on each concrete (except B and J, one specimen) but storage times do not diverge more then a week between the specimens
Table 7: Conditioning of small slabs.
Moulded and then stored
under water air storage fire test
Concrete start days start days* date age (days)* A 23-nov 230 11-jul 65 14-sep 295 B 17-nov 236 11-jul 56 05-sep 292 E 28-nov 225 11-jul 58 07-sep 283 F 06-dec 217 11-jul 70 19-sep 287 I 14-dec 209 11-jul 69 18-sep 278 J 09-jan 183 11-jul 65 14-sep 248 M 15-jan 177 11-jul 65 19-sep 247
N 17-jan 175 11-jul 71 20-sep 246 C 24-jan 168 11-jul 72 21-sep 240 D 26-jan 166 11-jul 75 24-sep 241 G 01-feb 160 11-jul 76 25-sep 236 H 01-feb 160 11-jul 77 26-sep 237 K 02-feb 159 11-jul 78 27-sep 237 L 02-feb 159 11-jul 79 28-sep 238 O 05-feb 156 11-jul 79 28-sep 235 P 05-feb 156 11-jul 97 16-okt 253 Q 06-feb 155 11-jul 85 04-okt 240 R 06-feb 155 11-jul 93 12-okt 248 S 08-feb 153 11-jul 96 15-okt 249 T 08-feb 153 11-jul 97 16-okt 250
* two tests were performed on each concrete but storage times do not diverge more then a week between the specimens
One of the aims of this project was to investigate the influence of the severity of the fire on the spalling behaviour. Therefore two different fire exposures were used, the RWS (Rijkswaterstaat) curve and the standard fire curve, EN 1363-1, see Figure 3. The RWS curve represents a very severe fire exposure trying to emulate a worst case fire scenario in a tunnel, and the standard fire curve represents the temperature development in room exhibiting a flash over.
0 200 400 600 800 1000 1200 1400 0 30 60 90 120 150 180 Time (minutes) T em per atu re ( o C) RWS-curve
Standard fire curve (EN 1363-1)
Figure 3: The Standard fire curve and the RWS fire curve.
Small slab tests
A small furnace was used in the test series performed as a part of this project. This furnace had the inner dimensions of 500 × 400 × 525 mm3 with an opening of
500 x 400 mm2. The specimens to be tested covered the opening, i.e. the fire
When testing according to the European standard, EN 1363-1, the furnace
temperature is measured using plate thermometers. The use of plate thermometers in furnace testing has revolutionised such testing relative to the use of ordinary shielded thermocouples. This has lead to smaller differences in fire exposure from different furnaces, i.e. the time a construction can survive a fire is less dependent of the furnace that is used.
In the small scale furnace used in this project, it was unfortunately not possible to include a plate thermometer because of the shadow effect on the specimen, as the dimensions of the plate thermometer are 100 x 100 mm2. However, in a attempt to adjust the thermal exposure to a standard fire test (EN 1363-1) the furnace was calibrated using a plate thermometer. The calibration was performed by regulating the furnace according to a plate thermometer and simultaneously measuring the temperature with a 1 mm shielded type K thermocouple, see Figure 4. The difference in temperature readings between the two temperature measurement devices is shown in Figure 5.
Figure 4: Comparison test between a 1 mm shielded typ K thermocouple and a plate thermometer
10 15 20 25 30 35 40 45 50 55 60
Standard time temperature curve
Thermocouple, 1 mm shielded type K
T(t) = 290*log(30*t+1)+20
Figure 5: Temperatures measured in the small furnace during calibration using different measurement devices.
The new fire curve for the small furnace to be used with 1 mm shielded thermocouples was calculated to be:
T(t) = 290 × log(30 × t + 1) + 20 [oC]
T = temperature t = time in minutes
The temperature was also measured inside the test specimens during the fire tests. This was accomplished using internal thermocouples with an welded tip as the measuring point. The thermocouples were placed in the concrete during moulding at a distance of 10 mm and 40 mm from the surface that was exposed to the fire. All temperatures were logged with a NetDaq system connected to a computer.
