Steel Fibers and Mechanical Behavior of Refractory ...americanceramicsociety.org/bulletin/2006_pdf_files/Peret.pdfSteel Fibers and Mechanical Behavior of Refractory Castables on Drying

©The American Ceramic Society American Ceramic Society Bulletin www.ceramicbulletin.org January 2006 9401

Steel Fibers and MechanicalBehavior of Refractory Castables

on DryingThe mechanical and drying behavior of refractory castablesthat contained short and long steel fibers was comparedwith castables that contained polyaramid fibers.

C.M. Peret and V.C. PandolfelliMaterials Microstructural Design Group, Dept. of Materials Engineering, Federal University of São Carlos, São Carlos, S.P., Brazil

T he heat-up of refractory castables has been the subject of extensive researchbecause of the importance manufacturers and users attribute to this stage of their pro-cessing and its impact on the productivity and quality of end products. The develop-ment of compositions capable of withstanding severe heating schedules minimizes therisk of spalling, which decreases the downtime of metallurgical equipment for mainte-nance and repair.

Polymeric fibers are an efficient way to enhance castable resistance to spalling upondrying, because they allow more severe heating schedules under a controlled risk. Twomechanisms are responsible for the benefits conferred by these fibers: an increase inpermeability caused by fiber melting, thermal degradation or shrinkage,1 as in the caseof polypropylene (PP) fibers; and mechanical reinforcement, which results especiallyfrom the increased energy dissipated during crack propagation,2 as when polyaramid(PAr) fibers are used.

Whenever fibers are added as a reinforcing particle, they should be selected in such away that their elastic properties are kept on a suitable level within the temperaturerange where the risk of damage is high. Previous studies1,3 have shown that, for low-cement high-alumina castables, this range is located between 150 and 200°C, taking asreference the surface temperature of the castable. Therefore, PAr fiber, a fiber that pre-sents thermal degradation only above 300°C, has shown positive results in preventingexplosive spalling on drying.2

In addition to polymers, many other materials have been successfully used asmechanical reinforcements to produce tough cementitious composites for applicationssuch as tunnels, pavements, monolithic and preshaped refractories, and advancedceramics. In the cases of the construction and refractories industries, carbon, naturaland stainless-steel fibers in varying sizes and shapes usually have been applied for rein-forcement at room and high temperatures.4

Steel fibers benefit mechanical reinforcement in many applications, and this mecha-nism has been identified as the determining factor of spalling resistance on drying.Therefore, the purpose of this work is to determine the efficacy of long (25 mm) andshort (2–6 mm) stainless-steel fibers as a drying reinforcement.

Castable PreparationA high-alumina ultra-low-cement refractory castablecomposition was designed for this study, followingAndreasen’s particle distribution model (q = 0.24). Theparticle size ranged from 0.1 to 4750 µm, and 2 wt%of calcium aluminate (CA) cement was added to thecomposition (Fig. 1). To disperse the particles in water(4.2 wt%), 0.05 wt% of citric acid was used.

Four types of metal fibers with varying lengths wereselected for this study. The fibers were made of AISI434 stainless steel (Gervois, Stax 130 and Stax 225) or 304 stainlesssteel (Daoli). However, the fibers were produced by differentprocesses: the Gervois, Stax 130 and Stax 225 fibers were made ofchopped drawn wire, whereas the Daoli fiber was obtained by meltextraction (Table 1 and Fig. 2).

For comparison, a 6 mm long PAr fiber with a 20 µm diameter alsowas used. These fibers have a high thermal stability, because theirelastic modulus is not decreased by degradation in the critical dry-ing temperature range (150–200°C), as shown in a previous report.2

All the fibers were added to the castable at a volumetric ratio of 0.36%, before the addition ofwater, to inhibit fiber agglomeration. The mixing and water addition steps took place in arheometer for castables developed by the authors’ research group (GEMM).5 It consisted of thefollowing procedure (Table 2): after 60 s of dry-mixing time, 3.6 wt% (12.5 vol%) of water wasadded at a constant rate; the remaining water, which amounted to a total of 4.5 wt% (15 vol%),was added only after a constant mixing torque (~160 s of mixing time) was attained.

Specimens were cast by vibration in the following geometries: 40 mm high � 40 mm diame-ter cylinders for drying and splitting tensile strength; 25 � 25 � 150 mm prismatic specimenswith an embedded notch 1 mm deep � 1 mm wide for work-of-fracture measurements; and75 mm diameter � 22 mm high cylinders for permeability measurements.

