Dynamic Data Reconciliation – What You Can Know And When You Can Know It

Dynamic data reconciliation allows for improved performance compared to steady state reconciliation because the dynamic model has accumulation terms built into into the model. Dynamic reconciliation models also allow for rejection of high frequency noise in an intuitive way. There are limitations on observing unmeasured variables. Two principal causes are correlation between unmeasured variables, and process lag between the unmeasured variables and the measured variables being used to infer values. The paper explores these issues, and to what extent one can mitigate them using variable transformations.