logo

Back

Nov-2013

Consider multivariate statistical modelling to maximise FCC unit profitability

Process models(a) and simulation(b) can be used to simulate the fluidised-bed catalytic cracking (FCC) unit process, kinetic or multivariate statistical models have been used.

Nikolay V Karpov, Lukoil Nizhegorodnefteorgsintes Refinery
Carl Keeley, Jeremy Mayol, Vasil Bozukov, Stefano Riva, Vasileios Komvokis and Stephen Challis
BASF Corporation, Refining Catalysts

Viewed : 3719


Article Summary

Rigorous non-linear reactor kinetic models have typically been used to develop project design yields, to support refinery planning etc. However, kinetic models have had limited application in support of FCC unit operations and optimisation, as they are relatively time consuming and expensive to build and run, and are difficult for the refinery engineer to maintain due to their complexity.

Multivariate statistical models based on operating data and using standard software, provide a suitable alternative for support of operations and optimisation. These models can be readily and cost-effectively developed. These models can be used for unit troubleshooting, optimisation and training purposes. Are also suitable for real-time process monitoring, to detect deviation from expected operation behaviour, and are suitable for evaluating changes in feed and operating variables.

This paper presents several generic examples to demonstrate the power of multivariate statistical process models. It also describes the procedure used to develop accurate unit specific models and perform simulations for the Lukoil NNOS Refinery(c) FCC unit. These models were successfully used at Lukoil NNOS to accurately estimate on-line FCC unit product yields, 
gasoline research octane number and regenerator bed temperature. This supported the refiner in meeting the market demand for a high 
yield of premium gasoline by enabling 
optimisation of catalyst, feed and operating conditions(7).

Introduction

In August 2005 Lukoil announced a major investment to upgrade the NNOS Refineryc. This investment was in response to a market trend for increased gasoline passenger car use3. The outlook for Russia and CIS is decreased demand for regular gasoline in favour of increased demand for premium gasoline working towards Euro V standards4.

In 2005 a licensor was selected to provide the design for a new FCC process unit, and the unit successfully commenced operation in 20116. This is a modern, short contact time riser design FCC unitd with: side-by-side vessel layout; Optimix feed distribution system; Vortex Separation System (VSS) riser termination; Advanced Fluidisation (AF) reactor stripper technology; RxCat riser technology; and combustor style high-efficiency regenerator7.

The licensor’s process technology and equipment, in combination with the appropriate catalyst, enabled Lukoil NNOS to maximise profitability by achieving best-in-class conversion, total liquid product yields and olefin selectivity7.

The catalyst in use is based on the award winning distributed matrix structures (DMS) technology platforme. DMS technology enables high bottoms conversion with low coke for higher yields of gasoline and light olefin products8.

In addition to catalyst, technical service was provided to help the NNOS Refineryc obtain the best value from the catalyst. A technical service requested was to build accurate FCC process unit models to provide simulation capability. Application of multivariate statistical modelling was judged to be the best approach to meet the refinery request.

This paper presents several generic examples to demonstrate the power of multivariate statistical process models. It also describes the procedure used to develop accurate unit specific models and perform simulations for the Lukoil NNOS Refineryc FCC unit.

Benefits of using Statistical Modelling on a FCC Unit
Rigorous non-linear reactor kinetic models have typically been used to develop project design yields and to support refinery planning, especially in defining linear programming (LP) sub-models, and evaluating new FCC feeds. However, application of kinetic models to support of FCC unit operations and optimisation are less common due to the significant effort required to produce accurate results for each operating scenario.1

Kinetic models, even if more accurate, require more specialised knowledge to build and calibrate. Thus, they are relatively time consuming/expensive to run and are difficult for the refinery engineer to maintain.

An attractive alternative is to use multivariate statistical models based on unit operating data. These models can be readily and cost-effectively developed by refinery engineers with the support of their catalyst supplier.

The supplier should have extensive knowledge of FCC operations to be able to guide selection of process parameters to be used in model correlations. In addition, the supplier may have the in-house capability to build statistical models for the refinery, as is the case with BASF.

Benefits of multivariate statistical models are:
• Easy to build using standard software, e.g. MS Excel and Statsoft Statistica.
• Does not require a detailed knowledge of the FCC hardware design to develop.
• Easy to maintain/update, e.g. by updating models if conditions change.
• Enables a detailed understanding of the impact of process variables on unit operations in a transparent and user-friendly way, e.g. can be used for unit troubleshooting, optimisation and training purposes.
• Suitable for real-time process monitoring to detect deviation from expected base operation behaviour, e.g. to support monitoring of a catalyst/additive trial, to identify potential hardware problems, etc.
• Easy to use in prediction mode, e.g. to examine the impact of a different feed quality, change in reactor temperature, etc.
Overall, there are a number of good reasons why a refiner should consider application of statistical modelling to their FCC unit. The power of using this approach to maximise FCC unit profitability is supported by the following generic examples from several FCC units.


Add your rating:

Current Rating: 4


Your rate: