Data Requirements and Validation

Boiling-Point Range Suggested Number of Hypothetical Components 

Any modeling exercise requires a reasonable set of input data to ensure that models remain valid and predictive over a wide range of operating scenarios. The complex nature and composition of crude feeds present additional modeling com-phcations when compared to a process that uses a well-defined component slate. 

Collecting the maximal possible process information is the best way to ensure that a model is vahd for a variety of operating scenarios. However, it is frequently too expensive or simply infeasible to collect detailed information during regular operation of the crude tower. Consequently, we must work towards building models that do not require detailed information but also remain valid and predictive over a variety of operating scenarios. 

The most important factor in the success of any model of crade columns is an accurate representation of the crude feed. There are two ways to quantify the crude feed to the unit The first method relies on the availability of cmde assays and knowledge of the ratios of the crude mixes fed to the unit This is particularly useful when a column only processes a few types of crude. The other method uses current column product yield and qualities to back-mix or back-blend these products. The goal o( back-blending is to recover the composition of the crude fed to the column. This method is very useful when we have little information of the crude fed to the column or the assays are too old and unreliable. Kaes (1) provides methods to estimate missing data when back-blending data for modeling. 

Once we have estimated the composition crude feed, we must also collect data that summarize the column operating conditions and profile. Table 2.44 is a basic list of data that we require to develop a reasonable model for the atmospheric unit 

---

Data Requirements and Validation 71 Table 2.6 Arab Heavy Density Distribution. 

---