Durability can be defined as the ability of a material to remain serviceable for at least the required lifetime of the structure of which it forms a part. Standards and specifications increasingly include requirements for a design life, which can typically be 50 or 100 years, but for many structures this is not well defined, so the durability should then be such that the structure remains serviceable more or less indefinitely, given reasonable maintenance.
For many years, concrete was regarded as having an inherently high durability, but experience in recent decades has shown that this is not necessarily the case. Degradation can result either from the environment to which the concrete is exposed, for example freeze–thaw damage, or from internal causes within the concrete, as in alkali– aggregate reaction.
Durability of Concrete
It is also necessary to distinguish between degradation of the concrete itself and loss of protection and subsequent corrosion of the reinforcing or pre-stressing steel contained within it. The rate of many of the degradation processes is controlled by the rate at which moisture, air or other aggressive agents can penetrate the concrete.
This penetrability is a unifying theme when considering durability, and for this reason we shall first consider the various transport mechanisms through concrete – pressure-induced flow, diffusion and absorption – their measurement and the factors that influence their rate. We shall then discuss the main degradation processes, firstly of concrete – chemical attack by sulphates, seawater, acids and the alkali– silica reaction, and physical attack by frost and fire – and then the corrosion of embedded steel.
In each case a discussion of the mechanisms involved and the factors that influence these will show how potential problems can be eliminated, or at least minimised, by due consideration of durability criteria in the design and specification of new structures. By way of illustration, some typical recommendations from current European specifications and guidance documents are included.
Ignorance of, or lack of attention to, such criteria in the past has led to a thriving and ever expanding repair industry in recent years; it is to be hoped that today’s practitioners will be able to learn from these lessons and reduce the need for such activities in the future. It is beyond the scope of this book to discuss repair methods and processes.
Transport Mechanisms Through Concrete
Hardened cement paste and concrete contain pores of varying types and sizes, and therefore the transport of materials through concrete can be considered as a particular case of the more general phenomenon of flow through a porous medium.
The rate of flow will not depend simply on the porosity, but on the degree of continuity of the pores and their size – flow will not take place in pores with a diameter of less than about 150 nm.
The term permeability is often loosely used to describe this general property (although we shall see that it also has a more specific meaning); Fig. 1 illustrates the difference between permeability and porosity.

Flow can occur by one of three distinct processes:
- permeation – i.e. movement of a fluid under a pressure differential
- diffusion – i.e. movement of ions, atoms or molecules under a concentration gradient
- sorption – i.e capillary attraction of a liquid into empty or partially empty pores.
Each of these has an associated ‘flow constant’, defined as follows:
1. In the flow or movement of a fluid under a pressure differential, flow rates through concrete pores are sufficiently low for the flow of either a liquid or gas to be laminar, and hence it can be described by Darcy’s law:
ux = -K∂h/∂x
where, for flow in the x-direction, ux = mean flow velocity, ∂h/∂x = rate of increase in pressure head in the x-direction, and K is a constant called the coefficient of permeability, the dimensions of which are [length]/[time], e.g. m/sec.
The value of K depends on both the pore structure within the concrete and the properties of the permeating fluid. The latter can, in theory, be eliminated by using the intrinsic permeability (k) given by:
k = Kɳ/ρ
where ɳ = coefficient of viscosity of the fluid and ρ = unit weight of the fluid. k has dimensions of [length]2 and should be a property of the porous medium alone and therefore applicable to all permeating fluids.
However, for liquids it depends on the viscosity being independent of the pore structure, and for HCP with its very narrow flow channels, in which a significant amount of the water will be subject to surface forces, this may not be the case.
Furthermore, comparison of k values from gas and liquid permeability tests has shown the former to be between 5 and 60 times higher than the latter, a difference attributed to the flow pattern of a gas in a narrow channel differing from that of a liquid (Bamforth, 1987).
It is therefore preferable to consider permeability in terms of K rather than k, and accept the limitation that its values apply to one permeating fluid only, normally water.
2. The movement of ions, atoms or molecules under a concentration gradient is described by Fick’s law:
J = -D∂C/∂x
where, for the x-direction, J = transfer rate of the substance per unit area normal to the x-direction, ∂C/∂x = concentration gradient and D is a constant called the diffusivity, which has the dimensions of [length]2 /[time], e.g. m2/sec.
Defining diffusivity in this way treats the porous solid as a continuum, but the complex and confining pore structure within concrete means that D is an effective, rather than a true, diffusion coefficient. We are also interested in more than one type of diffusion process, for example moisture movement during drying shrinkage, or de-icing salt diffusion through saturated concrete road decks.
Furthermore, in the case of moisture diffusion (in, say, drying shrinkage) the moisture content within the pores will be changing throughout the diffusion process. There is, however, sufficient justification to consider D as a constant for any one particular diffusion process, but it should be remembered that, as with the permeability coefficient K, it is dependent on both the pore structure of the concrete and the properties of the diffusing substance.

