Search

• State-of-the-art modeling and optimization of industrial-size refinery scheduling. • Integrated refinery scheduling network with advanced process-unit modeling. • Optimization decision-making with advanced re-optimization mechanisms. • Nonconvex mixed-integer nonlinear programming (MINLP) model for refinery scheduling. • Refinery scheduling optimization leads to higher profitability and more efficient operations. Modeling and optimization of large-scale refinery scheduling problems is challenging because of their complexity and size. Herein, we propose a mathematical model to represent such problems more accurately and realistically, and a state-of-the-art optimization framework for its solution. The framework leverages the use of mathematical optimization and algorithmic methods by combining modeling approaches (process design, model decompositions), solving strategies (rescheduling, heuristics), and machine learning regression (surrogate models). An industrial-size refinery scheduling problem (2 blenders, 4 feed tanks, distillation network with 5 towers, processing network with FCC, hydrotreaters, debutanizers, superfractionator, catalytic reformer) is formulated as a hierarchical nonconvex mixed-integer nonlinear programming (MINLP) model and is successfully optimized, providing higher profitability and more efficient scheduling operations considering 12 feedstocks, 10 products and multiple scenarios for time horizon and step. Results highlight the importance of tuning scheduling parameters and employing an enhanced computer-aided framework to enable the solution of industrial refinery scheduling operations. [ABSTRACT FROM AUTHOR]

• A new MIQCP model for scheduling of olefin plant furnace operations. • Simultaneously optimize furnace system manufacturing and decoking schedules. • Grouping decoking operations together to reduce product throughput upsets. • Well balancing chemical plant profitability and operability. Olefin plants employ multi-type cracking furnaces in parallel to convert various hydrocarbon feeds into products such as ethylene and propylene. The cracking process of each furnace is a performance-decaying batch operation, which needs periodic shutdowns for decoking and thus involves inherent and frequent upsets to the downstream process. Thus, the furnace feed allocation, batch processing time, and decoking sequence of the entire furnace system must be optimally scheduled to maximize the plant profitability as well as minimize the process upset induced by furnace decoking for the sake of the plant operability. In this paper, a novel MIQCP (mixed-integer quadratically constrained programming) model has been developed for the optimal scheduling of such parallel and performance-decaying unit system with consideration of inherent upset reduction. The idea is to group decoking operations of different furnaces together, so that the downstream has less frequency to suffer the product throughput upsets from the furnace system. The MIQCP model also addresses major scheduling issues of furnace systems, such as semi-continuous operation, non-simultaneous shutdowns, and secondary ethane cracking. The MIQCP model can be further converted to the MILP model by piece-wise approximation to speed the solving time. Case studies have demonstrated the efficacy of the developed methodology with significant economic potential and operational benefits. [ABSTRACT FROM AUTHOR]

With the potential to decrease operating costs and improve energy efficiency, model predictive control (MPC) has been proposed as a replacement for traditional heuristic, PID, and other conventional control strategies for heating, ventilation, and air conditioning (HVAC) systems in commercial buildings. Due to the size of large commercial HVAC systems, implementing MPC as a single monolithic optimization problem is not practical nor desirable given real-time operating requirements. In this paper, we present a hierarchical decomposition for economic MPC in large-scale commercial HVAC systems using a two-layer approach. We show a sample optimization for a campus of 25 buildings with 500 total zones and a central plant consisting of eight chillers. Then, we discuss an application of the ideas presented here in the recently completed $485-million replacement of the Stanford campus heating and cooling systems and conclude with some of the control theory challenges presented by this new class of applications. [ABSTRACT FROM AUTHOR]

The notion of Internet of Things (IoT), as well as related topics such as Cyber-Physical Systems, Industrie 4.0 and Smart Manufacturing are currently attracting a lot of attention within the process and manufacturing industries. Clearly, IoT offers many potential applications for automation, ranging from engineering installation of new plants to production management and more intelligent maintenance schemes including novel sensor technologies. The focus of this paper is, however, on the control and operations. Most likely IoT leads to new system architectures where open standards play a significant role. Through better connectivity, information will be much more easily available, which could result in that previously isolated functions will become more closely integrated. Here modeling at the right level of fidelity will be absolutely key. It can be expected that the importance of optimization will increase and this paper discusses some aspects related to the opportunities, challenges and changes triggered by IoT. [ABSTRACT FROM AUTHOR]

