assessment of spalling risk in concrete subjected to fire

TADEUSZ KOŚCIUSZKO CRACOW UNIVERSITY OF TECHNOLOGY FACULTY OF CIVIL ENGINEERING INSTITUTE OF BUILDING MATERIALS AND STRUCTURES CHAIR OF BUILDING MATERIALS TECHNOLOGY AND STRUCTURE PROTECTION

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE - DISSERTATION OUTLINE PHD CANDIDATE:

KATARZYNA MRÓZ, MSC SUPERVISOR:

IZABELA HAGER, PHD, DSC

CRACOW, POLAND 2016

POLITECHNIKA KRAKOWSKA IM. TADEUSZA KOŚCIUSZKI WYDZIAŁ INŻYNIERII LĄDOWEJ INSTYTUT MATERIAŁÓW I KONSTRUKCJI BUDOWLANYCH KATEDRA TECHNOLOGII MATERIAŁÓW BUDOWLANYCH I OCHRONY BUDOWLI

OCENA RYZYKA EKSPLOZYJNEGO ODPRYSKIWANIA BETONU W POŻARZE - KONSPEKT PRACY DOKTORSKIEJ KANDYDAT DO STOPNIA DOKTORA:

MGR INŻ. KATARZYNA MRÓZ OPIEKUN

NAUKOWY:

DR HAB INŻ. IZABELA HAGER

KRAKÓW, POLSKA 2016

LIST OF PUBLICATIONS OF PHD CANDIDATE

This dissertation outline is based on and related to work presented in the following papers: Hager, I., Carre, H., Krzemień, K., 2013. Damage assessment of concrete subjected to high temperature by means of ultrasonic pulse velocity method. Studies and Research (Studi e Ricerche), Scuola di Specializzazione per le Costruzioni in C.A, Flli Pesenti, Politecnico di Milano. Hager, I., Krzemień, K., 2013. Metoda impact-echo - ocena przydatności w diagnozowaniu działania wysokiej temperatury na beton. Przegląd Budowlany 12, pp. 57-63. Hager, I., Krzemień, K., 20th-23rd April 2015. An overview of concrete modulus of elasticity evolution with temperature and comments to European code provisions. IFireSS – International Fire Safety Symposium Coimbra, Portugal, pp. 703-712. Hager, I., Tracz, T., Krzemień, K., 2014. Usefulness of selected non-destructive and destructive methods in the assessment of concrete after fire. Cement Wapno Beton 3, pp. 145-151. Hager, I., Tracz, T., Śliwiński, J., Krzemień, K., 2015. The influence of aggregate type on the physical and mechanical properties of high-performance concrete subjected to high temperature. Fire and Materials, Issue DOI: 10.1002/fam.2318. Hager, I., Zdeb, T., Krzemień, K., 2013. The impact of the amount of polypropylene fibres on spalling behaviour and residual mechanical properties of Reactive Powder Concrete. MATEC Web of Conferences, 6 (DOI: 10.1051/matecconf/20130602003), p. 02003. Krzemień, K., Hager, I., 2015. Assessment of concrete susceptibility to fire spalling: A report on the stateof-the-art in testing procedures. Procedia Engineering 108, pp. 285-292. Krzemień, K., Hager, I., 2015. Post-fire assessment of mechanical properties of concrete with the use of the impact-echo method. Construction and Building Materials 96, pp. 155-163. Krzemień, K., Pimienta, P., Pinoteau, N., Hager, I., 8-9 October 2015. Moisture effect on mechanical behaviour of concrete at high temperature and its implication on fire spalling. 4 th International Workshop on Concrete Spalling due to Fire Exposure Leipzig, Germany, pp. 165-176.

TABLE OF CONTENTS LIST OF PUBLICATIONS OF PHD CANDIDATE ............................................................................................... 5 TABLE OF CONTENTS .................................................................................................................................. 7 1

INTRODUCTION ................................................................................................................................... 9

2

LITERATURE STUDY ............................................................................................................................10 2.1 CONCRETE SPALLING IN THEORY ............................................................................................................... 10 2.1.1 General description and types of spalling ............................................................................... 10 2.1.2 Concrete spalling theories ....................................................................................................... 11 2.1.3 Parameters influencing spalling .............................................................................................. 13 2.2 PREVENTING OF FIRE SPALLING IN CONCRETE STRUCTURES ............................................................................ 16 2.2.1 PP fibres .................................................................................................................................. 16 2.2.2 Thermal insulation barriers ..................................................................................................... 16 2.2.3 Endothermic building materials including concrete and gypsum ........................................... 17 2.3 EXPERIMENTAL METHODS FOR ASSESSMENT OF CONCRETE SUSCEPTIBILITY TO FIRE SPALLING ............................... 18 2.3.1 Introduction ............................................................................................................................ 18 2.3.2 Small-scale concrete spalling tests ......................................................................................... 18 2.3.3 Medium-scale furnaces ........................................................................................................... 21 2.3.4 Full scale fire tests ................................................................................................................... 25 2.4 INFLUENCE OF STRESS STATE OF CONCRETE SPALLING PROPENSITY................................................................... 25 2.4.1 Specimens loaded in compression .......................................................................................... 26 2.4.2 Specimen restrained with cold rim .......................................................................................... 27 2.4.3 Specimen restrained with external restraint ........................................................................... 28 2.5 SUMMARY........................................................................................................................................... 29

3

OBJECTIVES, HYPOTHESES AND SCOPE OF THESIS ..............................................................................30 3.1 3.2 3.3

4

JUSTIFICATION FOR TACKLING SPECIFIC SCIENTIFIC PROBLEMS......................................................................... 30 THESIS’ HYPOTHESES ............................................................................................................................. 30 SCOPE OF WORK ................................................................................................................................... 30

RESEARCH PROGRAM AND METHODOLOGY ......................................................................................32 4.1 NUMERICAL MODELLING ........................................................................................................................ 33 4.2 EXPERIMENTAL TESTS ............................................................................................................................ 34 4.2.1 Materials and specimens ........................................................................................................ 34 4.2.2 Heating and loading conditions and testing setup ................................................................. 35 4.3 SUMMARY OF RESEARCH PROGRAM.......................................................................................................... 36

5

PRELIMINARY RESEARCH ....................................................................................................................37 5.1 EXPERIMENTAL TESTS ............................................................................................................................ 37 5.2 PARAMETRICAL STUDIES ......................................................................................................................... 40 5.2.1 Input data and preliminary results .......................................................................................... 41 5.3 CONCLUSIONS ON PRELIMINARY STUDIES ................................................................................................... 45

6

REFERENCES ........................................................................................................................................46

7

PLAN ROZPRAWY DOKTORSKIEJ .........................................................................................................50 7.1 TEMAT I CEL PRACY ............................................................................................................................... 50 7.2 PRZEGLĄD LITERATURY ........................................................................................................................... 50 7.3 OGÓLNY PLAN BADAŃ............................................................................................................................ 50 7.3.1 Analiza numeryczna ................................................................................................................ 51 7.3.2 Badania doświadczalne .......................................................................................................... 51

| Chapter 1: Introduction

1 INTRODUCTION Thermal instability of concrete concerns its behaviour in fire conditions when subjected to high temperature material is presenting explosive behaviour. Another term for this phenomenon employed frequently in the literature is fire spalling. In general, fire spalling is defined as violent or non-violent breaking off of layer pieces of concrete from the surface of a structural element when it is exposed to high and rapidly rising temperatures as experienced in fires (Khoury & Anderberg, 2000). The sensitivity of concrete to spalling phenomenon during fire exposure is one of today’s issues of scientific interests. Real fires (ex. Channel Tunnel - 1996, Mont Blanc Tunnel – 1999), Figure 1.1, have indicated that spalling of concrete can have serious consequences and is a phenomenon that should be taken into account while designing of the susceptible to fire condition structures. The common result of exposition of concrete to fire is a detachment of concrete cover and thus uncovering of reinforcement in RC structures.

Figure 1.1: Real accidents of fire and observed fire spalling cases: left: Mont Blanc Tunnel fire (1999), right: Channel Tunnel fire (1996) (tunneltalk.com, 2015)

Recent achievements in concrete mix design have led to new types of concrete which, besides an increased performance during loading and exposure, also have shown enhanced sensitivity towards spalling. However, the sensitivity towards spalling of a concrete is until now not fully understood and more research is needed to predict risk of spalling. One of the issues that seems to be of particular importance and this issue which divides scientists is a question if spalling is a material property or spalling propensity is governed by structural effect? There is a wide range of factors that may induce spalling among which we can distinguish both mechanical and physical ones. Since concrete is a composite material, in which dispersed phase is represented by aggregate and continuous one by cement paste, both aggregate and cement paste may have an influence on thermal instability of concrete. It is already known that aggregate and moisture play an important role in behaviour of concrete during fire. Moreover, the concrete structures work in different conditions (curing, load, size, environment, etc.), and each of these factors may have a great impact on the thermal stability of concrete. This phenomenon is attributed, among others, to the combined action of high temperature and applied load. Recent research programs carried out by number of scientists indicate the influence of different stress states provided in tested specimen on observed concrete spalling propensity. The stress state varies with chosen test configurations. We can distinguish the following test configurations: unloaded slabs, uniaxially or bi-axially slabs loaded in compression, specimens with external restraint or specimens restrained with, so called, cold rim. In the following paragraphs, the state of recent research program on the mentioned test configurations is presented. Katarzyna Mróz | Dissertation Outline | 9

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | Current state of knowledge about the research investigations devoted to the phenomenon of fire spalling in concrete permits to conclude that so far there was no comprehensive scientific program on the influence of stresses induced in different manners on concrete susceptibility to fire spalling. Despite the fact that there is a wide range of results from various approaches, there is no connection between them and it is impossible to put an evident conclusion about the stated problem. Additionally, the survey of numerical modelling of each type of load applications with the use of the same finite element method and unified parameters, enriched with the experimental tests performed at one testing setup, are totally unique and valuable.

2 LITERATURE STUDY 2.1

CONCRETE SPALLING IN THEORY

2.1.1

GENERAL DESCRIPTION AND TYPES OF SPALLING

Spalling, in general, is defined as the violent or non-violent breaking off of layer pieces of concrete from the surface of a structural element when it is exposed to high and rapidly rising temperatures as experienced in fires (Khoury & Anderberg, 2000). In literature few types of spalling is distinguished: aggregate spalling, explosive surface spalling, explosive corner spalling, destructive spalling, local spalling, sloughing off, gradual spalling of a cross section and explosive spalling. The most important of the mentioned types is explosive spalling, however, spalling may become also as the combination of several or, in certain case, all of listed types. In order to illustrate a multitude of spalling phenomena, in further part of this paragraph, previously given types of concrete spalling defined by researchers with their short description are presented. Aggregate spalling This type of spalling is a failure of aggregate near the concrete surface. It is characterized by a popping sound. It is caused by thermal expansion of the aggregate and splitting of pieces of aggregate’s grains close to the surface because of physical or chemical changes, which occur at high temperatures in aggregates. The main cause for thermal expansion is the β – α conversion of quartz at 573°C. It has little impact on structural performance as the majority of the cover remains intact and insulates the reinforcement. In addition, as an aggregate spalling leads only to a superficial damage of concrete, the insulation function of the structural members in fire is little affected (Debicki, et al., 2012). It has been observed that an aggregate spalling does not occur in basalt aggregate concrete (Meyer-Ottens, 1972). Explosive surface spalling This phenomenon occurs when small pieces, up to about 20 mm in size, fly off the surface of a concrete element during the early part of its exposure to a fire event or a fire test. Surface spalling may result in uncovering the reinforcement, in case of exposition during long time (Debicki, et al., 2012). Another definition of this type of spalling describes detachment of surfaces of 100 cm² up to several square meters, especially in walls and columns loaded in compression, whereby the reinforcement was partially uncovered. Spalling craters generally have a depth of 25 to 50 mm (Connolly, 1995). Explosive corner spalling This type of spalling occurs during later stages of a fire exposure (after 30 minutes) (Connolly, 1995) when the concrete has become weak and cracks develop as a result of tensile stress along 10

| Chapter 2: Literature study edges and corners where the reinforcement is typically located. Pieces of concrete fall off from the element, and may be followed by pieces coming away from the faces as cracks develop further. Because of the advanced stage at which such spalling occurs, the strength of the element may have already been reduced significantly, and therefore this type of spalling may be of limited significance to structural stability because it has already been lost (Khoury & Anderberg, 2000). Destructive spalling This is a violent form of spalling occurring at an early stage of heating and may result in spalling of a few large pieces of concrete from the surface of the member (Schneider, 1982) and thus in extensive damage, and finally in complete destruction of the concrete element (Bazant & Kaplan, 1996). The detailed caused for such behaviour are not well described. Sloughing off This is a progressive form of breakdown which involves partial separation of material from the concrete element and may continue slowly through the later stage of heating (Bazant & Kaplan, 1996). Explosive fire spalling Explosive behaviour of concrete is the most dangerous type of spalling. It occurs in the first 20-30 minutes of a fire when the temperature in concrete reaches values of 150-250 ˚C. It is characterized by large or small pieces of concrete being violently spalled from the surface, accompanied by a loud noise. The pieces may be as small as 100 mm or as large as 300 mm in length and 15 - 20 mm deep. The phenomenon can occur just once or remain progressive at intervals even from the previously spalled parts (Khoury & Anderberg, 2000). Explosive spalling is immediately recognizable as it is accompanied by a large release of energy and produces a typical explosive noise. It may results in a sudden and complete failure of the concrete member which, as a result, does not sustain longer its load-bearing function (Connolly, 1995). 2.1.2

CONCRETE SPALLING THEORIES

There are two main mechanisms that cause concrete spalling. The first one relates to the increase of water vapour pressure in the pores located in surface layers of concrete. It is believed that the second mechanism of the explosive behaviour of concrete is the creation of tensile stresses caused by thermal deformation of concrete (Khoury, 2008). Pore pressure While the temperature increases in cross-section, the water contained in concrete goes into the vapour state. As a result of the temperature difference, and thus the different pressure in successive layers, the vapour is transported towards cooler section of the interior. Then, in cooler parts, water returns again to the liquid state and creates the saturated membrane, which hinders the further transport of water. In this area there is a sudden increase of water vapour pressure, which contributes to the formation of tensile stresses in concrete section. When the water pressure exceeds concrete tensile strength the explosive spalling of concrete occurs. The (Jansson & Boström, 2009) presented an investigation of development of a saturated moisture layer (a so called “moisture clog”) and its role in the fire spalling in concrete. Tested specimens, with 70 mm notch were used for splitting the specimen to view the saturated moisture layer. After heating with the standard fire curve, ISO 834-1, for 10, 15 or 20 minutes, the specimen was removed from the furnace and split along the notch. The development of a saturated moisture layer was clearly seen after 15 and 20 minutes, while any changes were seen after 10 minutes of fire exposure. Figure 2.1 Katarzyna Mróz | Dissertation Outline | 11

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | presents the splits surfaces from tests on concrete without PP fibres after 10, 15 and 20 minutes of fire exposure. Fire exposure was on the top of the specimens.

