I’ve talked about this before, but time to bring it up again. Engineering is not science or applied science. There are, in my opinion, two main areas where engineering differs from those two fields: the idea of design, and the importance of empirical fact. The picture above is an excellent example of why empirical fact matters.
That’s an old rubble-masonry retaining wall and it looks fine. It could maybe use a little pointing, but if that doesn’t happen for five years, it should still be fine. Retaining walls are designed to resist two forms of failure – sliding at the base and overturning as a whole – but a wall like this could also fail by rupture in the middle if the masonry is not of equal consistency. The internal shear and moment are greatest at the base, so a modern (i.e., reinforced concrete or reinforced concrete block) wall will fail only at the base. The wall above shows no signs of sliding forward, tilting forward, or bulging in the middle, the damage associated with the three failure modes. Good, right?
The ordinary analysis used for retaining walls says this wall doesn’t work. A rule I live by is that when analysis and reality disagree, it’s the analysis that’s wrong. But how? If the unreinforced rubble isn’t strong enough to carry the load imposed by standard analysis, which is basically a triangular wedge of earth behind it, how is it working? There are a number of assumptions in the standard analysis of a wall like this that may be wrong: the soil may have real cohesion, there may be bedrock outcroppings or boulders that reduce the amount of moveable soil in the wedge, the mortar may be capable of carrying tension, the wall may be thicker than it appears to be, and so on. In other words, the standard analysis is a worst-case scenario not an absolute. Looking in more detail at the conditions is not wrong, just more difficult.