Back in post #2473, Pinkman says:
This is essentially a refinement of interim report 1, tidying up some rough experiments. It's easy to criticize the slow progress but it's really difficult to simulate - think of the early wind tunnels.
At the end of the day, there needs to be a dedicated interagency fuel research centre to physically model these issues accurately, using a combination of actual engines and actual fuel systems in environmental test chambers (imagine!) and maybe even including extended duration high altitude flight testing. Irrespective of the current case, this kind of facility will anyway increasingly be needed when biomass derived fuels start to penetrate the market.
Using actual engines and fuel systems is probably neither the least expensive or most efficient way to go about investigating fuel system icing. Small-scale lab bench and workshop rigs, with a bit of mathematical modelling, will be a much more effective way of identifying the relevant parameters and their interaction.
On a lab scale, it should be relatively easy to identify the dependency of ice accreation rates on pipe external temperature, pipe size, internal surface characteristics, fuel temperature, fuel flow, fuel composition, including water content, effect of flow rate on the stability of the ice, and so on. Only when these elements are reasonably well understood will it be possible to predict what will happen in a full-size system, and reasonable to verify those predictions.
Perhaps such investigations are already under way in the industry. If not, it seems likely that they are a potential source of PhD theses in accademia.
On a slightly different tack, the AAIB report suggests that it was difficult or impossible in many cases to observe ice build-up inside their test rigs. I should have thought that techniques such as ultrasound would have been effective, and that it would not have been too difficult to send a miniature camera through the system.