1

2nd Int. PhD Symposium in Civil Engineering 1998 Budapest

HIGH STRENGTH LIGHTWEIGHT-AGGREGATE CONCRETE

Dipl.-Ing. Thorsten Faust1, Prof. Dr.-Ing. Dr.-Ing. e.h. Gert König2

University of Leipzig, Institute for Structural Engineering and Building MaterialD-04109 Leipzig

SUMMARY

One research field in Leipzig deals with the design of high strength lightweight-aggregateconcretes (HSLWAC). Due to the huge waveband of this concretes with all possiblecombinations of lightweight-aggregates and mortar matrix the formulation of generalrules is difficult. Therefore both components lightweight-aggregates and mortar matrixwill be investigated in detail in order to achieve a deeper understanding of the internalstress transfer and the failure mechanisms. Moreover the results of the components testsserve as input data for a computer simulation. This instrument should support theprediction and optimization of the compressive strength of LWAC.

Keywords: high strength lightweight-aggregate concrete, lightweight-aggregates, mortar matrix, internal stress transfer, brittleness

1 INTRODUCTION

Lightweight-aggregate concrete withclosed structure is often called structuralLWAC with regard to the applications inbuildings, bridges and offshore structures.The development of new concretingmaterials in the last years expanded thewaveband of these concretes with the aimto increase the strength or to decrease thedensity, respectively. Fig 1 shows the hugespectrum considering the strengths of 15 to100 MPa and oven dry densities of 1,0 to2,0 kg/dm³. Structural LWAC with aminimized density at a definite strength level is called HSLWAC. Thus the term “highstrength” in case of LWAC is not related to the strength, but to the relation of strength todensity. For all HSLWAC’s high strength mortar matrix are used so that in general theconcrete strength will be limited by the efficiency of the aggregates. This fact induces avery brittle behaviour, which is independent of the strength. Consequently, the validity ofthe existing design rules and calculation models have to be proven with regard to the useof HSLWAC in order to adjust possible limits of applications. 1 PhD student2 Professor in Structural Engineering

0

10

20

30

40

50

60

70

80

90

100

1 1,1 1,2 1,3 1,4 1,5 1,6 1,7 1,8 1,9 2

Oven-dry density [kg/dm³]

Spectrum of structural lightweight aggregate concrete

Cu

be

com

pre

ssiv

e st

ren

gth

[N

/mm

²]

Fig.1 Spectrum of structurallightweight-aggregate concrete

2

2 LIGHTWEIGHT-AGGREGATE CONCRETES IN COMPRESSION

The stress/strain behaviour of LWAC in compression is, compared to normal densityconcretes (NC) of the same compressive strength, generally characterized by a linearascending branch, a lower E-modulus and less ductility in the post-failure region. Thesecharacteristics are usually more pronounced with increasing compressive strength anddecreasing oven dry density. Therefore HSLWAC not only comprises LWAC with a highstrength, but also LWAC with a low density.

Fig.2 shows a stress-straincurve of a strain controlledcompressive test with acylindrical specimen. Afterlongitudinal cracking(Fig.3a) pieces of concretespall off the cylinder(Fig.3b). This phenomenonwas particularly observed incase of ALWA-concretes(All Light WeightAggregates). The reducedcross sectional area requires aquick unloading to avoid asudden failure. Thedecreasing load allows an elastic expansion of the undestroyed regions outside thedamage zone. In case of brittle concretes combined with a small damage zone in relationto the regions outside this fact causes a so called snap-back. In consideration of the lowmodulus of elasticity this effect is valid for all HSLWAC.

Due to the snap-back effect the failure phenomenaof the specimen must be controlled by other suitablemeans. The first experiments were performed with acombination of axial and circumferential straincontrol. This test setup was already usedsuccessfully with HSC. But in case of LWAC,partly unstable softening occurred as well, whichprobably was caused by the low modulus ofelasticity (Fig.4). For this reason a control method,where a considerable portion of the E-modulus willbe compensated, had to be taken into consideration.This seems to be a promising solution in case ofLWAC [1,2].

Fig.5 shows the principle of the control method. A linear combination of deformation andforce is chosen as the feedback signal for a closed-loop servo-controlled system:

K

FFS −= δ .