All tests were performed on specimens loaded in compression. It is known from previous experience and from the international literature that the presence of a compressive load is an important factor influencing the spalling process. In particular when the specimen is small and unreinforced, cracking of the specimen will be restricted by the load and the results from small scale specimens will be more in correlation with reinforced specimens of larger sizes. Previous experience indicate that it is the presence of a load and not the size that is important in the used type of specimens. The load level chosen was 10 % of the cube compressive strength as tested prior the fire test. Post stressing was performed using four Dywidag bars, thickness 28 mm, inserted in the concrete. The applied load was monitored using four load cells, two on each side of the specimen. The loading system is shown in Figure 6.
Figure 6: The loading system for small slabs.
The tensile stress in the bars was measured with load cells connected to an MGC Plus system and a computer for data collection.
Vapour pressure measurements
During the fire tests performed on 12 of the 40 tested small slabs, the internal vapour pressure was measured. Pressure pipes open at the following depths: 5, 10, 15, and
20 mm from the fire exposed side were inserted during moulding (Jansson, 2006). Just before each fire test the pipes were filled with oil. The oil translated the internal pressure out to external pressure gauges connected to an MGC Plus system and a computer for data acquisition. See chapter 3.3.4 for a more detailed description of the pressure measurement system.
The weight loss of the specimens due to fire exposure was recorded which can be seen as a rough estimation of the spalling degree. The roughness of this method comes from the fact that it is complicated to compensate for the weight loss due to moisture evaporation and the size of the boundary zone. A more precise method that was also used was to measure the spalling profile in 42 evenly spread points on the 400 x 500 mm2 fire exposed surface. Using this method, the maximum and mean
spalling depth along with a spalling profile, can be determine. With this type of measurement you can also take away the measurements close to the unheated zone of the concrete, i.e. close to the slab edges were the specimen is resting on the furnace.
Large slab tests
The furnace used in the large slab test series had the inner dimensions of
5 × 3 × 2 m3 with a horizontal opening of 5 × 3 m2. The specimens to be tested, with
the dimensions 1200 × 1700 × 300 mm3, covered the opening, i.e. the fire exposure
was one side only. The large slabs were tested in two separate furnace runs, one with the RWS fire exposure and one with the standard fire exposure specified in EN 1363-1. The location of the test specimens on the furnace in the two tests are shown in Figure 7 and Figure 8. The specimens were connected two and two with the internal loading system described in chapter 3.3.3, i.e., in Figure 7 A and B, E and F etc. are connected.
Figure 7: The location of test specimen during the RWS test as seen from above.
The temperature was measured inside the test specimens during the fire tests. This was accomplished using internal thermocouples with an welded tip as the measuring point. The thermocouples were placed in the concrete during moulding at a distance of 10, 20, 40, 80 and 120 mm from the surface exposed to the fire. Two
thermocouples were present at each depth.
All tests on large slabs were performed on specimens loaded in compression. The specimens were connected in pairs, see Figure 9. The load was close to 10 % of the cube compressive strength of that concrete with the lowest strength in the pair. Post stressing was performed using eight Dywidag bars with a diameter of 28 mm, inserted into the concrete. The applied load was monitored using four load cells during the fire test. A drawing of the loading system is shown in Figure 9. Figure 10 shows the loading system during fire test.
Figure 10: The loading system on the side without load cells during fire test.
Vapour pressure measurements
During 12 of the total of 40 fire tested small slabs, the internal vapour pressure was measured. Pressure pipes open at depths of 5, 10, 15, 20 mm from the fire exposed side were inserted during moulding (Jansson, 2006). Just before the fire test the pipes were filled with oil. The oil transferred the internal pressure inside the concrete to external pressure gauges connected to a MGC Plus system and a computer for data acquisition.
During casting of the concrete thin steel pipes, with an inner diameter of 2 mm and wall thickness of 0.2 mm, were inserted into the concrete. One end of each steel pipe was placed near the surface to be fire tested. The pipe extended from the
measurement point through the test specimen, exiting on the cold side. The
measurement depths used were: 5, 10, 15 and 20 mm from the fire exposed surface. To ensure that no concrete, i.e. cement paste, would fill the pipes, thin welding bars were inserted into the pipes during casting. The welding bars have been shown in previous projects to be effective in preventing cement paste from filling the pipes, which was also the case in the present study. There was, however, another problem with the welding bars inside the pipes. During the long conditioning in water some of the welding bars, which from the beginning had a very tight fit inside the steel pipes, were stuck by corrosion inside the pipes and it was impossible to remove them. That is the reason for the lack of data from tests on some of the specimens, i.e. in some of the specimens all the pressure pipes could be used and in some of them none of the pipes could be salvaged.