Figure 2.1: Split surfaces of concrete after fire exposure of a) 10 minutes, b) 15 minutes, c) 20 minutes (Jansson & Boström, 2009)

The presence of movement of moisture affects its properties and the physical behaviour of the materials. The presence of moisture clog leads to the increasing of Young’s modulus (Krzemień, et al., 2015), while a compressive strength of concrete presents a significant decreasing between temperature of 60 ̊C and 300 ̊C in concrete that contains initially a free water (Hager, 2004). Thermal stresses The second mechanism of the explosive behaviour of concrete is the creation of tensile stresses caused by thermal deformation of concrete. The layers located deeper are cooler and do not experience such large deformations, which means that between these layers there exist the temperature gradients and thus tensile stresses enlarge. If they exceed the tensile strength of concrete, spalling occurs. It should be noted that the difference in strains is observed between the aggregate and cement paste, as well as between concrete and reinforcing steel. The presence of a significant thermal gradient has been calculated by model of heat transfer in concrete element. The results indicate that after 10 minutes of fire according ISO 834-1 a surface experience a temperature of ca. 500 °C, while the concrete layer place at 2 cm from surface has ca. 115 °C. This analysis clearly confirm that the difference of 300 °C within 2 cm of concrete is of a particular importance and might affect a significant difference in thermal deformation of concrete. Combined theory Both of mentioned mechanisms are shown in Figure 2.2. It can be seen that the regions of creation of peak caused by temperature gradient and peak caused by increasing of pore pressure are very close to each other. Therefore, it is assumed that explosive spalling occurs as a result of a combination of those two mechanisms. Figure 2.2: Mechanisms of spalling (Khoury, 2008) Recent studies confirmed developing of both pore pressure and thermal gradient in concrete crosssection. (Kalifa, et al., 2000) developed a system for measuring of pore pressure in concrete crosssection. An original experimental set-up was used to perform pore pressure measurements in HPC and OC specimens subjected to high temperatures, in conjunction with temperature and mass loss. His analysis indicates that the pore pressure peak corresponds to the drying-dehydration front passing 12

| Chapter 2: Literature study through, and let suppose that this front is preceded by a quasi-saturated layer that acts as a moisture clog. The temperature gradients between successive layers are also clearly marked Figure 2.3. There is significant gradient of temperature what results in additional thermal stresses. Figure 2.4 presents the distribution of vapour pressure on the various depths of specimen. The pore pressure peak moves with time and depth and what is also important, rises with deeper layer due to cumulating of more and more volume of water. In fact, what is in accordance with expectations, temperature corresponding to maximum pressure peak in each layer is similar, (ca. 230 °C). It can be therefore assumed that spalling is caused by combining both thermal stresses between layers and pore pressure peak caused by migration of water to the cooler parts of concrete.

Figure 2.3: M100 (HPC, fc = 100 MPa), heated at 600°C. Pressure fields vs. time (Kalifa, et al., 2000)

Figure 2.4: M100 (HPC, fc = 100 MPa), heated at 600°C. Temperature fields vs. time (Kalifa, et al., 2000)

2.1.3

PARAMETERS INFLUENCING SPALLING

Since, spalling is a combined phenomenon, the parameters affecting a susceptibility of concrete towards spalling is a multiparameter issue gathered in Table 2.1. Table 2.1: Parameters influencing spalling Material effect

Geometrical and boundary condition effect

Concrete composition

Concrete properties

Element

Load

Boundary condition

cement type w/c ratio aggregate type aggregate size

age moisture content permeability porosity concrete class

size shape reinforcement cover thickness

type load

fire scenario no of heated surfaces duration of fire heating rate

Katarzyna Mróz | Dissertation Outline | 13

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | Heating rate (Connolly, 1995) showed that increasing the rate of surface heating promotes the likelihood of explosive spalling. Moreover, the higher is heating rate, the lower is the temperature in which spalling occurs, Figure 2.5. Author indicated also, that if spalling occurs at lower temperature a higher tensile strength has to be overcome and thus he confirmed that higher heating rate result in higher pore pressures within concrete.

Figure 2.5: Relationship between heating rate and explosive failure temperature (Connolly, 1995)

Concrete class As a result of increased density and better compaction of microstructure in high performance concrete, it is particularly more susceptible to spalling, whereas in normal concrete, in most cases, this phenomenon is not observed. Due to low permeability of HPC it comes to explosive spalling as a result of developing a vapour pressure. Figure 2.6 shows the relationship of the pressure in the concrete’s pores with temperature in normal concrete (NSC) and HPC (HSC) and indicates that in approximately 250˚C, when the pressure exceeds 2 MPa, HPC (HSC) experiences spalling (Phan, 2007). (Hager & Tracz, 2015) have confirmed that for concrete above 60 MPa, spalling occurrence is observed, Figure 2.7

Figure 2.6: Pore pressure in a function of temperature (Phan, 2007)

Figure 2.7: Max spalling depth in a function of concrete compressive strength (Hager & Tracz, 2015)

Moisture content Moisture content is one of the main factors causing spalling. In the absence of moisture no destructive spalling occurs. Increasing moisture content increases simultaneously the probability of occurring spalling. Moisture distribution in cross-section is of minor importance (Fèdèration Internationale du Bèton, 2007). 14

| Chapter 2: Literature study Free water and moisture gradients in concrete must be considered as being the main reason of fire spalling. Traditional concrete does not spall if it is dry, and all other reasons mentioned in this chapter may contribute to the effect of spalling, but cannot cause spalling without moisture (Hertz, 2003). (Shorter & Hermathy, 1961) noted that only concrete of a certain humidity will suffer from spalling, and the same concrete which originally was susceptible to spalling did not spall if it was dried to a depth of 20 – 30 mm from the surface. They have also noted that dry materials did not spall even if large temperature gradients were introduced, and they concluded that thermal gradient and thermal stresses cannot contribute to spalling without the presence of moisture. (Mayer-Ottens, 1974; Majorana, et al., 2010) claim that spalling can occur if the moisture content of ordinary strength concrete is more than 2% by weight (5% by volume). Explosive spalling is less likely for concretes with moisture contents less than 3% by weight (Meyer-Ottens, 1972), however, very dense high strength concrete can experience spalling in fire even with low moisture contents of 2.3 – 3.0 % by weight. This is due to the low porosity and permeability of such concrete, whereby even the release of chemically bound water can contribute significantly to pore pressures (Majorana, et al., 2010). Size of element According to tests by (Jansson & Boström, 2008), size of element being under investigation is of high importance. The authors presented the experimental program for small sized slab of 500 x 300 x 300 mm 3 and large slabs of 1200 x 1700 x 300 mm 3. The series of concrete were subjected to identical fire scenario ISO 834-1 an were simply supported and compressed uniaxially at level of 2.5 MPa. After fire test, max depth of spalling and its average were measured. For small element, the maximum spalling depth were approximately 89 mm, while for large slab it was 140 mm. Similar differences were indicated in average values. It was explained by differences in distribution of thermal stress and migration of water vapour between small and large slab. Boundary conditions – external load It has been tested by (Boström, 2004) that external compression can induce more spalling. It is explained by the mechanism of closing of microcracks that enable a water transport through specimen. While cracks are closed, concrete experience higher pore pressure and thus higher susceptibility to spalling. In experimental test, concrete slabs were made of four different composition. Each serie of concrete were tested in fire scenario ISO 834-1 under compression of 2.5 MPa as well as without load. After fire test, mass loss and depth of spalling in mean and maximum value were measured. Results shows that more spalling were observed while testing slab under compression. Differences in obtained results are presented in Table 2.2. Table 2.2: Spalling parameter for each tested concrete series (Boström, 2004)


ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE |

2.2

PREVENTING OF FIRE SPALLING IN CONCRETE STRUCTURES

2.2.1

PP FIBRES

There exist an effective and well-known techniques to prevent and reduce concrete spalling that consist in use of polypropylene (PP) fibres. Even in standards one can find the recommendation concerning polypropylene fibres dosage that enable to prevent and reduce the spalling occurrence. At a temperature of about 170 ˚C, PP fibres begin to melt (Kalifa, et al., 2001) and polymer blends into the concrete matrix. It produces a network of ducts in places previously occupied by fibres. Additionally, thermal expansion of the PP fibre is 8.5 times higher in comparison to concrete thermal strains (Sullivan, 2001). This feature leads to rising of tensile stresses, which provides to creating numerous of microcracks in concrete structure. Both mechanisms lead to increase of permeability of concrete, which reduces the water vapour pressure in the pores. There are a number of studies relating to the effective amount and type of PP fibres that affects positively on spalling behaviour of concrete. The different amounts (Hager & Tracz, 2010), diameters (Tatnall, 2002; Jansson & Boström, 2008), lengths (Hager & Tracz, 2010) and types were employed to find the one universal one recipe. (Connolly, 1995) found that in concrete with w/c of 0.4 the addition of 0.05% (by weight) completely eliminate the occurrence of spalling. Similar conclusions have been reached by (Han, et al., 2005). When the mixture contained PP fibre above 0.05% by volume, no spalling occurred, so that spalling resistance was significantly improved. (Hager, et al., 2013) performed testes on RPC with different amount of PP fibres (1 kg/m 3 and 2kg/m3) and different heating rate (0.5, 1, 2, 4 and 8 ̊C/min). The addition of PP fibres added in the amount of 2.0 kg/m3 PP fibres allows limiting the spalling risk of RPC cement material. Spalling was efficiently limited even when the relatively high heating rate (8 °C/min) was applied. In scientific community, however, it is known that PP fibres can limit spalling risk only to some extent. In case of HPC concrete and very severe fire scenario, spalling can be observed, regardless of PP fiber dosage. Therefore, the further research aimed at understanding the key factors influencing spalling are highly needed. 2.2.2

THERMAL INSULATION BARRIERS

There is a wide variety of the thermal insulation materials that can be used for a basic purpose of insulation from heat transfer. However, while testing a fireproofing of thermal insulators, one can find only few materials that can resist a real fire conditions. Mineral wool, expanded aggregate and cellulose are representatives of fireproof material for thermal insulation. Mineral wool, also known as rock wool or slag wool is one of the oldest types of insulation composed of non-combustible, naturally fire resistant stone wool. It can withstand temperature up to 1000 °C and does not burn. Over 1000 °C a mineral fibres start to melt. Mineral wool can be used as: the thermal and fire insulation between living area and non-heated roof spaces, a fire-resistant core for sandwich panels, a fireproof barrier for structural members in steel structures (Figure 2.8a), and as the fireproof cover for industrial pipes and ducts as well. Well designed and tightly built-in insulation barrier can be therefore an efficient passive thermal and fire protection. Other mineral materials are expanded perlite, shale, clay, slate and vermiculite those are recognized aggregate for fireproof cover manufacturing which offers the effective solution for life safety for both occupants and firefighting personnel. The non-combustible nature combined with high thermal insulation offers inherent structural integrity following exposure to fire make it the obvious choice for passive protection of building construction. Aggregate types affect fire ratings of cementitious composite material on the basis of heat transfer and on the basis of aggregate moisture absorption. Highly porous aggregates absorb moisture in varying degrees depending upon its type. 16

| Chapter 2: Literature study The presence of moisture in the aggregate during a fire test extends the fire duration by the time it takes for the moisture to be turned to steam and evaporated from the material. Finally, the cellulose insulation is made in a loose form from a recycled paper, newspaper, cardboard or other similar materials, it is considered as one of the most eco-friendly thermal insulation materials. Although the composition of the material is associated with the high flammability, the chemical treatment with ammonium sulfate and borate provide its incombustibility. What is more, because of a high compactness of the cellulosic fibres, the material contains almost no oxygen and effectively chokes wall cavities of combustion air and thus can minimize the spread of fire. As cellulose insulation is a loose material, it can only be used as filling of roof, floor and wall space, so the external part of structure is directly subjected to fire.

a)

b)

Figure 2.8: (a) Passive fire protection of steel structure, a fireproofing material sprayed onto steel structure elements; (b) endothermic reaction of concrete with dolomite aggregates, (DTA, 20 mg, 20°C/min)

2.2.3

ENDOTHERMIC BUILDING MATERIALS INCLUDING CONCRETE AND GYPSUM

Concrete is commonly known as fire resistant and incombustible material, so it has been used as a basic material for fire resistant structures for last decades. It protects a structure from fire in two ways. Concrete itself contains free water but also cement paste is made of significant quantity of hydrated crystals, so it contains a large amount of bound water. In case of fire, free water evaporates from a heat exposed surface and in this way it absorbs a great part of heat, leading to minimizing of temperature in internal part of structural member. In the next step, the dehydration process of CSH gel takes place, as well as portlandite decomposition when concrete is heated to temperature of 500550 °C. Those processes also absorb heat. The endothermic reaction can be even higher if the calcareous aggregates are used (Figure 2.8b). Due to its low thermal conductivity, concrete protects underlying part of structure for a sufficient period enabling to take a preventive action in case of fire. However, recent technological development and the increasing demand for high-strength structures caused also the development of concrete technology. As a result of increased density and better compaction of microstructure in high performance concrete, it is particularly more susceptible to fire spalling, whereas in normal concrete, in most cases, this phenomenon is not observed. Therefore, as far as normal concrete is used to protect steel in reinforced concrete (RC) structures, it provide its expected fire resistance. On the other hand, the cementitious coatings, ex. shotcrete, used as fire protection of steel structural members (beams, columns) are not recommended because of the risk of spalling, cracking or delamination in the contact layer between concrete and steel. Moreover, concrete-based coatings, as dense and massive materials, add a significant component of load to a load-bearing capacity design of steel structure. Katarzyna Mróz | Dissertation Outline | 17

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | Gypsum (calcium sulfate dihydrate) is a crystalline formed mineral found in sedimentary rock, but can also be a synthetic gypsum (Flue Gas Desulphurization gypsum or desulphurised gypsum) that is derived from coal-fired electrical utilities which are able to remove sulfur dioxide from flue gasses. Gypsum wallboards are an effective passive fire protection. As gypsum contains ca. 20 % of chemically bounded water, it can be evaporated in case of fire and help to minimize the temperature in the interior of protected structure and spread of fire, as described before. Moreover, gypsum boards are completely incombustible material and even after evaporation of entire amount of water, it remains a thermal insulation barrier. Producers of gypsum boards offer a wide variety of products for range of applications, including: wallboards for surface assembling on walls and ceilings, as well as in the interior of elevators or similar kind of shafts. Gypsum board can also be used to construct a fire separators between two areas or can be mounted directly of structural members, ex. steel beams, to provide a fire-resistant layer. However, in case of gypsum fireboard, tightness of coating is at highest importance.

