Comparison of MPC and PI Controller Performance in Preflash and Pipestill Column System

CATEGORY: ENERGY CONSERVATION
Proceedings of the 6th International Conference on Process Systems Engineering (PSE ASIA)
25 - 27 June 2013, Kuala Lumpur.
Comparison of MPC and PI Controller Performance in Preflash and Pipestill Column System
Renanto Handogo (a) and Indra Lesmana (a)
renanto@chem-eng.its.ac.id
a Department of Chemical Engineering, Institut Teknologi Sepuluh Nopember, Surabaya 60111, Indonesia
ABSTRACT
Preflash and Pipestill Column are two main units of petroleum processing industry which are interconnected each other. Recently the main unit used by this industry is Crude Oil Refinery Unit. As time goes by, the use of preflash and pipestill column becomes more popular because of its ability to reduce energy cost. The operation guidelines for these units always aim to safety and certain product flowrate and quality. In this work, assumptions are made where, process variables always change as the process goes. Model Predictive Control was used because it had better response compared to that of a conventional controller.
The steady state and dynamic simulation was conducted using Aspen DynamicTM, and MatlabTM .A ± 15% feed flowrate disturbances and ± 25% OIL-1 mass fraction disturbances were applied to the process. The dynamic performance of the controllers was compared between MPC and conventional controller using Integral of The Absolute Value of The Error/Set Point (IAE/Set Point), as a criteria how good the performance of the controller was. The result showed that the dynamic performance of MPC controller were more superior than that of conventional controller.
Introduction
After the energy crisis in 1970s the revamping of plant becomes very interesting since it can improve energy recovery and lowering operation cost. The typical case for revamping plant is revamping of an oil refinery plant. Following the revamping of plant, many efforts in optimizing energy within preflash and pipestill system were done Control configuration and control method using conventional controller was reported. Modifying the product composition controller using the temperature at the second stage as a manipulated variable only made small difference in the product composition control performance. Conventional controller showed a good performance when the prelash and pipestill system were given ±1% feed mass flowrate disturbances. There was a small offset in LCT quality, when disturbance is given. When the system was given ±2% disturbances, it exposed unstability (Handogo et al, 2010). Pipestill control using conventional controller was declared to have a good performance (Seborg, 2004), however people still demanding better controller performance. Since the first generation of MPC system were develop in 1970s the use of MPC become very popular. In the survey (Qin et al, 2003) more than 4600 application of MPC around the world at the end of 1999 were reported. This amount is twice as the result of five years before. This survey also reported that the majority of the user of MPC were from oil refineries and petrochemical plants.
GPC (Generalized Predictive Control) which is an adaptive MPC has developed. GPC was considered have a better performance compared to the adaptive controller using conventional controller. The comparison of QDMC and conventional controller to control pipestill was studied, QDMC is a new generation of MPC. It was reported that QDMC have a better performance. Furthermore QDMC was more easy to be applied in the plant compared to the conventional PI controller.
Aspen Plus has been widely applied in industrial and academic process simulation and design. It is also supported by regularly updated data from the US National Institute of Standards and Technology which provide the access to the best available experimental property data. In addition Aspen Tech have accumulated sufficient market and and size of sales to support an infrastructure for the continuous development and marketing. However, a control method which can handle feed composition disturbance and the usage of MPC in the preflash and pipestill system have not been studied yet. In this work MPC and conventional controller is used and compared to give a new sight about a control method which can control the feed composition disturbance and give a suggestion whether to use MPC or conventional controller.
Decoupling control was not used as the time constant of the processes differ significantly and imperfect model can not improve the performance of the system.
Free Full Text Source: http://sps.utm.my/wp-content/uploads/2013/09/PSEAsia2013-75.pdf  