When the fire tests were conducted the welding bars inside the steel pipes were removed and the pipes were filled with high temperature silicone oil, Sil 300, produced by Haake. The filling of the pipes was conducted by inserting a thin needle and syringe and carefully injecting oil from the bottom of the pipe to ensure that no air was trapped. Outside the concrete the steel pipes were connected to a pressure
gauge using a T-junction. The pressure gauges that were used were of the type P8AP/100bar from Hottinger Baldwin Messtechnik GmbH. The pressure gauges and the T-junctions are shown in Figure 11.
Figure 11: The pressure gauges and the T-junctions used in the pressure measurement system.
The spalling profile was measured in 528 evenly spread points on the
1700 × 1200 mm2 fire exposed surface. A frame used for spalling measurement is
shown in Figure 12. Using such a frame the maximum, minimum and mean spalling depth along with a spalling profile can be determine. This method is more accurate as one can choose a smaller area towards the centre of the fire exposed surface to study the spalling. This ensures that boundary effects are minimised. Further, it is difficult to adjust for loss of water from the test specimen using weight loss to quantify spalling due to the inherent uncertainty in the estimation of water loss in the part of the specimen that has not spalled.
Figure 12 Frame used for spalling depth measurements.
Results from fire tests
The test results from fire tests can be seen in the Appendices. The data on furnace temperatures during the two furnace tests on large slabs are shown in Appendix B. Results from measurements on large slab specimens can be found in Appendix C and all data from tests on small slabs can be seen in Appendix D.
Large slabs, 1700 × 1200 × 300 mm3, were tested both relative to standard fire
exposure (EN 1363-1) and RWS exposure. During the standard test, the test time was one hour; but during the RWS exposure, the concrete slabs broke in the longitudinal direction so the test was terminated after 30 minutes for safety reasons. The spalling in all specimens had stopped before the furnace was turned off. Further, during the RWS test the spalling was so severe between 4 and 12 minutes that it was not possible to maintain the furnace temperature curve as when spalling is rapid new cold surfaces are created inside the furnace effecting the temperature control.
The compressive strength of all the concrete included in the fire tests was determined according to EN 12390-3 using cubes of 28 days. Tests were also performed on cubes just prior to the fire tests. All cubes used in the tests were conditioned together with the specimens used for the fire tests. Test results are shown in Table 8. The results for 28 days are an average of three tested cubes and the results from the tests prior the fire tests are an average of two cubes.
Table 8 Measured compressive strength on cubes. Concrete Compressive strength [MPa]
28 days When fire tested A 54 79 B 45 69 C 55 80 D 58 85 E 47 68 F 60 87 G 54 78 H 58 86 I 53 79 J 52 77 K 59 82 L 55 80 M 58 78 N 66 84 O 56 73 P 58 87 Q 72 99 R 62 94 S 61 87 T 54 76 U 39 59
The cubes used for compressive strength tests just prior the fire test were dried at 105 °C directly after the compression test. The cubes had the same conditioning history as the specimens that were to be fire tested, see chapter 2.3. The moisture content measured in the cubes is shown in Table 9. The mean moisture content in cubes is not directly representative of that of larger specimens like the ones used in the fire testing. This can be illustrated by the following simple calculation with fictive values. Assume that the first outer centimetre of the specimens has the
moisture ratio of 3 % and that the inner core of the specimens has a moisture ratio of 5 %. This assumptions means that the boundary zone (3% moisture zone) of a 15 × 15 × 15 cm3 cube is 34.9 % of the volume. In a larger specimen,
170 × 120 × 30 cm3, the boundary zone will only be 9.3 % of the volume. If the (fictive) moisture ratio for the whole specimen is calculated it will be 4.3% based on the test results from the small cube while that for the whole larger slab is actually 4.8 %.
Table 9: Moisture content.