Geometrically this control corresponds to a rotation of the axis with the angle Φ=1/K,which allows a steadily increasing feedback signal, if the stiffness value K is chosen

0

50

100

150

200

250

300

350

400

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0

Strain [mm/m]

Co

mp

ress

ive

Fo

rce

[kN

]

LC 45/55 ρ=1,7 kg/dm³cylinder h/φ=300/100

circumferential expansion ab

Fig. 2 Stress-strain curve Fig.3 Different fracture statesusing strain control a) longitudinal cracks b) spalling of concrete pieces

Fig. 4 Stress-strain curve usingaxial and circumferential straincontrol

3

between Kpre and Kpost. In this context I would like toexpress my gratitude to Professor Marti Marti fromthe technical university in Zürich (ETHZ) and hisscientific assistants, Heinrich Schnetzer and KyrillSokolov, who conducted the first tests with twodifferent LWAC mixtures, for their outstandingsupport. Fig.6 shows stress-strain diagrams ofconcretes with different densities at three differentstrength levels. All of these stress-strain curvesindicate a stable descending branch as a sign of acontrolled post-failure region. Therefore the diagramsdemonstrates that the described test method presents asuccessful way to obtain the full stress-strain curves ofHSLWAC.

Fig.6 also illustrates a very brittle behaviour, also forLWAC of moderate strength, if aggregates ofmoderate density are combined with a high strengthcement matrix. This observation confirms thedefinition of HSLWAC given in the introduction.

The different densities in Fig. 6 represent LWAC with lightweight finematerial (ALWAC, lower density), LWAC with natural fine material

(middle density) and normalconcrete (ρ=2,3kg/dm³). Typicalfor the LWAC with natural finematerial and a strength over 35MPa was a sliding mode offailure (Fig.7). In these casesonly the undestroyed regionsoutside the damage zoneexpanded due to the decreasingload (moderate snap-back) incontrast to Fig.3b, where ahorizontal damage zone can’t bedefined in that way and where

the extension over nearly the whole specimen length causes an extremely snap-backeffect. Moreover the characteristics of the different mortar-matrix also influence theproperties of the respective concrete with regard to the linearity of the ascending branch,the ductility in the post-failure region, and finally the failure mode (Fig.8).

Fig.5 Principle of tests using alinear combination of force anddeformation as control variable

0

5

10

15

20

25

30

35

40

45

0 0,5 1 1,5 2 2,5 3 3,5 4

Strain ε [mm/m]

Co

mp

ress

ive

stre

ngth

[MPa

]

ρ=1,5 kg/dm³

ρ=2,3 kg/dm³

cylinder h/φ=300/100

ρ=1,7 kg/dm³

05

1015202530354045505560

0 0,5 1 1,5 2 2,5 3 3,5 4

Strain ε [mm/m]

Co

mp

ress

ive

stre

ngth

[MPa

]

ρ=2,3 kg/dm³

Zylinder h/φ=300/100

ρ=1,7 kg/dm³

ρ=1,9 kg/dm³

0

5

10

15

20

25

0 0,5 1 1,5 2 2,5 3 3,5 4

Strain ε [mm/m]

Co

mp

ress

ive

stre

ngth

[MPa

]

ρ=2,3 kg/dm³

cylinder h/φ=300/100

ρ=1,25 kg/dm³

Fig.6 Stress-strain diagrams of concretes with different densities and strength

Fig. 7 Sliding Fig.8 Stress-strain diagram ofmode of failure mortar-matrix with natural and lightweight fine aggregate

0

5

10

15

20

25

30

35

40

45

0 1 2 3 4

Strain ε [mm/m]

Co

mp

ress

ive

stre

ngth

[MPa

]

Normalweight matrixρ=2,0 kg/dm³

Leightweight matrixρ=1,7 kg/dm³

4

The European standard ENV 1992-1-4 EC2 [3] recommends a bi-linear stress-straindiagram (Fig.9) with a constant strain at the break point(2‰) and a constant ultimate strain of εlcu=3.5 ‰. Fig.6demonstrates that particularly in case of ALWAC thelength of the yield plateau seems to be overestimated in[3]. In the next months, additional tests will follow toinvestigate the behaviour under sustained load. Thepurpose of this study is to find a realistic formula whichtakes into account the dependence of εlcu on compressivestrength and oven dry density.

Contrary to the German standard DIN 4219, most of the international standards take intoaccount that the modulus of elasticity ofLWAC depends on the compressivestrength and the density. The testsconducted so far with differentaggregates confirm both influences. Fig.10 shows one evaluation of these resultscompared to values calculated by meansof the E-modulus given in EC 2 T.1-1multiplied with a reduction factorη=(ρ/2,4)², where ρ denotes the oven drydensity. Furthermore, investigations willfollow in order to find the reductionfactor with the best possible agreement.

In general, the development of strengthwith time is faster in case of LWACcompared with NC, particularly above thelimit strength of the LWAC (see Fig.14).Fig.11 shows an evaluation of self testsand various publications according to theequation given in Model Code 1990.Therefore the post-hardening is smaller forLWAC.