2.3

EXPERIMENTAL METHODS FOR ASSESSMENT OF CONCRETE SUSCEPTIBILITY TO FIRE SPALLING

2.3.1

INTRODUCTION

Laboratory tests of concrete which aim to investigate its susceptibility to fire spalling employ various testing procedures carried out on specimens of different sizes and shapes. Due to the lack of standardized testing guidance, there is a wide range of approaches to concrete fire spalling assessment. During such tests, different measurements are carried out in order to better describe the processes taking place in concrete during heating. These parameters are temperature and vapor pore pressure development, thermal strains or specimen deflection. Numerous experimental tests results can be found in the literature, all attempting to investigate the fire spalling phenomenon by indicating different parameters that may enhance spalling risk, such as the mix composition, the heating scenario, the initial water content and the geometry of specimen or mechanical boundary conditions. Unfortunately, the influence of specific spalling parameters is difficult to assess and compare with others due to differences in the testing procedures used to obtain these results. Generally, we can distinguish three main categories of testing methods for spalling behavior investigations: small-scale, medium-scale and full-scale tests. The subsequent paragraphs describe the diversity of furnaces, their components, testing instrumentations and procedures. 2.3.2

SMALL-SCALE CONCRETE SPALLING TESTS

Small-scale tests examine the material's behavior when exposed to elevated temperature. These tests are carried out on small concrete specimens: prisms, cubes or cylinders with the volume not exceeding ca. 4000 cm3. High temperature conditions are mostly provided by electrical heating coils. In these tests only the spalling occurrence is detected and the number of specimens which spall is recorded. In more elaborate techniques, the specimens are mechanically loaded during heating. In most cases, the furnace is placed in a loading ram along with a concrete specimen which is stressed during heating, ex. (Connolly, 1995; Phan & Carino, 2002; Hager & Pimienta, 2004; Phan, 2005; Mindeguia, 2009; Huismann, et al., 2012). Following the recommendation of (RILEM TC 200-HTC, 2007), two test methods can be distinguished in which the material's behavior is studied in stressed, and unstressed conditions. The stressed test method corresponds to conditions in which the specimen is uniaxially loaded during heating. A compressive load of 10–50 % of ultimate stress is applied to the specimen at room temperature, which remains constant during heating to the target temperature level (T). All spalling events are recorded as well as the temperature at its occurrence. Although these methods are 18

| Chapter 2: Literature study mainly used to determine the material properties at the hot stage, the concrete's susceptibility to spalling can be also assessed using these procedures. The testing set developed by (Connolly, 1995) employs cylindrical specimens of Ø 150 mm, H 100 mm. The specimen is mounted in steel ring and is loaded peripherally by hydraulic arms in both vertical and horizontal directions. The loading arms are also designed to restrain thermal expansion. Both the load and thermal expansion are recorded within the event by a load cell. Structural stability is obtained by the stiff restraint frame, which provides support for hydraulic jacks. Heating is delivered by electrical radiative heating elements by Kanthal Electrical Ltd., producing a heat flux level of 150 kW/m2. The temperature was measured with the use of thermocouples cast at different depths of the specimen, which was also equipped with a ceramic pipe connected to pore pressure transducers capable of reading the pressure within concrete up to 10 N/mm2. The assessment of spalling, if such occurred, was determined by counting the number of steel grid squares (10 x 10 mm) placed over the concrete surface which were damaged more than 50%. By this procedure the authors obtained the extent of spalling expressed in a percentage of the total surface area. The PTM test developed by (Kalifa, et al., 2001) is so far one of the most referenced and cited test methods. The examination of specimens rely on recording prisms with additional pressure (P), temperature (T) and mass loss (M) measurements. These specimens can be instrumented with six gauges made of a sintered metal round plate, which are placed at casting. The last one is brazed to a thin metal tube (inner diameter 1.6 mm), which comes out of the rear face of the specimen. At the time of testing, a tight connector is placed at the free end of the tube. Firstly, it connects the gauge to a piezoelectric pressure transducer by means of a flexible tube filled with silicone oil. Secondly, a thermocouple is inserted in the tube through the connector down to the metal plate. The free volume of the gauge is around 250 mm3. The specimen, a prism (300 mm x 300 mm x 120 mm), is positioned horizontally. Its upper surface is directly exposed to elevated temperature, with a heating rate of 100 °C / hour, from a radiant electric heater (quasi-unidirectional thermal load), whereas the other surfaces are thermally insulated, Figure 2.9.

Figure 2.9: The experimental setup of PTM test, after (Kalifa, et al., 2001)

Similarly, (Phan, 2005) proposes measuring pore pressure development inside a concrete specimen carried out on concrete blocks (100 x 200 x 200 mm3) which are molded with pressure gauges and thermocouples at different depths (13 mm, 25 mm, 50 mm and 75 mm from the heated surface). The blocks are insulated on all sides, except for one which is subjected to heating and placed inside the same electric furnace. One-sided heating is believed to reflect one-dimensional heat flow. The specimen is then exposed to a heating rate about 5 °C/min or 25 °C/min. In Kalifa's (Kalifa, et al., 2001) and Phan’s (Phan, 2005) tests the heating rates were designed in a way that the spalling did not occur during the test. Thus, the main goal of test was to measure internal pore pressure development along with temperature at different depths. Another approach to concrete spalling assessment is to mold a specimen in a steel mantle or steel ring and perform the restrained conditions for concrete surface, ex. (Hertz & Sørensen, 2005; Tanibe, et al., 2011). The spalling investigations performed by (Hertz & Sørensen, 2005) consist in tests carried out on cylindrical specimens Ø 150 mm, H 300 mm that are placed in a steel mantle, Figure 2.10. The two parts of a steel mantle (upper and bottom) are connected with the use of 12 bolts Ø 36 mm. The thin Katarzyna Mróz | Dissertation Outline | 19

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | space between mantle and specimen is filled with a neoprene in order to reduce irregularities of the concrete surface. One plane side of specimen is exposed to a temperature of 1000 °C from electrical oven by a 100 mm hole. The steel mantle is believed to counteract the thermal stresses that occur in concrete specimens. By the use of a steel mantle, the concrete cylinder is restrained in its peripheral, while the cylindrical base of the specimen is exposed to fire. Thus, the conditions taking place in the cylindrical specimen reflect that of a concrete wall exposed to fire. Susceptibility to spalling is assessed by measuring the area of spalling. In addition, the specimen's acoustic emission is recorded in order to determine the time of spalling event.

Figure 2.10: The experimental setup of testing restrained concrete, after (Hertz & Sørensen, 2005)

In (Tanibe, et al., 2011) the perimetral setup was designed to monitor the internal temperature, vapor pressure and restrained stress of the concrete specimen. A circular specimen of 284 mm in diameter and 100 mm in height was molded in two steel rings of 300 mm in external diameter, 50 mm in height and 8 mm in thickness. The modulus of elasticity Es and yield strength of the steel ring fy are known (Es = 210 GPa, fy = 295 MPa). The steel rings provide restraint for concrete and enable its thermal strains to be measured by exposing concrete surface to fire at a heating curve of RABT 30. The thermal strains are then monitored by recording strains of a steel ring, two strain gauges being attached to its external surface (the thermal strain of concrete equals the measured stress reduced by the thermal strain of steel at a particular temperature). In order to measure the temperature profile inside a concrete specimen, six thermocouples are molded at different depths from the surface exposed to fire. Additionally, two thermocouples are placed in the outer surface of steel ring, where the thermal strains are examined. Vapor pressure measurement is provided by pore pressure setup, which consists of a steel pipes molded into the concrete specimen at depth of 10 mm and 20 mm and is connected to a pressure transducer located outside the furnace. The aim of this experimental test is to measure the thermal strains of concrete and assess any spalling that might occur. Spalling assessment is carried out by measuring the depth of spalling in the fire-exposed area and monitoring the moment of spalling at the time of testing. In order to compare the cited small-scale tests, the Table 2.3 collect all their technical data. Table 2.3: Summary of reviewed small-scale experimental setups. Reference

Specimens type and dimensions [mm]

Set

Heating method

Heating curve

External load

additional measurements

(Connolly, 1995)

cylindrical Ø 150 H 100

vertical

electrical

linear

vertical horizontal

T, P, thermal stresses, area of spalling

(Kalifa, et al., 2001) (Hertz & Sørensen, 2005)

prism 300 x 300 x 120

horizontal

electrical

100 °C/hour

-

T, P, mass loss


horizontally

electrical

linear max 1000 °C

restrained

T, AE, thermal stresses area of spalling

(Phan, 2005)

prism 100 x 200 x 200

vertical

electrical

5 °C/min 25 °C/min

-

T, P

(Tanibe, et al., 2011)


vertical

electrical

RABT 30

restrained

T, P, restrained stress thermal strains, depth of spalling

T – temperature, P – vapour pore pressure, AE – acoustic emission

20

| Chapter 2: Literature study 2.3.3

MEDIUM-SCALE FURNACES

Medium-scale tests are performed as a screening tests in order to verify the specific concrete mixture's behavior in fire conditions and select the one which does not tend to spall. These types of tests are cost effective when compared to the full scale tests performed on structural elements. The results of a medium scale test allow one to preselect the concrete mixes that are not susceptible to spall. This approach is intended to limit the number of specimens employed in full scale testing and thus to increase its cost-effectiveness. Concrete spalling due to fire is mainly evaluated in two ways. The first consists of tests carried out during high temperature exposure where particular measurements are made during the spalling event, i.e. its beginning, duration and frequency. Acoustic emission methods, presented by (Huismann, et al., 2011), allow the spalling to be characterized by counting the number of acoustic events, whereas the method reported by (Carré, et al., 2013) employs a digital camera to register the volume and time of spalling. Moreover, there are other experimental techniques that aim to explain the resulting physical phenomena, such as internal vapor pore pressure or moisture profiles evaluation. These can be conducted during heating of specimen. The second approach to spalling effect characterization relies on measurements taken after the test's completion, i.e. the size, shape and amount (mass, volume) of the spalled concrete material. The extent of the spalling and its quantification may be presented as a spalling map, a diagram of maximum and average spalling depth or even spalling topography. Such medium-scale experiments are mostly carried out on a concrete slab whose average size, 1000 mm in length x 800 mm in width x 200 mm in thickness, does not exceed a volume of 1.5 m3. Depending on the testing set, specimens are placed horizontally on the top of furnace chamber, (Carré, et al., 2013; Heel & Kusterle, 2004; Jansson & Boström, 2008) or mounted vertically in the furnace wall (Huismann, et al., 2011). In most cases, the furnace along with specimen is placed in a loading frame and concrete specimen is stressed during heating. Both unloaded tests by (Mindeguia, 2009; Huismann, et al., 2011; Iglesias & Wetzig, 2009; Hager, et al., 2014) and loaded tests by (Carré, et al., 2013; Heel & Kusterle, 2004; Boström & Jansson, 2008) can be found in the literature. Compression can be applied in different ways. The first is to use post tensioning bars molded into specimen or mounted on the external perimeter of the slab. Another solution employs flat jacks or a prestressing clamping frame aimed at loading the slab in a uniaxial or biaxial compression. Example of testing sets and procedures are described below, distinguishing between unloaded and loaded tests.

Figure 2.11: Fire scenarios used in laboratory testing

During the test, temperature development is measured with the use of thermocouples molded inside concrete specimen, as well as placed near the fire exposed surface. The high temperature conditions are provided mostly by propane or oil burners. In general, in laboratory testing we can Katarzyna Mróz | Dissertation Outline | 21

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | distinguish few types of time-temperature curves that are followed by researchers in order to reflect expected fire conditions, Figure 2.11. While testing the specimens of construction purposes, the ISO 834-1 curve is used in the laboratory testing. It reflects the burning rate of the general building materials and contents. In buildings of risk of burning the car fuel tankers, petrol or other chemical tanker, the development of fire is faster and the temperature in such case exceed the ISO 834-1 curve. In that case the Hydrocarbon (HC) curve shall be used. Also in laboratory testing of tunnel concrete another curve, RWS, is used, as a result of studies carried out in Netherlands and Norway. In some research we can find also reference to RABT curve that was developed on the basis of series of research program performed in Germany. 2.3.3.1

Unloaded tests

A specimen placed horizontally on the top of the furnace is subjected to a dead load, without additional external loading. The temperature in the slab is measured by thermocouples molded during casting and in the furnace with a flat thermocouple. This test does not provide external load and the concrete specimen is molded without reinforcement. Such a solution was employed in VersuchsStollen Hagerbach by (Iglesias & Wetzig, 2009), in Cracow University of Technology by (Hager, et al., 2014) and in Politecnico di Milano1. Figure 2.12, consists of a steel shell and an internal fireproof lining. The specimen, a concrete slab (1200 mm x 1000 mm x thickness 300 mm), is placed horizontally on the top of furnace and exposed to a fire action surface of ca. 750 mm x 950 mm. The furnace has two ventilation pipes 120 mm in diameter with draft regulators. The fire conditions are provided by gas burners (140 kW) fueled with propane-butane. Specimens are subjected to a time – temperature scenario, which is presently an ISO 834-1 curve.

Figure 2.12: Dragon furnace developed and used in Cracow University of Technology, fot. I.Hager

The BAM furnace developed by (Huismann, et al., 2011) used for vertical investigation of a concrete slab 1000 mm x 1000 mm x 1000 mm runs with two oil burners that provide a fire scenario on the hydrocarbon (HC) fire curve. The fire exposed surface of the vertically placed concrete specimen is 500 mm x 500 mm. The furnace temperature is controlled by insulated thermocouples. During fire tests the temperature is measured inside the furnace as well as inside the specimen, an AE analysis is also being performed. Sensors and amplifiers are connected at five points on the unexposed surface. Additionally, the pore pressure is measured by pressure gauges which have been cast into the concrete. These are connected to the pressure sensors and the thermocouples. The entire setup does not provide a mechanical load. The spalling phenomenon, if it occurs, is described at the time of spalling and is presented in three ways – by means of Acoustic Emission within the test, by a diagram 1

Personal communication

22

| Chapter 2: Literature study of the maximum and average spalling depth and by spalling topography performed after completing the test. 2.3.3.2

Loaded tests

The experimental setup proposed by (Heel & Kusterle, 2004) provides tests on flat slabs with dimensions of 1400 mm x 1800 mm x 500 (or 300) mm under an RWS fire scenario and loading conditions. Load is provided by selecting one of two available systems: unbounded prestressing tendons or a prestressing clamping frame. The two-stage oil burner placed in the central part of the furnace housing's shorter side is controlled on the basis of digitally recorded temperatures. The fire exposure area is 800 mm x 1200 mm. Externally applied compressive force is performed as follows: the main loading force of 1.16 MPa or 9 MPa in a transverse direction and an accompanying force of 0.5 MPa in a longitudinal direction in order to reduce slab deformation, Figure 2.13. The tensile stress in the tendons induces compressive stress in the specimens, and is controlled by a readout of the prestressing force indicated on the prestressing jacks. Additionally, a manometer is used to measure increase of pressure in tendon ducts caused by thermal expansion of anti-corrosion grease protecting tendons. The mix composition, reinforcement and curing conditions are consistent with the design of a real structure. During the test, temperature is measured by thermocouples placed centrally at 10 mm and 40 mm from the fire exposed surface. For spalling assessment, temperature and pore pressure development is recorded at time of testing. When a spalling event takes place, the values of temperature, pore pressure and loading force are recorded. Additionally, after test completion, the maximum and average spalling depth is measured.

Figure 2.13: Left: Loading with the use of unbounded prestressing tendons; Right: Loading with the use of prestressing clamping frame (Heel & Kusterle, 2004)

The setup developed by (Jansson & Boström, 2008) allows testing of concrete slab specimens molded without reinforcement with dimensions of 600 mm x 500 mm x 200 mm. The slab is placed centrally on the top of furnace and is subjected to a hydrocarbon (HC) fire curve or linear heating with the rate of 10 °C/minute. Three aluminum pipes are placed in each specimen into which post-stress bars can be placed after the casting Figure 2.14a. Post-tensioning bars are used to apply the external compressive load. For the purpose of measuring temperature, thermocouples are centrally placed in each specimen at a depth of 10 mm and 40 mm from the fire exposed surface. The loading system is performed with a post tensioning system made of four Dywidag bars, Ø 28 mm, placed horizontally inside the concrete slab. The load level of the post-stressing force is 10 % of the cube compressive strength tested prior to the fire test. The resulting compressive force is transferred to the concrete specimen by the steel plates fixed to its loaded side. The tensile stress in the bars, which induces compressive stress in the concrete specimens, is controlled with load cells connected to the MGC Plus system and computer for data acquisition. Load can also be applied by externally mounted post-stressed bars, Figure 2.14b, which allows the influence of steel pipes' thermal expansion to be limited. In such cases, the concrete slab is molded Katarzyna Mróz | Dissertation Outline | 23

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | without reinforcement and aluminum pipes. During the test, temperature development and internal pore pressure are recorded. When spalling occurs, the corresponding values of T and P are registered, further spalling quantification being performed afterwards. To assess tendency to spall, average and maximum spalling depth is measured and a 3D spalling profile is prepared.