Concrete Moisture content [%] A 5.1 B 5.2 C 5.2 D 5.4 E 6.1 F 5.4 G 6.0 H 5.1 I 5.8 J 6.0 K 5.5 L 5.4 M 5.8 N 4.9 O 4.5 P 4.8 Q 4.5 R 4.6 S 5.1 T 5.3 U 4.3
Degree of capillary saturation
The transport of moisture and the possible occurrence of a moisture clog (see chapter 5.2) may be important for the behaviour of concrete at high temperature. The degree of capillary saturation at room temperature, i.e., how much of the capillaries that are initially filled with water, is a factor that influences this transport.
Method and preparation
Concrete filled plastic pipes were used to represent the two sided drying of large slabs. The plastic tubes were 300 mm long, had a diameter of 45 mm and were open at both ends. They were stored under the same conditions as the large specimens that
they represented. Close to the day of the fire test, the cylinders inside the plastic tubes were sliced to determine the capillary saturation profiles. The pipes with concrete were sliced in 2 cm thick slices. The size 2 cm was chosen because it was the smallest size possible without meeting technical difficulties with the slicing method chosen. To minimize the heating of the specimen during slicing the specimen was first notched a couple of mm with a diamond disc, see Figure 13, and then cut using an axe, see Figure 14.
The sliced test specimens were then used to determine of the degree of capillary saturation according to the method described by Hedenblad and Nilsson (1985). The specimen is first weighted, then the under side of the specimen is placed in contact with a water surface and evaporation protection is attached over the sample before the capillary suction starts. The specimen is kept in contact with the water surface until it is capillary saturated, i.e. when no change in weight over time can be seen. Then the weight is recorded and the specimen is dried in 105 °C. The degree of capillary saturation is determined using the following formula:
0 0 m m m m S cap w cap − − =
Scap = the degree of capillary saturation
mw = the mass of the sample in the moist state (initial state)
mcap= the mass of the sample after capillary suction
m0 = the mass of the sample in the dry state
The results of the degree of capillary saturation measurements, shown in Figure 15, indicate that the specimens containing PP-fibres show a higher degree of capillary saturation close to the surface to be fire tested (0-20 mm). It can also be seen that close to the surface to be fire tested, the sequence from lowest to highest degree of capillary saturation is the same for those concrete types with maximum aggregate size 16 mm as for those with maximum size 25 mm, see Figure 16 and Figure 17. In both cases the order from lowest to highest is: concrete without PP-fibres, 1 kg/m3 32
Degree of capllary saturation 0,8 0,82 0,84 0,86 0,88 0,9 0,92 0,94 0,96 0,98 1 1 2 3 4 5 6 7 8
20 mm slices from surface inwards
D e gr e e of c a pi la ry s a tur a ti on [ -] A 16A noPP B 25A noPP M 16A 1kg32PP N 25A 1kg32PP E 16A 1kg18PP F 25A 1kg18PP I 16A 1.5kg18PP J 25A 1.5kg18PP
Figure 15: The degree of capillary saturation for different concretes (16A and 25 A is the maximum aggregate sizes)
Degree of caplliary saturation close to the surface (0-20mm)
Max aggregate siz e 16mm
0,7 0,75 0,8 0,85 0,9 0,95 1 A, no PP M, 1kg 32PP E, 1kg 18PP I, 1.5kg 18PP [-]
Figure 16; The degree of capillary saturation at the depth 0-20 mm for concretes with maximum aggregate size 16 mm.
Degree of caplliary saturationclose to the surface (0-20mm)
Max aggregate size 25mm
0,7 0,75 0,8 0,85 0,9 0,95 1 B, no PP N, 1kg 32PP F, 1kg 18PP J, 1.5kg 18PP [-]
Figure 17: The degree of capillary saturation at the depth 0-20 mm for concretes with maximum aggregate size 25 mm.
Tests of the free thermal expansion of concrete types A, E and I were performed at Centre Scientifique et Technique du Bâtiment (CSTB) in France. The thermal expansion up to 600 °C and stress strain measurements at 20 °C and 600 °C, see chapter 4.5, were performed on the same type of specimens and the same experimental setup.
The equipment used is built to fulfil the requirements outlined in the RILEM TC 129 MHT “Test methods for mechanical properties of concrete at high temperatures- Part 6: Thermal strain”, (1997). The thermal expansion was measured on cylinders with a diameter 104 mm and length 300 mm. The cylinders were placed in a cylindrical furnace and a strain measurement system was attached. Both the longitudinal and radial expansion were measured each with 3 strain gauges. The heating rate was 1 degree per minute up to the maximum temperature 600 °C. The measurement system, shown in Figure 18, is described in more detail by Gaweska Hager (2004).
