Excellence in Concrete Construction through Innovation – Limbachiya & Kew (eds)© 2009 Taylor & Francis Group, London, ISBN 978-0-415-47592-1

Performance-based durability testing, design and specification in

South Africa: latest developments

M.G. Alexander & H. Beushausen

University of Cape Town, Cape Town, South Africa

ABSTRACT: Over the last decade, an approach to improving the durability of reinforced concrete constructionhas been developed in SouthAfrica. The philosophy involves the understanding that durability will be improvedonlywhenunambiguousmeasurements of appropriate cover concrete properties canbemade. Suchmeasurementsmust reflect the in situ properties of concrete, influenced by the dual aspects ofmaterial potential and constructionquality. Key stages in formulating this approach were developing suitable test methods, characterising a rangeof concretes using these tests, studying in-situ concrete performance, and applying the results to practicalconstruction. The paper discusses the latest developments in durability specification practice in South Africaand attempts to show a sensible way forward for practical application of the DI approach. The approach isan integrated one in that it links durability index parameters, service life prediction models, and performancespecifications. As improved service life models become available, they can be implemented directly into thespecifications. Concrete quality is characterised in-situ and/or on laboratory specimens by use of durability indextests, covering oxygen permeation, water absorption, and chloride conduction. The service life models in turnare based on the relevant DI parameter, depending on whether the design accounts for carbonation-induced orchloride-induced corrosion. Designers and constructors can use the approach to optimise the balance betweenrequired concrete quality and cover thickness for a given environment and binder system.

1 INTRODUCTION

Deterioration of reinforced concrete is often associ-atedwith ingress of aggressive agents from the exteriorsuch that the near-surface concrete quality largelycontrols durability. The bulk of durability problemsconcern the corrosion of reinforcing steel rather thandeterioration of the concrete fabric itself (Figure 1).The problem is then cast in terms of the adequacy ofthe protection to steel offered by the concrete coverlayer, which is subjected to the action of aggressive

Figure 1. The bulk of durability problems concern thecorrosion of reinforcing steel.

agents such as chloride ions or carbon dioxide fromthe surrounding environment.

For concrete structures, durability is generallydefined as the capability of maintaining the ser-viceability over a specified period of time withoutsignificant deterioration. In general, design conceptsfor durability can be divided into prescriptive conceptsand performance concepts. Prescriptive concepts arebased onmaterial specification from given parameterssuch as exposure classes and life span of the structure.However, durability is a material performance con-cept for a structure in a given environment and as suchit cannot easily be assessed through intrinsic mate-rial properties. Performance concepts, on the otherhand, are based on quantitative predictions for dura-bility from exposure conditions andmeasuredmaterialparameters.

As in other countries, durability problems in SouthAfrica derive mainly from inadequate attention todurability with regard to both design and construc-tion. This has resulted in extensive deterioration ofconcrete, which is mainly related to reinforcementcorrosion. In response to this situation, 3 durabil-ity index tests, namely oxygen permeability, watersorptivity and chloride conductivity were developed

429

(Alexander et al, 2001, Mackechnie & Alexander2002, Beushausen et al, 2003, Streicher & Alexander1995,Mackechnie 2002).The concrete surface layer ismost affected by curing initially, and subsequently byexternal deterioration processes. These processes arelinkedwith transportmechanisms, such as gaseous andionic diffusion and water absorption. Each index testtherefore is linked to a transport mechanism relevantto a particular deterioration process.

Material indexing of concrete requires quantifi-able physical or engineering parameters to characterisethe concrete at early ages. The 3 index tests havebeen shown to be sensitive to important material,constructional, and environmental factors that influ-ence durability. Thus, the tests provide reproducibleengineering measures of the microstructure of con-crete. The tests characterise the quality of concreteas affected by choice of material and mix propor-tions, placing, compaction and curing techniques, andenvironment.

In the South African approach, ‘durability indexes’are quantifiable physical or engineering parameterswhich characterise lab or in-situ concrete and are sensi-tive to material, processing, and environmental factorssuch as cement type, water: binder ratio, type anddegree of curing, etc. Increasingly, design specifica-tions for structures for which durability is of specialconcern include limiting values for chloride conduc-tivity (marine environment) and oxygen permeability(risk of carbonation-induced corrosion). The paperdiscusses the latest developments in durability specifi-cation in South Africa, using the Durability Index testmethods linked to oxygen permeability and chlorideconductivity.

2 DURABILITY INDEX TEST METHODS

The Durability Index test methods comprise oxygenpermeability, chloride conductivity and water sorp-tivity. As mentioned above, this publication concernsthe application of the former two test methods fordesign specifications. Test equipment and test proce-dures are described in detail in the literature developed(Alexander et al, 2001, Mackechnie & Alexander2002, Beushausen et al, 2003, Streicher & Alexander1995, Mackechnie 2002) and basic principles arediscussed in the following.

