Structural Behaviour and Movement of Masonry
Like any structural material, masonry must resist loads or forces due to a variety of external influences (or actions) and in various planes. Figure 1 illustrates the various forces that can arise and the likely actions. In this chapter we will discuss the behaviour of masonry under all these types of action, an understanding of which is an essential prerequisite for successful structural design.
Like plain concrete, unreinforced masonry is good at resisting compression forces, moderate to bad at resisting shear and bending but very poor when subjected to direct tension. Masonry structures that are required to resist significant tensile forces should be reinforced by adding steel or other tension components.
Unlike concrete, however, masonry is highly anisotropic because of its layer structure and this must always be borne in mind in design. Masonry is quite effective at resisting bending forces when spanning horizontally between vertical supports but it is somewhat less effective at resisting bending forces when spanning vertically or cantilevering from a support (Fig. 2) because the resistance of a lightly loaded wall in that direction is dependent solely on the mass and the adhesion of the units to the mortar.
Much of the resistance to bending and collapse, especially of simple cantilever masonry structures, is simply due to self-weight. Masonry is a heavy material, usually with a density in the range 500–2500 kg/m3 , i.e. between 0.5 and 2.5 tonnes per cubic metre. In relatively squat structures such as some chimneys, parapets or low or thick boundary walls the force needed to overcome gravity to rotate the wall to a point of instability is sufficient to resist normal wind forces and minor impacts.
Any masonry under compressive stress also resists bending since the compressive pre-stress in the wall must be overcome before any tensile strain can occur. There is much empirical knowledge about how masonry works, and many small structures are still designed using experience-based rules.
The Limit State Codes of Practice in the UK – BS EN 1996-1-1 (2005) supported by BMS (1997) or BS 5628: Part1 (2005) give a calculated design procedure, but much of this code is based on empirical data such as given by Davey (1952), Simms (1965), de Vekey et al. (1986, 1990) and West et al. (1968, 1977, 1986).
Broadly, it predicts the characteristic strength of masonry elements, such as walls, from data on the characteristic strength or other characteristics of the materials using various engineering models for the different loading conditions. A check is then made that the predicted strength is greater than the expected worse loading based on data about wind, dead, snow and occupancy loads.
To allow for statistical uncertainty in loading data a safety factor, gf , is applied, which increases the design load level, and to allow for uncertainty about the strength of the masonry a further factor, gm, which reduces the design strength value for the masonry, is used. The combination of these partial factors of safety (FOS) gives an overall safety factor against failure of the structure that is usually in the range 3.5–5. This relatively high FOS is used because of the high variability in the properties of masonry and its brittle failure mode, which gives very little warning of failure. An alternative simplified design procedure is also available for smaller buildings such as houses (BS EN 1996-3, 2006).
To decrease the size of this web page, content of this section is palced on an another web page, to access that page, please follow the link.
Shear Load in the Bed Plane
If a wall is loaded by out-of-plane forces, e.g. wind, impact, or seismic (earthquake) action, the force will act to try to slide the wall sideways (like a piston in a tube). In practice the action can be at any angle in the 360° plane, although it is most commonly parallel or normal to the wall face. This is a very complex loading condition and the result is rarely a pure shearing failure.
For small test pieces measured in an idealised test, the shear strength fv can be shown to follow a friction law with a static threshold ‘adhesion’ fv0 and a dynamic friction term K dependent on the force normal to the shearing plane, sa. The formula is simply:
fv = fv0 + sa.K
Measurement of pure shear is very difficult because of the tendency to induce rotation in virtually any physical test arrangement. The simple double-shear test of the type shown in Fig. 3 is suitable for measuring shear resistance of damp-proof course (DPC) materials as described in EN 1052-4 (1999) but is unsatisfactory for mortar joints, where a much shorter specimen is preferred of the type described in BS EN 1052-3. Table 1 gives some typical shear data.
In the example sketched in Fig. 4 the ends are supported so the wall tends to adopt a curved shape. In this case it is shown as failing by shear at the line of the DPC. If a wall is loaded by lateral forces acting on the end as in Fig. 5 (which can arise from wind load on a flank wall at right angles) the force will initially tend to distort the wall to a parallelogram shape as in the top right hand of the diagram and induce compression forces where shaded. Failure if it occurs can be by a number of alternative mechanisms, as shown and listed in Fig. 5.
Traditionally, masonry was made massive or made into shapes that resisted compression forces. Such structures do not depend to any great extent on the bond of mortar to units. Much of the masonry built in the last few decades has, however, been in the form of thin walls, for which the critical load condition can result from lateral forces, e.g. wind loads.
This phenomenon was largely made possible by the use of ordinary Portland cement (OPC) mortars, which give a positive bond to most units and allow the masonry to behave as an elastic plate. There are two distinct principal modes of flexure about the two main orthogonal planes:
- The vertically spanning direction shown in Fig. 2, which is commonly termed the parallel (or p) direction because the stress is applied about the plane parallel to the bed joints.
- The horizontally spanning direction, shown in Fig. 6, which is commonly termed the normal (or n) direction because the stress is applied about the plane normal to the bed joints.
Clearly the strength is likely to be highly anisotropic since the stress in the parallel direction is only resisted by the adhesion of the units to the mortar while the stress in the normal direction is resisted by:
- the shear resistance of the mortar beds (a)
- the elastic resistance of the head joints to the rotation of the units (b)
- the adhesion of the head joints (c)
- the flexural resistance of the units themselves (d).
Generally the limiting flexural resistance will be the lesser of (a)+(b) or (c)+(d), giving two main modes of horizontal spanning failure – shearing, shown in the lower part of Fig. 6 and snapping, shown in the upper part of Fig. 6.
