bimetallic steam trap

As the name implies, bimetallic steam traps are constructed using two strips of dissimilar metals welded together into one element. The element deflects when heated. (Figure 1):

There are two important points to consider regarding this simple element:
Operation of the steam trap takes place at a certain fixed temperature, which may not satisfy the requirements of a steam system possibly operating at varying pressures and temperatures (see Figure 2).

Because the power exerted by a single bimetal strip is small, a large mass would have be used which would be slow to react to temperature changes in the steam system.
The performance of any steam trap can be measured by its response to the steam saturation curve. The ideal response would closely follow the curve and be just below it. A simple bimetal element tends to react to temperature changes in a linear fashion.

Figure 2 shows the straight line characteristic of a simple bimetal element relative to the steam saturation curve. As steam pressure increases above P1, the difference between steam saturation temperature and trap operating temperature would increase. Waterlogging increases with system pressure, highlighting the trap's inability to respond to changing pressure conditions.

It needs to be noted that at pressures below P1, the steam trap operating temperature is actually above the saturation temperature. This would cause the steam trap to pass steam at these lower pressures. It may be possible to ensure the steam trap is adjusted during manufacture to ensure that this portion of the saturation curve is always above the operating line. However, due to the linear action of the element, the difference between the two would increase even more with system pressure, increasing the waterlogging effect.

Clearly, this is not a satisfactory operation for any steam trap, and various attempts have been made by manufacturers to improve upon the situation. Some use combinations of two different sets of bimetal leaves in a single stack, which operate at different temperatures (Figure 3).

The typical result is the split response line similar to that shown in Figure 4. This is an improvement on Figure 2, but still does not exactly follow the saturation curve. One set of bimetal leaves deflect to give the response P1 to P2. At a higher temperature a second set of bimetal leaves contributes to give response P2 to P3. Clearly, although an improvement from the former design, this is still unsatisfactory in terms of following the saturation curve.

A more innovative design is the disc spring thermostatic element shown in Figure 5. The thermostatic element is made up of a set of bimetal discs. These discs, if acting directly between the valve stem and the seat (as with some thermostatic steam traps), cause the discharge temperature of the condensate to change linearly with changing pressure (curve 'A', Figure 6). By incorporating a spring washer between the discs and a recess in the seat, this absorbs some of the bimetal expansion at low pressure so that a greater temperature change must occur with changing pressure. The spring washer shape is preferred over a coil spring because it develops force in an exponentially increasing rate, rather than in a linear rate. This effect takes place up to 15 bar g until the spring is deflected to the bottom of the recess, and means that the discharge temperature of the condensate will follow the steam saturation curve more accurately (curve 'B', Figure 6). Discharge rates are also improved by the dynamic clack which tends to produce a blast discharge.