I. Facilities Design
- Cellular Manufacturing: I proposed a new material handling based formulation for manufacturing cell formation in 1990. It departed from the then prevalent matrix or similarity coefficient-based methods. The paper received very large citations (322 in Google Scholar) and impacted future research in cellular manufacturing for the next two decades. Additional papers are IJPR 1990b, 1994; Annals of OR 1996.
- Location Theoretic Approaches to Facility Layout: This is a highly innovative alternative to the traditional space discretization and quadratic assignment approach to facility layout. Simultaneously, location theory is advanced to consider finite-sized new facilities. Along with a co-PI, two NSF awards were granted in this area, and high impact papers were published (about 18 in total, including OR, 2018 (in R4), 2002; Mg. Sc. 2009; IIE Trans 2011, 2005, 2003; EJOR 2009, 2007, 2007; NRL 2005; C&OR).
- Hybrid/Integrated Facility Design: Most facility redesign situations do not lend themselves to pure cellular structures. Aisle design, Input/Output point selection, and Material Handling System design are intertwined problems. A notable highlight that integrates layout and aisle design under congestion is: Zhang, M., Batta, R. and Nagi, R., “Designing Manufacturing Facility Layouts to Mitigate Congestion,” IIE Transactions on Design and Manufacturing, 2011, Vol. 43(10), pp. 689-702. [IIE Transactions Best paper award from journal issues from July 2011 through June 2012.]
II. Operation of Production Systems
- Introduced a new concept of Just-In-Time Production of Large Assemblies: While the academic literature is rich in scheduling theory, the scheduling of large multi-level assemblies has not attracted much attention. This problem is very relevant in practice and most companies use MRP (Materials Requirement Planning) systems to plan/schedule orders. Since the ground-breaking publication in IIE Transactions in 1996, I have been working on incorporating many realistic issues like lot-sizing and material handling, in this framework. This problem is of significant industrial impact (even though the solution methods are often heuristics). For instance: (i) ‘Contrary to popular belief, we have shown that lot-for-lot production in small batches may not always be the best JIT strategy. Even when the set-up times are as small as double the unit operation processing times, a judicious lot merging may achieve the goals of JIT better.’ (ii) In an empirical study integrated schedules provided improved makespans over sequential scheduling of work-center and then the material-handling transporters in 93% of the cases. Newer papers included formal Lagrangian optimization methods IEEE-TASE 2013; IJPR 2013.
- Integrated Production Logistics and Supply Chain Design. I have worked with two doctoral students in integrated production and distribution systems and supply chain design problems. I published a survey paper in IIE Transactions in 1999 (441 citations in Google Scholar) and contributed to supply chain design in response to a new market opportunity (3 papers; 2 [ScienceDirect’s Top 25 Hottest Articles.]). The paper on “Economics of Managerial Neglect in Supply Chain Delivery Performance” in the Engineering Economist received much publicity through various news releases (incl. New York Times and Washington Post).
III. Agile Manufacturing
- NSF CAREER award 1996: Developed a workflow-based information system paradigm to keep engineering information consistent in an agile/distributed partner manufacturing environment: Software prototype Virtual Information System for Agile Manufacturing Enterprises VISAME. IGES to STEP AP202 design translation research resulted in commercial implementation in STEPTools software http://www.steptools.com (leading STEP solutions provider). (211 and 77 citations.) Educational impact: Developed one of its kind Graduate Course in Agile Manufacturing integrating research results from CAREER award (presented in ASEE).
- Data Mining in Engineering Design: Novel application of data mining research in the domain of Engineering Design. Developed new unordered tree mining approach suitable for clustering Bills-Of-Materials. This can aid in search for similar products or parts or help develop product configurators. C&OR 2006; IEEE SMC 2005; ASME JCISE 2004.
- Mass Customization of Assemblies (Variant Design): This work (published as a two-part paper in IIE Transactions) has the potential to impact mass-customization of assembled products (e.g., automobiles). The approach is based on feature-level mating graph to capture assembly constraints and on hierarchical object models to capture compositional relationships, searching similar parts, and finally modifying them cost effectively to proliferate mass-customized variants.
IV. Sensor Networks, Military Applications, and Social Networks
- A new method for clustering sensors in wireless ad hoc networks operating in a threat environment was published in OR 2005. Cross pollination with location theoretic methods for a new Dynamic Max Expected Covering Location Problem was invented.
- Unmanned Aerial Vehicle Routing under threat, or joint routing and sensor selection; I have received as co-PI and PI two ONR awards on this topic. OR 2017; C&OR 2012; J Sim 2014; IEEE-TASE 2018.
- High-Level Information Fusion through graph theoretic models/algorithms is an area of recent interest. I have received funding from IARPA and many DoD agencies on these topics. I led as PI an Army Research Office MURI project ($6.25M) on Hard-Soft Information Fusion and another IARPA/AFRL project on Knowledge Discovery ($500k/yr).
- New algorithms for influence maximization in social networks. New parallel cascade model and optimal seed selection and scheduling (OMEGA 2018, 2016; IEEE CSS 2016). New graph/centrality algorithms (SNAM 2016, 2014). DARPA Graph Challenge Honorable Mention 2017.
V. Accelerated GPU Computing, Big Data and Large-scale Combinatorial Optimization
- Fastest Linear Assignment Problem Solver using Graphics Processing Units (GPU). PARCO 2016.
- Multi-Dimensional Assignment Problem using Map-Reduce and GPUs. PARCO 2013; 2018 (subm.)
- Quadratic Assignment Problem using Blue Waters Supercomputer. Tightest known lower-bounds and solved previously unsolved problems. J of Computing 2018 (subm.)
- Graph Association Problem introduced to literature. NRL 2013. Entity Resolution: Info Fusion 2018.
- Graph Matching (approximate, stochastic and dynamic) Info Fusion 2010; 2014; 2016.