Data analysis, modeling and control performance enhancement of an industrial fluid catalytic cracking unit

Considerable fluctuations were observed in the riser temperature of one of the fluid catalytic cracking (FCC) unit of a Southeast Asian refinery. This undesired occurrence has an adverse effect on the performance of the process unit. In the present study, several statistical tools are developed and then used, for the first time, for analyzing routine operating data in order to characterize the dynamics of the riser temperature and other critical variables that may be affecting the riser temperature. Subsequently, a first-principles-based dynamic model of the FCC unit is implemented to closely simulate the FCC unit under investigation. The model is validated by predicting the measured operating data of the FCC unit. This facilitated an in-depth study of the FCC unit, leading to the identification of several strategies for improving the control loop performance of the riser temperature.     

Five-lump kinetic model with selective catalyst deactivation for the prediction of the product selectivity in the fluid catalytic cracking process

The effective simulation of the fluid catalytic cracking (FCC) operation requires a good understanding of many factors such as, reaction kinetics, fluid dynamics, and feed and catalyst effects. The different product slates that can be obtained are the consequence of a complex reaction scheme including cracking, isomerization, hydrogen transfer, oligomerization, etc. Furthermore, the catalyst deactivation may affect each one of the reactions in different ways, which creates an additional reason for different variation with time-on-stream of the yield to each product. On the basis of the experimental data of the FCC pilot plant operated in Chemical Process Engineering Research Institute (CPERI, Thessaloniki, Greece), a lumping model was developed for the prediction of the FCC product distribution. The lumped reaction network involved five general lumps (gas oil, gasoline, coke, liquefied product gas, and dry gas) to simulate the cracking reactions and to predict the gas oil conversion and the product distribution. The paths of catalyst deactivation were studied and a selective deactivation model was adopted that enhances the fundamentality and accuracy of the lumping scheme. The hypothesis of selective catalyst deactivation was found to improve the product slates prediction. Models with different assumptions were examined, regarding the behavior of the catalyst, as deactivated, and its effect on the reactions of the lumping scheme. A large database of experiments, performed in the FCC pilot plant of CPERI was used to verify the performance of the models in steady state unit operation. The simulation results depict the importance of incorporating selective catalyst deactivation functions in FCC lumping models.     

应用经遗传算法优化的BP神经网络预测催化裂化装置焦炭产率

焦炭是催化裂化装置的主要副产物,准确预测催化裂化焦炭产率对提高装置的操作平稳度和经济效益具有重要意义。人工神经网络(ANN)具有强大的自学习和自适应能力,在非线性预测方面具有明显的优势。本研究将遗传算法(GA)与BP神经网络相结合,基于某炼厂催化裂化装置的生产数据,分别从原料、催化剂和操作条件3个方面选取28个关键影响参数建立了催化裂化焦炭产率预测模型,分别将BP神经网络和经遗传算法优化的BP神经网络(GA-BP)的预测结果与工业数据进行对比。结果表明,经遗传算法优化的预测模型无论在预测结果的准确性还是稳定性方面效果更好。最后,本研究还通过考察原料残炭、反应温度等单一关键参数对焦炭产率的影响,进一步证明了经遗传算法优化的BP神经网络预测模型的准确性。     

Modeling of industrial nylon‐6,6 polymerization process in a twin‐screw extruder reactor. II. Neural networks and hybrid models

This article describes the application of neural networks and hybrid models to the finishing stage of nylon‐6,6 polycondensation in a twin‐screw extruder reactor. A planned experiment in the industrial and in the pilot plant was employed to build the neural network and the hybrid model. The hybrid model combines information calculated from the phenomenological model with the neural network model. The comparison of experimental with calculated data shows good agreement. During two years, industrial data were collected. The comparisons of the models' prediction with these data were performed and reasonable results are achieved from the industrial point of view. These models help an increase of industrial production of about 20%. © 1999 John Wiley & Sons, Inc. J Appl Polym Sci 72: 905–912, 1999     

