Alternative API Plan 32

A double seal arrangement with glycol as the barrier fluid?

It seems like tracing would be the most simple solution.

Does a vertical pump need a seal at all?

If ethylene glycol is objectionable in this case (small amounts of this would leak into wastewater), and assuming flush rates are low, how about a propylene glycol - water mix as the seal flush liquid? For example, a 50:50 w/w mix would enable operation at ambient temp of down to -25degC or so. You could charge up a gas filled accumulator every now and then. If flush rates are too high for an accumulator, then a tandem or double seal solution would be obvious, but more difficult to retrofit.

Keep in mind that I wasn't suggesting glycol as a flush fluid but as a barrier fluid. Think of a pump for hazardous fluids that uses a separate barrier fluid for seal lube. That could eliminate many of the problems associated with freezing as well as an off the shelf solution. There should be near zero contamination or the wastewater stream.