The specimens were cured at 8°C for 7 d in a moisturized environment. However, the speci-mens for permeability testing were cured for the same time and at the same humidity, but at50°C. The temperature of 8°C was chosen because hydration of the cement phases leads to the

formation of CAH10 (CaO·Al2O3·10H2O),which has a higher molar volume. As aconsequence, the castable permeabilityis decreased and a critical condition fordrying is attained. The purpose of thepermeability measurements in thiswork was only to verify the effect of theaddition of various fibers and the rela-tive difference between the composi-tions before and after a heat treatment.Therefore, curing of the specimens forpermeability tests was conducted at50°C for 7 d.

A mechanical evaluation was madebased on splitting tensile strength(ASTM C406-96) and work-of-fracture(γwof) measurements, both at room

©The American Ceramic Society American Ceramic Society Bulletin www.ceramicbulletin.org January 2006 9402

Table 1. Castable Fiber Length, Diameter, Number

Concentration Diameter Length N/VC

Fiber (vol% [wt%]) (mm) (mm) (cm–3)

Polyaramid† 0.36 [0.12] 20 6 1909.9

Gervois‡ 0.36 [0.66] 70 2 467.7

Stax 130§ 0.36 [0.66] 130 4 67.8

Stax 225§ 0.36 [0.66] 180 6 23.6

Daoli¶ 0.36 [0.66] 450 25 0.9†DuPont AFS (USA). ‡Gervois S.A. (France). §Stax AG (Germany). ¶Daoli Ltda (Brazil–China).

Table 2. Castable Mixing Procedure

Mixing time Event

0 s Start

60 s First addition of water

(3.6 wt% at constant rate)

150–170 s Addition of remaining water

(total of 4.5 wt%)

60 s after finishing End of mixing

water addition (total of 210–230 s)

Figure 2 Length statistical distribution for the steel fibers.

temperature. These tests were conducted immediately after curing, without previous drying, tobetter simulate the true conditions of the material when saturated with water.

The splitting tensile strength (in MPa) is given by the following equation:

σf = (2/π)(P/dh) (1)

where P is the ultimate load (in newtons), and d and h the specimen diameter and height(both in millimeters), respectively.

The work-of-fracture was calculated by integrating the curve of load, P, as a function of theactuator displacement, δ, in a three-point bending test, under a constant displacement rate of20 µm/min:

γwof = (1/2A) ∫P dδ (2)

where A is the projected fracture surface area.

The non-Darcyan permeability constant, k2, was calculated by measuring the flow rate ofcompressed air through a cylindrical specimen of the material, as a function of the inlet airpressure (which varied from 0 to 5 bar (0 to 500 kPa)). The data were treated according toForchheimer’s equation:6

[(Pi2 –Po

2)/2PoL] = (µ/k1)VS + (ρ/k2)VS2 (3)

where Pi and Po are the inlet and outlet (atmosphere) air pressures, respectively; L the height ofthe specimen; µ and ρ the air viscosity and density, respectively; and VS the speed of the airthrough the material, calculated by the measured air flow rate and the cross-sectional area ofthe specimen.6,7 Unlike the mechanical measurements, the permeability was measured inspecimens previously dried at 50°C, and then repeated with the same specimens after theywere heat-treated at 800°C for 12 h.

The mass loss on drying was evaluated using a thermogravimetric apparatus developed bythe authors’ research group.8 This device consisted of an electrical furnace in which a speci-men holder was connected to a balance. The furnace heating elements were shielded againstspattering fragments by enclosing the specimen holder in a stainless-steel cage. The specimenwas heated at 20°C/min, from room temperature to 600°C.

The test can be interpreted using the W(t) evaluation, which is a measure of the cumulativefraction of the water that already has been released from the specimen at time t, in compari-son with the total water initially present:

W(t) = 100[(m0 – m(t))/(m0 – mf)] (4)

where m(t), m0 and mf are the speci-men mass at time t, the initial mass andthe dry-specimen mass, respectively.For the analysis, the instant drying ratewas calculated by deriving W(t) as afunction of time and plotted againstthe furnace temperature.

