We have developed a systematic procedure for a limited class of chemical processes that includes control problems at the early stage of a flow sheet development. The procedure decomposes the control problem into a set of sub-problems. For the economic evaluations it is assumed that the raw-material and the recycle costs dominate the process economics, and therefore the variables which affect the input-output and the recycle material and energy flows are considered as the optimization variables (process-flow optimizations). The results obtained from this procedure will be helpful in the following areas: (1) Identifying potentially inoperable flow sheets due to the presence of trace component impurities in the feed streams or produced in the reactor, (2) Estimating the economically justified modifications (both in the flow sheet structure and the sizes of process units) to the optimum base-case design, (3) Generating alternative sets of process-flow, control structures (a set of controlled and manipulated variables, along with their pairing, that can drive the input-output and the recycle flows to their desired steady-states), (4) Estimating the magnitude of the overshoot of the manipulated variables during the transients, and changing the structure of the flow sheet, the equipment sizes, or the structure of the steady-state control structure to accommodate disturbances in an optimum way. Based on the process economics and the relative gain analysis, the optimum control structure is synthesized that would minimize the total operating cost in the face of disturbances. The optimum values of the sizes of the constrained unit designs (which restrict the process-flow optimization in the face of disturbances), and the optimum values of the process variables (i.e., the design variables and the process flows) are determined by solving a two-stage optimization problem. A method for developing approximate, dynamic models for the process flows for continuous chemical plants with recycle streams is described. (Abstract shortened with permission of author.)

A large scale interconnected system consists of several subsystems that interact dynamically with one another through an interconnection network. Because of the constraints on information flow, a decentralized control system usually provides a more practical control solution than a centralized control system. Previous studies on the control of large scale interconnected systems often ignore the important issue of modelling uncertainties in the subsystems and the interconnection dynamics, and thus result in control systems that are inadequate. This thesis develops a design procedure for designing decentralized control systems to achieve robust performance of large scale interconnected systems. Based on a set of subsystem interface specifications, the global design problem is first decomposed into a set of subsystem design problems. Each subsystem then attempts to solve the subsystem problems independently. The decomposition procedure has the property that if all the subsystem design problems are solved, then the original global system design problem is also solved. A number of issues arise from the decentralized design procedure: the problem of interface selection; the problem of robust performance characterization for systems with external signals which are modelled using independent bounds on subsets of the signals (component bounded signals); and the problem of selecting a single compensator to simultaneously satisfy the design objectives for each of several design models (multiple design models). The first issue arises from the problem decomposition procedure. In the context of this thesis, the latter two issues arise from the solution of the subsystem design problems. In addition, these problems can arise from other practical problem formulation and are of interest in their own right. These issues have been addressed in this thesis, and solution or algorithmic approaches have been provided.

Algorithmic and heuristic-based approaches are proposed for synthesizing integrated chemical systems. The former is used in the synthesis of reactor network and reactor-recycle-separator systems; the latter in the synthesis of integrated crystallization systems. In the algorithmic method, a network, or superstructure, which embeds all possible equipment to be used in the process and the potential interconnections among the equipment is generated. The procedure for generating the reactor network and the reactor-recycle-separator flow sheet structure is described. A nonlinear programming (NLP) problem is then formulated for the network. The optimal flow sheet and accompanying operational conditions are obtained by solving the NLP problem. For integrated crystallization process synthesis, a heuristic-based systematic procedure is developed. In a step-by-step manner, the procedure guides the user to generate alternative flow sheets for a given crystallization task. First, the required unit operations are determined by comparing the product specifications (production rate, product purity, and others) with the crystallizer effluent characteristics (occlusions, inclusions, crystal size, and others). Second, the destinations of the reaction solvent, mother liquor, wash liquid, recrystallization solvent, and drowning-out solvent are assigned. Then, the solvent recovery system is considered to recover the solvents and unconverted reactants, and to remove impurities from the system. Downstream system problems such as excessive filtration time and/or filter size are often caused by unfavorable crystal size. Various crystallizer designs to improve the crystal size distribution are discussed; short-cut equipment models are used to evaluate the alternatives for potential improvement. Issues related to minimization of inclusion impurities and heat integration are also examined. Guidelines are provided to help the user to add more details to the flow sheet at each level.