3. Adsorption and absorption of a liquid into empty or partially empty pores occur by capillary attraction. Experimental observation shows that the relationship between the depth of penetration (x) and the square root of the time (t) is bi- or trilinear (Fig. 2), with a period of rapid absorption in which the larger pores are filled being followed by more gradual absorption (Buenfeld and Okundi, 1998).
A constant called the sorptivity (S) can be defined as the slope of the relationship (normally over the initial period), i.e.:
x = S.t0.5
As before, S relates to a specific liquid, often water. It has the dimensions of [length]/[time]0.5, e.g. mm/sec0.5. Different mechanisms will apply in different exposure conditions.
For example, permeation of seawater will occur in the underwater regions of concrete offshore structures, diffusion of chloride ions will occur when de-icing salts build up on concrete bridge decks and rain water falling on dry concrete will penetrate by absorption.
Measurement of Flow Constants for Cement Paste and Concrete
Permeability
Permeability is commonly measured by subjecting the fluid on one side of a concrete specimen to a pressure head, and measuring the steady-state flow rate that eventually occurs through the specimen, as illustrated in Fig. 24.3.

The specimen is normally a circular disc, the sides of which are sealed to ensure uniaxial flow. If the fluid is incompressible, i.e. it is a liquid such as water, the pressure head gradient through the specimen is linear, and Darcy’s equation reduces to:
ΔQ/ΔA = -K.ΔP/l
where ΔQ = volumetric flow rate, ΔA = total crosssectional area of flow perpendicular to the z-direction, ΔP = pressure head and l = length of flow path.
Much of the fundamental work on the permeability of cement paste to water was carried out by Powers and colleagues (Powers et al., 1954; Powers, 1958).
As the cement hydrates, the hydration products infill the skeletal structures, blocking the flow channels and hence reducing the permeability. As might be expected from cement hydration, the reduction of permeability is high at early ages, when hydration is proceeding rapidly. In fact, as shown in Fig. 4, it reduces by several orders of magnitude in the first 2 – 3 weeks after casting.

Although, as discussed above, permeability and porosity are not necessarily related (Fig. 1) there is a general non-linear correlation between the two for cement paste, as shown in Fig. 5.
The greatest reduction in permeability occurs for porosities reducing from about 40 to 25%, where increased hydration product reduces both the pore sizes and the sizes of the flow channels between them. Further hydration product, although still reducing porosity significantly, results in much lower changes in permeability.

This explains the general form of Fig. 5, and also accounts for the effect of water:cement ratio on permeability shown in Fig. 6 for a constant degree of hydration.

and water:cement ratio of mature cement paste (93% hydrated) (after Powers et al., 1954).
At water:cement ratios above about 0.5 the capillary pores form an increasingly continuous system, with consequent large increases in permeability. We shall see later in the chapter that many recommendations for durable concrete limit the water:cement ratio to a maximum value below this. It is apparent from the above arguments that high strength and low permeability both result from low porosity, and in particular a reduction in the volume of the larger capillary pores.
In general, higher strength implies lower permeability, although the relationship is not linear, and may be different for different curing histories and cement types.
The permeability of a concrete will also be influenced by the permeability of the aggregate. Many of the rock types used for natural aggregates have permeabilities of the same order as that of cement paste, despite having relatively low porosities. Lightweight aggregates, which are highly porous, can have much higher permeabilities.
However, values for the permeability of the composite concrete, despite considerable variation in reported values from different sources, are normally in the range 10-8 to 10-12 m/sec (Lawrence, 1985), i.e. between two and four orders of magnitude higher than that of either the cement paste or aggregate.
This is primarily owing to the presence of defects or cracks, particularly in the weaker interface or transition zone between the HCP and aggregate, which we saw in preceding chapters are present in the concrete before any load is applied. Permeability testing of concrete by fluid penetration under pressure (as in Fig. 3) can have considerable experimental difficulties, such as avoiding leaks around the specimen and the protracted timescales necessary for measuring flow rates through low permeability concrete.
An alternative indirect method of measuring permeability more rapidly that has become increasingly popular in recent years is the rapid chloride permeability test (ASTM C1202).

ASTM C1202).
The test, illustrated in Fig. 7, involves the application of a voltage between two sides of a concrete specimen with solutions of sodium hydroxide and sodium chloride on opposite sides. The chloride ions are driven through the concrete, and as they penetrate it the conductivity of the pore water and the current readings increase.
The test is continued for six hours and the total charge passed (current × time) determined. Some results from this test that show the effect of water:cement ratio and the incorporation of additions are shown in Fig. 8.

