A theoretical minimum amount of oxygen is needed to react with the combustible elements (mainly carbon and hydrogen) in a given quantity and type of fuel. In practice an additional (or excess) quantity of air is required to ensure adequate contact between fuel and oxygen to guarantee efficient combustion. There are no hard-and-fast rules defining the amount of excess air that is required in any given situation, but 20% is a generally accepted minimum. For instance, in his paper “Two Point Four Pounds per Ton and The Railway Revolution“, Doug Landau defines the Front End Limit as occurring “when the available excess air falls below about 20% [i.e. when] complete combustion can no longer be achieved”.
The quantity of air – and therefore the quantity of excess air – passing through the firebox is governed by the smokebox vacuum and the flow resistance through the ashpan, firebox and boiler tubes/flues. Thus the quantity of excess air passing through the firebed is dependent on the smokebox vacuum and therefore the exhaust system performance. In the normal situation, where 100% of combustion air passes through the firebed, increasing the excess air is likely to result in fuel loss caused by the entrainment of coal particles that are carried away with the combustion gases. However where GPCS is applied, only a small proportion of the combustion air (perhaps 30%) passes through the firebed, thereby allowing much higher levels of excess air to be applied.
Fig 27 from Wardale’s book “The Red Devil and Other Tales from the Age of Steam” illustrates the reduction in fuel carry-over that can be achieved with GPCS, from which the possibilities for increased airflow may be deduced. 
As might be expected, increasing the amount of excess air above the 20% nominal minimum has the beneficial effect of reducing smoke emissions, however it penalizes performance because the higher gas flow through the firebox and boiler require the exhaust system to work harder, thereby raising cylinder back-pressure. Wardale also points out (in a letter to Chris Newman dated 5 Apr 2001) that “loss of heat from excessive amounts of air can be far greater than that carried away in soot (smoke)”. He followed this up in a second letter with the observation that “excess air should be kept to a minimum for best boiler efficiency …. Diluting of pollutants [is] a worthy goal [but] I would go for both.”
In his design for the 5AT exhaust system, Dave Wardale opted for 30% excess air at maximum firing rate, saying that “this is a ‘safe’ value. The [oil fuel] combustion equipment must be designed to allow adequately complete combustion with the minimum of excess air.” In his combustion calculations (FDC 11.2 item 17) he adds that: “from experience [30% excess air] is an appropriate value for locomotive operation at maximum evaporation, when the combustion time is extremely short. A general goal is the highest combustion efficiency with the lowest excess air, and if the combustion equipment can give adequately complete combustion with lower excess air then boiler efficiency will be higher”