Figure 2.14: Testing sets in medium scale: a) with the inner post-tensioning bars, b) right with the outer posttensioning bars, after (Jansson & Boström, 2008), c) with 4 flat hydraulic jacks, after (Carré, et al., 2013)

In CSTB Laboratory in France, tests are carried out according to an ISO 834-1 time-temperature scenario, (Carré, et al., 2013). The external part of furnace is made of steel whereas internal lining is made of high temperature resistant bricks. A fire source consists of 8 gas burners using propane. This setup tests concrete slab specimens (680 mm x 580 mm x thickness (ex. 150 mm)) placed horizontally on the furnace and subjected to fire from beneath, the fire exposed surface being 600 mm by 420 mm. The examined slab is loaded horizontally with flat jacks (type Freyssinet), which enable uniaxial or biaxial stress state application during the test (2 or 4 flat jacks respectively, Figure 2.14c). Thermocouples inside the furnace, located 10 cm from the exposed part of the slab, are hand controlled and monitored during the test, while pressure gauges are employed at various depths - 10 mm, 20 mm and 30 mm from the exposed surface. During the fire test, spalling events are observed through a small hole and recorded using a digital camera. In such tests, acoustic events are recorded using a microphone located near the specimen. In order to secure a stress state provided by flat jacks, a reference test without thermal load is made using strain gauges placed on the specimen's surface. The summary of presented loaded tests is presented in Table 2.4. Table 2.4: Summary of reviewed medium-scale experimental loaded sets Reference

Fire exposed area

Fire curve

Reinforce ment

Location of load

Type of load

Fuel

Stress state

Additional measurements

[mm x mm] (Heel & Kusterle, 2004)(A)

1400 x 1800

RWS

yes

internally

post stressing bars

oil

compression

(Heel & Kusterle, 2004)(B)

1400 x 1800

RWS

yes

externally

prestressing clamping frame

oil

biaxial compression

(Jansson & Boström, 2008)

600 x 500

no

internally

post tensioning bars

gaspropane

compression

spalling depth and profile

(Carré, et al., 2013)

no

externally

flat jacks

gaspropane

uniaxial and biaxial compression

T,

580 x 680

ISO 834-1 10 °C/min

ISO 834-1

T – temperature, P – vapour pore pressure, AE – acoustic emission

24

T, P, spalling depth T, P, spalling depth T, P,

digital camera, acoustic events

| Chapter 2: Literature study 2.3.4

FULL SCALE FIRE TESTS

Full scale test setups allow tests to be performed on concrete elements (beams, columns walls, tunnel linings) by applying external load and choosing various fire scenarios. Full scale fire tests which permit real structural members to be tested are carried out on the fully sized concrete elements in which the boundary conditions, external load and conditioning correspond to design assumptions. Full scale tests, being the most representative, directly present the qualitative and quantitative behavior of structural concrete element subjected to fire. Since, the description of testing procedure of a particular setup strongly depends on the testing configuration, author describes two testing setups as examples of full scale tests. The furnace described by (Richter, 2004) is used for examination of tunnel segments. The internal dimensions of the furnace chamber were: 2600 mm in width, 3900 mm in length and 1300 mm in height. The specimen is placed on the top of furnace and subjected to any fire curve. The specimen is heated with seven oil burners, two on each wall except for one longitudinal wall, where only one is installed. The boundary conditions as well as the load can be adjusted to needs - restrictions of elongations and additional vertical (6 hydraulic jacks) or horizontal (2 hydraulic jacks) loads can be applied. During the tests temperature development can be measured 100 mm from the fire-exposed surface at three points. The vertical deflections of specimen and horizontal displacement at the support can be recorded continuously. Spalling and cracking behavior as well as the emergence of moisture on the unexposed surface of the tested element can be checked visually during the heating. If spalling occurs, the temperature accompanying this event is noted. After the test, spalling is quantified visually by observations of micro-cracks on the fire-exposed surface and detailed photographic evidence of particular specimen. The furnace used in SP Technical Research institute of Sweden by (Boström & Jansson, 2008) is a large horizontal furnace with a clear opening of 5000 mm x 3000 mm. It is used for the examination of different structural members, beams, columns, slabs, etc. Oil is used for fire development with examinations done according to ISO 834-1 and HC fire curves. Specimens, such as slabs, are located horizontally and supported on the walls from two sides. During the test for slab elements, flat thermometers are placed 10 mm below the fire exposed concrete surface in order to measure temperature development. Additionally, in the cited work the measurements of internal pore pressure was performed along with temperature measurements at different element depths. The specimens were loaded with compressive load by post tensioning threaded bars (Dywidag 36 mm) in aluminum pipes, with load cells mounted in order to ensure that a correct load level was applied, thus enabling continuous measurement during the fire tests. Such full scale tests are the most representative, directly showing the qualitative and quantitative behavior of structural concrete elements subjected to fire. The disadvantage is that tests performed using large furnaces are expensive and unsuitable for testing a great number of different concrete mixes. To overcome this limitation, and thanks to the dimensions of a large furnace (Mindeguia, 2009; Boström & Jansson, 2008; Taillefer, et al., 2013), many small-sized slabs can be tested at the same time, thereby making it possible to examine different concrete mixes, or various specimen sizes during a single test.

2.4

INFLUENCE OF STRESS STATE OF CONCRETE SPALLING PROPENSITY

The recent research programs carried out by number of scientists indicate the influence of different stress states provided in tested specimen on observed concrete spalling propensity. The stress state varies with chosen test configurations. We can distinguish the following test configurations: unloaded slabs, uniaxially or bi-axially loaded slabs in compression, specimens with


ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | external restraint or specimens restrained with, so called, cold rim. In the following paragraphs, the state of recent research program on the mentioned test configurations is presented. 2.4.1

SPECIMENS LOADED IN COMPRESSION

Concrete spalling propensity is generally tested on slab specimen that is placed horizontally on the furnace. The bottom fire-exposed side of the slab is approximately in area of 1 m 2. Recent developments employ additional loading frame that is aimed to introduce a uniaxial or biaxial compressive stresses in slab during heating. The (Boström & Jansson, 2008) performed tests with either unloaded specimens or with specimens loaded in compression. In loaded test, concretes were tested using different compressive load levels provided by three internal post-tensioning bars. The maximum tested load level was chosen to 10 % of the compressive strength. The specimens of small slabs had the dimensions 600 x 500 x 200 mm 3 exposed to ISO 834-1 temperature curve. There was no reinforcement, except the post-stress bars used for applying the external compressive load. The results from tests on the slabs showed that load level up to 10 % of the compressive strength did not affect the maximum spalling depth, Figure 2.15. The unloaded specimens, however, showed slightly less spalling. Therefore, it may be concluded that higher load levels result in more severe spalling.

Figure 2.15: Effect of applied compressive stress on the spalling depth (Boström & Jansson, 2008)

On the other hand, in studies by (Carré, et al., 2013) a steel frame had been developed to apply uniaxial and biaxial stresses on slabs during fire tests. The slabs had the dimensions of 580 × 680 × 150 mm3. Tests were carried out on an ordinary concrete (fc28 = 37 MPa) exposed to ISO 834-1 temperature curve with several levels of uniaxial loading: 0 MPa (unloaded specimen), and 5, 10 and 15 MPa. No spalling was observed when slabs were loaded at 0, 5 and 10 MPa. In the opposite, spalling was observed when the compressive stress was increased to 15 MPa, Figure 2.16. It was a further evidence on spalling propensity enhanced by increased compressive stresses in concrete. Another research representing the tests aimed at determining the influence of restraint on concrete susceptibility to spalling was presented by (Connolly, 1995). The author had designed the restraint system that enabled accurate measurement of restraint level. The experimental setup consisted of steel frame equipped with the hydraulic jacks that were employed to provide different levels of restraint. In the cited study, the tests were carried out on normal strength cylindrical concrete specimens. The load aimed at restraining the thermal expansion of specimen during test were distributed to the peripheral of concrete cylinder with the use of four curved loading shoes. The author tested four 26

| Chapter 2: Literature study series of specimens under different load level: unloaded, 10, 20 and 30 N/mm2. It was observed that spalling occurred in higher levels of compressive stresses: 20 and 30 N/mm 2. However, the repeatability of results were more convergent for higher level of heating rate (140 kW/m 2) that in case of lower one (80 kW/m2).

Figure 2.16: Effect of applied compressive stress on the spalling depth (Carré, et al., 2013)

2.4.2

SPECIMEN RESTRAINED WITH COLD RIM

On the other hand, compressive stresses can be introduced in concrete with the specimen itself. The unheated part of tested material does not experience thermal expansion and hence limits the thermal expansion of the heated central part of specimen. Therefore, the cold rim of specimen creates a kind of restraint to inner part of concrete and contribute to development of compressive stresses in heated part. High compressive stresses, caused by restraint to thermal expansion, develop when the rate of heating is such that the stresses cannot be relieved by creep quickly enough (Khoury, 2005). The research program performed by (Hertz, 2003) significantly showed that thermal stresses can be decisive for spalling to occur in concrete. In the experiment 10 specimens were made of dense concretes of various compositions close to the one used for the Great Belt tunnel project (Gunnarsson, 1998). The specimens were small slabs of 600 x 600 x 200 mm 3 heated in a limited area of 200 x 200 mm2 at the center of a side, so that the unheated part of specimen around a fire exposed area could resist the thermal expansion, according to what is described before. In all tested specimens, progressive spalling was observed for about 20 minutes until the thermal stresses caused cracking of cold part. When the cracks occurred, spalling stopped immediately, Figure 2.17.

Figure 2.17: Spalling of a tile and thermal cracks causing the spalling to stop (Hertz, 2003)


ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | It seems, that as soon as the internal thermal cracks were developed, the thermal compression stresses at the surface were unloaded, and consequently spalling were prevented unless the concrete external load or hindered thermal expansion of the entire cross-section caused additional stresses. For example in circular tunnel walls the geometry itself causes the hindrance necessary, and these restraints must be modeled if the fire tests are made on tunnel elements. The presented research provide a significant evidence for spalling susceptibility under compressive load. 2.4.3

SPECIMEN RESTRAINED WITH EXTERNAL RESTRAINT

Another approach to provide a limitation of thermal expansion in concrete is to cast concrete in stiff steel frame that is aimed to provide a restrained boundary conditions in tested element. Such solution also introduce compressive stresses in heated concrete that are believed to enhance a susceptibility to fire spalling. The influence of internal compressive stresses on the observed extent of fire spalling were presented by (Tanibe, et al., 2013). In the research program, the experimental setup was designed, Figure 2.18. It consisted of high performance (Fc: 80 MPa), small (of size 300 × 100 mm), concrete disc casted in steel ring aimed at limit the thermal expansion and work as the external restraint. In order to provide different levels of restraint, the authors used steel rings with various thicknesses of 0.5, 8 and 18 mm. Also the test without a steel ring were performed. The specimen was heated from one side with the use of gas burners. The heating scenario followed RABT curve. On the basis of measured strain of the ring, the authors were able to calculate the restraining stresses in concrete specimens.

Figure 2.18: Schema of specimen and steel ring (Tanibe, et al., 2013)

Figure 2.19: a) Depths of spalling (Tanibe, et al., 2013), b) Damage to heated surfaces (Tanibe, et al., 2013)

After 10 minutes of heating, the thermal stresses were 2.2, 7 and 9 MPa for 0.5, 8 and 18 mm of ring thickness, respectively. Spalling depth measured after the test correspond to the level of stresses. 28

| Chapter 2: Literature study In case of unrestrained and restrained specimen with 0.5 mm thick ring, the spalling depth reached 11 and 8 mm, while for higher stresses provided by 8 and 18 mm steel ring, the spalling depths were greater, 24 and 26 mm, Figure 2.19. Therefore, it can be concluded that compressive stressed induced by steel restraint are sufficient to spalling occurrence. However, the specimen size in the cited research correspond rather with material testing that with structural one.

2.5

SUMMARY

Most experimental tests used for concrete spalling attempt to explain its causes and define the factors influencing its occurrence. However, for the purpose of assessing concrete's tendency to spall, researchers have considered different approaches. Experimental tests carried out on small specimens with the use of an electrical heater mostly provides information regarding the material's response to fire, while the results of medium and large-scale tests take into consideration structural effects, such as loading, restraint and boundary conditions. Currently, there are no guidelines for medium-scale tests and the procedures for large-scale fire tests do not provide a means to quantify spalling and assess the damage extent. Medium scale tests seem to be a good way to screen specific concrete mixtures for their susceptibility to fire spalling. Unfortunately, there is a lack of specified testing procedures, which means that the results from different setups and test procedures are often inconclusive and not easily comparable. Therefore, guidelines for optimal test conditions, equipment and specimen instrumentation should be introduced. Furthermore, testing procedures (i.e. loading application) should be described in detail, providing criteria to evaluate spalling. Specifically, the methods employed to quantify the spalling effect need to be more unified, which would enable the results of different research to be compared. Any guidelines established for medium and full scale tests should arise from the experience of researchers in this field. Laboratory testing of concrete aiming at investigation of its susceptibility to fire spalling is carried out on specimens of different size and shape and using various testing procedures. Due to lack of standardized testing guidance concerning concrete spalling measurement, there is a wide range of approaches to concrete fire spalling assessment. In the current literature, the susceptibility of concrete to spalling is demonstrated in small scale, intermediate scale and full scale tests. During and after the test, different measurements are carried out in order to give a better description of the processes taking place in concrete during heating. The measurements which are taken during the tests are: temperature and vapour pore pressure development, thermal strains or specimen deflection. After the tests usually the size, shape and amount (mass, volume) of spalled concrete is evaluated, but also maximum and average depth of spalling with its topography is determined. Depth of spalling could be measured by hand with callipers (sometimes steel grids are placed on the sample to make it more precise) in regular distance measure points from which the grid is created. In the literature, a numerous experimental test results can be found that attempt to highlight the fire spalling phenomenon, and try to indicate different parameters that may enhance spalling risk, such as the concrete mix composition, heating scenario, initial water content, geometry of specimen or mechanical boundary conditions. However, the influence of each individual parameter on spalling is difficult to assess and compare with other results due to differences in testing procedures used to obtain these results.



