Goal: Development of methodologies and mathematical tools for RMS operations in a multi-part production mode including part re-routing scheduling and responsive maintenance policies.

The vision for this thrust area is to enhance the economic responsiveness of reconfigurable manufacturing systems to market demand fluctuation by developing a methodology that optimizes the system throughput and minimizes production downtime. In practice, temporary market demand downward fluctuation may create new challenges for operations of a RMS, including resource allocation, responsive scheduling and maintenance, and integrated decision making. Therefore, RMS requires a smart operational strategy in order to achieve responsiveness and high productivity when market demand fluctuates and system configuration changes.

TA4 research addresses the following policies through four coherent research tasks (see figure below):

  • A maintenance priority decision-making support system – instruct the maintenance crews to maintain and repair the machine that has the largest impact on throughput reduction and can be fixed at the shortest estimated time.
  • A systematic methodology to correlate machine/process performance and part quality deterioration to forecast production performance with proactive maintenance actions to achieve the maximum yield.
  • Reallocation of resources to reduce makespans in production scheduling in response to the demand fluctuations.
  • A part re-routing and scheduling policy to maximize throughput in case of machine(s) break down.

Thus, the operations of a RMS should be responsive to machine information, maintenance information as well as customer demand information. Ultimately, these decision-making tools would provide factory wide information modeling, from production batch size allocation to machine to maintenance to production enabling optimized, reconfigured and responsive schedules.

TA4 Overview

Description of TA 4 research

TA4-1 Responsive Scheduling of Maintenance and Reconfiguration Operations in Reconfigurable Manufacturing Systems

Maintenance is a critical factor in contemporary manufacturing systems and smart, strategic maintenance operations could greatly improve the overall operations of a manufacturing system. The role of maintenance is even more prominent in highly dynamic and quality critical Reconfigurable Manufacturing System, where manufacturers need to be responsive to fluctuations in market demand, while in the same time maintaining the highest possible level of product quality and process productivity.

The goal of this project is to utilize the on-line available information about the current and predicted equipment behavior, buffer status and system configuration in order to facilitate optimal maintenance scheduling that takes into account maintenance costs, production benefits, maintenance resource constraints and market demand fluctuations. The unified approach will result in an optimal system-level schedule, which will include possibilities of reconfiguration as well as traditional maintenance actions of repair and replacement and possible reactions to equipment degradation. The added freedom of reconfigurability will result in improved maintenance operations and better usage of reconfigurable resources. This project will develop, validate and demonstrate methods for proactive and responsive maintenance in Reconfigurable Manufacturing Systems (RMS) by integrating the predicted equipment reliability information and system configuration into a maintenance and reconfiguration schedule that is cost-effective with respect to the production goals and maintenance resource constraints.

TA4-2 Integration of Product Quality with Tooling System Data for Proactive Quality-Ensured Maintenance

Most of the existing maintenance decisions are made either based on reliability of the production system (reliability centered maintenance – RCM), or on-line sensing and monitoring of tooling degradations (Condition-based maintenance – CBM). However, few efforts have been made to systematically integrate the product quality measurements, either obtained on-line or off-line, into maintenance considerations. Furthermore, interactions between the production tooling and the product quality were not fully investigated for maintenance decision making. The goal of this project is to develop systematic methodologies to integrate multi-stream product quality information with the process information for effective proactive maintenance decision making for RMS. Two integration approaches are followed in this study: (1) expend the Stream of Variation (SoV) model by adding tooling degradation process to propose product quality and tooling reliability co-effect concept and associate model. An optimal maintenance policy is investigated to consider the production life-cycle cost and profits. (2) develop a new causal discovery and modeling algorithm from observational data for complex manufacturing processes where a physical model cannot obtained from design or DOE. In this model, the quality features and process variable features are extracted, and modeled into a Probabilistic Network model. The model is further used to analyze the process, predict quality and make maintenance decisions.