Figure 18: The equipment for measurements of mechanical properties at high temperature at CSTB in France.
Test results from the thermal expansion measurements on concrete types A, E and I show, as seen in Figure 19 to Figure 22, quite similar result for the different
concretes. This is expected because the only difference between the concretes is the amount of PP-fibres added and the major driving force for expansion is the aggregate which is the same in all mixes. There is one interesting difference although it is small. At around 200-250 °C the thermal expansion decreases to zero or close to zero for concretes E and I. After this decrease there is an increased expansion rate
between 250-300 °C compared to concrete A, so at 300°C the total strain is almost the same as before the strain plateau. Concrete A is free from PP-fibres while concrete E contains 1 kg/m3 18 μm PP-fibres and I includes 1.5 kg/m3 18 μm
PP-fibres. Further, the decrease is highest for the highest content of PP-PP-fibres. The behaviour can be seen both in the longitudinal and radial direction of the tested cylinders.
Free thermal expansion, longitudinal direction 0 2000 4000 6000 8000 10000 12000 0 100 200 300 400 500 600 T emperature [oC] St ra in [ µ m/ m] A E I
Figure 19: The free thermal expansion in the longitudinal direction.
Free thermal expansion, radial direction
0 2000 4000 6000 8000 10000 12000 14000 0 100 200 300 400 500 600 T emperature [oC] St ra in [ µ m/ m] A E I
Free thermal expansion, longitudinal direction 1000 1500 2000 2500 3000 3500 150 200 250 300 350 T emperature [oC] St ra in [ µ m/ m] A E I
Figure 21: The free thermal expansion in the longitudinal direction, the strain plateau.
Free thermal expansion, radial direction
1000 1500 2000 2500 3000 3500 150 200 250 300 350 T emperature [oC] St ra in [ µ m/ m] A E I
After the thermal expansion tests where the samples were heated from 20 °C to 600 °C, at an increase of 1 degree per minute as described in section 4.4.2, the temperature was maintained for one hour to reach thermal equilibrium in the specimen before determination of the stress strain curve at high temperature. Additional tests to determine the stress strain curves at room temperature (20 °C ) were also performed.
Figure 23 and Figure 24 show measurements from the strain gauges relative to the load applied. The load was applied in the longitudinal direction. The results illustrate, as expected, that the material is much stiffer at 20 °C than at 600 °C. The
compressive strength determined directly from the loading machine and the weight loss during thermal cycling, are shown in Table 10.
Stress/strain at 20 oC, load in longitudinal direction
-60 -50 -40 -30 -20 -10 0 -3000 -2000 -1000 0 1000 2000 Strain [µm/m] St ress [ MPa] A longitudinal A radial E longitudinal E radial I longitudinal I radial
Figure 23: Stress in longitudinal direction and strain in longitudinal and radial direction at 20 °C.
Stress/strain at 600 oC, load in longitudinal direction -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 -10000 -5000 0 5000 10000 15000 Strain [µm/m] St ress [ MPa] A radial A longitudinal E longitudinal E radial I longitudinal I radial
Figure 24: Stress in longitudinal direction and strain in longitudinal and radial direction at 600 °C.
Table 10 The compressive strength from the test machine and the weight loss during thermal cycle. Compressive strength 20 °C [MPa] Compressive strength 600 °C [MPa] Weight loss d uring heating to 6 00 ° C [%] A 60 20 7.76 E 56 17 7.88 I 55 18 7.64
Drying tests at high temperature
To investigate the drying behaviour of concrete when exposed to heat, a series of tests were performed on cylinders. The main goal was to explore possible differences between concrete with and without addition of PP-fibres. The most widespread theory concerning the action of PP-fibres is that they promote fast drying by melting and providing channels for the internal moisture to escape (Khoury, 2000). This
theory is, however, not the only explanation that has been promulgated. According to
Schneider and Horvath (2003) the following theories presently dominate this field:
Improvement of the permeability due to formation of capillary pores when the fibres melt and burn.
Improvement of the permeability caused by the development of diffusion open transition zones near the fibres.
