Srinivasan,Rajagopalan

Email Address

chergs@nus.edu.sg

Organizational Units

Organizational Unit

ENGINEERING

faculty

Organizational Unit

CHEMICAL & BIOMOLECULAR ENGINEERING

dept

Publication Search Results

Now showing 1 - 10 of 180

Negotiation-based approach for order acceptance in a multiplant specialty chemical manufacturing enterprise
(2011-05-04) Behdani, B.; Adhitya, A.; Lukszo, Z.; Srinivasan, R.; CHEMICAL & BIOMOLECULAR ENGINEERING
Order acceptance under uncertainty is a challenging problem in supply chain management, especially in make-to-order plants, such as in a specialty chemical manufacturing enterprise. In this work, we consider the effect of rush orders and design a negotiation-based policy for order acceptance. Rush orders pose a complex challenge primarily because the different actors involved (customer and enterprise) have different interests and asymmetric information. In addition, the interaction involves uncertainties both in the short- and long-term. To capture the rich complexity of the problem and provide a conceptual basis for developing agent-based models for such negotiation situation, this paper presents a general framework for negotiation and illustrates it using the customer-enterprise rush order due date negotiation case. On the basis of this framework, an agent-based model is developed. Various experiments are conducted to evaluate the effect of due date negotiation on profit and long-term customer behavior. © 2011 American Chemical Society.
A Graphic Processing Unit (GPU) Algorithm for Improved Variable Selection in Multivariate Process Monitoring
(2012) Chan, L.M.; Srinivasan, R.; CHEMICAL & BIOMOLECULAR ENGINEERING
Process monitoring is extremely important for producing high quality product and at the same time ensuring safe working environment in chemical process industry. Recently, it has been shown that selection of an appropriate subset of variables can improve the monitoring performance. The main contribution of this work is the development of a parallel version of the Genetic Algorithm-Principal Component Analysis algorithm which was proposed by Ghosh et al. [2] for variable selection. The developed algorithm has been implemented using NVIDIA's Compute Unified Device Architecture, CUDA parallel computing platform. Experimental results show that the proposed parallel approach is 12 times faster than the original serial code when applied to the Tennessee Eastman challenge problem. © 2012 Elsevier B.V.
From PSE to PSE2-Decision support for resilient enterprises
(2009-12-10) Naraharisetti, P.K.; Adhitya, A.; Karimi, I.A.; Srinivasan, R.; CHEMICAL & BIOMOLECULAR ENGINEERING
In recent years, the process systems engineering (PSE) community has recognized the need to address chemical enterprises comprising globally distributed, but strongly interacting, facilities. We examine this extension of PSE, which we call the PSE of enterprise (PSE2), as it relates to the five traditional PSE areas of system representation, modeling and simulation, synthesis and design, planning and scheduling, and control and supervision. We illustrate the strong structural, operational, and methodological parallels between PSE and PSE2 in this study. © 2009 Elsevier Ltd. All rights reserved.
Neural network systems for multi-dimensional temporal pattern classification
(2005-04-15) Srinivasan, R.; Wang, C.; Ho, W.K.; Lim, K.W.; ELECTRICAL & COMPUTER ENGINEERING; CHEMICAL & BIOMOLECULAR ENGINEERING
Classification of multivariate, multi-class, temporal patterns is an important yet challenging problem that arises during state identification in agile processes. In this paper, we propose two new neural network structures that overcome the curse of dimensionality and the complexities in temporal pattern recognition. The one-variable-one-network (OVON) architecture decomposes the problem in the variable dimension - univariate temporal patterns, called sub-states, are first identified in each variable; subsequently, the process state is inferred using the variable sub-states. The one-class-one-network (OCON) architecture uses a problem decomposition in the class dimension - the presence of specific temporal patterns in a high-dimensional space is first established using a bank of two-class classifiers. The process state is then identified by aggregating the results of the different classifiers. Both the architectures use a set of neural networks - in OVON there is one network for each variable, while in OCON, one network is used for each pattern class to be identified. In comparison to traditional monolithic neural networks, both the proposed architectures improve classification accuracy and minimize the training complexity. In addition, OVON is robust to sensor failures and OCON is well suited for addition of new pattern classes. The structures and training methodologies of the two architectures are presented and their performance compared against traditional neural networks using patterns arising during transitions in a simulated fluidized catalytic cracking unit. © 2004 Elsevier Ltd. All rights reserved.
