Freight operations proceed on the railway because they pay virtually nothing for the infrastructure they use or have reserved for them. Total freight access charges are around ~£43-45m a year, only 30% higher than the track access charges paid in total by the two Open Access Operators.
That's a very interesting statistic.
I worked in rail freight marketing for, broadly, the last three decades of the 20th century. Profitability was measured by operating ratio and expressed as a percentage. It was the proportion of revenue represented by costs, so that an O.R. of 20% represented traffic where costs represented 20% of revenue. The most remunerative sector was electricity coal with an O.R. of around 20% - 25%. Construction, in which I spent my early years, usually weighed in at 45% - 55%. No bulk traffic was less than 60%, it would not be carried.
Costs were then much more specific; for instance, if a bridge on a freight-only line needed replacement the traffic had to bear the cost.
The profits from freight were used to fill the black holes in the passenger businesses' accounts. We even subsidised highly profitable Inter City by moving, free, infrastructure materials for the line upgrades they benefited from.
When the model moved to Railtrack infrastructure, the charge was raised as a rate per tonne-mile (of the total train load). They had considerable inconsistencies but were in my day far from nominal. It was difficult to appreciate the basis sometimes; for instance, a track-bashing loco would generate a very high price per tonne-mile, which of course applied not only to the loco. but also to the wagons, and the relationship between apparently similar movements was opaque. Obviously, Railtrack charges only represented the infrastructure element of costs, human and mobile resource costs still being with the freight operator.
From what you say, things changed considerably after I left; but possibly it was to give a more even playing field with road haulage. As an example of the old days, if the cost of bridge replacement (as mentioned above) made the traffic uneconomic (rate too high for the customer) it went to road. The road haulier would then have heavy vehicles driving in the same area and, if this resulted in road bridges needing replacement because of damage by heavy vehicles, the road haulier was not charged but the cost fell on the domestic property tax payer or general taxation. These taxpayers, generally, only used the roads with light vehicles such as cars, so the cost of bridge-strengthening was of no direct benefit to them.