The influence of the specimen-shape onthe compressive strength tend to besmaller in case of LWAC. Typical valuesfor the ratio between the compressive strength of cylinders and cubes are slightly above0,9. The reason for that is the smaller effect of the lateral expansion restraint due to thesmaller lateral strain of LWAC at the maximum load.

0

5

10

15

20

25

30

0 5 10 15 20 25 30

measured E-Modulus [GPa]

calc

ula

ted

E-M

od

ulu

s [G

Pa]

η=(ρ/2,4)^2

EC 2 T.1-1

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

Strain εεlc [mm/m]

Str

ess

σσ lc

[M

Pa]

α·flcd=0,8·flck/γc

flck

Fig.9 Bi-linear stress-straindiagram given in [3]

Fig.10 Comparison between calculated andmeasured values of E-modulus of LWAC

Fig.11 Development of compressive strengthwith time for NC and LWAC

5

Fig.12 FE-mesh with geometricaljoints

Fig.13 Relation between concentration ofaggregate volume and LWACcompressive strength

3 MODELLING OF A LWAC-CUBE UNDER COMPRESSION

An axial compression test of a 100mmLWAC-cube should be simulated bymeans of SBETA, a program fornonlinear finite element analysis. Severalpublications confirmed the strong bondforming between the matrix and theexpanded aggregates such that failurewould be either through the paste orthrough the aggregate. Thus, the concretewas modelled as a two-componentmaterial for a two-dimensional computersimulation. In this simulation a plate wascut from the cube. The laterial strain onthe upper and lower surface of the platewas eliminated. The load was beingapplied on a steel bearing plate controlledby the strain. The volume of the grainswas chosen with Vp=40% of the concretevolume according to self experiences. Thesize and the distribution of the 121 pelletswere determined by a statisticalevaluation. Fig.12 shows the FE-meshwith all geometrical joints. To overcomethe influences of the mesh and three-dimensional effects, previous analyseswere carried out to calculate the necessaryinput value of the compressive strength ofthe matrix in a one-component system. Inthis way increased matrix strength datawere added to the program to reach the realvalue of the axial compression tests withvarious matrixes.

At the beginning of the simulation the lack of information about the properties of bothcomponents was indicated in consideration of the stress-strain relation in the compressionand tensile region. In particular, the compressive strength of the aggregates was adominate element with regard to the strength of the LWAC-cube. Fig. 13 shows, that inspite of a high strength matrix the strength of the cube was equal to the strength of thegrains, if the aggregate volume exceeded 40% of the concrete volume. The other wayaround, this result can be use to define the compressive strength or efficiency of LWArespectively.

Fig.14 shows the evaluation of the analysis. The known form demonstrates therelationship between the compressive strengths of the matrix and the LWAC andadditionally in comparison with laboratory results. The simulation also demonstrates thedifferent internal stress transfer and the failure mechanisms of LWAC and NC. In generalthe aggregates in LWAC have a lower stiffness than the mortar matrix in opposition toNC. Thus, compressive loads are mainly carried by the stiffer mortar matrix

0

10

20

30

40

50

60

70

80

90

100

0% 5% 10% 15% 20% 25% 30% 35% 40%

Concentration of aggregate volume

Co

mp

ress

ive

stre

ngth

[MPa

]

Matrix

Lightweight aggregate

Lightweight aggregate concrete

6

Fig.14 Relation between the strengths of Fig.15 Crack propagation in LWACmatrix and LWAC

corresponding to the stiffness relationship between matrix and aggregates, which causestransverse stresses in the aggregates and in the matrix. Finally, failure occurs afterexceeding the tensile capacity of the aggregates. The cracks usually propagate straightthrough the aggregate particles (Fig. 15). The smooth fracture surfaces transfer less stressand initiate a brittle failure.

4 PROPERTIES OF THE COMPONENTS

The important meaning of the component properties with regard to the simulation wasalready mentioned. Furthermore, their knowledge can be helpful for a betterunderstanding of the internal stress transfer and for the optimization of the strength. Fig.16 shows for instance the result of wedge splitting tests with LWAC’s consisting ofdifferent components. Obviously there seems to exist a strong connection between thecombination of the components and the fracture energy.

naturalsand

50% naturalsand; 50%lightweight

sand

lightweightsand

expandedglass

0

10

20

30

40

50

60

70

80

90

100

Fra

ctu

re e

ner

gy

[Jo

ule

/m²]

Matrix

Par

ticl

e d

ensi

ty

ρp=1.74 kg/dm³ ρp=1.22 kg/dm³ρp=0.83 kg/dm³

0

10

20

30

40

50

60

70

80

90

100

110

0 10 20 30 40 50 60 70 80 90 100 110

Compressive strength of the lightweight matrix[MPa]

MPa

]