The Oxygen Permeability Index (OPI) test methodconsists of measuring the pressure decay of oxygenpassed through a 25mm thick slice of 68mm diametercore of concrete placed in a falling head permeameter(Figure 2). The oxygen permeability index is definedas the negative log of the coefficient of permeabil-ity. Common OPI values for South African concretesrange from 8.5 to 10.5, a higher value indicating ahigher impermeability and thus a concrete of poten-tially higher quality. Note that oxygen permeability

Figure 2. Test set-up for the Oxygen Permeability Indextest (OPI).

Figure 3. Test set-up for the chloride conductivity test.

index is measured on a log scale, therefore the dif-ference between 8.5 and 10.5 is quite substantial. Anempirical prediction model for carbonation was for-mulated using the oxygen permeability test. Using thisapproach, 50 year carbonation depthsmaybe predictedfor different environments.

The chloride conductivity test apparatus (Figure 3)consists of a two cell conduction rig in which concrete

430

core samples are exposed on either side to a 5MNaCl chloride solution. The core samples are precon-ditioned before testing to standardize the pore watersolution (oven-dried at 50◦C followed by 24 hours vac-uum saturation in a 5M NaCl chloride solution). Themovement of chloride ions occurs due to the appli-cation of a 10V potential difference. The chlorideconductivity is determined by measuring the currentflowing through the concrete specimen.The apparatusallows for rapid testing under controlled laboratoryconditions and gives instantaneous readings.

Chloride conductivity decreases with the addi-tion of fly ash, slag, and silica fume in concrete,extended moist curing and increasing grade of con-crete. Portland cement concrete for instance generallyhas high conductivity values with only high-gradematerial achieving values below 1.0mS/cm. Slag orfly ash concrete in contrast has significantly lowerchloride conductivity values. While the test is sen-sitive to construction and material effects that areknown to influence durability, results are specificallyrelated to chloride ingress into concrete. Correlationsbetween 28-day chloride conductivity results and dif-fusion coefficients after several years marine exposurehave shown to be good over a wide range of concretes(Mackechnie &Alexander 2002).

3 APPLICATION OF THE DURABILITYINDEXAPPROACH

3.1 General

The sensitivity of the South African index tests tomaterial and constructional effects makes them suit-able tools for site quality control. Since the differenttestsmeasure distinct transportmechanisms, their suit-ability depends on the property being considered.Durability index testing may be used to optimisematerials and construction processes where specificperformance criteria are required. At the design stagethe influence of a range of parameters such asmaterialsand construction systems may be evaluated in termsof their impact on concrete durability. In this way, acost-effective solution to ensuring durability may beassessed using a rational testing strategy (Ronnè et al,2002).

The durability indexes, obtained with the abovetest methods, have been related to service life predic-tion models. Index values can be used as the inputparameters of service life models, together with othervariables such as steel cover and environmental class,in order to determine rational design life. Limitingindex values can be used in construction specifica-tions to provide the necessary concrete quality for arequired life and environment. Thus, a framework hasbeen put in place for a performance-based approach toboth design and specification.

Table 1. Environmental Classes (Natural environmentsonly) (after EN206).

Carbonation-Induced Corrosion

Designation Description

XC1 Permanently dry or permanently wetXC2 Wet, rarely dryXC3 Moderate humidity (60–80%)

(Ext. concrete sheltered from rain)XC4 Cyclic wet and dry

Corrosion Induced by Chlorides from Seawater

Designation Description

XS1 Exposed to airborne salt but not in directcontact with seawater

XS2a∗ Permanently submergedXS2b∗ XS2a+ exposed to abrasionXS3a∗ Tidal, splash and spray zones

Buried elements in desert areas exposedto salt spray

XS3b∗ XS3a+ exposed to abrasion

∗These sub clauses have been added for SouthAfrican coastalconditions

3.2 Service life prediction models

Two corrosion initiation models have been devel-oped, related to carbonation – and chloride – inducedcorrosion. The models derive from measurementsand correlations of short-term durability index val-ues, aggressiveness of the environment and actualdeterioration rates monitored over periods of up to10 years. The models allow for the expected lifeof a structure to be determined based on considera-tions of the environmental conditions, cover thicknessand concrete quality (Mackechnie &Alexander 2002,Mackechnie 2001). The environmental classes arerelated to the EN 206 classes as modified for SouthAfrican conditions (Table 1), while concrete qual-ity is represented by the appropriate durability indexparameter. The oxygen permeability index is used inthe carbonation prediction model, while the chloridemodel utilises chloride conductivity. The service lifemodels can also be used to determine the requiredvalue of the durability parameter based on predeter-mined values for cover thickness, environment, andexpected design life. Alternatively, if concrete qualityis known from the appropriate DI, a corrosion-free lifecan be estimated for a given environment.

