FCC and Hydrocracker/Hydrotreater Feed Characterisation Issue

Please be advised that an issue has been found around feed characterization for FCC and Hydrotreating/Hydrocracking reactor models. We are fixing this issue but want you to be aware of the impact that this may have on case results.

Affected Versions for FCC:

  • Petro-SIM 4.0 and service packs
  • Petro-SIM 4.1 and service packs 1 and 2
  • Petro-SIM 5

Affected Versions for Hydrotreating and Hydrocracking (HXX):

  • Petro-SIM 4.0 and service packs
  • Petro-SIM 4.1 and service packs 1 and 2

The impact of this issue is largely isolated to refinery-wide cases and potentially a handful of single unit model cases. We can summarize the impact as making the feed characterization more conservative than it should be:

  • For FCC-SIM it tends to make bad feeds (heavy and aromatic) worse than they actually are, with little difference on light feeds
  • For HXX-SIM it tends to make naphtha, diesel and LVGO feeds more aromatic and residue feeds less aromatic

Background

In Version 4 and onwards, we introduced a new feed characterization method in the FCC and HXX models that is a function of feed gravity, UOP K factor and viscosity at 100 C, which is estimated from an API correlation where absent. The issue is a bug between the temperature at which Petro-SIM calculates the viscosity and transfers it to the reactor models and the temperature which the reactor model expects from Petro-SIM, where this bug can cause a deviation in the feed characterization and specifically in the estimated core aromatics value.

Because viscosity is not a required input to either reactor model, many cases will not be affected. Cases where viscosity was not directly supplied will not be impacted. Cases where viscosity was supplied as a feed property or can be calculated from assay information for feeds are subject to this issue. The largest impact of this issue will be to refinery wide models or integrated models in which the viscosity is an assay property that is passed from Petro-SIM to the reactor model. The impact will be larger on the FCC because of the importance of feed aromatics on conversion; our comparison testing has shown little relative difference on hydroprocessing models. In the FCC the impact is to over-predict core aromatics, causing the FCC model to under-predict conversion. The effect will be most noticeable on cases with heavy viscous feeds.

Solution

This issue has been corrected in 4.1 sp3 build 707 and will be corrected in 5.0 sp1 build 1040 which will be available shortly (remember it is not an issue for HXX there). As part of this fix, Petro-SIM will check the extent of the impact of the fix on the results of the reactor model and advise users when the impact is worth checking into. Furthermore, the corrected builds add a new option to the Feed CA Characterization so that the reactor models always uses the internal API estimate to address situations where viscosity assay values are not reliable. This will mitigate potential negative situations where poor viscosity measurements can have a negative impact on reactor results and is also the recommended method where you have a mixture of feeds with known and unknown viscosity.

The new option for Feed CA Characterization will become the default for all new cases built in 4.1 sp3 build 707 and 5.0 sp1 build 1039 upwards, reflecting concerns raised within the consulting community on the reliability of crude assay viscosity data.

Please note that the FCC and HXX reactors from 4.1 sp1 onwards support the original RI-based feed characterization method in addition to the new methods and direct entry of COA and CON.

FCC - Workarounds for existing builds

There are two approaches you can take depending on the availability and reliability of feed viscosity measurements and predictions for each feed. We recommend the first option which has the advantage of being easy to implement. Our testing has shown results from it to be very consistent for the most part with the second method where viscosity values are available and reliable. The first option is the approach we will take for all new cases with the corrected versions of the program.

We strongly recommend using the first option if the FCC model has a mixture of feeds where viscosity is only available for some of those feeds.

  1. Allow FCC-SIM to characterize from feed density and UOP K, estimating viscosity as part of this
    • Go to the FCC unit operation property view in Petro-SIM and locate the Feed Data page on the Operating Data tab
    • Enter a zero value in the Viscosity row for each feed. This overrides any calculated feed viscosity and triggers the FCC model to estimate it internally
  2. Override the viscosity calculated from feed with the correct value
    • Add a spreadsheet unit operation to your case
    • Import the Viscosity (Kinematic) at 100 C for each feed into cells in the spreadsheet
    • Set up a formula to create a copy of each imported value that you can export
    • Export the formula result values to the FCC by dragging and dropping the cell value onto the appropriate cell in the Feed Data page of the FCC Operating Data tab

The issue affected the handling of viscosity in calibration and prediction so we recommend you look at how a specific FCC model was calibrated. If the calibration feeds included viscosity then the model will be consistently handling things, albeit using the wrong viscosity value and as such applying one of these workarounds should be done in both calibration and prediction. If the calibration feeds did not include viscosity then we recommend the first option here as your best path, where it can be applied to predict uses of the model only.

Please contact the software helpdesk if you have any questions regarding this alert. You may also wish to read the supplementary information that provides guidance on how to check for the likely impact of the issue on existing cases you may have.