Process boundaries and difficult regions for separation units limit the feasible products and recovery of those products. When process boundaries are encountered, a separating agent and or combinations of different types of equipment are used. In this manner, a number of steps are used collectively to meet the separations objective. One type of equipment configuration, the distillation-membrane hybrid, has been studied for binary and multicomponent systems. In this hybrid, the distillation column performs the bulk of the separation, because of the favorable economics of distillation. The membrane is used to bypass the process boundaries and difficult regions. Methods for applying and screening these hybrids were developed. Several configurations were examined conceptually. Case studies were performed on the following systems: water-acetic acid, ethanol-water, propylene-propane, benzene-heptane-octane, methanol-ethanol-water. Separations synthesis using all possible separation units (crystallization, membranes, distillation, decantation, extraction, etc.) was then examined. A design methodology for generating flowsheets of process alternatives to separate multicomponent systems was developed based on representing process boundaries with linear hyperplanes. This approximation allowed the generation of process alternatives using relatively simple calculations.

This dissertation studies several security issues in data networks, to reveal the vulnerability, to propose defense mechanisms, to provide better tools for analysis, and to develop good security architecture. This dissertation consists of the following three parts: (1) Internet-like topologies which capture the inherent properties are desirable for studying the resilience of Internet against malicious attack or normal failure. A novel hierarchical Internet topology generator is proposed to capture the inherent properties of Internet topology: power law degree distribution and hierarchical structure. (2) An analysis on the inherent trust built in TCP shows that the client could stretch a TCP connection tens of times and keep occupying the resource in the server with little abnormality to be detected. This could be potentially used in denial of service attack. (3) Some security setting like Bounded Storage Model calls for high-speed random number generating, while the current real random number generator would not be able to offer. A hybrid random-bit sequence generated by a pseudo-random number generator with the parameters specified randomly might be useful in this setting. A study on hybrid system using Linear Congruential Recurrence is presented, and hopefully it will provide insight for the study on hybrid system using one-way function.

Plantwide control refers to the control of entire plants, consisting of many interconnected unit operations. Synthesizing a plantwide control system requires evaluating numerous alternatives involving controlled variables, control structures, controller designs and tunings, etc. A hierarchical procedure for systematically synthesizing a plantwide control has been proposed by Zheng et al. [82]. In this procedure, the plantwide control problem is decomposed into six steps along which decisions are made based on economics. While many tools (e.g., a short-cut method for controlled variable selection, quantification of dynamic operability, etc.) have been developed by these authors to ease the implementation of the procedure on industrial processes, more tools need to be developed. For example, we need to develop systematic procedures for ensuring feasibility of the control structures, for selecting primary and secondary controlled variables and for designing controllers for systems with recycles. Furthermore, model uncertainty, which is important practically, needs to be taken into account to make these tools useful. The goal of this thesis is to accomplish these tasks. To this end, we address how model uncertainty affects the steady-state as well as the dynamic control structure.

The essential characteristics of large systems is their high dimensionality due to which conventional control techniques fail to give reasonable solutions with reasonable computational efforts. A number of large systems encountered in practice are composed of subsystems with similar dynamics interconnected in a symmetrical fashion. The analysis and control of a large system with these particular features must take advantage of the existing structural properties to achieve computational simplifications of the overall problem. The focus of this thesis is the feedback design and analysis of large systems possessing the property of spatial symmetry. Specifically, the problems of controller design and analysis for infinite dimensional toeplitz systems and their finite dimensional analogs, circulant systems, are studied. These spatially symmetric systems are special classes of large systems. The first part of this thesis is focused on the development of formal controller design methodologies which take advantage of the properties of the circulant matrices. The key to this development is the use of the FFT algorithm to diagonalize circulant matrices. The resulting controller design methodologies are computationally attractive and easily applicable to large systems with circulant symmetry. More specifically, the H$\sb2$ and H$\sb{\infty}$ controller synthesis problems are studied in detail and are shown to decompose into lower order independent problems. The second part of this work concentrates on proving that certain finite order toeplitz systems are asymptotically equivalent in an appropriate sense to circulant systems. This result justifies the use of circulant control design techniques for certain toeplitz systems. Moreover, the closed loop effects of controlling a toeplitz system with a controller designed for its asymptotically equivalent circulant system are analyzed. The application of the developed theoretical results to a realistic example is the focus of the last part of the thesis. The adaptive optics system used in this example is modeled by a transfer function matrix with toeplitz symmetry. The computational efficiency of the controller design methodologies developed in this thesis is illustrated by designing a series of controllers for this system.