3 OBJECTIVES, HYPOTHESES AND SCOPE OF THESIS 3.1

JUSTIFICATION FOR TACKLING SPECIFIC SCIENTIFIC PROBLEMS

The current state of knowledge about the research investigations devoted to the phenomenon of fire spalling in concrete permits to conclude that so far there was no comprehensive scientific program on the influence of stresses induced in different manners on concrete susceptibility to fire spalling. Despite the fact that there is a wide range of results from various approaches, there is no connection between them and it is impossible to put an evident conclusion about the stated problem. What is more, the PhD candidate is an active member of the RILEM Technical Committee 256-SPF: Spalling of concrete due to fire: testing and modelling, established in 2013. The state-of-the-art on fire spalling of concrete prepared by the scientific group shows a significant heterogeneity of the experimental studies. Indeed, no clear consensus is made with regard to critical parameters, ex. geometry of the specimen, the type of heating, or the mechanical boundary conditions. The differences between collected experimental data and within the testing procedures make the analysis of the causes of spalling not so evident. Recent problem of high importance for the group is the unified research program that provide a sufficient results on the influence of different types of load on fire spalling risk in concrete. The comparison between an effect of external biaxial or uniaxial load application, external restraint or restraint by cold rim on the stress propagation in concrete specimen conducted within unified procedures (i.e. repeatable concrete mix, furnace volume, geometry of specimen, fire scenario that would not provide an unexpected scatter of results) is highly desirable by RILEM Technical Committee and may enrich fire designing process for engineers.

3.2

THESIS’ HYPOTHESES

The aim of the thesis is to prove the following hypotheses:  The compressive stresses induce fire spalling in concrete of higher classes exposed to fire; stress level can effects the nature of spalling;  External restraint of thermal expansion of heated concrete is sufficient to create compressive stresses in concrete interior that result in concrete fire spalling;  Unheated part of specimen (cold rim) is a source of limitation of thermal expansion of heated part of concrete and therefore can be considered as additional internal restraint enhancing spalling risk;  There is a wide range of similarities in spalling observations that can be obtained by different manners of loading applications;  It is possible to propose a unified testing procedure in order to study the influence of stress state on spalling propensity in concrete.

3.3

SCOPE OF WORK

The research methodology consists in numerical analysis and experimental investigation. The aim of the numerical simulations will be to determine the stress state which corresponds to different mechanical loads of slabs, taking into account the method of their heating and the impact of boundary conditions. The numerical simulations that will be carried out will also help in selection of load levels to be applied to the specimens during the experiments and in design of the experiment. In the next stage of the project, tests on 1.0 m by 1.0 m slabs, 0.15 m thick, will be carried out in order to evaluate the influence of load type (restraint, external in-plane load) and load level on spalling intensity. An important aspect of the project will be demonstration of the impact of the slab thickness and the unheated edge of concrete (cold rim effect) on the element susceptibility to spalling. Cold rim plays the role of a restraint of thermal expansion of the heated concrete element that brings about an 30

| Chapter 3: Objectives, hypotheses and scope of thesis additional compressive stress component. The numerical analysis, along with the experimental investigations, will make it possible to determine the stress state promoting the occurrence of different spalling forms, i.e. the popcorn effect and explosive destruction, but also the relations between the stress state and the concrete spalling intensity due to fire. The scope of work would consist of literature studies, numerical analysis and experimental tests. The proposal scope has been divided into 6 stages, namely: 1° literature study, 2° FEM model, 3° design of experiment, 4° experimental tests, 5° experimental results, 6° validation of FEM model. Detailed outline of workplan is presented in Table 3.1. Table 3.1: Proposal outline of work

§

STAR on procedures of testing in intermediate scale, type of load and its influence

§

comparative state of the art on concrete mixture and its susceptibility to fire spalling

§

gathering data for concrete numerical model under thermal exposure

§

Literature study

stage 1°

Key tasks / outline of workplan

gathering data for modelling of different boundary conditions

§

§

heat transfer analysis Numerical analysis of effect of load, slab thickness and dimensions of unheated edge of slab on stress state in heated element thermomechanical analysis

§

FEM model

stage 2°

Implementation of numerical simulations aimed to determine heating and impact of boundary conditions.

conclusion on state and distribution of stresses in concrete specimen

§ § § § §

§ §

Execution of concrete slabs 1.0 x1.0 x 0.15 m with installation of temperature and strain sensors

§ § §

Design of experiment

stage 3°

§ §

§ §

§

Design of concrete mix, sample production for determining material properties (density, compressive and tensile strength, modulus of elasticity, thermal properties) and their changes resulting from heating design concrete mixture on basis of previous research project carried out at CUT concrete in fire batching cubic and cylindrical specimens with the chosen concrete mixtures curing of concrete specimens Design and measurement of stress state in heated slabs in different load configurations design of stress and strain measurements under elevated temperature development of measurement setup, instrumented with tensometers Development of explosive spalling measuring techniques during heating and methods of spalling intensity and its arrange evaluation after test decision on NDT techniques that should be implemented during the test development of procedure of measurements spalling after test

batching concrete slabs instrumented with tensometers and thermocouples curing of concrete specimens Experiment design based on numerical model assumptions for cases A (external loading), B (cold rim effect) and C (restraint) on basis of FEM analysis, decision on: I: levels of compressive load for external loading setup II: extent of cold rim for test of concrete restrained with unheated part of specimen III: type of external restrained with use of steel frame Adaptation of Dragon furnace to carry out research on 1.0 by 1.0 m slabs; burner system calibration calibration of burners to provide standard fire scenario in furnace chamber instrumentation of Dragon furnace according to the decisions taken in previous point validation of stress state from external load on unheated specimens with use of tensometers


§

§ §

Implementation of experiments on influence of the load level on intensity and course of explosive behavior of concrete performing 'A' experimental approach on slabs performing 'C' experimental approach on slabs Realization of tests on influence of cold rim dimensions on intensity and course of explosive spalling behavior of concrete performing 'B' experimental approach on concrete slabs

§

§

§ § §

§ § §

Execution of damages map of elements; determination of parameters characterizing spalling intensity measurement of spalling topography measurement of maximum spalling depth, volume and scatter decision on parameters that sufficiently describe spalling intensity Research results analysis and interpretation comparison between results obtained in the A, B or C approach conclusion on influence of different manner of stress state application on spalling conclusion on the most severe of the most appropriate experimental approach: A, B or C

§ §

Experimental tests Experimental results Validation of FEM model

stage 6°

stage 5°

stage 4°


Validation of model on the basis of experiment comparison between heat transfer FEM analysis and experimental data of temperature distribution in concrete specimen validation of state of stresses in different approach A, B or C Comprehensive analysis of experimental results and numerical simulations drawing conclusions on performed tests state of knowledge and perspectives

4 RESEARCH PROGRAM AND METHODOLOGY The scientific methodology consists of two fields: numerical modelling and experimental test. The main research field is an experimental test while numerical modelling is aimed to reduce the number of attempts of laboratory tests and to provide an input data for mechanical boundary conditions: supports and level of external loading. The block diagram presented in Figure 4.1 illustrates the scope of experimental tests that are detailed in the following paragraphs.

Figure 4.1: Block diagram of scope of experimental tests

32

| Chapter 4: Research program and methodology

4.1

NUMERICAL MODELLING

Numerical analyses will be performed in order to evaluate both the thermal field inside the specimen during heating and the thermal stresses induced by the constrained thermal dilation (taking into account the variation with the temperature of the physio-mechanical properties of concrete). The commercial software ABAQUS will be used. The numerical modelling of thermo-mechanical behaviour of concrete specimen under different stress state will be conducted by Finite Element Method (FEM) in two stages. The first stage will consist of heat transfer analysis of concrete unit subjected to heat source from one side. The main goal of thermal analysis is the evaluation of the temperature T in all the points belonging to the domain under investigation. The thermal analyses are based on a purely conductive model, according to Fourier's differential equation. Considering an isotropic, homogeneous and continuous body, Fourier equation can be used to work out the specific heat flux exchanged between two surfaces at a distance dx according to Eq. 4.1: 𝑞 = −𝜆

𝑑𝑇 𝑑𝑥

[

𝑊 𝑚2

]

(Eg. 4.1)

,where q is the heat flux per unit [W/m2], λ is thermal conductivity [W/mK], T is temperature [K] or [°C], and x is a distance in the direction of heat flow [m], and the sign minus indicates that the heat flux has the same direction of temperature decrease. In proposal modeling approach, a transient problem in which both the boundary conditions and the solution are time-dependent is going to be carried out. This procedure can be shown for material limited within a spatial domain Ω by a contour Γ, where the temperature T is fixed on ΓD and the heat flux 𝑞 is fixed on ΓN. The classical formulation of the transient problem is: to find T = T(x,t) in order that Eqs. 4.2, 4.3, 4.4 and 4.5 are satisfied: 𝛻𝑞 = 𝛻(𝜆𝛻𝑇) = 𝜌𝑐

𝛿𝑇 𝛿𝑡

𝑖𝑛 𝛺, 𝑡 > 0

𝑇 = 𝑓(𝑥, 𝑡) 𝑜𝑛 𝛤𝐷 , 𝑡 > 0 𝑞 = −(𝜆𝛻𝑇)𝑛 = −𝜆

𝛿𝑇 𝛿𝑛

𝑜𝑛 𝛤𝑁 , 𝑡 > 0

𝑇 = ℎ(𝑥, ) 𝑖𝑛 𝛺, 𝑡 = 0

(Eq. 4.2) (Eq. 4.3) (Eq. 4.4) (Eq. 4.5)

2

, where ρ is density of concrete [kg/m ], c is temperature-dependent specific heat of concrete [J/kgK], λ is thermal conductivity [W/mK] and h is convection coefficient [W/m2K]. The problem is regarded as a one-dimensional transient occurring along the thickness of the wall, with no influence of the perimetral boundaries of the element. The thermal properties of concrete, i.e. thermal conductivity, density, specific heat are taken accordingly to (EN 1992-1-2, 2004). The coefficient of exchange by convection and radiation to the exposed face and on the unexposed face are adopted from standards (EN 1991-1-2, 2002) and (EN 1992-1-2, 2004). The surface emissivity of the concrete specimen is taken as εm = 0.7, while coefficient of heat transfer by convection = 25 W/m2K for heat-exposed side and at non-exposed side a reduced convection coefficient is 9 W/m2K). The gas temperature in the boundary follows a temperature-time ISO 834-1 scenario for 2 hours. The ambient temperature is assumed as 20 °C. The second stage of numerical analysis is to model the thermomechanical case of concrete specimen subjected to fire scenario and different loading states. Four different manners of load applications are going to be considered, namely:  unloaded specimen exposed to fire from one surface in uniform manner – this case is meant to be the specimen under no stress state;  unloaded specimen exposed to elevated temperature in central region of the surface – this case is meant to consider the influence of cold rim (cold part) of concrete specimen that is believed to Katarzyna Mróz | Dissertation Outline | 33

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | restrain a thermal expansion of internal, heated part of material and thus to introduce a stress state in specimen;  specimen restrained with external steel stiff frame – this case is meant to compare the influence of cold rim on stress state between the real restraint that is provided by stiff frame in boundary zone;  externally loaded specimen in uniaxial or biaxial manner with the use of flat jacks – this case is meant to introduce different compressive stress level into specimen subjected to fire and to calculate the sufficient load level for reflecting a natural restraint. Mechanical behaviour of concrete under elevated temperatures will be modelled accordingly to the experimental data obtained in the tests carried out in the previous scientific project “Multiparameter diagnostics of condition of cement concretes subjected to a fire temperature impact”, financed by the National Science Centre N N506 045040. In the project, a comprehensive studies on decrease of compressive strength, modulus of elasticity, stress-strain curve, splitting tensile strength and density were performed and were published in (Hager, et al., 2015). Since, the conclusions made in the mentioned project will be the basis for design of a concrete mix that is susceptible to spalling, it is justified to utilize the already existed results on mechanical properties of concrete made of this specific mixture. It should be emphasized that those results were obtained for unloaded specimens and therefore the data can be used only to model the behaviour of unloaded specimen. For loaded cases, the data for behaviour of concrete under complex thermal and stress condition are required. Mechanical properties of loaded concrete under fire exposure are assumed according to the results obtained under the collaboration with researchers from universities and laboratories in France (Hager, 2004; Carré, et al., 2013), the state-of-the-art and the experimental data gathered from other research program performed at French laboratory Centre Scientifique et Technique du Bâtiment – France (CSTB). In order to reflect real behaviour of concrete under stress state, the following temperature-dependent mechanical properties are taken into account: thermal strains (TS) of concrete under loading and unloading conditions, stress-strain curves, compressive strengths and modulus of elasticity of concrete, average residual cracking energy and constant Poisson ratio of concrete.

4.2

EXPERIMENTAL TESTS

4.2.1

MATERIALS AND SPECIMENS

On the basis of results obtained in the recent project “Multiparameter diagnostics of condition of cement concretes subjected to a fire temperature impact”, described in point 5.1, it has been chosen a one concrete mix that was considered as the most susceptible to spalling. The tests were carried out on different concrete mixes. The research aimed at determining the influence of different parameters: w/c ratio (0.30; 0.45; 0.60), cement type (CEM I and CEM III) and type of aggregates (riverbed, granite and basalt) on concrete spalling. After the test several measurements were conducted enabling spalling severity evaluation and comparison between different concrete types. The comparative results within test performed on seven unloaded concretes slabs made of different mixtures, indicated that concrete made of a riverbed aggregate, cement CEM I 42.5 R and water to cement ratio w/c = 0.3 showed the most explosive behaviour under fire exposure. The maximum spalling depth and the maximum volume of spalling measured after cooling phase were obtained for that specific concrete. Also, it had been shown that the distribution of spalling events was the most homogenous in comparison to the remaining concretes. In the proposal project the High Performance Concrete HPC slabs will be manufactured using the following components: Portland cement CEM I 42.5R, riverbed quartz sand 0/2 mm and riverbed gravel 2/16 mm, of proportion indicated in Table 4.1. Plasticizer and superplasticizer will be used, and 34

| Chapter 4: Research program and methodology the water/cement ratio is going to be equal to 0.3. As it can be seen in Table 4.1, the mix composition of the concrete will consist of constant paste and mortar volume, 300 dm/m3 and 550 dm/m3, respectively, and will be the same for all mix batches. Table 4.1: Composition of designed concrete Component

Unit

Cement CEM I 42,5 R Małogoszcz Water w/c ratio Aggregates: Dwudniaki river sand 0-2 mm Dwudniaki gravel 2-8 mm Dwudniaki gravel 8-16 mm

kg/m

O/0,30/CEM I 3

482

dm3/m3

145

-

0,30

kg/m3 kg/m3 kg/m3

663 610 558

% mc % mc

0,90 2.1

dm3/m3 dm3/m3

300 550

Admixtures: plastyfier BASF BV 18 superplastyfier BASF SKY 591 Content: Cement paste Mortar

The tested series will consist of slabs of the following dimensions: 100 cm x 100 cm x 15 cm placed horizontally on the top of the furnace. The size of specimen is chosen in order to provide a symmetry in stress state induced by external loading and by restraint as well. The number of specimen for each individual case is chosen to be two. 4.2.2

HEATING AND LOADING CONDITIONS AND TESTING SETUP

The specimens will be tested in different loading configurations under fire conditions. The tests will consist in exposing one surface of concrete slab to fire in the DRAGON furnace, Figure 4.2a equipped with a 140 kW capacity gas burner. The furnace chamber was design to allow heating of concrete slabs with an intensity corresponding to the conditions of a standard fire scenario ISO 834-1. The temperature development inside the furnace will be measured by a plate thermocouple close to the heated surface of specimen and it follows the ISO 834-1, given with formula dependent on time t [min]: T(t) = 20 + 345log (8t +1) [°C], Figure 4.2b. In the experimental tests, four different configurations of load are going to be used, similarly to those described in paragraph for modelling:  unloaded specimen exposed to ISO 834-1 fire scenario from the bottom;  unloaded specimen exposed to ISO 834-1 fire scenario from the bottom that is going to be insulated in the region of 10 – 30 cm from the external region of surface – the fire exposed area will be only a central part of the bottom surface and cold part of specimens is aimed to work as restraining cold rim;  specimen restrained with external steel stiff frame and exposed to ISO 834-1 fire scenario from the bottom;  externally loaded specimen in uniaxial or biaxial manner with the use of flat jacks and the load level decided on the basis of prior numerical analysis. The stress conditions by external loading will be provided with the use of four flat jacks with capacity of 750 kN each. The load will be distributed in two directions by the loading shoes fitted with steel brushes in order to reduce friction between loading shoes and concrete surface and to obtain pure compression. Twelve steel brushes will be located at boundaries – 3 brushes per each of four Katarzyna Mróz | Dissertation Outline | 35

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | sides, Figure 4.3a. Each single brush is made of 1692 (36 x 47) steel bars of diameter 4 mm and length 132 mm, Figure 4.3b. The free unfixed part of bars equals 105 mm. Each bar is welded into steel pillow of dimensions 200 x 155 x 8 mm and it enables the load distribution between flat jacks and set of brushes.

a)

b)

Figure 4.2: a) DRAGON furnace and unloaded concrete slab placed horizontally (Hager, 2014), b) ISO 834-1 fire scenario

a)

b)

Figure 4.3: a) External loading frame fitted with flat jacks and brushed shoes, b) set of three brushes

During the test, the following measurement are going to be provided:  temperature on the heated surface with the use of plate thermocouple and the internal temperature at different depth of specimens with the use of thermocouples type K introduced to specimen during casting;  stress state with the use of tensometers mounted externally and internally;  the recording of sound with the use of Acoustic Emission recorder in order to specify the initial time of spalling (if observed), intensity of spalling and duration of this phenomenon;  in case of restraining with the use of steel frame, the deformations of steel are going to be measure during the fire test;  the impact-echo measurement in order to conclude about internal deterioration of concrete specimen exposed to fire.