• A novel methodology for scheduling of energy intensive industries under the uncertainty of spot electricity market. • Development of a scenario generator tool using ARIMA for electricity price forecasting. • Integration of deterministic optimization with multi scenario approach. • Risk metric analysis to facilitate decision making in choosing the best suited scenario. • Implementation of the methodology in a real-life cryogenic air separation plant. The planning of energy-intensive processes is intrinsically uncertain due to their dependence on the volatile energy market, with scheduling having a vast impact on the final production cost of these plants. Traditional stochastic methods are mathematically very complex, which translates into a significant computational effort that might prevent a timely response to varying electricity prices. To encounter this uncertainty, we develop a reliable hybrid simulation-optimization approach for optimizing the production plant scheduling, combining scenario analysis with risk analysis. The proposed methodology is demonstrated with real data from a cryogenic air separation plant in Tarragona (Spain). This approach also informs decision-makers about risk or expected shortfall associated with the implied scenario. The generic methodology used here can be easily adapted to schedule facilities in other energy-intensive sectors such as cement, metallurgy or pulp and paper. [Display omitted] [ABSTRACT FROM AUTHOR]

Continuous-time scheduling models for multiproduct pipelines typically view the contents of a pipeline as a set of batches. It facilitates tracking batch coordinates but prevents rigorously enforcing forbidden product sequences when dealing with intermediate dual-purpose nodes. Such models require the initial characterization of the pipeline and total number of batches as inputs, decisions that can be non-trivial. To overcome these limitations, this paper presents a new mixed-integer linear programming (MILP) formulation for straight pipelines with a single tuning parameter, the number of event points in the grid. It is derived from Generalized Disjunctive Programming (GDP) followed by a convex hull reformulation. Compared to a product-centric formulation based on the Resource-Task Network (RTN), it is much simpler, smaller in size and can be up to two orders of magnitude faster. Comparison to its batch-centric counterpart is not as favorable, the highlight being an 8% improvement in makespan for a benchmark problem. [ABSTRACT FROM AUTHOR]

Scheduling of production and maintenance plays a fundamental role in the effective operation of process plants. Frequently the two decision processes are independently addressed, overlooking the tight relation existing between the way the plant is operated to produce the required goods and the appearance of maintenance requirements. The presence of degradation phenomena affecting the performance of plant units and limiting the operational choices makes the integration of the two decision processes even more important. In this paper an industrial framework for the integration of maintenance and production scheduling of process plants is presented as well as some considerations on how the presence of plant unit degradation impacts on the scheduling problem. The proposed approach ties industrial key-components such as asset management and production scheduling closer together and represents therefore a contribution to the smart manufacturing revolution. The integrated maintenance and production scheduling problem is formulated as a mixed integer linear program (MILP) and tested on a generic process plant described as a State Task Network (STN). [ABSTRACT FROM AUTHOR]

The design of multi-purpose batch plants is a challenging task, because the number of degrees of freedom for optimization is high. Important optimization variables are the scheduling, operating conditions and the sizes of the equipment items. Since all factors interact, a simultaneous consideration would be beneficial in order to reduce capital and operating costs. In this article, this complex task is tackled for a new case study of a sequential plant for protein production. The case study contains comprehensive models of the unit operations to evaluate equipment dimensions, mass balances and operating times. Variable changeover times and semicontinuous unit operations are considered. For optimization, a MINLP model is used that consists of smaller NLP and MILP submodels in order to simplify modeling. Simulation runs for different product demands are performed and it is shown that good solutions can be found in an adequate time. [ABSTRACT FROM AUTHOR]

