IECM 12.0 beta User Manual > Introduction to Uncertainty Analysis >

A Technical Example

A second example focuses on a performance parameter for an advanced pollution control system. This parameter has an important effect on system performance and cost.

The example focuses on an assessment of uncertainty in the performance of an innovative emission control system for coal-fired power plants. In this system, a chemical sorbent circulates between a fluidized bed reactor, where SO2 in the flue gas is removed by chemical reaction with the sorbent, and a regenerator, in which SO2 is evolved in a reaction of the sulfated sorbent with methane. There is no commercial experience with this system; the largest test unit has been sized to handle 100 scfm of flue gas. Furthermore, the test units have used batch, rather than continuous, regeneration.

One of the key parameters affecting the performance and cost of this system is the regeneration efficiency, which is defined as the fraction of the spent sorbent which is converted for reuse. In small-scale tests in which the regeneration efficiency has been estimated, the efficiency was found to be roughly 30 to 50 percent. In a more recent test, the regeneration efficiency was not measured due to instrumentation difficulties; however, it may have been lower than the previously obtained values. Regeneration residence times were typically greater than 30 minutes.

A detailed modeling study of the regenerator estimated that a properly sized and designed unit, coupled with heating of the sorbent to a sufficiently high reaction temperature, would result in a regeneration efficiency of just over 99 percent at a 30-minute residence time.

A potential problem that may be occurring in the test units is that regenerated sorbent in the regenerator may be reabsorbing some of the evolved SO2. However, this was not considered in the modeling study of the regenerator.

Based on this information, it appears that it may be possible to achieve the design target of over 99 percent regeneration efficiency. Clearly, however, it is possible that the actual efficiency may be substantially less than this target value. As a worst case, we might consider the known test results as a lower bound. Thus, there is a small chance the regeneration efficiency may be less than 50 percent. We expect the regeneration efficiency to tend toward the target value of 99.2 percent. Thus, to represent the expectation that the efficiency will be near the target value, but may be substantially less, we can use a negatively skewed distribution. In this case, we assume a triangle with a range from, say, 50 to 99.2 percent with a mode also at 99.2 percent. The triangle in this case gives us a distribution with a mean of about 83 percent and a median of about 85 percent. This type of triangular distribution, in which a minimum, maximum, and modal value are specified, is often a convenient way of expressing uncertainty distributions when a little information is available.