4.3

SUMMARY OF RESEARCH PROGRAM

After the experimental tests, the results of stress state and NDT measurements will be elaborated in order to clarify the conclusions about the influence of different manner of stress application on spalling severity and stress state in concrete subjected to fire exposure. The further aim will be to describe the possibility of result’s comparison obtained in different approaches. Finally, the author will 36

| Chapter 5: Preliminary research conclude about the influence of stress state and distributions on concrete susceptibility to spalling behaviour, what is of particular interest in engineering society.

5 PRELIMINARY RESEARCH 5.1

EXPERIMENTAL TESTS

The preliminary research was carried out owing to the recent scientific project “Multiparameter assessment of cement concretes exposed to fire temperature”, financed by the National Science Centre (N N506 045040). The aim of those studies was to analyse the influence of different parameters, such as: cement type, aggregate type, water to cement ratio, on the behaviour of concrete under fire attack. The comprehensive analysis of the obtained results were presented in the technical report (Hager, 2014). The tests were carried out on 7 concrete compositions that are presented in Table 5.1.

dm3/m3 -

O/0,60/ CEM I

482 145 0,30

kg/m3

O/0,45/ CEM I

O/0,30/CEM III

B/0,30/ CEM III

Component Cement CEM I 42,5 R Małogoszcz Cement CEM III/A 42,5 N Małogoszcz Water w/c ratio Aggregates Dwudniaki river sand 0-2 mm Dwudniaki gravel 2-8 mm Dwudniaki gravel 8-16 mm Gracze basalt 2-8 mm Gracze basalt 8-16 mm Strzegom granite 2-8 mm Strzegom granite 8-16 mm Admixtures: plastyfier BASF BV 18 superplastyfier BASF SKY 591 Content: Cement paste Mortar

O/0,30/ CEM I

Unit

G/0,30/ CEM I

Concrete

B/0,30/CEM I

Table 5.1: Composition of concretes in kg/m3 and basic properties (Hager, 2014)

388 175 0,45

325 195 0,60

kg/m3 kg/m3 kg/m3 kg/m3 kg/m3 kg/m3 kg/m3

662 709 648 -

663 635 580

663 610 558 -

662 709 648 -

663 610 558 -

663 610 558 -

663 610 558 -

% mc % mc

0,90 2,20

0,90 2,20

0,90 2,10

0,90 2,35

0,90 2,35

0,80 -

0,30 -

dm3/m3 dm3/m3

300 550

The concretes with various water cement ratio w/c = 0.30, 0.45 and 0.60, contained the river bed aggregate and CEM I 42.5 R cement. In B/0,30/CEM III and O/0,30/CEM III the CEM III cement was used. Also the influence of the different aggregate types were investigated. In concrete designated as B/0,30/CEM I - basalt (B) crushed aggregate was used, in O/0,30/CEM I and O/0,30/CEM III riverbed gravel (O) and G/0,30/ CEM I, granite crushed aggregate (G) was used. In the case of all seven concretes cement paste and mortar volumes remained the same. The basic properties determined after 90 days of maturation in standard conditions such as density, gas permeability (Cembureau method), water content (tested on cubic specimens stored in the same conditions as the slabs), compressive strength, splitting tensile strength and modulus of elasticity were determined and given in Table 5.2. The influence of the temperature on properties of those seven concretes was investigated in previous research studies (Hager, 2014). Those properties were tested on standard specimens cubes and cylinders after having them heated to temperature 200, 400, 600, 800 and 1000 °C and cooled down to room temperature. Katarzyna Mróz | Dissertation Outline | 37

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | Spalling observation were performed on slabs heated using DRAGON furnace and the same heating procedures, enabled the comparison of spalling severity in case of those seven concretes. From those tests the influence of the following parameters was investigated: water cement ratio, cement and aggregate type.

Property

Unit

B/0,30/CEM I

G/0,30/CEM I

O/0,30/CEM I

O/0,45/CEM I

O/0,60/CEM I

B/0,30/CEM III

O/0,30/CEM III

Table 5.2: Basic properties of tested concretes given as a mean values (Hager, 2014)

Density

kg/m3

2558,8

2376,7

2300,7

2268,6

2177,4

2533,2

2315,6

Water content

%

2,6

2,4

2,3

2,2

2,3

2,7

2,8

Permeability (kc)

m2

7,00E-18

2,58E-17

1,20E-17

3,39E-17

2,22E-16

5,23E-18

9,96E-18

Compressive strength MPa

84,9

73,3

77,0

64,4

51,5

96,2

87,4

Splitting tensile strength

MPa

6,2

4,9

6,0

4,7

4,0

6,9

5,6

Modulus of elasticity

GPa

44,4

30,6

29,7

27,3

24,7

48,9

30,9

The study consisted in the examination of concrete slabs of dimensions 1.2 x 1.0 x 0.3 m3, which were heated from one side in the DRAGON furnace, presented in Figure 4.2a. In the slabs no steel reinforcement was used. The slabs were placed horizontally on the top of the furnace and subjected to the dead load, without additional and external loading. The slabs were resting on the edges of the furnace. By consequence, the surface of the slab subjected to fire curve was of 0.95 x 0.75m2. The slabs were not loaded however, due to the slab dimensions (significant thickness of 0.3 m) and the presence of the cold concrete rim on the edge of the slabs could be considered as the restraint. The cold rim induced the stress development due to the restrained thermal dilation of the heated concrete centre. The furnace was equipped with a 140 kW capacity gas burner allowed heating of concrete slabs with an intensity corresponding to standard fire scenario ISO 834-1. Furnace casing consisted of steel sheet with a thickness of 4 mm, reinforced at the corners angles. The insulation of the furnace consisted of two layers of ceramic fibre boards with a total thickness of 100 mm. In the side walls of the cover two exhaust vents with a diameter of 70 mm were prepared. The exhaust gases were discharged through a conduit with a diameter of equipped with baffles to regulate the draft. The height of chimney conduits were of 1400 mm. This furnace was developed in the frame of research programme entitled "Multi-parameter assessment of cement concretes exposed to fire temperature". During two hours heating the temperature in the slab cross-section was measured with type K thermocouples, placed in the central part of the slab while it was cast at the depths increasing of 3 cm from 1 cm to 22 cm. The temperature was recorded using a LUMEL KD7 recorder with the frequency of 1 measurement/min. In addition, during the test the temperature in the furnace was measured, close to the surface of the slab, using a plate thermocouple (surface area 100 cm 2), in compliance with the PN-EN 1363-1:2012 standard. The slab was subjected fire scenario ISO 834-1. After heating and cooling down of slabs, the pictures of the slabs were taken, Figure 5.1. The measurements of spalling depth were performed in order to collect data for drawing a map of damage. The spalling depth was measured in a regular distance of 25 mm and from this results the damage map was drawn, Figure 5.2. 38

| Chapter 5: Preliminary research To assess spalling severity, the following parameters were defined and evaluated: maximum depth of spalling (dmax), the volume of the material that spalled (Vs), and the percentage of the volume of the slab that spalled (Vrel) relative to the whole volume to the slab. T [˚C] 1000

973 943

O/0,3/CEMIII

800 560

600

470 300

400

192

200

124

78

0 0

Figure 5.1: Example of spalling damage (O/0,3/CEMIII) (Hager & Tracz, 2015)

50

100

150

200

250

300 x [mm]

Figure 5.2: Maximum temperature in the slab cross section and spalling map, mesh 25x25mm (O/0,3/CEMIII), (Hager & Tracz, 2015)

After completion of all 7 tests all the results were compared in order to highlight the differences and similarities in tested concretes behaviour. The only concrete that did not presented spalling behavior was the O/0.6/CEM I, concrete with w/c = 0.60 made with riverbed gravel. The other 6 concrete have spalled during the test. The spalling in all cases started from the first minutes (to) of the test as the popcorn like form. After first few minutes (t1) the explosive spalling events took place when the large patches of concrete were detaching from the slab surface. The consecutive explosive events took place during first 30-40 minutes of test duration afterwards the spalling stopped (t2).

Figure 5.3: Visual observations of slabs made of CEM I after fire test

After 120 minutes the gas burner was turned off. The next day the slabs were lifted and the visual observations were performed, Figure 5.3. On the lateral side of the slabs a large cracks were observed 2 or 3 at the longer side and 1 or 2 at the shorter slab side. No sloughing off due to the rehydration of concrete was observed during the post fire stage, most probably because no calcareous or dolomite aggregate grains were present in any of tested concretes. It should be emphasized that the surface of slab was not brushed or cleaned before the spalling measurements were taken because the surface of concrete remained firm and compact. The summary of results and observations made during the Katarzyna Mróz | Dissertation Outline | 39

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | heating and after cooling down to the room temperature was presented in Table 5.3. The results are presented in the increasing order from the smallest to the largest amount of spalling volume. The initial values of permeability of tested concretes was also presented in this table in order to assess if the relation between this property and material spalling could be determined. Table 5.3: Analyzed spalling parameters. The order from the smallest to the largest amount of spalling (Hager, 2014) Concrete O/0.60/CEM I B/0.30/CEM III G/0.30/CEM I B/0.30/CEM I O/0.30/CEM III O/0.45/CEM I O/0.30/CEM I

t0 [min] 0 3 3 3 4 1 1

t1 [min] 0 6 7 6 5 4 5

t2 [min] 0 34 38 30 33 34 40

dmax [mm] 0 40 35 45 50 55 65

Vs [cm3] 0 6 045 7 260 12 378 12 905 14 595 14 860

Vrel [%] 0 1.7 2.0 3.4 3.6 4.0 4.1

Initial permeability kc [m2] 2,22E-16 5,23E-18 2,58E-17 7,00E-18 9,96E-18 3,39E-17 1,20E-17

The results have confirmed that spalling is more likely to occur for denser and less permeable concretes, Table 5.3. With the decrease of water cement ratio the initial permeability decreased 2,22E-16, 3,39E-17 and 1,20E-17 for O/0.60/CEM I, O/0.45/CEM I and O/0.30/CEM I respectively. Spalling was not observed for O/0.60/CEM I, only the lateral cracks and surface cracking, with the cracks opening of about 0.5 to 2 mm were present. For O/0.45/CEM I and O/0.30/CEM I total measured volume of spalled concrete was of 14 595 cm3 and 14 860 cm3. Also the percentage of the volume of the slab that spalled (Vrel) relative to the whole volume of the slab was presented the similar values – in presented example – 4.0% and 4.1%. It should be emphasised that the difference in spalling severity expressed in volume was not so important and the difference in spalling maximum depth was only of 10 mm. From Table 5.3 it could be also observed that for concretes made with CEM III cements B/0.30/CEMIII (kc=5,23E-18 m2) and O/0.30/CEM III (kc=9,96E-18 m2) the spalling volume was smaller than for concretes made with CEM I cement despite the lower values of initial permeability B/0.30/CEM I (kc=7,00E-18 m2) and O/0.30/CEM I (kc=1,20E-17 m2) respectively. For those concretes also the maximum depth of spalling followed the same trend. The differences in spalling behaviour between concretes made with different types of aggregates showed that highest spalling volume as well as spalling maximum depth was observed for concrete with riverbed gravel (O/0.30/CEM I - 14 860 cm3). The other was concrete made with basalt aggregate (B/0,30/CEM I - 12 378 cm3) and the granite aggregate concrete (G/0,30/CEM I - 7 260 cm3). Thus, the preliminary study could form the basis for the design of concrete mix susceptible to fire spalling.

5.2

PARAMETRICAL STUDIES

The parametrical study aims at modeling of mechanical and thermal stresses in slab under fire in various configurations, mentioned in paragraph 4.1. The main goal of numerical simulation is to observe if the particular simulation being under consideration may deliver the expected results in experimental tests (ex. if the presence of cold rim of 10 cm and 30 cm in thickness are sufficient to observe the differences on the influence of cold rim). Author emphasizes that numerical analysis is not intended to model nor the spalling behaviour neither hydro-thermal phenomena occuring in concrete in fire. The numerical model aims only to facilitate the correct design of experimental tests. Parametrical studies are performed in Abaqus software. So far, author has completed the heat transfer analysis and considered the influence of cold rim (cold part) of concrete specimen that is believed to restrain a thermal expansion of internal, heated part of material and thus to introduce a stress state in specimen. In following paragraphs the models and results are presented. 40

| Chapter 5: Preliminary research 5.2.1

INPUT DATA AND PRELIMINARY RESULTS

The analysis is performed in two stages. In the first stage, the heat transfer analysis is calculated in unloaded specimen, while in the second stage the thermo-mechanical analysis in elastic range is considered. Heat transfer analysis The problem is regarded as one-dimensional transient task occurring along the thickness of the slab. The thermal properties of concrete, i.e. thermal conductivity, density, specific heat are taken accordingly to (EN 1992-1-2, 2004). The equations of the thermal properties proposed by standard are given below. 

