K.P. Madhavan

Pinch analysis for aggregate production planning in supply chains

Global competition has made it imperative for the process industries to manage their supply chains optimally. The complexity of the supply chain processes coupled with large computational times often makes effective supply chain management (SCM) difficult. Production system is an important component of a supply chain. This paper introduces a novel approach for aggregate planning of production in supply chains. The approach derives inspiration from pinch analysis, which has been extensively used in heat and mass exchanger network synthesis. By representing demand and supply data as composites, it gives planners greater insight into the SCM process and thus facilitates re-planning and quick decision-making. Two case studies are solved, one involving a single product and another involving multiple products on a single processor. For the first case study, optimal production plans are obtained and matched with the results obtained by solving equivalent optimization problems in GAMS®. For the second case study, an algorithm is proposed to determine the sequence of production of the multiple products. The initial guess obtained by following the algorithm reduces the computational time to one-sixth of the time otherwise taken by the solver. It may be concluded that plans obtained by pinch analysis provide either the best aggregate plans or excellent starting points to reduce the computational time for solutions by mixed integer programming formulations.

Optimal reflux rate policy determination for multicomponent batch distillation columns

Abstract This paper describes a procedure to obtain optimal reflux or optimal distillate rate policy for multicomponent batch distillation columns using the Pontryagins continuous maximum principle. For application to multicomponent systems, a reduced order model is used using short-cut procedure developed specifically for analysis and design of batch columns.

Nonlinear internal model control using quadratic prediction models

This study is aimed at development of a nonlinear IMC controller that can handle a larger class of nonlinear systems than hitherto, including the singular systems that exhibit change in the sign of the steady state gain. The construction of the model inverse is achieved through inversion of successive quadratic approximation of the nonlinear model operator. The controller synthesis problem is formulated as minimization of 2-norm of single step prediction error. Using the theory of solutions of the multidimensional quadratic equations, a closed form dead-beat type quadratic control law is evolved. The performance of the proposed control law is demonstrated by simulating control of two highly nonlinear systems and also a singular system. In the case of singular system, it has been shown that the proposed quadratic control law can be effectively used for handling a difficult robust control problem arising out of non-attainability of set point due to plant-model mismatch.

BATCH-DIST - A COMPREHENSIVE PACKAGE FOR SIMULATION, DESIGN, OPTIMIZATION AND OPTIMAL-CONTROL OF MULTICOMPONENT, MULTIFRACTION BATCH DISTILLATION-COLUMNS

Abstract BATCH-DIST is a general-purpose simulation package for the design, simulation and optimization of multicomponent, multifraction batch distillation columns operating under different modes (constant reflux, variable reflux and optimal reflux policy). The package includes simulation models of varying degrees of complexity and rigor; efficient but simplified models (based on short-cut methods) for preliminary design and rapid analysis of column behavior, and rigorous models (based on solution of transient heat and mass balance differential equations) for verification and detailed column design. Besides simulation and design, BATCH-DIST can also accomplish optimization and optimal control of columns. Coded in Fortran 77, the package is flexible and user-friendly. BATCH-DIST has been extensively tested with benchmark cases involving binary and multicomponent systems, with nonideal behavior and in columns with appreciable holdup effects. Such test cases have clearly demonstrated that predictions of the simplified models in the package compare well with those of the rigorous models. This powerful and comprehensive package is expected to be computationally more efficient than existing packages.

Nonlinear controller for a pH process

Abstract A nonlinear feedback—feedforward controller is synthesized for controlling pH processes. The performance of the proposed controller is evaluated by simulation both for regulatory and servo problems. The design procedure of the present method is much easier and the preformance of the controller is very good. The control system is robust to significant parameter variations and to unmeasurable disturbances.

Robust Controllers for Nonlinear Chemical Processes

Abstract The design of robust nonlinear feedback-feedforward controllers for state-variable nonlinear systems x =f(x) + g(x)u + d(t) and y = h(x) is presented. The controller is robust to modelling errors and disturbances occurring in the system. This is illustrated with a composition control of semi-batch copolymerization reactors.

Virtual air-fuel ratio sensors for engine control and diagnostics

Virtual air-fuel ratio sensors for an internal combustion engine using recurrent neural and wavelet networks have been developed. A nonlinear state-space modeling strategy is proposed for the architecture of the stated recurrent neural network which is trained using some variants of real time recurrent learning (RTRL) algorithm. A two-stage training approach is proposed for improving the accuracy of the RNN topology. Additionally, wavelets as activation functions have been employed to construct a single-layer network called wavenet. The wavenet is used to model the exhaust air-fuel ratio that has proved a more challenging task in a purely neural net-based architecture using sigmoid activation functions. The methodology has been implemented in a V8 spark ignition engine through rapid prototyping tools for the real time generalization and performance evaluation. Observations and comments are made on the test patterns used for the training. Some of the limitations of such a data driven approach are highlighted. Representative experimental results for the 8-cylinder engine test data are listed. The virtual sensor may be used for more precise average air-fuel ratio control and enhanced reliability engendered through the diagnostic capabilities of the sensor

Modeling of pH process using recurrent neural network and wavenet

This paper focuses on development of an advanced neural network architecture that has the capability to model the steady state and dynamic behavior of a complex nonlinear process. The pH process considered in the study shows variability in the nonlinear behavior and can also exhibit drastic changes in the gain over a small operating region. Two ANN architectures have been considered to provide the necessary flexibility and discriminatory features required to capture this behavior. The highly nonlinear state space model of pH process is captured by Recurrent Neural Network (RNN) trained using variant of Real Time Recurrent Learning (RTRL) algorithm. The RNN is found to model the behavior over narrower domains of operation. To capture the multi-scale character of the response, wavelets are used as activation functions to construct a single layer network called wavenet. Wavenet architecture shows promise in handling the complex non-linearity exhibited by the pH process.

Fuzzy control of a semi-batch copolymerization reactor

Abstract This study describes the development of a rule-based fuzzy controller for regulating the copolymer composition ( F 1 ) in a semi-batch solution copolymerization s. The membership functions, fuzzy ranges for errors and the controller output are defined. A fuzzy rule base has been constructed relating error to controller output based on the operators knowledge. The performance of the fuzzy controller has been evaluated by simulating the semi-batch copolymerization reactor by means of a mathematical model. The fuzzy controller developed at one operating condition has also been tested for other operating conditions.

Robust control of a batch-fed fermenter

Abstract The design of a robust non-linear feedback controller is analysed for product concentration tracking in a batch-fed fermenter. The simulation results show the effectiveness of the control scheme and its robustness to modelling errors and disturbances entering the system. The controller design is also analysed for the situation in which the kinetics are unknown and only limited measurements of state variables are available.