Diego Ruiz

On-line fault diagnosis system support for reactive scheduling in multipurpose batch chemical plants

Abstract In this work, a simple strategy for the development and implementation of a fault diagnosis system (FDS) that interacts with a schedule optimiser in batch chemical plants is presented. The proposed FDS consists of an artificial neural network (ANN) structure supplemented with a knowledge-based expert system (KBES) in a block-oriented configuration. The system combines the adaptive learning diagnostic procedure of the ANN and the transparent deep knowledge representation of the KBES. The information needed to implement the FDS includes a historical database of past batches, a Hazard and Operability (HAZOP) analysis and a model of the plant. Two motivating case studies are presented to show the results of the proposed methodology. The first corresponds to a fed-batch reactor. In this example, the FDS performance is demonstrated through the simulation of different process faults. The second case study corresponds to a multipurpose batch plant. In this case, the results of reactive scheduling are shown by simulating different abnormal situations. A performance comparison is made against the traditional scheduling approach without the support of the proposed FDS.

Neural network based framework for fault diagnosis in batch chemical plants

Abstract In this work, an artificial neural network (ANN) based framework for fault diagnosis in batch chemical plants is presented. The proposed FDS consists of an ANN structure supplemented with a knowledge based expert system (KBES) in a block-oriented configuration. The system combines the adaptive learning diagnostic procedure of the ANN and the transparent deep knowledge representation of the KBES. The information needed to implement the FDS includes a historical database of past batches, a hazard and operability (HAZOP) analysis and a model of the batch plant. The historical database that includes information related to normal and abnormal operating conditions is used to train the ANN structure. The deviations of the on-line measurements from a reference profile are processed by a multi-scale wavelet in order to determine the singularities of the transients and to reduce the dimensionality of the data. The processed signals are the inputs of an ANN. The ANNs outputs are the signals of the different suspected faults. The HAZOP analysis is useful to build the process deep knowledge base (KB) of the plant. This base relies on the knowledge of the operators and engineers about the process and allows the formulation of artificial intelligence algorithms. The case study corresponds to a batch reactor. The FDS performance is demonstrated through the simulation of different process faults. The FDS proposed is also compared with other approaches based on multi-way principal component analysis.

Fault diagnosis support system for complex chemical plants

Abstract A process fault detection and diagnosis system (PFD&D) is proposed for complex chemical plants. The system combines an artificial neural network (ANN) based supplement of a fuzzy system in a block-oriented configuration. A methodology for designing the system is described. As a motivating example, a chemical plant with a recycle stream is considered. Faults in the supply of raw materials and in controllers are simulated. The performance of the system in handling simultaneous faults is also analysed. A comparison of the proposed approach is made with a classification method (ANNs) and inference methods (knowledge-based system). Results of system implementation in a fluidised bed coal gasifier at pilot plant scale are also shown.

On-line process fault detection and diagnosis in plants with recycle

Abstract A process fault detection and diagnosis system is performed for the complex case of plant-wide control in processes with recycle in which the control system is the inventory control. It is considered an artificial neural network based supplement of a fuzzy system in a block oriented configuration. A methodology to design the system is described. As a case study, a chemical plant with a recycle stream is considered. Faults in supply of raw materials and in controllers are simulated. Performance of the system to handle simultaneous faults is also analysed. A comparison is made against a classification method (artificial neural networks) and an inference method (knowledge — based system).

Industrial Experience in the Deployment of Real Time Online Energy Management Systems

Abstract This paper presents real industrial examples in which the whole utilities system of a production Site (i.e., steam, fuels, boiler feed water and electricity) is optimized with a real time online, industrially well established software. Experiences gained during more than 20 years of industrial projects deployed worldwide are commented. Main project steps are explained and critical details to be taken into account to assure successful use and proper technology transfer are presented. The optimization objective is the overall utilities system cost reduction and takes into account the constraints associated with the existing equipment, fuels and electricity pricing and contracts, including emissions limits, quotas and rights. The energy management system models are executed and optimized at a scheduled frequency, fed with online, real time data, flowing into and out the program using the standard OPC protocol. Besides the optimization, Key Performance Indicators (KPIs) are also calculated and sent back to the Site Plant Information System or DCSs for Operations and Management use. Application examples and results corresponding to projects implemented worldwide in refineries and chemical plants are presented and commented.