Multi-recipe and multi-stage material handling (M3H) processes are widely employed by various industries where complex multi-product manufacturing and assembly tasks are required. Cyclic hoist scheduling (CHS) could significantly improve the productivity of an M3H process. In this paper, a novel CHS methodology has been developed, which considers employing multiple sub-cycles to efficiently deal with job duality issues associated with multi-capacity processing units. The methodology contains three modeling and solving stages. In Stage I, a mixed-integer linear programing (MILP) model is developed to obtain the optimal solution of a sub-cycle CHS problem with the tolerance of job duality. In Stage II, the obtained solution will be examined to see if the job duality exists. Once a job duality issue is identified, another MILP model in Stage III will be performed to schedule an additional sub-cycle CHS problem, which targets the minimum slot usage discrepancy caused by the identified job duality. After that, the combined CHS solutions from the previous and additional sub-cycles will be fed back to Stage II for the job duality examination again. Iteratively checking and rescheduling between Stages II and III, job duality issues will be eventually eliminated and a full scheduling cycle coupling multiple sub-cycles will be accomplished. The methodology can significantly increase the CHS optimality for M3H processes, which are demonstrated by two case studies. [ABSTRACT FROM AUTHOR]

The efficient and economic operation of processing systems ideally requires a simultaneous planning, scheduling and control framework. Even when the optimal simultaneous solution of this problem can result in large scale optimization problems, such a solution can represent economic advantages making feasible its computation using optimization decomposition and/or few operating scenarios. After reducing the complexity of the optimal simultaneous deterministic solution, it becomes feasible to take into account the effect of model and process uncertainties on the quality of the solution. In this work we consider those changes in product demands that take place once the process is already under continuous operation. Therefore, a reactive strategy is proposed to meet the new product demands. Based on an optimization formulation for handling the simultaneous planning, scheduling, and control problem of continuous reactors, we propose a heuristic strategy for dealing with unexpected events that may appear during operation of a plant. Such a strategy consists of the rescheduling of the products that remain to be manufactured after the given disturbance hits the process. Such reactive strategy for dealing with planning, scheduling and control problems under unforeseen events is tested using two continuous chemical reaction systems. [ABSTRACT FROM AUTHOR]

We provide a systematic review of current progress in integrated control and operations. Recently, an increasing number of integrated methods have been presented in the literature. These methods optimize the entire system by simultaneously solving the control and operation problems. Their advantages have been demonstrated in comparison to the conventional sequential method, which solves the subproblems independently and neglects the interaction among the subproblems. We give an in-depth review of these integrated methods in terms of the integration scopes, model formulations, solution algorithms, and implementation strategies to address uncertainty. Though significant progress has been made, research on integrated methods is still in its infant stage. It is still mathematically challenging to solve a complex integrated problem composed of binary/logic variables from operation problems and nonlinear differential-algebraic equations inherited from control problems. We discuss some major challenges concerning complexity, heterogeneity, and uncertainty, and suggest a number of promising directions for future research. [ABSTRACT FROM AUTHOR]

Highlights: [•] We present an improved continuous time model for multi-period scheduling of bio-pharmaceutical plants. [•] The specific features include better handling of changeovers, shelf life, and waste disposals. [•] Effective handling of intermediate due dates, with penalties for backlogs and late deliveries. [•] Improved tightening and sequencing constraints presented leading to better solution. [•] The model is tested on a literature example involving multistage multiproduct production. [Copyright &y& Elsevier]

Highlights: [•] A multi-objective scheduling model is developed for regional multi-plant start-up control. [•] The regional air quality impact and the total start-up time mismatch for all plants are simultaneously minimized. [•] The study can benefit government environmental agency, regional chemical plants, and local communities. [Copyright &y& Elsevier]

Abstract: We propose two mixed-integer linear programming based approaches for the 2D orthogonal strip packing problem. Using knowledge from the alternative forms of time representation in scheduling formulations, we show how to efficiently combine three different concepts into the x- and y-dimensions. One model features a discrete representation on the x-axis (strip width) and a continuous representation with general precedence variables on the y-axis (strip height). The other features a full continuous-space representation with the same approach for the y-axis and a single non-uniform grid made up of slots for the x-axis. Through the solution of a set of twenty nine instances from the literature, we show that the former is a better approach, even when compared to three alternative MILP models ranging from a pure discrete-space to a pure continuous-space with precedence variables in both dimensions. All models are available in www.minlp.org. [Copyright &y& Elsevier]

