The art and science of boiler drum water level measurement has long been the source of questionable results. A low-tech examination of the boiler drum reveals where the drum level has been controlling — there is a water mark inside the drum. The problem is obvious; the mark in the drum and that in the control system do not agree. The level is off by a wide margin. This paper seeks to explain how poor assumptions are the source of the error. Then two design approaches are considered to eliminate the sources of error. Consider that boiler drum water is not saturated; feedwater entering from the economizer is below saturation. Also consider the constant head chamber. Typically, the reference leg is considered to be at ambient temperature, but it is really a gradient starting at near drum temperature. The effect of these two poor assumptions is a relatively large error such that the true level is different from indicated level.
Once this situation is understood, it is possible to redesign instruments which take the actual situation into account. One such instrument is nearly the same as the original DP-type instrument, but the constant head chamber is located inside the drum. This design change deals with both errors. It eliminates the differences in temperature and density between the constant head leg and the water inside the drum. It also exposes the constant head leg to the exact same temperature as the boiler water, thus eliminating the need for complicated (and incorrect) formulae to compensate for the differences in density. A second design approach is to “steam jacket” an instrument located outside the drum, thus eliminating the differences in temperature and density. This is an appropriate way to deal with the Conductivity Probe and the Boiler Water gauge glass. By steam jacketing these instruments, we can make the water temperature between the indicator and the drum almost the same.