Aggregates that are chemically stable will neither react chemically with cement in a harmful manner nor be affected chemically by normal external influences. Reactive aggregates may result in serious damage to the concrete by causing abnormal expansion, cracking and loss of strength.

Alkali-aggregate Reactions

Some aggregates containing reactive silica will react with the alkalies in cement - sodium and potassium oxides - to form an alkali-silica gei which takes up water and swells. This causes abnormal expansion and map-cracking of the concrete. The situation in New Zealand is that considerable investigations were carried out by the DSIR (now IRL) over a long period of time to establish rock types that were prone to alkali reaction.

The following categorises the principal rock type into non-reactive and reactive types.

Aggregates known to be non-reactive from field experience and testing:

• Greywacke

• Schist

• Basalt <50% SiO₂

• Quartz Sands

• Phonolite

• Rhyolitic pumice

• Granite

• Perlite

• Vermiculite

• Limestone

Aggregates or minerals known to be potentially reactive either from field experience or laboratory testing:

• Basalt >50% SiO₂

• Christobalite
• Andesite
• Tridymite
• Dacite
• Quartzite
• Rhyolite
• Amorphous and Criptocrystalline silicas
  (including Opal & Chalcedony)
• Volcanic glass

From experience gained by examining a limited number of structures that had experienced the problem, it was concluded that if the alkali content in the concrete could be kept no higher than 2.5 kg/m³ then the risk of expansive reactions was significantly lowered when potentially reactive rocks are used. New Zealand cement manufacturers assist with this requirement by voluntarily keeping the alkali level of the cement to below 0.6%.

Figure 1. Results of testing rhyolite, dacite, and some alluvial materials containing these rocky types by ASTM C289




Figure 2. Results of testing Egmont andesite from Taranaki by ASTM-C289

One key factor that has often been overlooked is that it is the sand that can be an important trigger mechanism in the reaction. This has been particularly so in New Zealand which perhaps explains why the mortar test method set out in NZS 3111. Section 11 has been successful in predicting problems. Typical traces of results for rhyolite and andesite compared with non-reactive aggregates clearly demonstrates the relative reactive risks between materials, see Figures 1, 2 & 3.

Figure 3. Results of testing greywacke samples by ASTM C289

 

A Whakamaru rhyolite B Tongariro andesite C Egmont andesite

Figure 4. Typical pessimum proportion curves for three New Zealand rock types (expansion at 12 months at 1.5% Na₂0 equivalent)

One topic not easily understood is that different reactive aggregates produce different amounts of expansion in concrete as their proportional content changes. Figure 4 shows how, for example, having just 12% of rhyolite causes 0.55% expansion yet using a concrete that has been made with 100% rhyolite materials the expansion is less than 0.1 %. It is important in any analysis of structural expansion to consider whether pessimum levels of reactive materials were in use.

Tests and Testing

Field service records, when available, provide information for the selection of non-reactive aggregates. If an aggregate has no service record, a petrographic examination can be useful by providing a description of its mineralogical and chemical constituents. It involves an examination of the aggregate particles with a microscope, together with other procedures for determining the constituents present, and in the hands of an experienced person can identity potentially reactive materials.

Physical tests are also available to measure potential reactivity. These include NZS 3111, Section 11. The potential reactivity of cement-aggregate combinations is determined by measuring the expansion of mortar bars (25 x 25 x 250 mm) during storage at 38 ± 2°C and a relative humidity not less than 90%. Cement aggregate combinations which show expansions greater than 0.10% at six months should usually be considered capable of harmful reactivity. When six-month results are not available, combinations should be considered potentially capable of harmful reactions if they show expansions greater than 0.05% at three months. The chemical method described in ASTM C289 on AS 1141, Section 39 is a rapid method used to obtain in 24 hours an assessment of potential reactivity.

Because of the influence of certain minerals on the test results, the chemical test should always be accompanied by a petrographic analysis. Indeed, the testing of aggregates for potential reactivity, and the interpretation of the results obtained, requires skill and experience. The mortar-bar test appears to give the best correlation with the behaviour of concrete, but, for improved assurance, it also should be conducted in conjunction with petrographic examination of the aggregate.

For more information:

Alkali Aggregate Reaction [TR 03] 2.5 Mb