Using small walls (wallettes), either as shown in Fig. 6 for measuring horizontal bending or 10 courses high by 2 bricks wide for vertical bending, and tested in four-point bending mode, the flexural strength of a large range of combinations of UK units and mortars has been measured for the two orthogonal directions; typical ranges are given in Table 2, with further data in Hodgkinson et al. (1979), West et al. (1986), de Vekey et al. (1986) and de Vekey et al. (1990).
The ratio of the strength in the two directions, expressed as p-direction divided by n-direction, is termed the orthogonal ratio and given the symbol m. In cases where only the bond strength (p-direction) is required a simpler and cheaper test is the bond wrench.
The flexural strengths given in Table 2 are for simply supported pieces of masonry spanning in one direction. If the fixity of the masonry at the supports (the resistance to rotation) is increased the load resistance will increase in accordance with normal structural principles (Fig. 7). Again, if one area of masonry spans in two directions the resistances in the two directions are additive. Seward (1982), Haseltine et al. (1977), Lovegrove (1988), Sinha (1978) and Sinha et al. (1997) cover some aspects of the resistance of panels.
Further, in cases where the edge supports for a masonry panel allow no outward movement arching occurs and may be the dominant flexural resistance mechanism for thicker walls.
Masonry made with conventional mortars has a very limited resistance to pure tension forces and, for the purposes of design, the tensile strength is usually taken to be zero. In practice it does have some resistance in the horizontal direction and somewhat less in the vertical direction. In an attempt to make a viable prefabricated masonry panel product for use as a cladding material, polymer latex additives can be used to improve the tensile strength.
Panels of storey height and a metre or more wide have been manufactured and could be lifted and transported without failure. The horizontal tensile strength of masonry has been measured but there is no standard test and virtually no data has been published in the public domain. Tensile bond strength is usually measured using a simple two-brick prism test, as illustrated in Fig. 8.
Data from such tests indicate that the direct tensile strength across the bond is between one-third and two-thirds of the parallel flexural strength (see Table 2). Other tests have been developed along similar lines including one in which one unit is held in a centrifuge and the bond to another unit is stressed by centrifuging. A useful review is give by Jukes and Riddington (1998).
The stiffness or elastic modulus of masonry is an important parameter required for calculations of stresses resulting from strains arising from loads, concentrated loads, constrained movement and also for calculating the required area of reinforcing and post-stressing bars.
The most commonly measured value is the Young’s modulus (E), but the Poisson’s ratio (n) is also required for theoretical calculations using techniques such as finite element analysis. If required the bulk (K) and shear (G) moduli may be derived from the other parameters.
Young’s modulus is normally measured in a compression test by simultaneously measuring strain (εp) parallel to the applied stress (s) whereupon:
E = s/εp (35.3)
If the strain (εn) perpendicular to the applied stress is also measured, Poisson’s ratio may be derived:
v = εn/εp (35.4)
Masonry is not an ideally elastic material because it is full of minor imperfections such as microcracks in the bond layers and because the differences between the unit and mortar stiffnesses and Poisson’s ratios produce high local strains at the interface, which results in non-linear behaviour.
This means that the stress–strain curve is typically of a broadly parabolic form with an early elastic region, i.e. it is similar to that of concrete for similar reasons.
An instantaneous value of E can be obtained from the tangent to the curve at any point but for some calculations, such as creep loss of post-stressed masonry, the effective E derived from the secant value is required. Figure 9 illustrates this behaviour. Data on elastic properties in compression are given in Davies and Hodgkinson (1988).
The elastic modulus is also important in estimating the deflections of walls out of plane due to lateral loads such as wind. In this case the modulus can be measured by using load deflection data for small walls tested in fourpoint bending, with E given by:
E = 8Wa(3L2 − 4a2 )/384Id (35.5)
where W is the applied force, L is the support span, a is the distance from the supports to the loading points, I is the moment of inertia and d is the deflection.
In compression tests the value of E has generally been found to be in the range 500–1000 times the compressive strength. For typical materials this is likely to be around 2–30 GPa. In flexure the early tangent modulus has been found to be in the range 2–4 GPa for tests in the strong (normal) direction and 1–2 GPa for equivalent tests in the weak (parallel) direction. Some more data are given in Table 3.
Movement and Creep
Unrestrained masonry is subject to cyclic movement due to moisture and temperature changes, irreversible creep due to dead loads or post stress loads, irreversible shrinkage due to drying/carbonation of mortar, concrete and calcium silicate materials and expansion due to adsorption of moisture by fired clay materials. Table 3 contains some typical ranges for common masonry components derived mainly from Digest 228 (1979).
In simple load-bearing masonry elements vertical movements are accommodated by the structure going up and down as required and are no problem. Problems can arise, however, where materials with different movement characteristics are joined or where thick elements have differential temperature/ moisture gradients through them.
Elements with a vertical pre-stress much larger than the differential stresses will tolerate the movement, but lightly stressed and unrestrained elements will need a slip plane or soft joint between the elements to avoid problems. A classic case is (expanding) clay masonry on (shrinking) concrete frames where soft joints are used to stop stress transfer to the masonry.
Where restraint is present horizontal shrinkage will be converted into tensile forces and expansion into compressive forces. Since walls are probably two orders of magnitude weaker in tension than in compression the result is that restrained walls in tension usually crack while those in compression usually just build up stress.
Where walls are unrestrained the reverse is usually the case: the shrinking wall simply contracts, but the expanding wall interacts with flank walls (those at right angles) and causes cracking by generating rotation at the corner. Design strategies to cope with movement and avoid cracking are contained in BS EN 1996-2 and BS 5628-3. Because there is nearly always differential restraint the pure tensile or compressive forces will also usually give rise to some associated shear forces.