Integrated Optimization Strategies for Dynamic Process Operations

Process systems engineering faces increasing demands and opportunities for better process modeling and optimization strategies, particularly in the area of dynamic operations. Modern optimization strategies for dynamic optimization trace their inception to the groundbreaking work Pontryagin and his coworkers, starting 60 years ago. Since then the application of large-scale non-linear programming strategies has extended their discoveries to deal with challenging real-world process optimization problems. This study discusses the evolution of dynamic optimization strategies and how they have impacted the optimal design and operation of chemical processes. We demonstrate the effectiveness of dynamic optimization on three case studies for real-world reactive processes. In the first case, we consider the optimal design of runaway reactors, where simulation models may lead to unbounded profiles for many choices of design and operating conditions. As a result, optimization based on repeated simulations typically fails, and a simultaneous, equationbased approach must be applied. Next we consider optimal operating policies for grade transitions in polymer processes. Modeled as an optimal control problem, we demonstrate how product specifications lead to multistage formulations that greatly improve process performance and reduce waste. Third, we consider an optimization strategy for the integration of scheduling and dynamic process operation for general continuous/batch processes. The method introduces a discrete time formulation for simultaneous optimization of scheduling and operating decisions. For all of these cases we provide a summary of directions and challenges for future integration of these tasks and extensions in optimization formulations and strategies.     

An approximate mathematical framework for resource-constrained multistage batch scheduling

A rigorous representation of the multistage batch scheduling problem is often useless to even provide a good feasible schedule for many real-world industrial facilities. In order to derive a much simpler scheduling methodology, some usual features of multistage batch plants should be exploited. A common observation in industry is that multistage processing structures usually present a bottleneck stage (BS) controlling the plant output level. Therefore, the quality of the production schedule heavily depends on the proper allocation and sequencing of the tasks performed at the stage BS. Every other part of the processing sequence should be properly aligned with the selected timetable for the bottleneck tasks. A closely related concept with an empirical basis is the usual existence of a common batch sequencing pattern along the entire processing structure that leads to define the constant-batch-ordering rule (CBOR). According to this rule, a single sequencing variable is sufficient to establish the relative ordering of two batches at every processing stage in which both have been allocated to the same resource item. This work introduces a CBOR-based global precedence formulation for the scheduling of order-driven multistage batch facilities. The proposed MILP approximate problem representation is able to handle sequence-dependent changeovers, delivery due dates and limited manufacturing resources other than equipment units. Optimal or near-optimal solutions to several large-scale examples were found at very competitive CPU times.     

Integrated campaign planning and resource allocation in batch plants

Operational planning in batch plants has been studied extensively in the literature. However, the focus has largely been on developing better formulations and solution approaches for large examples. This work attempts to capture the key aspects of the industrial planning activity such as interactions among the planner and other stakeholders and the effect of resource allocation on process performance. We present a simple mathematical formulation for integrated resource allocation and campaign planning in multiproduct batch plants. Our model enables decision support pertaining to campaign scheduling, sequence-dependent changeovers, key resource allocations, scheduled maintenance, inventory profiles with safety stock limitations, and new product introductions. To demonstrate the performance of our mathematical model, we consider a case study from a typical specialty chemical plant from the lube industry. We validate our approach using a series of dynamic business and market scenarios with planning horizons of up to 2 years.     

Reactor modeling and recipe optimization of polyether polyol processes: Polypropylene glycol

The reactor modeling and recipe optimization of conventional semibatch polyether polyol processes, in particular for the polymerization of propylene oxide to make polypropylene glycol, is addressed. A rigorous mathematical reactor model is first developed to describe the dynamic behavior of the polymerization process based on first‐principles including the mass and population balances, reaction kinetics, and vapor‐liquid equilibria. Next, the obtained differential algebraic model is reformulated by applying a nullspace projection method that results in an equivalent dynamic system with better computational performance. The reactor model is validated against plant data by adjusting model parameters. A dynamic optimization problem is then formulated to optimize the process recipe, where the batch processing time is minimized, given a target product molecular weight as well as other requirements on product quality and process safety. The dynamic optimization problem is translated into a nonlinear program using the simultaneous collocation strategy and further solved with the interior point method to obtain the optimal control profiles. The case study result shows a good match between the model prediction and real plant data, and the optimization approach is able to significantly reduce the batch time by 47%, which indicates great potential for industrial applications. © 2013 American Institute of Chemical Engineers AIChE J, 59: 2515–2529, 2013     

Mathematical modeling for simultaneous design of plants and supply chain in the batch process industry

Most supply chain design models have focused on the integration problem, where links among nodes must be settled in order to allow an efficient operation of the whole system. At this level, all the problem elements are modeled like black boxes, and the optimal solution determines the nodes allocation and their capacity, and links among nodes. In this work, a new approach is proposed where decisions about plant design are simultaneously made with operational and planning decisions on the supply chain. Thus, tradeoffs between the plant structure and the network design are assessed. The model considers unit duplications and the allocation of storage tanks for plant design. Using different sets of discrete sizes for batch units and tanks, a mixed integer linear programming model (MILP) is attained. The proposed formulation is compared with other non-integrated approaches in order to illustrate the advantages of the presented simultaneous approach.     

