Professor Andrew Schofield has developed arguably the two greatest advances in soil mechanics of the last 30 years - critical state soil mechanics and the use of the geotechnical centrifuge. As he retires, he still has a mission to accomplish. He talks to Paul Wheeler.
Andrew Schofield, a professor of engineering at the University of Cambridge and a Fellow of the Royal Society, could at 68 be excused for retiring with a sense of smug satisfaction. His contributions to the development of theoretical soil mechanics are unquestionable. Instead, I find a man with a missionary zeal.
He is initiating a campaign, although he denies it, that could reshape geotechnical design. If it is successful the C-word will only remain in the engineers' vocabulary a 'apparent cohesion'.
The peculiar thing is, this sudden vigour does not rely upon new data. It is essentially the same message made in Critical State Soil Mechanics, co-authored with the late Peter Wroth, and published in 1968.
'It's a pity I didn't make these points as strongly 20 years ago, but I didn't appreciate the importance,' he says. 'It's only gradually that certain things impact upon you and this insight is one of those.
'I am asking, does the industry really want its engineers to be taught something that is fundamentally wrong? Must Terzaghi textbooks from 1943 still be gospel in 2000? I don't think so.'
The argument, which Professor Schofield presents in abridged form on the following pages, is essentially that behaviour of remoulded soil (be it sand, silt or clay) is governed by friction and particle interlocking. The Mohr Coulomb equation, popularised by Terzaghi, and underpinning developments in soil mechanics since the 1930s, is simply wrong. Terzaghi, the grandfather of modern geotechnics, the man who made soil mechanics a science, made a mistake when he said soil's strength is provided by cohesion and friction.
'The key piece of data is the work of Hvorslev. Hvorlsev and Terzaghi misunderstood the data, and I think that is a very serious question. If a similar fundamental error was identified in, say, fluid mechanics, the senior designers in the industry would immediately want to know. If you told the designers of the Boeing company that their design was based on an erroneous concept they wouldn't bullshit. They would say let's have the details, let's have the data, we need to know.
'I find it extraordinary that people can be told there is an error in the Mohr Coulomb equation, and do next to nothing about it.
'You talk to people in industry and they say 'we don't use cohesion very much, we just put a bit in.' It's like saying we have a drink every now and then at a wedding, but we're really teetotal. The reason, I suspect, is because every textbook is based on the Mohr Coulomb equation and all the professors teach it.
'If I speak to a professor and say, you shouldn't teach that because it's wrong, they say you don't appreciate that my students have got to be employed in industry. So I go to the consulting companies and say you are hiring engineers who all believe the Mohr Coulomb equation, and when they arrive you are going to tell them that actually you hardly use any cohesion. The consultants say we are just working for clients and our clients' instructions are they want a building that is going to satisfy the codes.'
That is the nub: the people who write the codes, who are always established figures, can't turn round and say 'oops we've got it wrong'.
Advancing mechanical science
'If you followed Terzaghi all that mattered was experience in the field, gained through consultancy,' says Schofield. 'But in my view the only way to make advances is from a purely academic approach.
'I don't think you learn anything unless you can validate it with repeated experiments under laboratory controlled conditions with the mechanics correctly evaluated, and with the data present for examination and analysis. This way you can put together a piece of work that turns understanding from being purely empirical - typical Terzaghi - into something that is closer to mechanics.
'What we have been able to do with the centrifuge at Cambridge, time and time again, is take an interesting complex practical situation, get into the mechanics of it and find the behaviour.
'The conclusion I have reached is that critical state soil mechanics is right. Of course you have cyclical loading and anisotropy and Cam Clay is just a very rudimentary first look at the problem, but what matters much more than the subtleties is that on the dry side peak strengths are not due to cohesion, they are due to interlocking. The famous Drucker Prager constitutive model that tells people how to represent soil behaviour by friction and cohesion are just wrong.
Just as cohesion is a wrong concept, so are Terzaghi's bearing capacity factors. Soil structure criteria, for example in jack-up spend fixity, needs to be approached through a centrifuge model first, and a yield locus.
'Terzaghi's view that if you are an expert 'you know', is just not good enough, not nowadays. It was OK for a certain stage in the development of soil mechanics, but if you can get a bit nearer to some science, then a more modern client just wants to get validated code, good scientific data, and get something more like mechanics.'
This ideal neatly ties together two of Cambridge's major contributions to soil mechanics: critical state theory and its incorporation in the finite element code CRISP, but above all physical modelling in the geotechnical centrifuge.
'Physical modelling has now advanced to a stage where a completely new generation of foundation engineers will be able to achieve a new more valid design and I am sure the people who are going to be the successful foundation engineers of the future will work in this way.'