Abstract: This paper presents a novel decomposition strategy for solving large scale refinery scheduling problems. Instead of formulating one huge and unsolvable MILP or MINLP for centralized problem, we propose a general decomposition scheme that generates smaller sub-systems that can be solved to global optimality. The original problem is decomposed at intermediate storage tanks such that inlet and outlet streams of the tank belong to the different sub-systems. Following the decomposition, each decentralized problem is solved to optimality and the solution to the original problem is obtained by integrating the optimal schedule of each sub-systems. Different case studies of refinery scheduling are presented to illustrate the applicability and effectiveness of the proposed decentralized strategy. The conditions under which these two types of optimization strategies (centralized and decentralized) give the same optimal result are discussed. [Copyright &y& Elsevier]

• Scheduling algorithm for optimizing the transportation of liquids in pipelines. • Two-stage decomposition with closely related continuous-time formulations. • Backtracking mechanism addresses flowrate conflicts by generating cuts. • Orders of magnitude improvements compared to state-of-the-art full space formulation. • Heuristic method able to find optimal solution of four benchmark instances tested. This article develops a novel two-stage algorithm for the detailed scheduling of branched multiproduct pipeline systems with a single refinery and multiple depots. It features continuous-time models allowing multiple batches to be injected/delivered over a slot. The first-stage model reduces the complexity by neglecting the lower flowrate limits, enforced by big-M constraints. The objective is to minimize the makespan. The second-stage model works with the first-stage assignments and with an extended time grid. Coupled with a backtracking mechanism, the proposed algorithm can significantly reduce computational time and improve the pipeline transportation capacity. We use four benchmark instances from the literature to test and validate the effectiveness and superiority of the method proposed in this paper. Compared to the literature, a new optimal solution is reported in one case, while the computational time is reduced by at least one order of magnitude in the other cases. [ABSTRACT FROM AUTHOR]

Abstract: In this work we address the simultaneous process control and design problem of polymerization reactors during dynamic grade transition operation. The problem is cast as a Mixed-Integer Dynamic Optimization (MIDO) formulation and, by using the full discretization approach for solving dynamic optimization problems [Kameswaran, S., & Biegler, L. T. (2006). Simultaneous dynamic optimization strategies: Recent advances and challenges. Computers & Chemical Engineering, 30 (10–12), 1560–1575], is transformed into a Mixed-Integer Nonlinear Program (MINLP). The resulting MINLP is solved using a full space nonconvex optimization formulation [Flores-Tlacuahuac, A., & Biegler, L. T. (2007). Simultaneous Mixed-Integer Dynamic Optimization for integrated design and control. Computers & Chemical Engineering, 31, 588–600]. The control and design formulation has been applied to two polymerization reactors featuring highly nonlinear behavior. In both cases, the proposed MIDO formulation was capable of finding optimal solutions. This amounts to finding optimal steady states, reactor designs, as well as open-loop and closed-loop dynamic optimal trajectories, control structures and controller parameters by specifying either the polymer molecular weight distribution or monomer conversion. Because CPU solution time tends to increase with system complexity, some strategies for lowering CPU time are discussed. [Copyright &y& Elsevier]