Kriging models for forecasting crude unit overhead corrosion

Crude unit overhead corrosion is a major issue in the refinery field. However, the corrosion models in the literature are difficult to apply to real refinery processes due to the characteristics of corrosion. We propose a Kriging model, an advanced statistical tool for geostatistics, to forecast the corrosion rate in a real refinery plant. Instead of spatial coordinates, the proposed model employs the non-spatial coordinates of six key corrosion variables: H₂S, Cl^?, Fe²⁺, NH₃, pH, and flowrate. The Kriging model is compared with two well-known forecasting models, multiple linear regression and an artificial neural network. To overcome the insufficiency of the number of data sets measured in the plant to use the six non-spatial coordinates, the significance probability is applied to reduce the dimensions from six to four. Among all the developed models in this paper, the Kriging model with four corrosion variables showed the best forecasting performance.     

Soft Sensor design for a Topping process in the case of small datasets

In this paper, a new strategy to cope with the identification of nonlinear models of industrial processes, when a limited number of experimental data is available, is proposed. The approach is intended to improve the generalization capabilities of the model and it is based on the integration of bootstrap resampling, noise injection and neural model stacking. A number of algorithms to stack the first level neural models are also compared. The method proposed has been applied to develop a Soft Sensor for the estimation of the Freezing Point of Kerosene in an atmospheric distillation unit (Topping) working in a refinery in Sicily, Italy. The improvements obtained thanks to the strategy proposed, with respect to a classical neural model, are shown in the paper.     

Integrated FCC riser—regenerator dynamics studied in a fluid catalytic cracking pilot plant

In this paper a dynamic simulator of the fluid catalytic cracking (FCC) pilot plant, operating in the Chemical Process Engineering Research Institute (CPERI, Thessaloniki, Greece), is presented. The operation of the pilot plant permits the execution of case studies for monitoring of the dynamic responses of the unit, by imposing substantial step changes in a number of the manipulated variables. The comparison between the dynamic behavior of the unit and that predicted by the simulator arise useful conclusions on both the similarities of the pilot plant to commercial units, along with the ability of the simulator to depict the main dynamic characteristics of the integrated system. The simulator predicts the feed conversion, coke yield and heat of catalytic reactions in the FCC riser on the basis of semi-empirical models developed in CPERI and simulates the regenerator according to the two-phase theory of fluidization, with a dilute phase model taking account of postcombustion reactions. The riser and regenerator temperature, the stripper and regenerator pressure drop and the composition of the regenerator flue gas are measured on line and are used for verification of the ability of the simulator to predict the dynamic transients between steady states in both open- and closed-loop unit operation. All the available process variables such as the reaction conversion, the coke yield, the carbon on regenerated catalyst and the catalyst circulation rate are used for the validation of the steady-state performance of the simulator. The comparison between the dynamic responses of the model and those of the pilot plant to step changes in the feed rate and preheat temperature reveals the ability of the simulator to accurately depict the complex pilot process dynamics in both open- and closed-loop operation. The dynamic simulator can serve as the basis for the development of a model-based control structure for the pilot plant, alongside its use as a tool for off-line process optimization studies.     

Long Term Pilot Plant Experience on Aromatics Extraction with Ionic Liquids

Since 2004, we have been conducting pilot plant trials with various contactors and different ionic liquids for petrochemical model feeds as well as real refinery feeds. Our pilot plant contains a Rotating Disc Contactor with a height of 6 m and a diameter of 60 mm. Up to 100 kg of ionic liquid and 200 L of feed are applied in experiments. In this paper, the hydrodynamic and mass transfer performance of a rotating disc contactor has been characterized for toluene/n-heptane, a model FCC feed and a real LCCS refinery feed using 3-methyl-N-butyl-pyridinum dicyanamide ([3-mebupy]N(CN)₂) as the ionic liquid. Experiments with the real LCCS feed demonstrated comparable extraction performance to the model FCC feed. Over the past 5 years, the same batches of the ionic liquid have been repeatedly regenerated by evaporation and reused. Our studies revealed that, although coloration occurs, the extraction performance of the ionic liquids has not changed even after several years of usage. After the real LCCS refinery feed experiments, validation with the toluene/n-heptane model feed confirmed unchanged extraction performance.

Supplemental materials are available for this article. Go to the publisher's online edition of Separation Science and Technology to view the free supplemental file.     

