Flexibility analysis of chemical processes considering overlaying uncertainty sources

Novel Reformulations for Modeling Uncertainty and Variations in a Framework for Chemical Process Design

“…Additionally, disturbances around the nominal point (current and future) may be present which are comparable to the traditional interpretation of operational uncertainty (which Swaney and Grossmann 19 aimed to analyze by means of the flexibility index). In line with Marton et al 33 and Langner et al, 34 we refer to this (more traditional) interpretation of uncertainty as short-term operational disturbances. Note that these short-term operational disturbances may also be affected when the nominal operating point changes, for example, due to a singular event.…”

“…To identify the maximum feasible change of the nominal value, the flexibility index problem needs to be formulated for different discrete values, that is, discrete points on the line illustrating the expected (maximum) change of the nominal value of θ 2 in Figure 4, and the resulting formulations can then be solved in an iterative scheme. To avoid such a time-consuming iterative scheme, we 34 have previously presented a reformulation of the original flexibility index problem (formulated by Swaney and Grossmann 19 ), which allows for considering uncertainty in the nominal value, and which yields the maximum feasible change of the nominal value such that all expected short-term operational disturbances are (exactly) feasible. The solution obtained when applying this reformulation to the flexibility index problem for the theoretical example is shown in Figure 4 and marked as the feasible maximum shift.…”

“…In comparison to the first and the second theoretical example, the example shown in Figure includes a new dimension of uncertainty which is the combination of the uncertain change of the nominal value and the short-term operational disturbances (see also Langner et al where this is referred to as overlaying uncertainty sources). Even for the case of a retrofit where historical measurement data may be available to describe current short-term operational disturbances, this kind of analysis relies on assumptions regarding how the operation of the process will be affected by the rare event; that is, uncertainty in data is added to the uncertainty in operation.…”

“…Another cause of independent operating periods can be that the nominal operating point changes due to a (future) uncertain singular/rare event, as discussed by Marton et al 33 and Langner et al 34 Examples of such a singular/rare event are a permanent switch of feedstock, a change of operational parameters required to comply with new emission legislation and/or a change in the production rate. All these events may have a lasting effect on the operation of the process in question, and a possible consequence is that the nominal operating conditions change temporarily or even permanently.…”

“…Additionally, we show that the application of traditional models may lead to a significant discrepancy between the modeled uncertainty space and the actual uncertainty space for problems characterized by the occurrence of (multiple) independent operating periods. To overcome these issues, we adopt new approaches to model the expected uncertainty space with high accuracy, based on our recently published studies on flexibility analysis considering parameter dependencies as well as multiple independent operating periods. , One consequence of inexact modeling of the expected uncertainty space can be that identified critical operating points represent combinations of uncertain parameter values which are not expected to occur simultaneously, which in turn can result in unnecessary overdesign of the equipment. To avoid this, new approaches to model the expected uncertainty space can also be used to identify critical operating points.…”