3.3 Specifying durability index values

Twopossible approaches to specifying durability indexvalues are a deemed-to-satisfy approach and a rig-orous approach. The former is considered adequate

431

for the majority of reinforced concrete constructionand represents the simpler method in which limit-ing DI values are obtained from a design table, basedon binder type and exposure class, for a given coverdepth (50mm for marine exposure and 30mm forcarbonating conditions).

The rigorous approach will be necessary fordurability–critical structures, or when the designparameters assumed in the first approach are not appli-cable to the structure in question. Using this approach,the specifying authority would use the relevant ser-vice life models developed in the concrete durabilityresearch programme in South Africa. The designercan use the models directly and input the appropriateconditions (cover depth, environmental classification,desired life, and material). The advantage of thisapproach is its flexibility as it allows the designer touse values appropriate for the given situation ratherthan a limited number of pre-selected conditions.

3.3.1 Examples for the deemed-to-satisfyapproach

This approach mimics structural design codes: thedesigner recommends limiting values which, if metby the structure, result in the structure being ‘deemed-to-satisfy’ the durability requirements.

The carbonation resistance of concrete appears tobe sufficiently related to the early age (28 d) OxygenPermeability Index (OPI) value, so that OPI can beused in a service life model. The environments thatrequire OPI values to be specified in the SouthAfricancontext are XC3 and XC4 (Table 1), with XC4 con-sidered the more critical because steel corrosion canoccur under these conditions. Two design scenarioswith standard conditions and required minimum OPIvalues are shown in Table 2.

Chloride resistance of concrete is related to its chlo-ride conductivity, and therefore this index can be usedto specify concrete performance in seawater environ-ments.Table 3 presents chloride conductivity limits forcommon structures (50 years service life). Differentvalues are given for different binder types, since chlo-ride conductivity depends strongly upon binder type.The horizontal rows give approximately equal perfor-mance (i.e. chloride resistance) in seawater conditionsfor the different binders. Binder types are restricted to

Table 2. Deemed to Satisfy OPI values (log scale) forcarbonating conditions.

CommonStructures Monumental Structures

Service Life 50 years 100 years 100 years

Minimum Cover 30mm 30mm 40mmMinimum OPI 9.7 9.9 9.7

blended cements for seawater exposure, since CEM Ion its own has been shown to be insufficiently resistantto chloride ingress.

3.3.2 Example for the rigorous approachAs an example of practical implementation of therigorous approach, consider the case of specifying amarine structure for a 50-year design life, subject to theenvironmental conditions given inTable 1. Combiningthe relevant durability index of chloride conductivitywith the appropriate service life model yields the datagiven in Table 4. It should be noted that the DI valuesare presented here for purposes of illustration only.The relative values are more important than the abso-lute values as these will vary in response to regionaland environmental variations.

Table 3. Maximum Chloride Conductivity Values (mS/cm)for Different Classes and Binder Types: Deemed to SatisfyApproach – Common Structures (Cover= 50mm).

Binder combination

EN206 Class 70:30 50:50 90:10

CEMI:FA CEMI:GGBS CEMI:CSFXS1 3.0 3.5 1.2XS2a 2.45 2.6 0.85XS2b, XS3a 1.35 1.6 0.45XS3b 1.1 1.25 0.35

Table 4. Limiting DI values based on rational predictionmodel: maximum chloride conductivity (mS/cm) (50 yearlife).

432

The table shows the trade-off betweenmaterial qual-ity (i.e. chloride conductivity) and concrete cover,with lower quality (represented by a higher conduc-tivity value) allowable when cover is greater. Thedependence of the conductivity on binder type isalso illustrated, with higher values permissible forblended binders at any given cover, based on theirsuperior chloride ingress resistance. These higher val-ues translate into less stringent w/b ratios. Therefore,a conservative approach is recommended at present,with mixes for which the concrete grade may be lessthan 30MPa, and/or the w/b may be greater than 0.55,not being recommended. However, in these cases,the particular cover and binder can be used, but theconductivity value will be over-specified.

3.4 Establishing limiting values forconcrete mixtures

To establish limiting DI values for concrete mixturesand evaluate compliancewith durability requirements,the following two aspects need to be considered:

– Statistical variability of test results (hence selectionof appropriate characteristic values for DurabilityIndexes)

– Differences between as-built quality (in-situ con-crete) and laboratory-cured concrete

– The consideration of the two above aspects is dis-cussed below and illustrated by an example inTable 5.