Dynamic optimization of semibatch vinyl acetate/acrylic acid suspension copolymerizations

This work performs the dynamic optimization of semibatch vinyl acetate (VAc)/acrylic acid (AA) suspension copolymerizations. The proposed dynamic optimization strategy is based on a direct search Complex algorithm and is used to control the copolymer composition along the batch. First, a sequential optimization procedure is used to determine the optimum AA concentration and feed rate profiles, required to provide the specified copolymer composition. In the second step, a sequential optimization procedure is coupled with a predictive controller to guarantee that the manipulation of feed flow rates can allow for attainment of the desired copolymer compositions. The optimization strategy is validated through simulation, by assuming that reactions are subject to perturbations of the reaction temperature, initiator, and VAc concentrations. It is shown that the proposed optimization strategy can be used successfully both for design of monomer feed rate profiles and removal of process disturbances during semibatch suspension copolymerizations, to keep the copolymer composition constant throughout the batch. POLYM. ENG. SCI., 2009. © 2009 Society of Plastics Engineers     

On the role of the necessary conditions of optimality in structuring dynamic real-time optimization schemes

In dynamic optimization problems, the optimal input profiles are typically obtained using models that predict the system behavior. In practice, however, process models are often inaccurate, and on-line model adaptation is required for appropriate prediction and re-optimization. In most dynamic real-time optimization schemes, the available measurements are used to update the plant model, with uncertainty being lumped into selected uncertain plant parameters; furthermore, a piecewise-constant parameterization is used for the input profiles. This paper argues that the knowledge of the necessary conditions of optimality (NCO) can help devise more efficient and more robust real-time optimization schemes. Ideally, the structuring decisions involve the NCO as follows: (i) one measures or estimates the plant NCO, (ii) a NCO-based input parameterization is used, and (iii) model adaptation is performed to meet the plant NCO. The benefit of using the NCO in dynamic real-time optimization is illustrated in simulation through the comparison of various schemes for solving a final-time optimal control problem in the presence of uncertainty.     

Integrated process design,scheduling, and model predictive control of batch processes with closed-loop implementation

Simultaneous evaluation of multiple time scale decisions has been regarded as a promising avenue to increase the process efficiency and profitability through leveraging their synergistic interactions. Feasibility of such an integral approach is essential to establish a guarantee for operability of the derived decisions. In this study, we present a modeling methodology to integrate process design, scheduling, and advanced control decisions with a single mixed-integer dynamic optimization (MIDO) formulation while providing certificates of operability for the closed-loop implementation. We use multi-parametric programming to derive explicit expressions for the model predictive control strategy, which is embedded into the MIDO using the base-2 numeral system that enhances the computational tractability of the integrated problem by exponentially reducing the required number of binary variables. Moreover, we apply the State Equipment Network representation within the MIDO to systematically evaluate the scheduling decisions. The proposed framework is illustrated with two batch processes with different complexities.     

Nichtlineare stochastische Optimierung unter Unsicherheiten

Nonlinear Stochastic Optimization under Uncertainty Robust decision making under uncertainty is considered to be of fundamental importance in numerous disciplines and application areas. In dynamic chemical processes in particular there are parameters which are usually uncertain, but may have a large impact on equipment decisions, plant operability, and economic analysis. Thus the consideration of the stochastic property of the uncertainties in the optimization approach is necessary for robust process design and operation. As a part of it, efficient chance constrained programming has become an important field of research in process systems engineering. A new approach is presented and applied for stochastic optimization problems of batch distillation with a detailed dynamic process model.     

