I’ve had a lovely year. I genuinely love talking to people about foam fractionation, and I’ve been doing a lot of talking this year because 1.
The staggering efficacy of foam fractionation
for the remediation of PFAS contaminated water streams has recently been demonstrated, and 2. The use of the
Hydrodynamic Theory of Rising Foam for understanding the mechanism of foam fractionation (as is expanded upon
in my book, jointly-authored with Bruce Li) is gaining traction amongst practitioners and academic researchers alike.
This week I had an enjoyable chat with a research student who is working on foam fractionation. The student demonstrated to me excellent mechanistic understanding combined with a desire to apply the process. Great! There was a little point about dimensional (dimensionless) analysis that wasn’t fully understood however, and so I thought I’d write something about this herein.
The keystone of the ‘mechanistic’ models that I developed for foam fractionation was the
foam drainage law, which is an expression for the liquid superficial drainage velocity in a Lagrangian frame of reference that is moving with the bubbles. This was developed using the method of dimensionless analysis which is often misunderstood. In essence, it shows that the drainage velocity, non-dimensionalised as a ‘Stokes number’ is a function of the volumetric liquid fraction only. It doesn’t say what that function is, but merely that such a function must exist. I decided to fit a power-law relationship to
drainage data
because many behaviours in nature demonstrate such dependency, and it works very well. The dimensionless analysis reveals the dependency of drainage upon bubble size, liquid viscosity and density and the acceleration due to gravity; if one wants to foam fractionate on the moon, then this is the way forward! The fit is empirical, so my method isn’t completely mechanistic. Let’s call it ‘semi-mechanistic’, for the sake of argument. (It is worth noting that, during my conversation this week, I discovered that I wasn’t, in fact, the first to propose a power-law dependency of drainage upon volumetric liquid fraction; Miles et al. (1945) beat me to it. Thanks for telling me.)
A better-known dimensional analysis is that leading to the Blasius relationship for the friction factor due to turbulent flow in a hydraulically smooth tube, where the friction factor is a power-law function of the Reynolds number. One can measure wall shear stress as a function of flow velocity, but correct application of Buckingham’s Pi-Theorem reveals the dependency upon liquid physical properties and internal pipe diameter too. This is long way from simply performing a curve-fit on an Excel graph.
Many academic researchers dislike empiricism. Some think it is a kind of ‘cop-out’ that is employed by those who can’t develop a completely mechanistic model. I guess that, in some respects, it is. If I could develop a completely mechanistic model that can be employed in practise for foam drainage I would, but I cannot for reasons that I now describe:
Leonard & Lemlich (1965) developed a foam drainage model that assumed that viscous losses were only in the Plateau borders and not the foam nodes. That is a good assumption for very dry foam, which fractionation foams generally are not. (Robert Lemlich of the University of Cincinnati was a titan of foam fractionation development but his work is often overlooked by some; when I was developing foam fractionation models I would always check if Lemlich had done it before, and often he had.) The viscous losses in the Plateau borders are a function of the surface shear viscosity, but
one cannot measure this surface property
to within two or three orders of magnitude (at best) and so the model doesn’t have any practical efficacy. Three decades later, Verbist & Weaire (1996) proposed ‘The Foam Drainage Equation’, but this was identical to Leonard & Lemlich save for the fact that they assumed that the gas-liquid interface was rigid. Stephan Koehler, when he was at Harvard, proposed an alternative drainage model in a series of papers that assumed that losses occurred entirely at the nodes, and then generalised his model to account for Plateau border losses. This was better, but still the problem of surface shear viscosity arose. (For wetter foams, the losses must be dependent upon surface dilatational viscosity as well, which is easier to measure in the lab.) For all the mathematical elegance of these drainage models, they are not of any practical use precisely because we do not know the surface shear viscosity. Thus, empiricism must be the way forward. My dimensionless analysis was not an attempt to combine the drainage models of Lemlich and Koehler, since Koehler had already done that. Instead, my drainage law simply offers an expression that is mathematically tractable (thereby facilitating its use in foam fractionation modelling) and works. When somebody can measure the surface shear viscosity of a gas-liquid interface with any certainty then we can talk; in the interim, I will stick with my empiricism for the purposes of practical process design.
A significant proponent of the power of dimensional analysis in the process industries was the late, great, Prof John Davidson FRS who was arguably the inventor of fluidisation. I was fortunate to have him as a teacher and mentor. I know that John didn’t think that my drainage law was a ‘cop-out’!
15 years ago I spotted that a research group in Australia were ‘doing’ empiricism incorrectly with respect to some froth flotation data, in that they were proposing dimensionally-inconsistent relationships gained by curve fits on Excel. Poorly-performed empiricism can send the practitioner down the garden path, since dimensionally-inconsistent relationships only apply for the specific kit upon which the data was obtained. Thus I was prompted to write a
‘paper’ on the use of dimensional analysis in minerals processing. This is essentially just a lecture, and I discuss the development of Blasius first, before considering the foam drainage law. If you have got this far in my blog post, then you may like to take a look.
The power of dimensional analysis in understanding complex process systems is immense, and foam fractionation is certainly a complex process system. It is unlikely that the performance of a foam fractionator within the Arctic Circle will replicate that operating in outback Australia, for instance. There will be a temperature dependency. Dimensional analysis has illuminated this dependency, such that we can estimate how foam fractionation will perform in extremely low temperatures, and we know how to make it work. And this was done without me trying to find my woolly hat.