3.4.1 Characteristic values versus target valuesThe values determined in Tables 2, 3 and 4 are charac-teristic values to be achieved in the as-built structure,

Table 5. As-built chloride conductivity values (mS/cm) vs.potential target values (hypothetical case).

not target (average) values. The material supplier mustaim at target values that will achieve the requiredcharacteristic values with adequate probability. Thevariability inherent in concrete performance needs tobe considered when evaluating the test results, similarto the approach that is adopted with strength cubes.Since durability is a serviceability criterion, the limi-tations may not need to be as stringent as for strength.It is proposed that a 1 in 10 chance be adopted at thisstage for the Durability Index tests with a margin of0.3 below for the OPI, and 0.2mS/cm above for thechloride conductivity test.

3.4.2 As-delivered concrete quality versusas-built concrete quality

A clear distinction must be drawn between materialpotential and in-situ construction quality. Althoughspecifications are usually only concerned with as-built quality, the processes by which such quality isachieved cannot be ignored. There are two distinctstages and responsibilities in achieving concrete of adesired quality. The first is material production andsupply,which could be froman independent party suchas a ready-mix supplier. A scheme for acceptance ofthe as-suppliedmaterial must be established so that theconcrete supplier can have confidence in the poten-tial quality of the material. The second stage is theresponsibility of the constructor in ensuring that theconcrete is placed and subsequently finished and curedin an appropriate manner. It is ultimately the as-builtquality that determines durability and the constructorhas to take the necessary steps and precautions in theconstruction process to ensure that the required qual-ity is produced. If the as-built quality is found to bedeficient, the specification framework must have aninternal acceptance scheme that is able to distinguishwhether the deficiency arises from the as-suppliedmaterial or the manner in which it was processed bythe constructor.To enable this, a two-level quality con-trol system has been proposed in South Africa, withtesting of both material potential and as-built qual-ity. Material potential is represented by as-suppliedconcrete specimens with a laboratory-controlledwet curing period (5 days), while as-built qualityis determined using in-situ sampling of concretemembers.

As a general rule, concrete in the as-built struc-ture may be of lower quality compared with the sameconcrete cured under controlled laboratory conditionsdescribed above. To account for the improved perfor-mance of laboratory concrete over site concrete, thecharacteristic values for the durability indexes of thelaboratory concrete should be:

– For OPI: a margin of at least 0.10 greater than thevalue determined in Sect. 3.3.

– For chloride conductivity: a factor no greater thanof 0.90 times the value determined in Sect. 3.3.

433

4 CLOSING REMARKS

The paper describes the development of the DurabilityIndex approach to addressing problems of reinforcedconcrete durability in the South African context. Theapproach is an integrated one in that it links durabilityindex parameters, service life prediction models, andperformance specifications. As improved service lifemodels become available, they can be implementeddirectly into the specifications. Concrete quality ischaracterised in-situ and/or on laboratory specimensby use of durability index tests, covering oxygen per-meation, water absorption, and chloride conduction.The service life models in turn are based on the rele-vant DI parameter, depending on whether the designaccounts for carbonation-induced or chloride-inducedcorrosion. Designers and constructors can use theapproach to optimise the balance between requiredconcrete quality and cover thickness for a given envi-ronment and binder system. More work remains tobe done, in particular generating correlations betweenindexes and actual structural performance. Only in thisway will the usefulness of the approach be assessed.

REFERENCES

Alexander, M.G., Mackechnie, J.R. and Ballim, Y. (2001),‘Use of durability indexes to achieve durable cover con-crete in reinforced concrete structures’, Materials Science

of Concrete, Vol. VI, Ed. J. P. Skalny and S. Mindess,American Ceramic Society, 483–511.

Beushausen,H.,Alexander,M.G., andMackechnie, J. (2003),‘Concrete durability aspects in an international context’,Concrete Plant and Precast Technology BFT, vol. 7, 2003,Germany, pp. 22–32.

Mackechnie, J.R. and Alexander, M.G. (2002), ‘Durabil-ity predictions using early age durability index test-ing’, Proceedings, 9th Durability and Building MaterialsConference, 2002. Australian Corrosion Association,Brisbane, 11p.

Mackechnie, J.R., (2001), Predictions of ReinforcedConcrete Durability in the Marine Environment –Research Monograph No. 1, Department of Civil Engi-neering, University of Cape Town, 28 pp.

Mackechnie, J.R., and Alexander, M.G., (2002), ‘DurabilityPredictions Using Early-Age Durability Index Test-ing,’ Proceedings of the Ninth Durability and BuildingMaterials Conference, Australian Corrosion Association,Brisbane, Australia, 11 pp.

Ronnè, P.D., Alexander, M.G. and Mackechnie, J.R. (2002),‘Achieving quality in precast concrete construction usingthe durability index approach’, Proceedings: Concretefor the 21st Century, Modern Concrete Progress throughInnovation, Midrand, South Africa, March 2002.

Streicher, P.E. andAlexander, M.G. (1995), ‘A chloride con-duction test for concrete’, Cement andConcrete Research,25(6), 1995, pp. 1284–1294.

434