First,
THANK YOU Erik for pointing out that study. I haven?t had time to look at it but it?s something that could be useful.

Brad,
You are spot on with your comment. Despite its name, I currently consider the ?MIN-RATIO? for the DOE2/eQ boiler models to really be the minimum PLR value at which the HIR f(PLR) curve is valid, and simulated hourly efficiency is held steady below that value (w/some adjustments if using DOE-2.2). If ?MIN-RATIO? was part of a mechanistic model of a boiler (instead of a curve fit model), it would certainly be set to 100% for a single stage boiler. Also note the eQUEST/DOE-2.2 standard boiler representation includes ?Start-up Time? and ?Standby Time? which affect hourly efficiency below the ?MIN-RATIO? and attempt to account for losses and mass of the boiler when in cycling mode. See the Misc tab in the Boiler forms.

The ?default? eQ curve, also cited in T24 ACM/COMNET, results in extremely low hourly efficiencies below 20% part load. The default curve for EnergyPlus (cited in the follow-on PNNL guide) uses a different representation Efficiency f(PLR) instead of HIR f(PLR) and it results in less degradation at very low loads. T24 ACM/COMNET/PNNL cites a 25% ?default? for the baseline input for MIN-RATIO, but it?s not clear exactly why. In my limited experience with older, atmospheric commercial boilers, most stage 33%, 66%, 100% (small burner, large burner, both burners).

Note that real, in-place boiler efficiency at low loads will depend on a great number of factors, including the boiler water temperature (which is not considered in these representations) and whether the water is pumped through the boiler continuously or only when the boiler cycles on. FORTUNATELY, all the nasty ?traditional? standby losses are minimized, but not eliminated, in decent modern lightweight forced-draft boiler and HW system design.

In the chart below, note the non-linear horizontal axis scale, for clarity of comparison below 30% PLR.
[cid:image001.png at 01D26B24.EA04E630]

Fred