Utilities systems on-line optimization and monitoring: Experiences from the real world

Abstract Utilities systems at oil refineries and other large industrial complexes such as pulp and paper mills or chemical plants are very big energy users that have many degrees of freedom. Manipulating these degrees of freedom with the advice of a cost based optimization program usually can result in significant savings in operating costs with small investment needs. This is particularly important within the electrical deregulation context. Since the electrical system is the main economic trade-off with a steam system, electrical deregulation provides many new challenges in order to operate the combined systems at the minimum overall cost. This paper will not describe just all the features of the software or fully explain on-line optimization technology. The objective of this work is to present some interesting facts and lessons from the experience of implementing a cost based optimization program at thirty oil refineries and petrochemical complexes, around the world, since 1997. This paper will focus on the key optimization variables and constraints in steam system optimization, how they should be handled and how the human and organizational aspects can be addressed. Several of the key optimization problems found in a typical oil refinery steam system such as boilers, extraction-condensing turbines, co-generation and turbine/motors spare drivers are discussed and how those problems can be handled properly is described.

Integration of Decision Tools in EMS

Abstract This paper introduces an Energy Management System (EMS) for industrial plants, which supports decisions in several levels: scheduling, real-time optimization and key performance indicators (KPI) monitoring. The examples that illustrate the application of the EMS include real industrial implementations in petrochemical plants and district heating and cooling complexes.

Use of Online Energy System Optimization Models

Modern industrial facilities operate complex and inter-related power systems. They frequently combine internal utilities production with external suppliers, including direct fired boilers, electric power generation with turbo alternators or gas turbines, heat recovery steam generators, have different drivers (i.e., turbines or motors) for pumps or compressors and several types of fuels available to be used. Tighter and increasingly restrictive regulations related to emissions are also imposing constraints and adding complexity to their management. Deregulated electric and fuels markets with varying prices (seasonally or daily), contracted and emissions quotas add even more complexity. Production Department usually has the responsibility for the operation of the facility power system but, although Operators are instructed to minimize energy usage and usually tend to do it, a conflict often is faced as the main goal of Production is to maintain the factory output at the scheduled target. The power and utilities system is seen as a subsidiary provider of the utilities needed to accomplish with the production target, whichever it takes to generate it. Big and complex industrial facilities like Refineries and Petrochemicals are becoming increasingly aware that power systems need to be optimally managed because any energy reduction that Operations accomplish in the producing Units could eventually be wasted if the overall power system cost is not properly managed. However, process engineers always attempted to develop some kind of tools, many times spreadsheet based, to improve the way utilities systems were operated. The main drawback of the earlier attempts was the lack of data: engineers spent the whole day at phone or visiting the control rooms to gather information from the Distributed Control System (DCS) data historian, process it at the spreadsheet and produce recommendations that, when ready to be applied, were outdated and not any more applicable. The evolution from plant information scattered through many islands of automation to unified and centralized Plant Information Systems was a clear breakthrough for the process engineering work. The long term, facility wide Plant Information System based historians constitute what is known as an enabling technology, because they became the cornerstone from where to build many other applications. Besides others, advanced process control, optimal production programming, scheduling and real time optimization technologies were built over them and flourished after data was stored for long terms and became easily retrievable.

Dynamic cross-functional factory-to-business links in the batch industry

Publisher Summary The lack of a cross-functional factory-to-business link between the shop floor and the necessary supply chain relation to e-business creates a gap. To bridge this gap, a web-business-plant route is developed in a pilot plant using the TicTacToe sequencing algorithm. A web-based order management system is created to generate optimal plans taking into account the factory logistic status and detailed information of the real plant through a fault diagnosis system (FDS) with rescheduling capabilities. A simple strategy for a web-business-plant connection between the shop floor and a sequence optimizer in a multiproduct batch plant is presented. When a new plan is developed to fulfill the customer orders received through the web, it is considered the actual status of the ongoing plan, which also includes the FDS amendment actions taken into account at previous time. This allows realistic and efficient plant performances to support the dynamics of the manufacturing market. Real-time optimization (RTO) problems and integrated systems are waiting for and must foster improved solutions to old, complex, and unsolved problems. New approaches with trade-off scopes must be developed to fill the gap between technology and science.

The use of process dynamic simulation for learning to design digital controllers

The discussion presented in this paper draws on a comparative study based on several disign techniques of digital feedback controllers implemented as SISO structure of a binary distillation column. By using rigorous process models, which serves as an authentic pilot plant, offers an attractive tool to learn more efficiently the “real” control problem. It must be noted that many textbooks have presented several design techniques about this subject but they did not compare over chemical processes. Generally the overall conclusions are based on linear transfer functions only. Chemical processes represent a real challenger for developing efficient digital control design techniques. The overall steps from the identification of the nonlinear, with dead time and inverse response system, including the design and tune of the controllers are presented. Hence, is easier to verify which is the “cost” of obtaining bad models which drive to bad controller designs. Several proofs, for load and reference changes are carried out by designing discrete controllers such as PID, Ragazzini and W transform methodologies by using MATLAB-SIMULINK software. The rigorous model of the distillation column is developed through an S-function of SIMULINK.