Rheological BehaviorThe torque profile with the mixing time(Fig. 3) presents some differences whencomparing compositions containingsteel or PAr fibers against compositionswithout fibers. The mixing profilechanges slightly with the addition ofsteel fibers. The Stax 130 and Stax 225

©The American Ceramic Society American Ceramic Society Bulletin www.ceramicbulletin.org January 2006 9403

Figure 3 Torque vs mixing time for the fiber-containing refractory castables(Ref. means reference, i.e., without fibers).

steel fibers present a decrease of ~13% in the time it took to reach the maximum torque, andan increase of ~10% in the final torque was attained during mixing. These changes are mini-mal in comparison with those observed from the addition of PAr fibers.9 For these polymericfibers, the torque is ~33% higher than that of the reference material.

The average mixing effort can be calcu-lated by integrating the torque profilecurves as a function of time (Fig. 4). Ahigher mixing effort value was verifiedwhen processing the PAr-fiber-containingcastable. These polymeric fibers were 6mm long—which is in the same order ofsome steel fibers studied—but presenteda diameter of ~20 µm.

The average number of fibers per unitvolume of castable, N/Vc, can be calculat-ed approximately by dividing the volu-metric ratio of fibers in the castable, p, bythe volume of a single fiber, estimated byconsidering it a perfect cylinder withdiameter d and length L:

N/VC = p/Vsingle fiber = 4p/πd2L (5)

According to their geometrical features, the average number of fibers per unit volume ofcastable, for a constant volumetric ratio of 0.0036 (0.36%), may be 468, 68, 24 and 0.9fibers/cm3 for the Gervois, Stax 130, Stax 225 and Daoli fibers, respectively.

On the other hand, with the PAr fiber, this value increases to ~1909 fibers/cm3. Thus, the dis-tance between the fibers is shorter for the polymeric fibers, and the probability of collisionsand interference during mixing and molding is increased.

The number of fibers, therefore, is an important factor determining the rheological behaviorof the castable during the mixing step. This statement can be extrapolated to material behav-ior during other steps of rheological processing, such as pumping and shotcreting.

Effect of Fibers on PermeabilityNo significant differences in the values of the permeability constants were found among thecastables (Fig. 5) when evaluated in the green stage. After a heat treatment at 800°C for 12 h,

the permeability increased in the fol-lowing order: Gervois < Stax 130 < Stax225 < Daoli. This result possibly wascaused by the mismatch of the thermalexpansion coefficient and the difficul-ties to accommodate these stresses,depending on the fiber geometry.

In the PAr-fiber-containing castable,the increase in permeability after ther-mal treatment was 10-fold greater thanthat of the steel-fiber-containing casta-bles, because the thermal degradationgenerated many permeable channelsthrough the microstructure.1

©The American Ceramic Society American Ceramic Society Bulletin www.ceramicbulletin.org January 2006 9404

Figure 4 Average mixing effort for the fiber-containing castables.

Figure 5 Non-Darcyan permeability constant (k2) of the green and firedfiber-containing refractory castables.

Effect of Fibers on Mechanical PropertiesThe splitting tensile strength of the materials varied from 0.8 to 1.4 MPa (Fig. 6). The Daoli fiber,which is the longest of the steel fibers evaluated, conferred the greatest improvement instrength on the material. However, the addition of PAr fiber had little effect on this property.These results are in agreement with thegeneral literature, which shows that thevolume of short fibers that can be addedto ductile- and brittle-matrix compositesis insufficient to produce a significantimprovement in their elastic modulus andstrength.4,10,11

On the other hand, the mechanicalbehavior of the castables after the begin-ning of crack propagation is greatly influ-enced by the fibers, as indicated by theγwof results (Fig. 7). Among the steel-fiberrefractories, the γwof scaled up with theaverage length of the fibers, although thevalue for the castable containing Gervoisfibers was not altered in comparison withthe reference material. The other steelfibers—Stax 130, Stax 225 and Daoli—produced composites apparently tougherthan the one with PAr fibers. The relativeincrease of this property, obtained withthe addition of Daoli fibers, was ~900%,whereas that of PAr fibers was only 250%.

The increase in energy consumption afterthe beginning of propagation in the fiber-containing composites is driven by thebridging phenomenon.This effect is causedmainly by a crack-wake compressive forcegenerated by fiber anchoring and pullout.12

The low γwof attained with Gervois fiberindicates that there may be a minimumlength for fibers to provide enough anchor-ing in the castable matrix, as well as suffi-cient interfacial friction during pullout to contribute to the dissipation of the stored elastic energy.