Agent-based decision support for failure-prone networked infrastructures
(2009-11) Van Dam, K.H.; Lukszo, Z.; Srinivasan, R.; CHEMICAL & BIOMOLECULAR ENGINEERING
The operation of existing infrastructures is often inefficient and subject to failures. When a failure occurs, various stakeholders need to make decisions that are specific to the failure type and bear little resemblance to decisions faced during normal operation. In this work, we demonstrate a model-based approach to making rational decisions in such situations. Agent-based models serve as a suitable paradigm for modelling complex sociotechnical systems. Given the broad similarities between different networked infrastructure systems, an ontology has been developed as the foundation for a 'model factory' for such systems. A specific application of this model factory to a refinery supply chain system is described. Further, the use of this simulation model for decision support to manage an abnormal situation in the supply chain is reported. Copyright © 2009 Inderscience Enterprises Ltd.
Supply chain redesignsmultimodal optimization using a hybrid evolutionary algorithm
(2009) Naraharisetti, P.K.; Karimi, I.A.; Srinivasan, R.; CHEMICAL & BIOMOLECULAR ENGINEERING
Supply chain redesign (SCR) involves decisions regarding the timings, amounts, and locations of the investment and disinvestment in facilities, production, material purchase, product sales, contracts, capital-raising loans and bonds, etc. such that the profit is maximized. SCR is a heavily constrained problem; hence as the problem size increases, the MILP formulations (Naraharisetti, P. K.; Karimi, I. A.; Srinivasan, R. Supply Chain Redesign through Optimal Asset Management and Capital Budgeting. Comput. Chem. Eng. 2008, 32, 3153-3169) become increasingly difficult to solve. In addition, MILP solvers typically give only one solution, while multiple optimal solutions may be desirable in practice. Hence, an alternative optimization technique is warranted. In this work, we propose a hybrid MILP-evolutionary algorithm strategy for supply chain redesign and present progress on three fronts: (a) a novel reformulation of the MILP in which most decision variables are unconstrained and the rest can be easily repaired to satisfy constraints, (b) a single-objective hybrid optimization algorithm that uses an evolutionary search and reaches 97% of the objective value attained by CPlex 9.0 on a small example, while outperforming CPlex 9.0 on a large SCR problem, and (c) a multimodal algorithm that identifies multiple supply chain networks with 90-95% of the objective value obtained by CPlex 9.0. Finally, we analyze the effect of uncertainty on each supply chain network identified by our multimodal algorithm. © 2009 American Chemical Society.
A decision support database for inherently safer design
(2003) Srinivasan, R.; Chia, K.C.; Heikkila, A.-M.; Schabel, J.; CHEMICAL & ENVIRONMENTAL ENGINEERING
An inherently safer process relies on naturally occurring phenomena and robust design to eliminate or greatly reduce the need for instrumentation or administrative controls. Such a process can be designed by applying inherent safety (IS) principles such as intensification, substitution, attenuation, limitation of effects, simplification, etc, throughout the design process, from conception until completion. While the general principles and benefits of IS are well known, a searchable collection of inherently safer designs that have been implemented in industry has not been reported. Such a database of inherently safer design (ISD) examples would assist the process designer in the early stages of the design lifecycle when critical design decisions are made. In addition to examples of IS design which have been successfully carried out, the database that we have developed contains process incidents which could have been averted by the application of ISD. In this paper, details of the database, the query engine, and potential applications are presented. © 2003 Elsevier B.V. All rights reserved.