Permeability test (afer Zhang et al., 1999).
These show that, as with cement paste, similar factors control both the permeability and strength of the concrete, and it is therefore possible to produce low permeability by attention to the same factors required to produce high strength.
More generally, these include using a low water:cement ratio and an adequate cement content, and ensuring proper compaction and adequate curing. Additions can preferentially improve the properties of the interface transition zone, although longer curing periods are necessary to ensure continuance of the pozzolanic reaction. The avoidance of microcracking from thermal or drying shrinkage strains and premature or excessive loading is also important.
Diffusivity
The principle of diffusivity testing is relatively simple. A high concentration of the diffusant is placed on one side of a suitable specimen (normally a disc) of HCP, mortar or concrete, and the diffusion rate calculated from the increase of concentration on the other side.
In the case of gas diffusion, the highconcentration side may be an atmosphere of the pure gas; in the case of salts, a high-concentration aqueous solution would be used. The test is therefore similar to the fluid permeability test without the complication of high pressure.
It is generally found that, after an initial period for the diffusant to penetrate through the specimen, the concentration on the ‘downstream’ side increases linearly with time. The diffusivity will change if the moisture content of the concrete changes during the test, and so the specimens must be carefully conditioned before testing.

concrete
Control of test conditions is therefore important, and diffusivity measurements from different test programmes are not entirely consistent. Table 1 shows values of chloride-ion diffusivity that were obtained on mature saturated pastes and concrete.
As with permeability the values are higher for concrete than for paste, but in both cases the beneficial effects of low water:cement ratios and the use of additions are clear.
Sorptivity
Sorptivity can be calculated from measurements of penetration depth, and tests are carried out on samples in which penetration is restricted to one direction only, such as cylinders with the curved surface sealed with a suitable bitumen or resin coating.
The penetration depth at a particular time can be measured by splitting a sample open, but this requires a considerable number of samples to obtain a significant number of results. It is often more convenient to measure weight gain, in which case the sorptivity is expressed as the amount of water absorbed per unit exposed surface/square root of time, e.g. with units of kg/m2/hr0.5 or similar.
Penetration calculations can be made if the concrete’s porosity is known (which can be conveniently found by drying the specimen after the test), and the results can be expressed in the normal way. Values of sorptivity at various distances from the surface of a concrete slab are shown in Fig. 9.

cast surface of concrete made with Porland cement and
additions (after Bamforth and Pocock, 1990).
These were obtained on slices of cores cut from concrete slabs 28 days old, which had been moistcured for 4 days and then air-cured for 24 days. The sorptivity decreases with depth, attributed to the air drying causing imperfect curing of the surface zone.
However, although the similar strength mixes containing additions had similar sorptivities in the 15-mm thick surface zone, they generally had lower values than the plain Portland cement concrete at greater depth, again indicating the advantages to be gained from these materials with sufficient curing.
A number of tests have been developed to measure the absorption and permeability characteristics of in-situ concrete while still in place, i.e. avoiding the need to cut cores. These all measure the penetration rate of a fluid (normally air or water) into the concrete, either through the concrete surface or outwards from a hole drilled into the concrete.

One popular test of this type is the Initial Surface Absorption Test (ISAT), shown in Fig. 10. It is covered by BS 1881-5. A cap is clamped to the concrete surface and a reservoir of water is set up with a constant head of 200 mm. The reservoir is connected through the cap to a capillary tube set level with the water surface.
At the start of the test, water is allowed to run through the cap (thus coming into contact with the concrete surface) and to fill the capillary tube. The rate of absorption is then found by closing off the reservoir and observing the rate of movement of the meniscus in the capillary tube.
Readings are taken at standard times after the start of the test (typically 10 mins, 20 mins, 30 mins, 1 hour and 2 hours) and expressed as flow rate per surface area of the concrete, e.g. in units of ml/m2/sec. The rate drops off with time and in general increases with the sorptivity of the concrete.
Typical results showing the effect of the water:cement ratio of the concrete and the duration of the initial water curing period on the 10-min ISAT value for tests carried out on 28-day-old concrete are shown in Fig. 11.

curing on surface absorption of concrete as measured
by the ISAT test (after Dhir et al., 1987).
Not surprisingly, decreasing water:cement ratio and increased curing time both decrease the ISAT values; the results clearly reinforce the importance of curing. In common with other tests of this type, the ISAT has two main disadvantages.
Firstly, the results depend on the moisture state of the concrete at the start of the test, which is particularly difficult to control if the test is carried out in situ.
Secondly, the flow-path of the fluid through the concrete is not unidirectional but diverges; a fundamental property of the concrete is therefore not measured and it is difficult to compare results from different test systems.
However, the tests all measure some property of the surface layers of the concrete and, as we shall see, this is all important in ensuring good durability.
Thanks for reading about “durability of concrete.”