Studies On Process Modeling And Its Application For Catalytic Cracking

Fluid catalytic cracking (FCC), the main secondary process of refinery, is a primary process for the profitable conversion of heavy petroleum fractions into valuable fuels such as gasoline, diesel and LPG. Process modeling and simulation is key research and application issue in petrochemical and refinery industry. This dissertation takes FCC as the research background, and focuses on the important typical theoretical and engineering issues of process, modeling and its application. The detailed contents are arranged as following,1. A novel eight lumps kinetic model is developed for the MIP-CGP technology of residual oil catalytic cracking, in which propylene separately as a lump for accurate prediction. According to the oil gas-particle mixing conditions and hydraulic characteristics in the riser reactor, a combination reactor model is developed by integrating ideal Plug Flow Reactor model and multi-stage Continuous Stirred-Tank Reactor model. Plant unit factors are introduced to correct the deviation between model prediction and plant data based on mechanistic analysis.2. The eight lumps kinetic model parameters are benchmarked with plant data by minimization of the sum of the squares of the deviations between plant and predicted values of products yields. In order to reduce the difficulty of parameters estimation because of huge number of parameters, some kinetics parameters from literatures which have similar kind of catalyst are adopted. Different parameter estimation algorithms are discussed and a hybrid particle swarm optimization (HPSO) algorithm is developed for the parameter estimation of the proposed FCC process model. The efficiency of the HPSO algorithm is validated by the successful parameter estimation.3. The off-line application of the developed model about process of FCC is discussed in detail including model verification, process simulation and analysis of reaction. First of all, plenty of plant data are simulated with the eight lumps kinetic model, and the comparisons of products yields between model prediction and plant measured data are made for the warranty of model maintenance.then the distribution of the lump concentration inside the riser reactor is simulated by the models and the rationality of reactor model is validated. Finally the effect of operating variables such as reaction temperature, weight hourly space velocity on the products yields is investigated. Off-line simulation involves model verification and process simulation, and process analysis incorporates simulation of the reactor and sensitivity analysis. The above-mentioned discussion and analysis are necessary for online application of model-based soft sensor and optimization software for FCC processes.4. A model-based soft sensor is developed to predict the product yields of an industrial catalytic cracking unit in real time. The prediction accuracy of the soft sensor is greatly improved by using the triple updating strategy, which is the integration of model kinetics parameter estimation, unit factors updating and bias updating. The soft sensor strategy was successful applied in an industrial FCC plant for advanced process control.5. An online process calculation based on kinetic model is carried out on industrial continuous catalytic cracking process to forecast the yield and distribution of the cracking products. The key problems on practical operation of online soft sensors system are discussed and the strategy of online model correcting is proposed for the warranty of model prediction accuracy. The practical online calculation results show that the model predictions agree well with plant data observed.6. Functions of APC-Sensor software are described. The lump kinetic model of catalytic cracking is taken as a mould of APC-Sensor. As engineering software, it has been applied upon one industrial catalytic cracking unit successfully.