Improvement of the permeability due to additional micro pores, which develop during the addition and mixing of fibres in the concrete mix. Improvement of the permeability due to additional micro cracks at the tip of
the PP-fibres which develops during heating up and melting
One should note, however, that all the above theories indicate that in some way the drying process, i.e. moisture transfer, will be facilitated by the presence of PP-fibres.
Cylinders with diameters 34 and 60 mm and length 300 mm were exposed to two different heating ramps, 2 and 5 degrees per minute. The heating was performed in a cylindrical furnace with a height 400 mm and diameter 200 mm. During the heating, the specimens were hanging on a wire connected to a weight measurement device, see Figure 25. The temperature, regulated by the system and shown in the diagrams, was measured a couple of mm from the vertical centre of the specimen. The readings concerning the weight of the specimen were recorded directly by a computer.
Results2 °C/min 34 mm core
Concrete A, B and U (reference) are the ones without addition of PP-fibres. When analysing the diagram in Figure 26 it can be seen that U, the reference concrete, exhibits the fastest total weight loss rate up to approximately 400 °C. The weight loss behaviour of concrete A is somewhere in the middle of that shown by all tested concretes while concrete B has the lowest total weight loss after approximately 230 °C. The derivative of the weight ratio curve is shown in Figure 27. With this type of presentation it is easier to discern a systematic difference between the concrete with and without PP-fibres. Concrete A and B have the lowest maximum weight loss rate and the peak in weight loss rate comes at a higher temperature for concrete with PP-fibres. The reference concrete, U, shows a different type of behaviour, i.e., it exhibits an early peak with a high weight loss rate.
Drying of cores, d=34 mm l=300 mm heat rate=2°C/min 0,91 0,92 0,93 0,94 0,95 0,96 0,97 0,98 0,99 1 0 100 200 300 400 500 600 Temperature [°C] W e ig h t r a ti o [-] A B E F I J M N U
Weight loss rate of cores, d=34 mm l=300 mm heat rate=2°C/min 0 0,0001 0,0002 0,0003 0,0004 0,0005 0,0006 0,0007 0 100 200 300 400 500 600 Temperature [°C] We ight l o s s r a te [ 1 /s ] A B E F I J M N U
Figure 27: The weight loss rate of 34 mm cores at heating rate 2 °C/minute.
5 °C/min 34 mm core
Concrete J, containing 1.5 kg/m3 18 μm PP-fibres shows a fast drying behaviour, see
Figure 28 and Figure 29. It is as fast as the reference concrete U, with a slightly higher peak in the maximum drying rate. As expected the peak in weight loss rate comes later when the heating rate is 5 °C/min compared to 2 °C/min. The peak in the weight loss, see Figure 29, is seen at approximately the same temperature for all concretes but the concretes without PP-fibres exhibits lower peaks with the exception of reference concrete, U, which also exhibits a high peak.
Drying of cores, d=34 mm l=300 mm heat rate=5°C/min 0,91 0,92 0,93 0,94 0,95 0,96 0,97 0,98 0,99 1 0 100 200 300 400 500 600 Temperature [°C] We ight r a ti o [ -] A1 A2 B E F I J M N U
Figure 28: The weight loss of 34 mm cores at heating rate 5 °C/minute.
Weight loss rate of cores, d=34 mm l=300 mm heat rate=5°C/min 0 0,0001 0,0002 0,0003 0,0004 0,0005 0,0006 0,0007 0 100 200 300 400 500 600 Temperature [°C] W e ig h t lo s s r a te [ 1 /s ] A1 A2 B E F I J M N U
Figure 29: The weight loss rate of 34 mm cores at heating rate 5 oC/minute. 1 °C/min 60 mm core
As shown in Figure 30 and Figure 31, up to approximately 180 °C, concrete A (no PP-fibres) and E (1 kg/m3 18 μm PP-fibres) show almost identical behaviour
although concrete E exhibits a larger total weight loss compared to concrete A. Up to approximately 250°C, concrete I (1 kg/m3 18 μm PP-fibres) is the type of concrete
that has the lowest total weight loss.
Drying of cores, d=60 mm l=300 mm heat rate=1°C/min 0,92 0,93 0,94 0,95 0,96 0,97 0,98 0,99 1 0 100 200 300 400 Temperature [°C] We ig ht r a ti o [ -] A E I
Weight loss rate of cores, d=60 mm l=300 mm heat rate=1°C/min 0 0,0001 0,0002 0,0003 0,0004 0,0005 0,0006 0,0007 0,0008 0 100 200 300 400 Temperature [°C] W e ig h t lo s s r a te [ 1 /s ] A E I
Figure 31: The weight loss rate of 60 mm cores at heating rate 1 °C/minute. 2 °C/min 60 mm core
The drying behaviour for 2 °C/min of 60 mm cores, shown in Figure 32 and Figure 33, is in general close to the behaviour shown for the same heating rate in 34 mm cores, i.e. the weight loss peak comes later and is higher for concrete containing PP-fibres.
Drying of cores, d=60 mm l=300 mm heat rate= 2°C/min
0,91 0,92 0,93 0,94 0,95 0,96 0,97 0,98 0,99 1 0 100 200 300 400 500 600 Temperature [°C] We ig ht r a ti o [ -] A B E F I J M N U
Weight loss rate of cores, d=34 mm l=300 mm heat rate=2°C/min 0 0,0001 0,0002 0,0003 0,0004 0,0005 0,0006 0,0007 0 100 200 300 400 500 600 Temperature [°C] We ight l o s s r a te [ 1 /s ] A B E F I J M N U
Figure 33: The weight loss rate of 60 mm cores at heating rate 2 °C/minute. 5 °C/min 60 mm core
The concretes without PP fibres, A and B, exhibit the highest total weight loss up to approximately 320 °C, see Figure 34. Then the concretes with PP-fibres exhibit a faster drying rate, in the case of concrete I only a slightly faster drying rate, see Figure 35. Drying of cores, d=60 mm l=300 mm heat rate=5°C/min 0,91 0,92 0,93 0,94 0,95 0,96 0,97 0,98 0,99 1 0 100 200 300 400 500 600 Temperature [°C] We ight r a ti o [ -] A B E F I J M N U
Weight loss rate of cores, d=60 mm l=300 mm heat rate=5°C/min 0 0,00005 0,0001 0,00015 0,0002 0,00025 0,0003 0,00035 0,0004 0,00045 0,0005 0 100 200 300 400 500 600 Temperature [°C] W e ig h t lo s s r a te [ 1 /s ] A B E F I J M N U
Figure 35: The weight loss rate of 60 mm cores at heating rate 5 °C/minute.
Speed of sound
The speed of sound is related to the dynamic elastic modulus of the material. For a stiff material the speed of sound is high. The elastic modulus for concrete decreases almost linearly from room temperature to between 800 °C to 1000 °C when it is almost zero. The purpose of the speed of sound measurements conducted in this study, was to see if the internal damage is dependent on the heating rate.
The cylindrical test specimens used in the drying tests (see chapter 4.6) were also tested in a speed of sound apparatus, see Figure 36 and Figure 37. When conducting these experiments, a contact gel is put on the sensors and the apparatus is calibrated against a reference material with a known speed of sound.
The speed of sound was tested before heating the specimen during the drying test and directly after cooling down to room temperature. Three heating rates were used: 1, 2 and 5 °C per minute.
Figure 36: The contact gel, reference material and speed of sound apparatus.
Figure 37: The speed of sound sensors.
Some clear trends can be discerned when comparing concrete with and without addition of PP-fibres with respect to the speed of sound, or more precisely the change in speed of sound as a result of heating. Although the heating rates used in this experiments are low, 1, 2 and 5 °C per minute, compared with the rapid heat exposure that leads to fire spalling during a fully developed fire, it is interesting to see how different heating rates influences the change in speed of sound. The main trend is that during the most rapid heating used in these tests, 5 °C per minute,
specimens without PP-fibres exhibit the greatest damage. But when the heating rate is lower, 1 and 2 °C per minute, this is not as clearly evident.
The speed of sound before and after 5 °C per minute heating of cores, with diameter 34 mm, are shown in Figure 38. It can be seen that the concrete types without PP-fibres, concrete A, B and the reference concrete U, are damaged most by the heat treatment. This is more clearly visible when looking at the ratio after heating divided by the virgin values, see Figure 39.