富乙烯气预分馏系统对裂解装置的影响分析

燕山石化于2005年建成一套回收催化裂化干气制乙烯系统,包括干气变压吸附与净化装置（在炼厂）和富乙烯气预分馏装置（在乙烯厂）.分析了富乙烯气预分馏装置的运行对乙烯装置的影响及解决措施.     

Effect of metals poisoning on FCC products yields: studies in an FCC short contact time pilot plant unit

In this work the effects of two metal poisons (Ni, V) on FCC products were investigated in an FCC pilot plant. The most important effects of metals were found on the gasoline, coke and H₂ yields. A comparison study of the metal distribution in the catalyst particles aged in the FCC pilot plant, a CDU and an industrial FCCU was also performed using a SEM–EDS method. SEM results showed that the metal profiles from CDU samples simulate more satisfactorily the profiles of the E-cat than that from the FCC pilot plant.     

Simulation of the ammonia oxidation process over an oxide monolith catalyst

A mathematical model of ammonia oxidation over an oxide monolith honeycomb catalyst has been developed. Experimental data from a pilot unit operating under process conditions equal to those in a type UKL-7 commercial nitric acid plant were used to refine the parameters of the mathematical model. The results from calculating the NO yield are in satisfactory agreement with the pilot unit data. There is now computer program that allows us to calculate the conditions of an industrial reactor functioning.     

一种混杂系统数据校正新方法

对于既包含连续生产过程又包含离散事件的混杂系统,尤其是对于带有生产方案切换的实际生产过程,通过在物料平衡模型中引入随机调度方程,从而构造出包含随机调度方程参数变量θ的新型协调模型,然后利用一种不确定模型的协调算法对此模型进行求解,最后,通过仿真研究证实了该方法的有效性和鲁棒性.     

Disaggregation-aggregation based model reduction for refinery-wide optimization

In this paper, reduced nonlinear refinery models are developed by generating and using input-output data from a process simulator. In particular, rigorous process models of continuous catalytic reformer (CCR) and naphtha splitter units are used for generating the data. To deal with complexity associated with large amounts of data, that is usually available in the refineries, a disaggregation-aggregation based approach is presented. The data is split (disaggregation) into smaller subsets and reduced artificial neural network (ANN) models are obtained for each of the subset. These ANN models are then combined (aggregation) to obtain an ANN model which represents all the data originally generated. The disaggregation step can be carried out within a parallel computing platform. Refinery optimization studies are carried out to demonstrate the applicability and the usefulness of the proposed model reduction approach.     

固相环流技术在催化裂化装置的工业应用

扬子石化炼油厂800kt/a催化裂化(FCC)装置引入固相环流技术,分别对反应器的预提升段及汽提器的结构进行了一系列改造,提高了油剂接触效率和油气与催化剂分离效果。工业应用结果表明,装置的生焦量大幅降低,提高了轻质油产品的收率。     

A new generic approach for the modeling of fluid catalytic cracking (FCC) riser reactor

A new kinetic model for the fluid catalytic cracking (FCC) riser is developed. An elementary reaction scheme, for the FCC, based on cracking of a large number of lumps in the form of narrow boiling pseudocomponents is proposed. The kinetic parameters are estimated using a semi-empirical approach based on normal probability distribution. The correlation proposed for the kinetic parameters’ estimation contains four parameters that depend on the feed characteristics, catalyst activity, and coke forming tendency of the feed. This approach eliminates the need of determining a large number of rate constants required for conventional lumped models. The model seems to be more versatile than existing models and opens up a new dimension for making generic models suitable for the analysis and control studies of FCC units. The model also incorporates catalyst deactivation and two-phase flow in the riser reactor. Predictions of the model compare well with the yield pattern of industrial scale plant data reported in literature.     