Aggregate C

Aggregate B

Aggregate A

cube 46

cube 49

cube 38

cube 50cube 51

cube 42

cube 52

cube 53

cube 54

cube 55

45°

Fig. 16 Fracture energy for LWAC with different components,investigated in wedge splitting tests

7

4.1 Properties of lightweight aggregates

There is a wide range of different lightweight aggregates, which differ in the rawmaterial, density, shape, outer skin and water absorption. In spite of this fact, theirproperties can be estimated with simply formulas, which in general depend on the particledensity. In this context wetested several aggregates(expanded clay, shale andglass, sintered fly ash,pumice) with regard to thecompressive and tensilestrength. Fig. 17 and 19demonstrate that there areobviously connections onlywith slight dependence on theaggregate type. Fig. 17 alsoshows the necessity to use ahigh-strength matrix in orderto determine the compressivestrength of the particles. Thetest methode, measuring thecrushing resistance, can not berecommend to achieve astatement about the efficiency of the aggregates.

For the tensile strength of pelletized aggregates, we developed a new testing arrangementshown in Fig. 18. Twelve grains were glued in the openings of two opposite plates andthe tensile force to the fracture area were related. In this way we achieved the lower limitof the tensile strength, since a premature failure of a single grain can’t be excluded.Fig.19 shows that the particle tensile strength increases exponential with increasingparticle density. Thus, this fact explains the increasing fracture energy with increasingparticle density.

The dynamic particle modulus of elasticity was estimated according to Schütz as:

Edyn,p = 8000·ρp2 .

0,0

0,5

1,0

1,5

2,0

2,5

3,0

3,5

4,0

4,5

0,00 0,25 0,50 0,75 1,00 1,25 1,50 1,75 2,00

Particle density ρp [kg/dm³]

Argex 550Argex 650

Argex RArlita F3

Arlita F5Arlita F7

EmbraLeca HDLeca NW

Liapor 4,25Liapor 5

Liapor 6Liapor 6,5

Liapor 7Liapor 8

Liapor 9,5Lytag UKPollytag

VasimPoraver

fpt = 1,25·ρp1,5

Parti

cle

tens

ile s

treng

th fp

t [N

/mm

²]

Fig.18 Testing arrangement for Fig.19 Relation between particle tensile strengththe particle tensile strength and particle density of 7mm grains

0

10

20

30

40

50

60

70

80

90

100

110

0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0

Particle density ρp [kg/dm³]

Com

pres

sive

cub

e st

reng

th [

MP

a]

NSM Blähton

LSM Blähton

LSM Blähschiefer

LSM Bims

LSM Blähglas

LSM gesinterteSteinkohleflugasche

Trendline: Natural sand matrixfck,matrix = 82 MPa

fck,cube = 45,6·ρp1,5

Trendline:Lightweight matrixfck,matrix = 51 MPa

fck,cube = 35,9·ρp1,25

Expanded clay(+ natural sand)Expanded clay

Expanded shale

Pumice

Expanded glass

Sintered fly ash

Fig.17 Relation between cube compressive strengthof LWAC and particle density in dependence on thematrix strength

8

4.2 Properties of mortar matrixes

The influences of the different matrixes on the shape of the stress-strain curves werealready mentioned. Additional important aspects are their differences with regard to themodulus of elasticity and the ratio between compressive strength and tensile strength. Forinstance, the internal stress transfer is influenced considerably by the stiffness ratio ofboth components. However, the properties of the lightweight matrixes are almostunknown. This is the reason that a study was started to investigate different matrixes.Intermediate results will be published at a later moment.

5. CONCLUSIONS

The formulation of general design rules for LWAC requires a deeper understanding ofboth components lightweight-aggregate and mortar matrix. There are numerous types ofeach component and their combinations, which have a decisive effect on the behaviour ofLWAC. Therefore the investigation of different components was started, whichcomplement the tests with LWAC’s in order to

- support the evaluation and explanation of the concrete tests

- optimize the properties of LWAC’s

- predict the concrete behaviour by means of a computer simulation.

6. ACKNOWLEDGEMENTS

The author would like to thank the sponsor, Lias Franken, for financial support for theseinvestigations.

7. REFERENCES

[1] Okubo, S., Nishimatsu, Y.: Uniaxial Compression Testing Using a LinearCombination of Stress and Strain as the Control Variable; Int. J. of Rock Mech.,Min. Sci. & Geomech., Abstract., Vol. 22, No.5, 1985, pp. 323-330

[2] Jansen, D.C., Shah, S.P.: Effect of Length on Compressive Strain Softening ofConcrete – Journal of Engineering Mechanics. Jan., 1997

[3] ENV 1992-1-4 EC2; “Design of LWAC with closed structure”