Specific heat (Cp) [J/Kg K]: Cp = 900

for T≤100°C

Cp =Cpeak =2020

for T˃100°C to ≤115°C (for moisture content of 3 % of concrete weight)

Cp =2020+

1000-2020 *(T-115) 85

for T˃115°C to ≤200°C

Cp =1000+

T-200 2

for T˃200°C to ≤400°C

Cp=1100 

for T˃400°C to ≤1200°C

Thermal conductivity (λ) [W/m K]: T +0.0107*(T/100)2 100

λ=5.324-0.02604*T

for T˃140°C to ≤160°C

T

Specific heat Cp [J/Kg K]

λ=1.36-0.136* 100 +0.0057*(T/100)2 2500 2000 1500 1000 500 Cp (moisture 3%), (EN 1992-1-2, 2004)

0 0

200

400 600 800 Temperature (°C)

for T≤140°C

1000 1200

Figure 5.4: Specific heat of concrete at temperature, (EN 1992-1-2, 2004)

for T˃160°C Conductivity λ (W/m K)

λ=2-0.2451*

3 2 2 1 1 λ, (EN 1992-1-2, 2004)

0 0

200


1000

1200

Figure 5.5: Thermal conductivity of concrete at temperature (EN 1992-1-2, 2004)

The density is assumed of 2300 kg/m3 and remains constant during heating. The coefficient of exchange by convection and radiation to the exposed face and on the unexposed face are adopted from standards (EN 1991-1-2, 2002) and (EN 1992-1-2, 2004). The surface emissivity of concrete specimen is taken as εm = 0.7, while coefficient of heat transfer by convection = 25 W/m2K for heatexposed side and at non-exposed side a reduced convection coefficient is 9 W/m2K). The gas temperature in the boundary follows the temperature-time ISO 834-1 scenario for 2 hours, Fig. 5.6. The ambient temperature is assumed as 20 °C. The C3D8T finite elements from coupled temperature displacement family with FEM mesh of # 0.02 m x 0.02 m are used, Fig. 5.8



Temperature (°C)

1200 1000 800 600 400 ISO 834-1

200 0 0

30

60 Time (min)

90

120

Figure 5.6: Fire scenario of ISO 834-1, (ISO 834-1, 1999)

The obtained results of temperature distribution at different depth over time, are compared to experimental data gathered from research program performed at French laboratory Centre Scientifique et Technique du Bâtiment – France (CSTB), (CSTB, 2014). The comparison of numerical results with experimental data at different depth is presented in Fig. 5.7 while visualization of result after 2 hours of heating is given in Fig. 5.8.

Figure 5.7: Temperature distribution in concrete slab over time at different depths. Comparison of numerical results with experimental data obtained in laboratory CSTB in France, (CSTB, 2014)

Figure 5.8: FEM mesh and results of heat transfer analysis for time t = 120 min. One-dimensional temperature distribution in concrete slab. NT11 - node temperature (°C).

Thermo-mechanical analysis of the influence of cold rim The problem of presence of cold rim in concrete slab subjected to elevated temperature is modeled. The fire exposed surface of 0.9 m x 0.9 m in area, are surrounded by unexposed concrete rim of various thicknesses: 10 cm, 20 cm and 30 cm. In such case, the expansion of inner surface subjected to fire load is limited by cold unexposed surface. Therefore, the additional stresses are believed to be present in inner part of concrete slab. The aim of such analysis is to confirm higher stresses in concrete specimen with cold rim and check the influence of its extent. In order to solve thermo-mechanical problem, the mechanical properties of concrete under fire load are implemented. For elastic problem, the following data are implemented: linear expansion of concrete according (EN 1992-1-2, 2004), modulus of elasticity and compressive strength, after (Hager, 42

| Chapter 5: Preliminary research 2014) and constant Poisson ratio of 0.30. The input data are presented in Fig. 5.9, Fig. 5.10 and Fig. 4.11 for coefficient of thermal expansion, modulus f elasticity and compressive strength at elevated temperature, respectively. The coefficients of thermal expansion at different temperature are calculated on the basis of free thermal strain for unloaded specimen given in (EN 1992-1-2, 2004).

2.50E-05 2.00E-05 1.50E-05 1.00E-05 5.00E-06 ΔL/(L*ΔT), (EN 1992-1-2, 2004)

0.00E+00 0

200

400 600 800 1000 1200 Temperature (°C)

E, (Hager, 2014)

0

200


1000

1200

Figure 5.10: Modulus of elasticity of concrete at temperature, (Hager, 2014) Compressive strength (MPa)

Figure 5.9: Coefficient of linear thermal expansion of concrete (EN 1992-1-2, 2004)

35 30 25 20 15 10 5 0

Modulus of Elasticity E (GPa)

Linear expansion coefficient ΔL/(L*ΔT)

The thermal problem is the same as described in paragraph Heat Transfer Analysis. The C3D8T finite elements from coupled temperature displacement family with FEM mesh of # 0.02 m x 0.02 m are used, Fig. 5.8

80 60 40 20 fc, (Hager, 2014) 0 0

200


1000

1200

Figure 5.11: Compressive strength of concrete at temperature, (Hager, 2014)

The parametrical studies consist of analysis of trajectories and level of stresses induced by cold rim of concrete surrounding the inner part of concrete that is subjected to heating. The three cases of different thickness of cold rim, namely 10 cm, 20 cm and 30 cm are taken under consideration and compared to the results obtained for concrete slab without cold rim (heated on the entire bottom surface). The boundary conditions assume a zero displacement in vertical direction for the area of cold rim and zero displacement in both horizontal directions in four edges of the slab. In case of slab without cold rim, only displacement in edges are blocked. In order to compare the results, the reference point is taken as the central point in the surface subjected to fire. The values of individual data are evaluated for temperature 200 ˚C what enables the homogenization of conclusions. The schemes of four cases are presented in Table 5.4 . For each individual case the reference point RP, the size of modeled specimen and heated area are indicated. Moreover, the maximum and minimum values of stresses and pressures are given in red and blue, respectively and also the obtained value for stress, pressure and strain at 200 ˚C is given in black. The obtained results for stress, pressure and strains can be assessed by comparing the relative values for each thickness of cold rim. In Table 5.4 it can be observed that the presence of cold rim affects significantly the development of stresses in the inner of concrete slab. While comparing Katarzyna Mróz | Dissertation Outline | 43

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | numerically the influence of different cold rim thicknesses on stress development, it can be concluded that the most effective restraint may be obtained with the use of the 20 cm cold rim. Table 5.4: Comparison of results for parametrical studies of cold rim effect Size of specimen (mm x mm x mm)

1000 x 1000 x 150

1000 x 1000 x 150

1200 x 1200 x 150

Pressure (MPa)

1400 x 1400 x 150

0.04 0.03 0.02 0.01 0 -0.01 0 -0.02 -0.03 -0.04 -0.05

Heated area (mm x mm)

1000 x 1000

900 x 900

900 x 900

900 x 900

Stress (MPa)

400

600

0.0231

1.00

-0.0340

1.00

-0.0278

1.00

-0.0234

1.00

0.0158

1.00

0.0557

1.32

0.0316

1.37

-0.0473

1.39

-0.037

1.33

-0.0335

1.43

0.0224

1.42

0.0577

1.37

0.033

1.43

-0.0494

1.45

-0.0384

1.38

-0.0362

1.55

0.0242

1.53

0.0581

1.38

0.0332

1.44

-0.0498

1.46

-0.0387

1.39

-0.0364

1.56

0.0243

1.54

800

1000

2.5 Strain (mm)

Strain (mm)

Relative Strains

1.4106

1.00

1.6903

1.20

1.7841

1.26

1.7851

1.27

0.08 0.06 0.04 0.02 0 -0.02 0

200

400

600

800

1000

-0.04

3 2 1.5 1 0.5 0 400 600 800 1000 Temperature (°C) Figure 5.14: Strain - temperature diagram at reference point RP

44

Relative Pressures

1.00

Temperature (°C) Figure 5.12: Pressure - temperature diagram at reference point RP

0

Pressure (MPa)

0.0422

No cold rim Cold rim 10 cm Cold rim 20 cm Cold rim 30 cm

200

Relative Stresses

Stress (MPa)

Scheme of case

200

-0.06

Temperature (°C) Figure 5.13: Stress - temperature diagram at reference point RP

| Chapter 5: Preliminary research Spalling phenomenon is commonly observed at temperatures under 200 ˚C. In the Figures 5.12 5.14 diagrams for pressure, stress and strain development at temperature are presented. It can be clearly observed that both pressure, stresses and strain changes significantly between 100 ˚C and 200 ˚C. The differences in pressure and stress development for slab without and with cold rim are due to the limited deformability in the second case, that can be seen in Figure 5.15. As it can be seen in Table 5.4 and Figures 5.12 - 5.14 the relative stresses increase in ca. 50% while strains in only 25% while considering cold rim presence. Therefore the additional impact in stress development shall be considered while the design of the spalling risk.

a)

b) Figure 5.15: Deformability at 200 ˚C for a) slab without cold rim - free deformations; b) slab with cold rim of 20 cm - limited deformations.

Such analysis is aimed to give the guidelines for experimental tests. In case of the cold rim effect, the general conclusion can be drawn as follow: there is no need to test experimentally different thicknesses of cold rim. The experiment is then limited to the slab without cold rim and with cold rim of 20 cm in thickness. The more detailed in case of cold rim and remaining parametrical cases are the subject of further consideration.

5.3

CONCLUSIONS ON PRELIMINARY STUDIES

The preliminary test results carried out on slabs heated from one side using DRAGON furnace has presented fire examination of spalling severity consisted in measuring spalling depth, spalled concrete volume and also the ratio of the fallen off concrete to sample size. On the basis of the results of the conducted research the authors can formulate the conclusions and observations. The feasibility of proposed research objectives can be summarized as follow:  The extent of preliminary research indicate that more dense and less permeable concrete, made of CEM I and w/c ratio of 0.3 is more susceptible to spalling behaviour;  While testing unloaded slab, the influence of cold rim was indicated. Also the similar form of crack propagation as in (Hertz, 2003) was noted experimentally;


ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE |  The results have confirmed that spalling is more likely to occur for denser and less permeable concretes however the differences in initial material permeability (kc) has no straightforward influence on material behaviour in fire, thus other parameters influencing spalling, such as stress state should be tested;  The authors has developed heating set up the Dragon furnace. However, in order to consider the effect of load type it would be sufficient to redesign the setup for enabling testing of concrete slab of size 100 m x 100 m x thickness. It will enable to provide a symmetrical stress state and limit the misinterpretation of the obtained results;  The Dragon furnace is about to be equipped with loading frame, described in paragraph 4.2. It will enable testing slab in compression with different level of load. The load can be applied uniaxially or bi-axially. Presently, the loading frame is designed to be instrumented with the flat jacks and loading shoes ending with brushes that provide a friction limitation;  The preliminary parametrical studies of cold rim indicate that the thickness of cold rim may induce additional stresses in the concrete part subjected to fire. Hence cold rim presence shall be taken into consideration while assessment of spalling risk.

6 REFERENCES Fire Desing of Concrete Structures - Materials, Structures and Modelling, Bulletin 38, Lausanne: Fédération Internationale du Béton, 2007 Bazant, Z. & Kaplan, M., 1996. Concrete at High Temperatures: Material and Mechanical Models. London: Longman (Addison-Wesley). Boström, L., 2004. Innovative self-compacting concrete - Development of test methodology for determination of dire spalling, SP Fire Technology. SP Sweden. Report 06. Boström, L. & Jansson, R., 2008. Self-Compacting Concrete Exposed to Fire, Borås: SP Technical Research Institute of Sweden. Boström, L. & Jansson, R., 2008. Self-Compacting Concrete Exposed to Fire, Borås: SP Technical Research Institute of Sweden . Carré, H. i inni, 2013. Effect of compressive loading on the risk of spalling. MATEC Web of Conferences, 6(01007). Connolly, R., 1995. The Spalling of Concrete in Fires. PhD Thesis. The University of Aston in Birmingham. CSTB, 2014. Synthesis Benchmark Vulcain tests on 3 Walls. Champs-Sur-Marne, France: Centre Scientifique et Technique du Batiment| Safety, Structures and Fire Department . Debicki, G., Haniche, R. & Delhomme, F., 2012. An experimental method for assessing the spalling sensitivity of concrete mixture submitted to high temperature. Cement and Concrete Composites, 34(8), pp. 958 - 963. EN 1991-1-2, 2002. Eurocode 1: Actions on structures - Part 1-2: General actions - Actions on structures exposed to fire, [Authority: The European Union Per Regulation 305/2011, Directive 98/34/EC, Directive 2004/18/EC]: (English). EN 1992-1-2, 2004. Eurocode 2: Design of concrete structures - Part 1-2: General rules - Structural firedesign. (English): [Authority: The European Union Per Regulation305/2011, Directive 98/34/EC, Directive 2004/18/EC]. Gunnarsson, J. G., 1998. Eksplosiv afskalning af beton (Explosive spalling of concrete). M.Sc. Thesis, Department of Buildings and Energy [in Danish].

46

| Chapter 5: Preliminary research Hager, I., 2004. Comportement à haute température des bétons à haute performance - évolution des principales propriétés mécaniques, l’Ecole Nationale des Ponts et Chaussées, Politechnika Krakowska. Hager, I., 2014. Wieloparametrowa diagnostyka stanu betonów cementowych poddanych działaniu temperatury pożarowej. Sprawozdanie merytoryczne z realizacji projektu badawczego własnego N N506 045040, Kraków: Katedra Technologii Materiałów Budowlanych i Ochrony Budowli. Politechnika Krakowska. Hager, I., Carre, H. & Krzemień, K., 2013. Damage assessment of concrete subjected to high temperature by means of ultrasonic pulse velocity method. Studies and Research (Studi e Ricerche), Scuola di Specializzazione per le Costruzioni in C.A, Flli Pesenti, Politecnico di Milano. Hager, I. & Krzemień, K., 2013. Metoda impact-echo - ocena przydatności w diagnozowaniu działania wysokiej temperatury na beton. Przegląd Budowlany, Tom 12, pp. 57-63. Hager, I. & Krzemień, K., 20th-23rd April 2015. An overview of concrete modulus of elasticity evolution with temperature and comments to European code provisions. IFireSS – International Fire Safety Symposium Coimbra, Portugal, pp. 703-712. Hager, I. & Pimienta, P., 2004. Mechanical properties of HPC at high temperatures. Milan, Italy, pp. 95-100. Hager, I. & Tracz, T., 2010. The impact of the amount and length of fibrillated polypropylene fibres on the properties of HPC exposed to high temperature. Archives of Civil Engineering, LVI(1), pp. 57 68. Hager, I. & Tracz, T., 2015. Parameters influencing concrete spalling severity - intermediate scale tests results. Leipzig, IWCS - 4th International Workshop on Concrete Spalling due to Fire Exposure. Hager, I., Tracz, T. & Krzemień, K., 2014. Usefulness of selected non-destructive and destructive methods in the assessment of concrete after fire. Cement Wapno Beton, Tom 3, pp. 145-151. Hager, I., Tracz, T., Śliwiński, J. & Krzemień, K., 2015. The influence of aggregate type on the physical and mechanical properties of high-performance concrete subjected to high temperature. Fire and Materials, Issue DOI: 10.1002/fam.2318. Hager, I., Zdeb, T. & Krzemień, K., 2013. The impact of the amount of polypropylene fibres on spalling behaviour and residual mechanical properties of Reactive Powder Concrete. MATEC Web of Conferences, 6(DOI: 10.1051/matecconf/20130602003), p. 02003. Han, C., Hwang, Y., Yang, S. & Gowripalan, N., 2005. Performance of spalling resistance of high performance concrete with polypropylene fiber contents and lateral confinement. Cement and Concrete Research, Tom 35, pp. 1747 - 1753. Heel, A. & Kusterle, W., 2004. Die Brandbeständigkeit von Faser-, Stahl- und Spannbeton [Fire resistance of fiber-reinforced, reinforced, and prestressed concrete] (in German), Tech. Rep. 544, Vienna: Bundesministerium für Verkehr, Innovation und Technologie. Hertz, K. D., 2003. Limits of spalling of fire-exposed concrete. Fire Safety Journal, Tom 38, pp. 103116. Hertz, K. & Sørensen, L., 2005. Test method for spalling of fire exposed concrete. Fire Safety Journal, Tom 40, pp. 466-476. Huismann, S., Korzen, M., Weise, F. & Meng, B., 2011. Concrete spalling due to fire exposure and the influence of polypropylene fibres on microcracking. Delft, The Netherlands: RILEM Publications S.A.R.L., E.A. Koenders, E.A. and Dehn, F. (Ed.), Concrete Spalling due to Fire Exposure - Proceeding of the 2nd International RILEM Workshop..


ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | Huismann, S., Weise, F., Meng, B. & Schneider, U., 2012. Transient strain of high strength concrete at elevated temperatures and the impact of polypropylene fibers. Materials and Structures, Tom 45, pp. 793-801. Iglesias, E. & Wetzig, V., 2009. Influence of porosity and specimen stiffness on the fire resistance of pp fibre mixed concrete. Budapest, ITA-AITES World Tunnel Congress. ISO 834-1, 1999. Fire-resistance tests -- Elements of building construction -- Part 1: General requirements. Jansson, R., 2013. Fire Spalling of Concrete. Theoretical and Experimental Studies. PhD Thesis. Borås: KTH Architecture and the Build Environment. Jansson, R. & Boström, L., 2008. Experimental study of the influence of polypropylene fibres on material properties and fire spalling of concrete. 3rd International Symposium on Tunnel Safety and Security (ISTSS), Stockholm, Sweden. Jansson, R. & Boström, L., 2008. Spalling of concrete exposed to fire, Borås: SP Technical Research Institute of Sweden. Jansson, R. & Boström, L., 2009. Fire spalling - the moisture effect, MFPA Institute Leipzig, Germany: 1st International Workshop on Concrete Spalling due to Fire Exposure - From Real Life Experiences to Practical Applications to Lab-scale Investigations and Numerical Modelling. Kalifa, P., Chene, G. & Galle, C., 2001. High-temperature behaviour of HPC with polypropylene fibres. From spalling to microstructure. Cement and Concrete Research , Tom 31, pp. 1487 - 1499. Kalifa, P., Menneteau, F. & Quenard, D., 2000. Spalling and pore pressure in HPC at high temperature. Cement and Concrete Research, Tom 30, pp. 1915-1927. Khoury, A. & Anderberg, Y., 2000. Concrete spalling - review, Fire Safety Design. Khoury, G. A., 2005. Spalling review, Fire Safety Design. Final report. Khoury, G. A., 2008. Polypropylene fibres in heated concrete. Part 2: Pressure relief mechanisms and modelling criteria. Magazine of Congrete Reserach, 60(3), pp. 189 - 204. Krzemień, K. & Hager, I., 2015. Assessment of concrete susceptibility to fire spalling: A report on the state-of-the-art in testing procedures. Procedia Engineering, Tom 108, pp. 285-292. Krzemień, K. & Hager, I., 2015. Post-fire assessment of mechanical properties of concrete with the use of the impact-echo method. Construction and Building Materials, Tom 96, pp. 155-163. Krzemień, K., Pimienta, P., Pinoteau, N. & Hager, I., 2015. Moisture effect on mechanical behaviour of concrete at high temperature and its implication on fire spalling. 4 th International Workshop on Concrete Spalling due to Fire Exposure. 8-9 October , Leipzig, Germany, pp. 165-176. Majorana, C., Salomoni, V., Mazzucco, G. & Khoury, G., 2010. An approach for modelling concrete spalling in finite strains. Italy, pp. 1684-1712. Mayer-Ottens, C., 1974. Behaviour of concrete structural members in fire consitions (in German). Beton, Tom 4, pp. 133 - 136. Meyer-Ottens, C., 1972. The question of spalling of concrete structural elements of standard concrete under fire loading, Germany: PhD Thesis, Technical University of Braunschweig. Mindeguia, J., 2009. Contribution expérimentale a la compréhension des risques d'instabilité thermique des bétons. L'Université de pau et des pays de l'Adour. PhD Thesis, pp. 115-120. Phan, L., 2005. High-strength Concrete at High Temperature - An Overwiev. Vancouver, International Conference on Construction Materials. Phan, L., 2007. Spalling and mechanical properties of high strength concrete at high temperature. France.

48

| Chapter 5: Preliminary research Phan, L. & Carino, N., 2002. Effects of Test conditions and Mixture Proportions on Behaviour of High-Strength Concrete Exposed to High Temperatures. ACI Mater J, 99(1), pp. 54-66. Richter, E., 2004. Fire test on single-shell tunnel segments made of a new high-performance fireproof concrete, pp. 261-270, Workshop: Fire Design of Concrete Structures: What now? What next?. RILEM TC 200-HTC, 2007. Mechanical concrete properties at high temperatures - modelling and applications. Part 2: Stress-strain relation. Materials amd Structures, Tom 40, pp. 855-864. Schneider, U., 1982. Behaviour of concrete at high temperatures. Paris: RILEM, Report to Committee 44-PHT, p. 72. Shorter, G. & Hermathy, T., 1961. Discussion on the Fire Resistance of Prestresses Concrete Beams. Proceedings of the Institution of Civil Engineering, Tom 20, p. 313. Sullivan, P., 2001. Deterioration and spalling of high strength concrete under fire, City University London: Report for UK Health & Safety Executive. Taillefer, N., Pimienta, P. & Dhima, D., 2013. Spalling of concrete: A synthesis of experimental tests on slabs. MATEC Web of COnferences, 6(01008). Tanibe, T. i inni, 2013. Thermal stress estimation in relation to spalling of HSC restrained with steel rings at high temperatures. MATEC Web of Conferences, 6(01004). Tanibe, T. i inni, 2011. Explosive spalling behaviour if restrained concrete in the event of fire. Delft, The Netherlands: RILEM Publications S.A.R.L., p. 319-326, E. A. Koenders, E.A. and Dehn, F. (Ed.), Concrete Spalling due to Fire Exposure - Proceeding of the 2nd International RILEM Workshop.. Tatnall, P., 2002. Shortcrete in Fires: Effects of fibers on explosive spalling. Shortcrete, pp. 10-12. tunneltalk.com, 2015. [Online], available at: http://www.tunneltalk.com/



7 PLAN ROZPRAWY DOKTORSKIEJ 7.1

TEMAT I CEL PRACY

Planowana rozprawa doktorska dotyczy określenia wpływu stanu naprężeń na występowanie eksplozyjnego odpryskiwania betonu w warunkach działania ognia. Program badawczy ma na celu wykazanie istnienia związku między poziomem obciążenia płyt betonowych oraz stanem naprężeń ściskających i intensywnością oraz charakterystyką obserwowanego zjawiska eksplozyjnego łuszczenia się powierzchni badanego elementu poddanego działaniu ognia. Celem projektu jest potwierdzenie następujących hipotez badawczych:   

 

7.2

Naprężenia ściskające przyczyniają się do powstania zjawiska łuszczenia w betonach wyższych klas poddanych działaniu ognia; poziom naprężeń może wpływać na charakter powstających złuszczeń; Zewnętrzne skrępowanie/ograniczenie odkształceń termicznych ogrzewanej płyty betonowej jest wystarczające do wywołania naprężeń ściskających we wnętrzu elementu betonowego, co skutkuje zjawiskiem łuszczenia ogrzewanej powierzchni; Nieogrzewana krawędź płyty betonowej (ang., cold rim) powoduje ograniczenie odkształceń termicznych ogrzewanej części betonu, a tym samym może być uważana za dodatkowy czynnik wewnętrznego skrępowania elementu, które potęguje ryzyko powstania zjawiska łuszczenia; Istnieje szeroki zakres podobieństw w obserwowanych zjawiskach łuszczenia wynikających z zastosowania różnych warunków brzegowych, powodujących wprowadzenie naprężeń ściskających; Możliwe jest, zaproponowanie ujednoliconej procedury prowadzenia badań podatności betonu na zjawisko eksplozyjnego łuszczenia.

PRZEGLĄD LITERATURY

Przegląd literaturowy dotyczący wpływu naprężeń na podatność betonu na odpryskiwanie wykazał, że różne stany naprężeń wprowadzone do elementu betonowego istotnie wpływają na intensywność i głębokość obserwowanych odprysków w czasie pożaru. Wśród licznych programów badań poświęconych zjawisku odpryskiwania betonu w pożarze można wymienić różne konfiguracje badań, które wpływają istotnie na rozkład naprężeń w badanych elementach. Można tu wymienić następujące podejścia badawcze: badanie nieobciążonych elementów, elementy jedno- i dwukierunkowo ściskane, elementy ze skrępowaniem zewnętrznym oraz elementy z tzw. zimnym brzegiem (ang. cold rim) powodującym wewnętrzne skrępowanie elementu. Przeprowadzony przegląd literatury pozwala stwierdzić, iż mimo licznych prac i programów badań poświęconych tematyce odpryskiwania betonu, jak dotąd nie przeprowadzono kompleksowego programu badawczego nad wpływem naprężeń na podatność betonu na odpryskiwanie w pożarze. Pomimo faktu, iż spośród przeprowadzonych dotychczas programów badań można wyszczególnić i opisać różne stany naprężeń oraz ich skutek, nie ma żadnego związku między nimi ze względu na różnorodność podejść badawczych. Z tego powodu umieszczenie wyraźnych wniosków dotyczących postawionego problemu nie jest możliwe bez przeprowadzenie kompleksowego programu badań.

7.3

OGÓLNY PLAN BADAŃ

Planowane badania przebiegać będą dwuetapowo. W pierwszym etapie zostaną zrealizowane symulacje numeryczne, których celem będzie określenie stanu naprężenia, który odpowiada różnym sposobom obciążenia płyt, przy uwzględnieniu sposobu ich ogrzewania oraz wpływu warunków 50

| Chapter 7: Plan rozprawy doktorskiej brzegowych. Przeprowadzone symulacje numeryczne mają na celu ułatwić dobór warunków realizacji badań eksperymentalnych i poziomu obciążeń, które zostaną zastosowane podczas badań eksperymentalnych. W kolejnym etapie projektu, zostaną zrealizowane badania na elementach płytowych o powierzchni 1 m2 i grubości 0,15 m . Wpływ stanu naprężeń na występowanie zjawiska eksplozyjnego zachowania się betonu w warunkach działania ognia zostanie określony poprzez wprowadzenie naprężeń ściskających do płyty na drodze zastosowania trzech podejść A, B lub C. Podejście A będzie stanowiło analizę wpływu nieogrzewanej krawędzi betonu na wartość naprężeń ściskających obecnych w ogrzewanej części elementu. W podejściu B naprężenia ściskające zostaną wprowadzone poprzez zewnętrzne skrępowanie płyty betonowej stalową ramą. Podejście C będzie stanowiło analizę wpływu różnych poziomów naprężeń ściskających na obserwowane zjawisko eksplozyjnego złuszczania się betonu poprzez wprowadzenie zewnętrznych sił ściskających. W każdym podejściu zostanie określona relacja między poziomem i naturą wprowadzonego stanu naprężenia, a obserwowanym charakterem i głębokością eksplozyjnych złuszczeń. 7.3.1

ANALIZA NUMERYCZNA

Modelowanie numeryczne zachowania się betonu w różnych stanach naprężenia w warunkach działania wysokiej temperatury będzie prowadzone przez wykorzystanie metody elementów skończonych (MES) w dwóch etapach. Pierwszy etap będzie polegał na analizie przepływu ciepła przez element betonowy poddany jednostronnemu działaniu ognia. W drugim etapie zostanie przeprowadzona analiza termomechaniczna w różnych stanach obciążenia w zakresie sprężystym. Oprócz płyty nieobciążonej modelowane obejmować będzie trzy sposoby wprowadzenia naprężeń ściskających do płyty, analogicznie do wspomnianych podejść A, B, C: A) element nieobciążony, ogrzewany w centralnej części powierzchni, B) element skrępowany zewnętrzną stalową ramą, C) element jedno- lub dwuosiowo ściskany obciążeniem zewnętrznym. Zmiany właściwości mechanicznych betonu nieobciążonego będą modelowane w odniesieniu do wyników badań uzyskanych w poprzednim projekcie finansowanym z NCN. Dla betonu obciążonego, dane do obliczeń zostaną zaczerpnięte z badań (Hager, 2004; Carré, et al., 2013) oraz wyników badań przeprowadzonych w laboratorium CSTB we Francji. Modelowanie numeryczne zostanie przeprowadzone w programie ABAQUS. 7.3.2

BADANIA DOŚWIADCZALNE

W kolejnym etapie projektu, zostaną zrealizowane badania doświadczalne. Badania realizowane będą za pomocą pieca wysokotemperaturowego DRAGON wykonanego ze środków uzyskanych w ramach zakończonego projektu NCN (Umowa N N506 045040). Istniejący w laboratorium Politechniki Krakowskiej system umożliwia jednostronne ogrzewanie nieobciążonych płyt betonowych o wymiarach 1,2 m x 1,0 m x grubość spoczywających na zewnętrznych krawędziach pieca. Płyta poddana jest działaniu termicznemu o przebiegu zbliżonym do nominalnej krzywej temperatura-czas, według zaleceń ISO 834-1. Piec wymagać będzie dostosowania do ogrzewania płyt kwadratowych 1,0 x 1,0 m i kalibracji palników, które to zadania zrealizowane zostaną w ramach projektu. Zastosowanie próbek badawczych o powierzchni kwadratowej pozwoli na uzyskanie warunków symetrii, co ułatwi zarówno odzwierciedlenie pracy płyty w modelu numerycznym, jak i instrumentację i przeprowadzenie pomiarów w badaniach eksperymentalnych. Jedno- lub dwuosiowe ściskanie płyty, w podejściu C, będzie zadawane poprzez system siłowników hydraulicznych opracowany przez zespół przy wsparciu finansowym ze strony Wydziału. Siłowniki zakończone zostaną szczotkami wykonanymi z prętów, które umożliwią ograniczenie tarcia w strefie stykowej i realizację czystego ściskania. Stan naprężeń w badanych elementach przed Katarzyna Mróz | Dissertation Outline | 51

ASSESSMENT OF SPALLING RISK IN CONCRETE SUBJECTED TO FIRE | rozpoczęciem ogrzewania zostanie pomierzony czujnikami tensometrycznymi a rozkład temperatury w przekroju płyty podczas ogrzewania termoparami typu K. W ramach badań doświadczalnych zostaną przebadane płyty wykonane z betonu wysokowartościowego oraz zrealizowane zostaną badania materiałowe zmierzające do określenia stałych materiałowych dla badanego betonu, jak również określone zostaną zmiany tych właściwości spowodowane ogrzewaniem. Badania eksperymentalne umożliwią obserwację intensywności i przebiegu eksplozyjnego zachowania się betonów dla wybranych stanów naprężeń.

52