TA4-3 Customer-Focused Production Control

In a deterministic world, it is theoretically possible to generate production sequences which minimize makespan. In reality, system disruptions, both internal (such as machine failures) and external (such as customer changes during order processing), can make the original sequence sub-optimal or even infeasible. Thus, additional research has focused on addressing such system

variability. Inherent in these research efforts is the need to establish metrics for evaluating solution quality. Metrics typically considered include: average makespan, holding and shortage costs, and excess inventory. It is no longer sufficient, however, to focus solely on these manufacturer-focused metrics. The competitiveness of today’s global economy necessitates focusing on customer responsiveness as well. In environments where capacity is tightly constrained relative to customer demand, and where customer orders are typically filled through new production rather than from inventory, disruptions can often prevent manufacturers from fully satisfying customer demand at the original due date. In such cases, it is not immediately clear what the relationship is between a given production sequence and how customer responsive it is. Given that a customer’s demand cannot be fully satisfied at the original due date, particularly if the order quantity has changed, what feasible alternatives are most desirable? How can this less easily quantifiable notion of customer responsiveness be incorporated in the decision making process?

The goal of this project is to develop production control strategies with focus on customer requirements in a RMS environment. In this research, we study how sequencing decisions can be used not only to decrease makespan, but also to improve customer responsiveness. A key element of our work is that we do not dictate how the system performance should be evaluated, but rather allow users flexibility in assessing solution quality according to their individual situations. The goal of our research is thus, not to develop optimal sequencing strategies for a single industrial situation, but rather to develop a framework – both a vocabulary and a set of analytical tools – for analyzing a broad class of problems.

TA4-4 Responsive Reconfigurable Resource Allocation with Reliability Considerations

In today’s market, more robust manufacturing processes are crucial. The moving bottleneck caused by demand changes and machine breakdowns or repairs has always caused companies difficulty. One solution to this problem is to keep excess machine and buffer capacity on hand for each of several stations. However, since product life-cycles are shorter and machine costs are higher, this solution is very expensive. Without additional capital investments on excess capacity, dynamic allocation of resources will allow factory managers to reduce recovery costs of machine failures by routing jobs from failed machines to reconfigurable machines. Compared to dedicated manufacturing systems, the same performance indices (throughput and average WIPs in the system) can be achieved by using less capacity in RMS.

The goal of this project is to develop methodologies to realize responsive reconfigurable resource allocation with reliability considerations. The research results will provide companies with estimates of how valuable a reconfigurable resource can be from an operational standpoint. That is to say, given inputs like mean time to failure and average downtime, our policy and cost benefit evaluation software can show the difference in the cost between systems with or without dynamically allocated reconfigurable resources.


TA4-1 Maintenance Scheduling in RMS

  • Developed the maintenance policies for incorporating reconfiguration into age-based and condition-based preventive maintenance actions
  • Investigated both Integrated Reconfiguration & Age-based Maintenance (IRABM) policy and the Integrated Reconfiguration & Condition-based Maintenance (IRCBM) policy through simulation modeling and analysis
  • Conducted the benchmarking study for ABM & IRABM Policies
  • Performed quantitative study for integrated reconfiguration and condition-based preventive maintenance (IRCBM) policy

TA4-2 Quality and tooling data integration for quality-ensured maintenance

  • Developed the methodologies for product quality and tooling data integration through both physical (SoV) and empirical (causal discovery) modeling approaches.
  • Conducted simulation study for modeling, analysis and decision making for quality ensuredmaintenance.
  • Developed new causal discovery modeling techniques to realize data fusion for complex manufacturing systems, and also conducted case studies from production data.
  • Developed quality-ensured maintenance strategies for Multistage Manufacturing Processes considering.

TA4-3 Customer Focused Production Control

  • Developed mathematical models and associated software to minimize makespan (job completion time) for N products in M stage pure serial configurations
  • Constructed a responsiveness framework to assess how sequencing impacts not only cost but responsiveness with regards to unexpected internal and external system disruptions
  • Explored industrial applications of the proposed techniques and identified gaps and limitations for further improvements.