In this paper, we address the integration of production planning and reactive scheduling for the optimization of a hydrogen supply network consisting of five plants, four inter-connected pipelines and 20 customers. We present multiperiod mixed integer nonlinear programming (MINLP) models for both the planning and scheduling levels. The planning model includes complex pricing functions resulting from deregulation, with a simplified pipeline description and determines feed and energy prices, as well as production levels, for a monthly horizon divided into 12-h time periods. Prices are fixed on the scheduling level, while a detailed pipeline model is included, to determine the on/off status and load steps of compressors on an hourly basis, in order to satisfy the actual demands as they become known while adhering to pressure constraints. In addition, we propose a solution methodology where the demand forecast is updated and the planning model is rerun every 12 h, while the scheduling model is run every hour and information is passed between the two levels to facilitate integration. We show that the planning model quickly becomes intractable and propose a heuristic solution method for this level based on Lagrangean decomposition. Results show that the proposed Lagrangean decomposition heuristic reduces the computational effort for solving the planning model by more than an order of magnitude compared to the commercial MINLP solver DICOPT++. It is also shown that for the majority of the plants, the power consumption and hydrogen production from the scheduling level agrees with the planning level. In some cases, however, the integration is hampered by the presence of nonlinearities, especially on the scheduling level, that lead to suboptimal or infeasible solutions. These nonlinearities need to be further addressed before the proposed methodology can be implemented in practice. [Copyright &y& Elsevier]

This paper addresses the problem of preventive maintenance (PM) strategy optimization in a semiconductor manufacturing environment, with the objective of minimizing maintenance costs. The approach developed takes into account the interaction of production and maintenance aspects. For this purpose, a discrete-event production-oriented simulator (MELISSA-C++) has been extended to incorporate equipment failures and maintenance operations, thus modeling residual breakdowns, occurring in a combined corrective/PM context. The usefulness of the simulation tool has also been demonstrated for the estimation of both direct and indirect maintenance costs, which are impossible to determine empirically due to the reentrant nature of product flows in a semiconductor manufacturing facility. The results obtained have confirmed the marked effect of equipment characteristics (bottleneck or non-limiting step) on maintenance cost evaluation. Following a tutorial example, typical results are presented and analyzed. [Copyright &y& Elsevier]

• Method to coordinate multiple production and utility systems. • Stackelberg game with incomplete information. • Only demand-dependent energy cost exchanged. • Significant cost reduction with practical potential. Large industrial sites commonly contain multiple production and utility systems. In practice, integrated optimization is often not possible because the necessary complete information cannot be exchanged between the systems. Often, industrial sites optimize the operation of production and utility systems sequentially without any feedback, which leads to suboptimal operation. In this paper, we propose a method to coordinate between production and utility systems in a multi-leader multi-follower Stackelberg game. The proposed method does not require complete information exchange. The only information exchanged in one feedback loop is the energy demand and demand-dependent energy cost. In two case studies, the method reduces the total production cost by 7.6% and 3.4% compared to the common sequential optimization. These cost savings correspond to 84% and 88% of the potential cost savings by an integrated optimization. In summary, the proposed method reduces cost significantly, while only incomplete information is exchanged between production and utility systems. [ABSTRACT FROM AUTHOR]

• Campaign-based discrete-time scheduling approach to deal with changeover costs. • Uninterrupted sequences of jobs of the same class modeled as individual campaigns. • Tighter formulation, reducing the computational burden by three orders of magnitude. Discrete-time, Mixed-Integer Linear Programming (MILP) models are frequently used for scheduling problems in which the utilization of resources and/or the determination of costs and revenues imply complex time-dependent functions that are directly related to the development of the tasks. One of the main challenges in these models is to deal with changeovers, which are one of the most complicating features. In this short note, we provide some insight to effectively model the scheduling of operations by means of the concept of campaigns. We show that this allows to reduce the computational burden by three orders of magnitude when compared to conventional approaches. [ABSTRACT FROM AUTHOR]

This work examines applying deep reinforcement learning to a chemical production scheduling process to account for uncertainty and achieve online, dynamic scheduling, and benchmarks the results with a mixed-integer linear programming (MILP) model that schedules each time interval on a receding horizon basis. An industrial example is used as a case study for comparing the differing approaches. Results show that the reinforcement learning method outperforms the naive MILP approaches and is competitive with a shrinking horizon MILP approach in terms of profitability, inventory levels, and customer service. The speed and flexibility of the reinforcement learning system is promising for achieving real-time optimization of a scheduling system, but there is reason to pursue integration of data-driven deep reinforcement learning methods and model-based mathematical optimization approaches. [ABSTRACT FROM AUTHOR]