Speed of sound, cores d = 34 mm Heating: 5 deg/min to 600 oC 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 A1 A2 B1 E1 F1 I1 I3 J1 M1 M2 N1 U1 S pe e d of s ound [ m /s ] Virgin After heating
Figure 38: Speed of sound , cores d =34 mm, heating: 5 °/min up to 600 oC
Ratio (after/virgin), cores d = 34 mm Heating: 5 deg/min to 600 oC 0,2 0,22 0,24 0,26 0,28 0,3 0,32 A1 A2 B1 E1 F1 I1 I3 J1 M1 M2 N1 U1 [-] Ratio
Figure 39: Speed of sound ratio (after/virgin), cores d =34 mm, heating: 5 °/min up to 600 oC
In Figure 40 and Figure 41, the values for cores with diameter 34 mm and the heating rate of 2 °C per minute are shown. It can be seen that concrete A is among the concrete types exhibiting the least damage in contrast to concrete B which exhibits most damaged, with the exception of the reference concrete U.
Speed of sound, cores d = 34 mm Heating: 2 deg/min to 600 oC 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 A3 B3 E2 F3 I2 J2 M3 N2 U 2 S pe e d of s oun d [m /s ] Virgin After heating
Figure 40: Speed of sound , cores d =34, heating: 2 °/min up to 600 oC
Ratio (after/virgin), cores d = 34 mm Heating: 2 deg/min to 600 oC 0,24 0,245 0,25 0,255 0,26 0,265 0,27 0,275 0,28 0,285 0,29 0,295 A3 B3 E2 F3 I2 J2 M3 N2 U 2 [-] Ratio
Figure 41: Speed of sound ratio (after/virgin), cores d =34 mm, heating: 2 °/min up to 600 oC
In Figure 42 and Figure 43 the values for cores with diameter 60 mm and heating rate 5 °C per minute are shown. It can be seen that concrete A is the concrete type
Speed of sound, cores d = 60 mm Heating: 5 deg/min to 600 oC 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 A1 B1 E1 E3 F1 I1 J1 J3 M1 N1 U1 S p e e d of s ound [ m /s ] Virgin After heating
Figure 42: Speed of sound , cores d =60 mm, heating: 5 °/min up to 600 oC
Ratio (after/virgin), cores d = 60 mm Heating: 5 deg/min to 600 oC 0,2 0,21 0,22 0,23 0,24 0,25 0,26 0,27 0,28 A1 B1 E1 E3 F1 I1 J1 J3 M1 N1 U1 [-] Ratio
Figure 43: Speed of sound ratio (after/virgin), cores d =60 mm, heating: 5 °/min up to 600 oC
In Figure 44 and Figure 45 no clear trends can be seen, but concrete F that was heated to 520 °C instead of 600 °C shows clearly that the temperature reached is the dominant factor for damage at these heating rates.
Speed of sound, cores d = 60 mm Heating: 2 deg/min to 600 oC 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 A2 B2 E2 F2 only to 520 deg. I2 J2 M2 N2 U2 S pe e d of s ound [ m /s ] Virgin After heating
Figure 44: Speed of sound , cores d =60 mm, heating: 2 °/min up to 600 oC Ratio (after/virgin), cores d = 60 mm
Heating: 2 deg/min to 600 oC 0,2 0,22 0,24 0,26 0,28 0,3 0,32 0,34 A2 B2 E2 F2 only to 520 deg. I2 J2 M2 N2 U2 [-] Ratio
Figure 45: Speed of sound ratio (after/virgin), cores d =60 mm, heating: 2 °/min up to 600 oC
At the heating rate 1 °C per minute to 400 °C of diameter 60 mm cores there is no clear difference between the damage in the three tested cores, Figure 46 and Figure 47.
Speed of sound, cores d = 60 mm Heating: 1 deg/min to 400 oC 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 A3 E4 I3 S p e e d of s ound [ m /s ] Virgin After heating
Figure 46: Speed of sound , cores d =60 mm, heating: 1 °/min up to 400 °C
Ratio (after/virgin), cores d = 60 mm Heating: 1 deg/min to 400 oC 0,2 0,25 0,3 0,35 0,4 0,45 0,5 0,55 A3 E4 I3 [-] Ratio
Figure 47: Speed of sound ratio (after/virgin), cores d =60 mm, heating: 1 °/min up to 400 °C
According to some theories dealing with the cause of fire spalling of concrete, the moisture transport is a key parameter (see also chapter 5.2). Restricted moisture transport can lead to failure due to rising internal pressure. If, however, the moisture