Multilevel surrogate modeling of an amine scrubbing process for CO2capture

Surrogate models provide a powerful method for simplifying calculations within complex simulations. While surrogate models are broadly applied within chemical engineering, little research exists investigating the level of surrogacy's impact on a simplified process model. In this work, artificial neural networks (ANN) and Kriging models are used as surrogate models at the process, process unit, and thermodynamic levels for a CO₂ amine scrubbing process. The surrogated models are evaluated against an Aspen Plus simulation for accuracy, convergence behavior, computational cost, and ability to extrapolate. The thermodynamic and process unit models can better handle discontinuous, non-smooth behavior, and convergence issues in the surrogated truth model, but poor conditioning in the final system of equations results in a lower accuracy and convergence rate than the process level surrogate. Beyond model accuracy, availability of diverse data, intended re-usability, and the desired outputs must be considered when selecting a level of abstraction.     

Dynamic modelling of an industrial R2R FCC unit

The aim of this study is to obtain a model that can simulate the performance of an industrial fluid catalytic cracking (FCC) unit in steady and dynamic state, and which will subsequently be used in studies of control and real time optimisation. In this paper, a dynamic model for a R2R type FCC unit is presented. The model includes the riser, the stripper/disengager, the regeneration system and the catalyst transport lines. Mass, energy and pressure balances are performed for each of these sections.Simulation results for steady state are presented and compared qualitatively to those obtained from previous FCC models. The dynamic behaviour of the system is explored through two perturbations in open loop, one on the fresh feed flow rate and one on the air flow rate to the first regenerator. The results illustrate the consistency of the model and are in agreement with what has been observed in studies available in the open literature.     

Product tri-section based crude distillation unit model for refinery production planning and refinery optimization

Accuracy of a crude distillation unit (CDU) model has a significant impact on refinery production planning. High accuracy is typically accomplished via nonlinear models which causes convergence difficulties when the entire refinery model is optimized. CDU model presented in this work is a mixed-integer linear model with a modest number of binary variables; its accuracy is on par with rigorous tray to tray CDU models. The model relies on the observation¹² that a line through the middle of the product true boiling point (TBP) curve depends on the crude feed properties and the yields of the adjacent products. Novelty of the product tri-section CDU model is that it does not require models of individual distillation towers comprising the CDU, thereby leading to a much simpler model structure. Significant reduction in the computational effort required for the optimization of nonlinear refinery models is illustrated by comparison with previous work.     

Dynamic neural network modeling for hydrochloric acid recovery process

This paper describes neural network models for the prediction of the concentration profile of a hydrochloric acid recovery process consisting of double fixed-bed ion exchange columns. The process is used to remove the Fe²⁺ and Fe³⁺ ion from the pickling liquor, resulting in increasing the acid concentration for reusing in the pickling process. Due to the complexity and highly nonlinearity of the process, the modeling of the process based on the first principle is difficult and involve too many unknown parameters. Therefore, an attractive alternative technique, neural network modeling, has been applied to model this system because of its ability to model a complex nonlinear process, even when process understanding is limited. The process data sets are gathered from a real hydrochloric acid recovery pilot plant and used for neural network training and validation. Backpropagation and Lenvenberg-Marquardt techniques are used to train various neural network architectures, and the accuracy of the obtained models have been examined by using test data set. The optimal neural network architectures of this process can be determined by MSE minimization technique. The simulation results have shown that multilayer feedforward neural network models with two hidden layers provide sufficiently accurate prediction of the concentration profile of the process.     

流化催化裂化(FCC)催化剂跑损机制及故障树分析

炼油厂中流化催化裂化（FCC）装置催化剂跑损的故障原因分析多数来自现场工程师,在故障机理方面少有报道。为了解决这一问题,本文利用故障树分析方法（FTA）,研究FCC装置催化剂跑损机制。采用催化剂跑损为顶事件,结合跑损途径和跑损机理,确定FCC装置故障、操作工艺异常和催化剂颗粒物性3个因素作为中间事件,并通过逐层向下深入分析,确定诸如翼阀磨损等21个因素作为底事件,建立催化剂跑损故障树模型。根据FCC装置故障树风险分析,得到任何一个底事件出现都有可能导致顶事件发生,且对FCC装置催化剂跑损的贡献度相同。研究结果表明：利用FTA方法可以更深层次了解装置跑剂原因,对考察FCC装置催化剂跑损机理具有指导意义,并提出了相应的故障判定流程和跑剂预防措施。