• The scheduling operations of a real-world pipeline network are addressed. • The network has reversible pipelines and transports heavy oil derivatives. • The NP-Complete problem is divided into assigning, sequencing, and timing blocks. • MILP and heuristics are combined to solve the assigning and sequencing steps. • Brazilian real network monthly schedules are obtained in low running time. To transport oil using pipelines is very efficient, but expanding the existing infrastructure is costly. Therefore, it is essential to optimize the use of the existing pipelines. One step to optimize their operations is to make a proper schedule, an NP-Complete problem. Researchers have developed approaches to solve this problem, and in some of these, to avoid computational burden, the problem is divided into roughly three main components: assigning, sequencing, and timing. We describe a new approach to the assigning and sequencing steps to schedule a Brazilian pipeline network used to convey heavy oil derivatives. An MILP model is used to generate the assigning solution, while both a heuristic algorithm and an MILP model are used to solve the sequencing problem. Computational tests considered the operational particularities of the real-world pipeline network, as inline oil blending. The proposed approach was able to obtain monthly schedules in relatively low running time. [ABSTRACT FROM AUTHOR]

The success of lumber manufacturing lies on the ability to maintain the quality of the wood along the production process. The drying stage in a sawmill is one of the more important steps in the lumber production, since wood value and characteristics required by customers depend on this process phase. In a sawmill, a set of drying kilns operating in parallel out of phase is available, and periodically, packages of boards enter to this drying sector in order to be processed. Drying kilns are filled with wagons loaded with packages of boards. Different operative conditions must be satisfied in order to assure an efficient processing. These tasks involve several decisions, which are addressed through a Generalized Disjunctive Programming model, and solved using the corresponding Mixed Integer Programming reformulation considering different objective functions. From the examples, two scenarios are assessed and the capabilities of the presented approach are highlighted. [ABSTRACT FROM AUTHOR]

• Hydrogen- and ammonia-based energy storage systems for renewable-only energy supply. • Optimal combined capacity planning and scheduling to determine system investment and operation. • Consecutive temporal clustering to efficiently aggregate renewable generation and demand data. • Ammonia suitable for renewable energy storage throughout the United States. Hydrogen and, more recently, ammonia have received worldwide attention as energy storage media. In this work we investigate the economics of using each of these chemicals as well as the two in combination for islanded renewable energy supply systems in 15 American cities representing different climate regions throughout the country. We use an optimal combined capacity planning and scheduling model which minimizes the levelized cost of energy (LCOE) by determining optimal unit selection and size along with unit commitments, production rates, and storage inventories for each period of system operation. These periods are aggregated from full year hourly resolution data via a consecutive temporal clustering algorithm. Ammonia is generally more economical than hydrogen as a single method of energy storage. Additionally, systems which use both hydrogen and ammonia outperform those which use only one storage option and have LCOE between $0.17/kWh and $0.28/kWh, including full investment in renewable generation infrastructure. [ABSTRACT FROM AUTHOR]

Due to increases in both connectivity of process systems and availability of computing power, there is growing interest in the integration of online scheduling and control. While significant progress has been made on formulations and optimization strategies for this integrated problem, obtaining a single solution is not sufficient for a practical implementation: because of disturbances and the inherent finite-horizon scope, it eventually becomes necessary to reoptimize considering the new state of the plant. When this loop is closed, there is an unfortunate gap in knowledge with respect to theoretical properties possessed by the resulting evolution of the system. Thus, to facilitate an analysis of the online behavior, we present a unified state-space formulation for integrated scheduling and control. We then use recent theory from economic model predictive control to derive terminal constraints that can be added to the problem to provide nominal recursive feasibility and a bound on closed-loop performance. Via examples, we demonstrate the improved performance of these techniques relative to naive closed-loop implementation strategies. These developments put closed-loop scheduling, economic-optimizing control, and the integrated problem within a common framework, and they can hopefully serve to guide the development of alternative decomposition strategies that still enjoy the desired nominal theoretical properties while avoiding unexpected pathological behavior. [ABSTRACT FROM AUTHOR]