Drying BehaviorExplosive spalling occurred in all the steel-fiber-containing castables tested at 20°C/min (Fig. 8).On the other hand, the material containing PAr fibers survived heating at that rate, attesting tothe greater efficiency of these fibers in enhancing refractory resistance to drying and explosion.

The thermogravimetric analysis revealed no significant differences among the mass loss rateprofiles of the materials. Indeed, the structure of the permeable channels through which thewater vapor flows was not greatly modified by the presence of the fibers studied. Neithermelting of PAr and steel fibers nor degradation of PAr fiber occurred below the critical temper-ature range involved in the drying process.

The greater resistance to damage on drying displayed by the PAr-fiber-containing materialindicates that a more severe heating rate can be applied to the material, thereby lessening theoverall time needed for the completion of heat-up.

©The American Ceramic Society American Ceramic Society Bulletin www.ceramicbulletin.org January 2006 9405

Figure 6 Splitting tensile strength for the refractory castables.

Figure 7 Work-of-fracture of the refractory castables containing steel orPAr fibers.

Effect of Distance between FibersThe mechanical effect of fibers upon the drying damage resistance is related to the increase inenergy dissipation with crack propagation through a bridging and pullout mechanism. Theextent of such nonlinear fracture behavior can be inferred by the castables γwof.

13

Nevertheless, the mechanical reinforcement potential can be activated only if many fiberscross the crack propagation front. Therefore, the distance between fibers should be designedto be compatible with the scale of the cracks formed during the first stages of damage.According to Betterman et al.14 closer fibers (microfibers) are capable of lessening the propa-gation of microcracks that precede catastrophic failure, because the mean distance that acrack propagates before being crossed by a fiber is, at most, half of that between the fibers.

The distance between fibers is determined mainly by their geometry and volumetric ratio inthe castable. Considering a homogeneous, cubic distribution of fibers in the microstructure,the shortest distance, S, between them can be estimated by

S = (VC/N)1/3 = (πd2L/4p)1/3 (6)

Equation (6) originates from a model that represents the fibers by their middle points andtakes into account neither the natural variation of diameter and length nor the typical entan-glement of more flexible, higher-aspect-ratio fibers, such as PAr fiber.

According to Eq. (6), the distance between the fibers in the present work follows the orderPAr < Stax 130 < Stax 225 < Daoli (Table 3). This trend was confirmed through visual evalua-tions of the fracture surfaces of the materials (Fig. 9).

According to Griffith’s theory, unstable crack propagation is followedby an increase of kinetic energy.To prevent catastrophic failure, thepropagating front must be crossed by the fibers while the strain ener-gy release rate is low.Therefore, to enhance the drying performance ofrefractory castables, in which only minor microcracking is allowedbefore the material explodes, the fibers must be distributed as close aspossible to each other to provide a useful reinforcement.

That is why PAr fibers were the only ones able to inhibit explosive spalling in the presentstudy. Our results show that 25 mm long steel fibers, which are efficient to enhance refractorythermal shock and fatigue resistance, are not suitable for protecting castables from vaporpressurization at low volume fractions.15 If there are points in the castable microstructure thatare “unprotected” by the mechanical action of fibers, they are vulnerable to explosion.

Fiber Selection FactorsAlthough steel fibers caused significant modifications in castable work-of-fracture and split-ting tensile strength, especially when compared with the polyaramid-containing material, thisbenefit could not ensure their survival under severe drying heating schedules.

Before explosive spalling occurs, microcracks grow and merge in the castable microstructure.Thus, to prevent such critical propagation, the fibers should be distributed as close as possibleto each other, thus increasing the drying resistance of the material. This is not achieved usinglonger or thicker fibers, such as the steel fibers studied here. However, better results may beobtained by increasing their volumetric content.

Our results indicate that values of mechanical strength and work-of-fracture are not the onlyproperties that should be considered when evaluating castables for their resistance to spallingupon drying.

©The American Ceramic Society American Ceramic Society Bulletin www.ceramicbulletin.org January 2006 9406

Table 3. Distance between Fibers†

Fiber Distance, S (mm)

Polyaramid 10.3Gervois 3.5Stax 130 2.5Stax 225 1.3Daoli 0.8

†Containing 0.36 vol% of steel or polyaramid fibers.

Fiber reinforcement can offer advantages provided selection takes into account the followingfactors:

• The fibers must have a minimum length to provide anchoring in the castable matrix.

• The unit volume and volumetric ratio of the fibers should be designed so that the distancebetween them is compatible with the spatial scale of the mechanical loads to which the mate-rial may be subjected.

• The mechanical strength and elasticity of the reinforcing particles (i.e., fibers) should be keptwithin a suitable range at the temperature at which the stress is at its greatest (in the case ofcastable drying, in the range 150–200°C, considering the surface temperature). ■

Acknowledgments

The authors are grateful to FAPESP, Magnesita S.A. and ALCOA-Brazil for their financial support of thiswork, and to Daoli Ltd. (Brazil–China), Stax AG (Germany), Gervois S.A. (France) and DuPont AFS (USA) forproviding the steel and polymeric fibers used.

References1M.D.M. Innocentini, C. Ribeiro, R. Salomão, V.C. Pandolfelli and L.R.M. Bittencourt,“Assessment of MassLoss and Permeability Changes during the Dewatering Process of Refractory Castables ContainingPolypropylene Fibers,” J. Am. Ceram. Soc., 85 [8] 2110–12 (2002).

2C.M. Peret, R. Salomão and V.C. Pandolfelli,“Polymeric Fibers as Additives for the Drying of RefractoryCastables,” J. Tech. Assoc. Refract., Jpn. (Taikabutsu Overseas), 24 [2] 88–92 (2004).

3M.D.M. Innocentini, M.F.S. Miranda, F.A. Cardoso and V.C. Pandolfelli,“Vaporization Processes and PressureBuildup during Dewatering of Dense Refractory Castables,” J. Am. Ceram. Soc., 86 [9] 1500–503 (2003).

4H.G. Jiang, J.A. Valdez, Y.T. Zhu, I.J. Beyerlein and T.C. Lowe,“The Strength and Toughness of CementReinforced with Bone-Shaped Steel Wires,” Compos. Sci. Technol., 60 [9] 1753–61 (2000).

5R.G. Pileggi, V.C. Pandolfelli, A.E.M. Paiva and J. Gallo,“Novel Rheometer for Refractory Castables,” Am.Ceram. Soc. Bull., 79 [1] 54–58 (2000).

6M.D.M. Innocentini, A.R.F. Pardo, V.R. Salvini and V.C. Pandolfelli,“How Accurate is Darcy’s Law forRefractories?” Am. Ceram. Soc. Bull., 78 [11] 64–68 (1999).

7M.D.M. Innocentini, A.R.F. Pardo and V.C. Pandolfelli,“Influence of Air Compressibility on the PermeabilityEvaluation of Refractory Castables,” J. Am. Ceram. Soc., 83 [6] 1536–38 (2000).

8M.D.M. Innocentini, C. Ribeiro, J. Yamamoto, A.E.M. Paiva, V.C. Pandolfelli, L.R.M. Bittencourt and R.P.Rettore,“Drying Behavior of Refractory Castables,” Am. Ceram. Soc. Bull., 80 [11] 47–56 (2001).

9R. Salomão, V. Domiciano, C. Isaac, R.G. Pillegi and V.C. Pandolfelli,“Mixing Step and Permeability ofPolymeric-Fiber-Containing Refractory Castables,” Am. Ceram. Soc. Bull., 83 [1] 9301–9308 (2004).

10E. Absi (org.),“Béton de Fibres: Synthèse dês Etudes et Recherches Réalisées au CEBTPæ; pp. 85–127,Annalles de l’Institut Technique du Batîment et des Travaux Publics, No. 250, Jan. 1994.

11A.M. Pallière (org.),“Le Béton de Fibres Métalliques: Etat Actuel des Connaissances”; pp. 37–68, Annalesde l’Institut Technique du Bâtiment et des Travaux Publics No. 515, July–Aug. 1993.

12R.W. Steinbrecht,“Toughening Mechanisms for Ceramic Materials,” J. Eur. Ceram. Soc., 10 [3] 131–42 (1992).

13S.M. Barinov and M.M. Sakai,“The Work-of-Fracture of Brittle Materials: Principles, Determination andApplications,” J. Mater. Res., 9 [6] 1412–25 (1994).

14L.R. Betterman, C. Ouyang and S.P. Shah,“Fiber–Matrix Interaction in Microfiber-Reinforced Mortar,” Adv.Cem. Based Mater., 2 [2] 53–61 (1995).

15P.C. Tatnall,“Shotcrete in Fires: Effects of Fibers on Explosive Spalling,” Shotcrete, [Fall] 10–12 (2002).

©The American Ceramic Society American Ceramic Society Bulletin www.ceramicbulletin.org January 2006 9407