There are some surprisingly tricky trade-offs in choosing the right anilox cell pattern. If you want a reasonable % transfer, say 40-50% then the choice of Print and Anilox LPC (multiply by 2.5 if you want LPI), Depth and ink Viscosity interact strongly. Wall thicknesses should be as small as you dare before they become too delicate. But your choice of which minimum dot you want to print also affects things. If you really want to print 1% dots then you will probably hit some problems.
Choosing the right anilox roller for flexo printing involves compromises. Although there is plenty of advice on the websites of those who manufacture such rollers (and I am grateful for their helpful information) it is hard to see what it all means.
This app brings the issues alive so that you can see everything in one place. Although there are simplifying assumptions, the results map convincingly onto the general advice found on those websites. For simplicity I am using a hexagonal cell and the lpc (lines per centimetre) (or lpi, lines per inch) values are specified along the flat edges (rather than the apexes) of the hexagons which, normally, are at 60° to reduce moiré, an effect I hope to discuss in another app.
The key fact is that we need to get ~2μm of ink onto the flexo plate, so our anilox roller must contain the equivalent of at least 4μm of ink, given that the split between anilox and plate is seldom more than 50%. For a coarse roller this is not a problem, but for a fine roller the cells have to be deeper and deeper in order to contain sufficient volume, but this quickly becomes self-defeating as the % transfer rapidly drops from 50%.
Why would anyone choose a fine roller? Because if any dot on the plate is smaller than the cell size, it can "dip" into the cell and get covered with extra ink, which shows up as a print defect. This forces another compromise. If the lpc for the print is lower, your smallest dots are larger. If you are forced to a high lpc print then your smallest dot is larger if you avoid trying to print all the way down to 1% dots. By choosing a lower limit of, say, 4% dot, your smallest dot does not require such a high lpc anilox.
Those who print UV flexo know that the higher viscosity ink leads to a low % transfer onto the plate. Smarter cell patterns (e.g. making elongated cells that are still narrow enough to prevent dot dipping) can help here but are not included in the app. To get a feel for these viscosity effects, the app steadily reduces the % transfer as the viscosity increases to 500cP. Although this is a typical viscosity for a UV ink if measured at room temperature, with a heated anilox roller and some shear thinning, the viscosity might be down in the 250cP region which will help a lot.
All this makes sense when you play with the app. Some more details are below
Because the key is to get ~2μm onto the plate we need to know how to calculate cell volume then translate that into the thickness this would cover at a given % transfer. The US love of bizarre units gives us BCM, an acronym that is doubly wrong. First, it is billion cubic microns, without stating that it is BCM per unit area. Second the unit area is square inches. Why any country chooses to mix a bizarre metric unit with an old-fashioned non-metric unit is a mystery. The real units are cm³/m² which, delightfully, work out as μm. So 4cm³/m² when split 50:50 will deliver 2μm to the plate which will print 1μm on the substrate after another 50:50 split.
The Plate and Anilox lpc are straightforward. The Depth input is a bit vague as the shape is not specified and the image uses a simplistic Bezier shape for programming simplicity: complexity is not justified as the shapes can vary considerably. The wall thickness becomes super-important at high lpc. Obviously a thin wall improves the printing of a fresh cylinder but is rapidly eroded with time. The viscosity input is used to estimate % transfer. The % dot, in combination with the Print lpc, gives the dot diameter which is shown as a number but also drawn to scale on the cell so you can see how close it might be to dipping.
The other outputs are lpi values from the lpc values, the volume/area of the cells, the Open width and the important Depth/Open parameter discussed below. The % Transfer is an estimate from the Depth ratio and the viscosity (print speed plays an important role too, but is not included here) and from the Volume/Area ratio and %T we have the thickness on the plate.
Large Depth/Open values are difficult to produce, inefficient in delivering ink, and are difficult to clean if ink dries in the bottom. Small Depth/Open ratios are difficult to control (a 1μM error is 10% in a 10μM depth) but provide efficient delivery and easy cleaning. The general agreement is that a ratio of 20-40% is the best compromise.