Recipe determination and scheduling of gasoline blending operations
(2010-02) Li, J.; Karimi, I.A.; Srinivasan, R.; CHEMICAL & BIOMOLECULAR ENGINEERING
Gasoline is a major contributor to the profit of a refinery. Scheduling gasolineblending operations is a critical and complex routine task involving tank allocation, component mixing, blending, product storage, and order delivery. Optimized schedules can maximize profit by avoiding ship demurrage, improving order delivery, minimizing quality give-aways, avoiding costly transitions and slop generation, and reducing inventory costs. However, the blending recipe and scheduling decisions make this problem a nonconvex mixed-integer nonlinear program (MINLP). In this article, we develop a slot-based MILP formulation for an integrated treatment of recipe, specifications, blending, and storage and incorporate many real-life features such as multipurpose product tanks, parallel nonidentical blenders, minimum run lengths, changeovers, piecewise constant profiles for blend component qualities and feed rates, etc. To ensure constant blending rates during a run, we develop a novel and efficient procedure that solves successive MILPs instead of a nonconvex MINLP. We use 14 examples with varying sizes and features to illustrate the superiority and effectiveness of our formulation and solution approach. The results show that our solution approach is superior to commercial solvers (BARON and DICOPT). © 2009 American Institute of Chemical Engineers.
Decision support for green supply chain operations by integrating dynamic simulation and LCA indicators: Diaper case study
(2011-12-01) Adhitya, A.; Halim, I.; Srinivasan, R.; CHEMICAL & BIOMOLECULAR ENGINEERING
As the issue of environmental sustainability is becoming an important business factor, companies are now looking for decision support tools to assess the fuller picture of the environmental impacts associated with their manufacturing operations and supply chain (SC) activities. Lifecycle assessment (LCA) is widely used to measure the environmental consequences assignable to a product. However, it is usually limited to a high-level snapshot of the environmental implications over the product value chain without consideration of the dynamics arising from the multitiered structure and the interactions along the SC. This paper proposes a framework for green supply chain management by integrating a SC dynamic simulation and LCA indicators to evaluate both the economic and environmental impacts of various SC decisions such as inventories, distribution network configuration, and ordering policy. The advantages of this framework are demonstrated through an industrially motivated case study involving diaper production. Three distinct scenarios are evaluated to highlight how the proposed approach enables integrated decision support for green SC design and operation. © 2011 American Chemical Society.
Novel solution approach for optimizing crude oil operations
(2004-06) Reddy, P.C.P.; Karimi, I.A.; Srinivasan, R.; CHEMICAL & BIOMOLECULAR ENGINEERING
Scheduling of crude oil operations is a complex nonlinear problem, especially when tanks hold crude mixes. We present a new mixed-integer nonlinear programming (MINLP) formulation and a novel, mixed-integer linear programming (MILP)-based solution approach for optimizing crude oil unloading, storage, and processing operations in a multi-CDU (crude distillation unit) refinery receiving crude from multiparcel VLCCs (very large crude carriers) through a high-volume, single-buoy mooring (SBM) pipeline and/or single-parcel tankers through multiple jetties. Mimicking a continuous-time formulation, our primarily discrete-time model allows multiple sequential crude transfers to occur within a time slot. It incorporates several real-life operational features including brine settling and tank-to-tank transfers, and is superior to other reported models. Notably our algorithm avoids concentration discrepancy and MINLP solutions by identifying a part of the horizon, for which its linear relaxation is exact, and then solving this MILP repeatedly with progressively shorter horizons. By using 8 h time slots and a hybrid time representation, an attractive approach to this difficult problem is presented. © 2004 American Institute of Chemical Engineers.

Srinivasan,Rajagopalan

Email Address

Organizational Units

Filters

Author

Subject

Date

Has files

Item Type

Types

Department

Settings

Sort By

Results per page

Citations

Publication Search Results