Steam reforming plant optimization with Model Predictive Control

CATEGORY: MPC – MODEL PREDICTIVE CONTROL
2013 IEEE 18^th Conference on Emerging Technologies & Factory Automation (ETFA), 10-13 Sept. 2013, Page(s): 1 – 8, Cagliari, Italy
Steam reforming plant optimization with Model Predictive Control
Zanoli, Silvia M. ; Orlietti, Lorenzo
DII - Dipartimento dell'Ingegneria dell'Informazione, Università Politecnica delle Marche, Via Brecce Bianche, 60131 Ancona AN, Italy
Abstract
Describes the optimization of a steam reforming unit located in a petrochemical plant. MPC, or Model Predictive Control , was used to address the need to enhance efficiency, to increase profitability, and to meet precise production standards. Implemented in an actual plant, the performance of the MPC system was compared with the performance obtained with previous PID controllers.
Hydrogen demand in refineries is increasing due to clean-fuels programs. Hydrocarbon steam reforming is currently the main technology used to produce hydrogen. Most reforming plants are controlled with standard PID control loops that assure safety requirements and good performances for single loops.  However, with these conventional methods, designing integrated solutions for controlling systems with interacting variables and constraints is difficult. In many cases the operators run the systems so as to guarantee the safety of the processes, neglecting those aspects related to profitability.
Authors have addressed the need to follow rapidly changing economic factory goals involving many variables and constraints. Their goal is development and implementation of a Model Predictive Control (MPC) for the optimization of a steam reforming plant located in a petroleum refinery plant.
Full Text Source (Subscription or Fee): http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=6647993&contentType=Conference+Publications 

Model Predictive Controller Performance Monitoring Based on Impulse Response Identification

CATEGORY: PREDICTIVE CONTROL
Ind. Eng. Chem. Res., 2013, 52 (23), pp 7803–7817, DOI: 10.1021/ie303593j
Model Predictive Controller Performance Monitoring Based on Impulse Response Identification
Zhong Zhao，Feng Song，Hailiang Yang
zhaozhong@mail.buct.edu.cn
College of Information Science & Technology, Beijing University of Chemical Technology, Beijing 100029,China
Abstract
While model predictive control has been widely applied, many such controllers cannot be operated for long durations. This is due to the lack of a model predictive controller performance monitoring system, which precludes the ability to self-heal.
Authors present a method of identification of impulse response for closed-loop sensitivity function and complementary sensitivity function based on Haar scale transform. Combining the principle of model predictive control and robustness analysis, a method of model predictive controller performance monitoring based on identified impulse response for closed-loop sensitivity function and complementary sensitivity function is proposed. Simulation results and industrial application results are provided to demonstrate the feasibility and effectiveness of the proposed method.
Full Text Source (Subscription or Fee): http://pubs.acs.org/doi/abs/10.1021/ie303593j  

Infinite horizon MPC applied to an industrial FCC converter

CATEGORY: FLUID CATALYTIC CRACKING
2013 9th Asian Control Conference (ASCC), 23-26 June 2013, Page(s): 1 – 6, Istanbul, Turkey
Infinite horizon MPC applied to an industrial FCC converter
Márcio A. F. Martins and Darci Odloak
odloak@usp.br
Department of Chemical Engineering, University of São Paulo
Abstract
Describes the application of a closed-loop stable MPC to an industrial FCC converter. Researchers produced nominal stability of the proposed controller by considering an infinite prediction horizon. Further, the state-space model used in the controller formulation was derived from the analytical form of the step response associated with transfer function models of the process, using data obtained from actual plant tests.
The fluidized-bed catalytic cracking (FCC) unit is one of the most profitable process units.  It is also perhaps the most complex and challenging operating process of an oil refinery. The process exhibits complex dynamic behavior, strongly interacting variables as well as economic and operating constraints on the process outputs and inputs. Consequently, it is prudent to employ advanced control strategies rather than the conventional PID-based decentralized control techniques to provide stable operating environment, to preserve mechanical integrity and, importantly, to reach the economic targets of the unit. Model predictive control (MPC) has become the standard method for control of the FCC process. Authors present a study of the implementation of a practical MPC with guaranteed nominal stability at an industrial FCC system.
Full Text Source (Subscription or Fee): http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=6606022&contentType=Conference+Publications