Communication networks have experienced tremendous growth in recent years, and it has become ever more challenging to design, control and manage systems of such speed, size and complexity. The traditional performance modeling tools include analysis, discrete-event simulation and network emulation. In this dissertation, we propose a new approach for performance modeling and we call it time-driven fluid simulation. Time-driven fluid simulation is a technique based on modeling the traffic going through the network as continuous fluid flows and the network nodes as fluid servers. Time is discretized into fixed-length intervals and the system is simulated by recursively computing the system state and advance the simulation clock. When the interval length is large, each chunk of fluid processed within one interval may represent thousands of packets/cells. In addition, since the simulation is synchronized by the fixed time intervals, it is easy to parallelize the simulator. These two factors enable us to tremendously speed up the simulation. For single class fluid with probability routing, we prove that the error introduced by discretizing a fluid model is within a deterministic bound proportional to the discretization interval length and is not related to the network size. For multi-class traffic passing through FIFO servers with class-based routing, we prove that the worst case discretization error for any fluid flow may grow linearly with the number of hops the flow passes but unaffected by the overall network size and the discretization error of other classes. We further show via simulation that certain performance measures are in fact quite robust with respect to the discretization interval length and the path length of the flow (in number of hops), and the discretization error is much smaller than that given by the worst case bound. These results show that fluid simulation can be a useful performance modeling tool filling the gap between discrete-event simulation and analysis. In this dissertation, we also apply another technique, rational approximation, to estimate the cell loss probabilities for an ATM multiplexer fed by a self-similar process. This is another method that compensates the analysis and simulation techniques.

Data communication networks have been experiencing tremendous growth in size, complexity, and heterogeneity over the last decade. This trend poses a significant challenge to the modeling, simulation, and analysis of the networks. In this dissertation, we take the fluid model as the way to attack the issue and apply it to network simulation, and to the analysis of queues. Traditional discrete-event packet-based approaches to simulating computer networks become computationally infeasible as the number of network nodes or their complexity increases. An alternative approach, in which packet-based traffic sources are replaced by fluid sources, has been proposed to address this challenge. We quantitatively characterize the amount of computational effort needed by a simulation scheme using the notion of a simulation's event rate, and derive expressions for the event rate of a packet and fluid flow at both the input and output sides of a queue. We show that the fluid-based simulation of First In First Out (FIFO) networks requires less computational effort when the network is small. However, the so-called “ripple effect” can result in fluid-based simulations becoming more expensive than their packet-based counterparts. Replacing FIFO with weighted fair queuing reduces the ripple effect, however the service rate re-distribution process incurs extra event rate. We then propose time-stepped hybrid simulation (TSHS) to deal with the scalability issue faced by traditional packet-based discrete event simulation method and fluid-based simulation methods. TSHS is a framework that offers the user the flexibility to choose the simulation time scale so as to trade off the computational cost of the simulation with its fidelity. Simulation speedup is achieved by evaluating the system at coarser time-scales. The potential loss of simulation accuracy when fine time-scale behavior is evaluated at a coarser time-scale is studied both analytically and experimentally. In addition, we compare an event-driven TSHS simulator to the time-driven version, and find out that the time-driven TSHS simulator out-performs event-driven simulator due to TSHS simulation model's time-driven nature and the simplicity of time-driven scheme. In this dissertation, we also apply the fluid model, together with the theory of stochastic differential equations, to the queueing analysis. We formulate and solve a number of general questions in this area using sample path methods as an important part of the process. Relying on the theory of stochastic differential equations, this approach brings to bear a heretofore ignored but quite effective problem solving methodology.

Experiments indicate the applicability and potential of adaptive systems for chemical process control. Stability results based on ideal conditions show that these systems asymptotically converge to the desired behavior. However, unpredictable behavior can result due to the nonlinear relationships between model uncertainty, operating conditions and the tuneable parameters of adaptive control system. The purpose of this thesis is to study the effects of model uncertainty on adaptive control systems. Global input-output and local asymptotic stability analyses are used to quantify the range of operating conditions and tuneable parameters which allow good performance for a given degree of model uncertainty. The results are used to develop better adaptive control and optimization algorithms. A nonlinear adaptive control scheme is developed which combines nonlinear model-based compensation with adaptive estimation techniques, and its stability is analyzed for a specific class of nonlinear systems. Conic sector bounds on the mismatch between the compensator and process nonlinearities are developed which guarantee global input-output stability of the feedback system. A design method is proposed which uses approximate steady-state process models. Distillation column and CSTR simulation studies reveal improved performance which results from the nonlinear compensation scheme. The concept of coupling nonlinear compensation with adaptive estimation is used to develop a novel approach to optimization based on the construction of a locally valid static relationship from approximate models and local geometric characteristics. Simulations and experiments illustrate the performance of the adaptive extremum controller on a continuous fermentor. The results from global input-output stability analyses are limited because boundedness properties can be preserved despite poor performance; however, local asymptotic analysis based on linearization and bifurcation theory provides a means of studying the effects of model uncertainty on the performance of adaptive controllers. Two simple examples are analyzed and it is shown that adaptive controllers can perform poorly in the presence of model uncertainty for certain operating conditions and tuneable parameters. Local stability boundaries are computed from the analysis; various routes to global instability and chaos are identified; and design guidelines and algorithmic modifications for a simple class of model-reference adaptive controllers are developed.