T. Mcavoy et R. Miller, Incorporating integrating variables into steady-state models for plantwidecontrol analysis and design, IND ENG RES, 38(2), 1999, pp. 412-420
In many processes level loops are important in terms of the operability of
plantwide control schemes. Arkun and Downs have presented an approach from
which a steady-state gain matrix which includes the rate of change of level
s can be calculated. Their approach makes use of a dynamic process model, w
hich is often not available in practice. Steady-state process simulators as
sume that levels are perfectly controlled. Thus, these simulators cannot be
used directly to obtain gain matrices that include level variables. This p
aper presents an approach to using steady-state simulators to obtain approx
imately the same gain matrix as that in the Arkun and Downs approach. Once
this matrix is calculated, it can be used to assess the operability of plan
twide control schemes. The methodology is developed using a distillation to
wer system, and then it is applied to analyze the "snowball effect" in a th
ree reactor/three distillation tower plant. The methodology is also applied
to-the Tennessee Eastman process where level dynamics are more difficult t
o model and accuracy issues arise.