PDPSO优化多阶段AR-PCA间歇过程监测方法

针对间歇过程固有的多阶段特性和动态性,提出基于种群多样性的自适应惯性权重粒子群算法（PDPSO）优化的多阶段自回归主元分析（AR-PCA）间歇过程监测方法。该方法引入了PDPSO算法指导AP聚类偏向参数的选取,避免了传统方法依据聚类评价指标选取参考度时的盲目性。对PDPSO优化AP聚类的多阶段发酵过程的数据样本建立AR-PCA模型能够消除各阶段的动态性及变量之间的自相关和互相关影响。最后,对自回归（AR）模型的残差矩阵建立主成分分析（PCA）模型用于发酵过程监测。将该方法应用到青霉素发酵过程,并与传统方法进行对比,结果表明,该方法能够有效进行间歇过程阶段划分并降低故障的漏报和误报。     

Dynamic nonlinear batch process fault detection and identification based on two‐directional dynamic kernel slow feature analysis

The batch process generally covers high nonlinearity and two‐directional dynamics: time‐wise dynamics, which correspond to inherently time‐varying dynamics resulting from the slowly varying underlying driving forces within each batch duration; and batch‐wise dynamics, which are associated with different operating modes among different batches. However, most existing dynamic nonlinear monitoring methods cannot extract the slowly varying underlying driving forces of the nonlinear batch process and rarely tackle the batch‐wise dynamic characteristics among batch runs. In order to address these issues, a new monitoring scheme based on two‐directional dynamic kernel slow feature analysis (TDKSFA) is developed by combining kernel SFA with a global modelling strategy. In the TDKSFA method, kernel SFA is integrated with the ARMAX time series model to mine the nonlinear and time‐wise dynamic properties within a batch run due to its capability of extracting the slowly varying underlying driving forces. Furthermore, the global modelling strategy is presented to handle the batch‐wise dynamics among batches by calculating the total average kernel matrix of all training batches. After the slow features are extracted, Hotelling's T² and SPE statistics are built to detect faults. To solve the issue of fault variable nonlinear identification, a novel nonlinear contribution plot inspired by the pseudo‐sample variable projection trajectories in the TDKSFA model is further proposed to identify fault variables. Finally, the feasibility and effectiveness of the TDKSFA‐based fault diagnosis strategy is demonstrated through a numerical system and the penicillin fermentation process.     

Analysis and simulation of the sludge profile dynamics in a full‐scale clarifier

A one‐dimensional clarifier model was assessed for its capability to describe dynamic full‐scale sludge concentration profiles by using the settling properties calibrated with batch settling curve data collected by a SettloMeter^®. These sludge concentration profiles and batch settling tests formed part of a detailed one‐month measuring campaign on a full‐scale wastewater treatment plant; the measurements showed a daily variation in settling properties. Using the settling properties obtained from batch settling tests and a one‐dimensional model without dispersion, the dynamics of the full‐scale clarifier were analysed and the need for dispersion clearly shown. The parameters of the dispersion model were estimated from the full‐scale sludge concentration profiles. The settling properties of activated sludge can be automatically determined by fitting the model to the on‐line batch settling curve measurements and are needed as input to the one‐dimensional model. This model can therefore be used for operation and control. The dispersion model parameters have to be determined from dynamic sludge concentration profiles but are assumed to be constant for a specific clarifier. Copyright © 2005 Society of Chemical Industry     

一种基于状态估计的间歇过程模型实时更新预测控制策略简

Nonlinear model predictive control （NMPC） is an appealing control technique for improving the per- formance of batch processes, but its implementation in industry is not always possible due to its heavy on-line computation. To facilitate the implementation of NMPC in batch processes, we propose a real-time updated model predictive control method based on state estimation. The method includes two strategies： a multiple model building strategy and a real-time model updated strategy. The multiple model building strategy is to produce a series of sim- plified models to reduce the on-line computational complexity of NMPC. The real-time model updated strategy is to update the simplified models to keep the accuracy of the models describing dynamic process behavior. The method is validated with a typical batch reactor. Simulation studies show that the new method is efficient and robust with respect to model mismatch and changes in process parameters.     

Multiproduct batch plants design using linear process performance models

In this contribution, a novel linear generalized disjunctive programming (LGDP) model is developed for the design of multiproduct batch plants optimizing both process variables and the structure of the plant through the use of process performance models. These models describe unit operations using explicit expressions for the size and time factors as functions of the process variables with the highest impact. To attain a linear formulation, values of the process variables as well as unit sizes are selected from a set of meaningful discrete values provided by the designer. Regarding structural alternatives, both kinds of unit duplications in series and in parallel are considered in this approach. The inclusion of the duplication in series requires different detailed models that depend on the structure selected. Thus, in a new approach for the multiproduct batch plant design, a set of potential structural alternatives for the plant is defined. © 2010 American Institute of Chemical Engineers AIChE J, 2011     

Batch-to-batch optimal control of a batch polymerisation process based on stacked neural network models

A neural network based batch-to-batch optimal control strategy is proposed in this paper. In order to overcome the difficulty in developing mechanistic models for batch processes, stacked neural network models are developed from process operational data. Stacked neural networks have enhanced model generalisation capability and can also provide model prediction confidence bounds. However, the optimal control policy calculated based on a neural network model may not be optimal when applied to the true process due to model plant mismatches and the presence of unknown disturbances. Due to the repetitive nature of batch processes, it is possible to improve the operation of the next batch using the information of the current and previous batch runs. A batch-to-batch optimal control strategy based on the linearisation of stacked neural network model is proposed in this paper. Applications to a simulated batch polymerisation reactor demonstrate that the proposed method can improve process performance from batch to batch in the presence of model plant mismatches and unknown disturbances.     

Model-based control strategies for a chemical batch reactor with exothermic reactions

Batch reactor control provides a very challenging problem for the process control engineer. This is because a characteristic of its dynamic behavior shows a high nonlinearity. Since applicability of the batch reactor is quite limited to the effectiveness of an applied control strategy, the use of advanced control techniques is often beneficial. This work presents the implementation and comparison of two advanced nonlinear control strategies, model predictive control (MPC) and generic model control (GMC), for controlling the temperature of a batch reactor involving a complex exothermic reaction scheme. An extended Kalman filter is incorporated in both controllers as an on-line estimator. Simulation studies demonstrate that the performance of the MPC is slightly better than that of the GMC control in nominal case. For model mismatch cases, the MPC still gives better control performance than the GMC does in the presence of plant/model mismatch in reaction rate and heat transfer coefficient.     

基于多阶段多向核熵成分分析的间歇过程故障检测方法

针对间歇过程的非线性、多阶段特性,提出一种基于多阶段多向核熵成分分析（multistage-MKECA,MsMKECA）的故障检测方法。针对间歇过程的多阶段特性,建立一种时序核熵主元关联度的矩阵相似性阶段划分方法,实现对间歇生产过程的多阶段划分;针对传统批次展开方式在线监控需要预估批次未来值的缺陷,进一步引入一种批次-变量三维数据展开方式建立每个阶段的MKECA非线性统计模型,实现对间歇过程的分阶段监控。最后对盘尼西林发酵过程开展仿真研究,结果表明所提方法能够比传统MKECA方法更为快速地进行故障检测。     

A COMBINED MILP AND LOGIC-BASED APPROACH TO THE SYNTHESIS OF OPERATING PROCEDURES FOR BATCH PLANTS

Synthesizing a set of operating procedures for the safe and efficient transient operation of chemical plants is a difficult problem owing to the enormous number of possible combinations of actions in a typical plant. In most current industrial plant design practice, there are no formal methods for systematically transforming process specifications into operating procedures for the plant operators and into sequence control instructions for the control computers. There is much scope for a formalized computer-based procedure synthesis methodology to assist the design engineer/plant operator with both the formulation and assessment of procedures off-line and, eventually, with the on-line problem of procedure synthesis in response to unexpected situations

A recently developed approach for operating procedure synthesis for multipurpose batch plants is considered. The modelling formalism used includes the separate definition of process operations, as State Task Networks, and of physical plant, at the level of detail of a piping and instrumentation diagram. In this paper, a subgoaling procedure is developed using the State Task Network representation which decomposes the procedure synthesis goals into simpler subgoals by means of an efficient Mixed Integer Linear Programming (MILP) technique. Detailed control sequences are then generated for each subgoal using a set of rules and algorithms specific for each type of subgoal. The procedure sequences thus generated are validated by simulation on a plant model with checking of physical and operational constraints at each new plant state. We have found that this hierarchical approach to the procedure synthesis problem greatly reduces the problem complexity

The usefulness of the general approach and of the subgoaling procedure in particular are demonstrated through a multiproduct batch plant example.     

Optimization and nonlinear control of a batch crystallization process

Crystallization process has been widely used for separation in many chemical industries due to its capability to provide high purity product. To obtain the desired quality of crystal product, an optimal cooling control strategy is studied in the present work. Within the proposed control strategy, a dynamic optimization is first preformed with the objective to obtain the optimal cooling temperature policy of a batch crystallizer, maximizing the total volume of seeded crystals. Two different optimization problems are formulated and solved by using a sequential optimization approach. Owing to the complex and nonlinear behavior of the batch crystallizer, the nonlinear control strategy which is based on a generic model control (GMC) algorithm is implemented to track the resulting optimal temperature profile. The optimization integrated with nonlinear control strategy is demonstrated on a seeded batch crystallizer for the production of potassium sulfate.