Developing a prescriptive decision support system for shop floor control
| DOI | https://doi.org/10.1108/IMDS-09-2021-0584 |
| Published date | 15 July 2022 |
| Date | 15 July 2022 |
| Pages | 1853-1881 |
| Subject Matter | Information & knowledge management,Information systems,Data management systems,Knowledge management,Knowledge sharing,Management science & operations,Supply chain management,Supply chain information systems,Logistics,Quality management/systems |
| Author | Minakshi Kumari,Makarand S. Kulkarni |
Developing a prescriptive decision
support system for shop
floor control
Minakshi Kumari
Operations and Supply Chain Area, Indian Institute of Management Jammu,
Jammu, India, and
Makarand S. Kulkarni
Department of Mechanical Engineering, Indian Institute of Technology Bombay,
Mumbai, India
Abstract
Purpose –The reported study aims at connecting the two crucial aspects of manufacturing of future, i.e.
advanced analytics and digital simulation, with an objective to facilitate real-time control of manufacturing
operations. The workputs forward a framework for designingprescriptive decision support system for a multi-
machine manufacturing environment.
Design/methodology/approach –The schema of the decision support system design begins with the
development of a simulation model for a manufacturing shop floor. The developed model facilitates prediction
followed by prescription. As a connecting link between prediction and prescription mechanism, heuristics for
intervention have been proposed. Sequential design and simulation-based demonstration of activities that span
from development of a multi-machine shop floor model; a prediction mechanism and a scheme of intervention
that ultimately leads to prescription generation are the highlights of the current work.
Findings –The study reveals that the effect of intervention on the observed predictors varies from one
another. For a machine under observation, subject to same intervention scheme, while two of the predictive
measures namely penalty and desirabilitystabilize after a certain point, a third measure, i.e. complexity, shows
either an increase or decrease in percent change. The work objectively establishes that intervention plans have
to be evaluated for every machine as well as for every environmental variable and emphasizes the need for
dynamic evaluation and control mechanism.
Originality/value–The proposedprescriptive controlmechanism hasbeen demonstratedthrough a case of a
high pressure die casting (HPDC) manufacturer.
Keywords Heuristics, Industry 4.0, Multi-machine, Prescriptive analytics, Predictive control, Simulation
Paper type Research paper
1. Introduction
Digital transformation and data-driven decision-making are an evident trend across all
sectors. The latest edition of industrial revolution, Industry 4.0, represents a major
application area for advanced analytics due to the enormous amount of data generated across
its value chain (Evans and Lindner, 2012). At the core of Industry 4.0 is the notion of
developing “Resilient Factory”by leveraging the advantage of emerging information
technologies like Internet of Things (IoT), wireless sensor networks, big data, cloud
computing, embedded system and mobile Internet services. These smart facilities aim at
creating data intensive environments which if used rationally may facilitate effective and
accurate decision-making in real time (Monostori et al., 2016;Kusiak, 2018). With a rise in
demand for data-driven decision-making mechanisms, smart analytics is gaining the center
stage and a gradual migration from just “descriptive”to “descriptive”,“predictive”and
“prescriptive analytics”is being observed (Kiron and Shockley, 2011;Xu and Duan, 2019;
Bertsimas and Kallus, 2020). While descriptive analysis techniques have been used widely for
drawing insights through patterns and trends of available information, use of predictive
techniques has gained significant interest as a means to identify the potential risk to
Prescriptive
decision
support system
1853
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/0263-5577.htm
Received 22 September 2021
Revised 15 April 2022
11 June 2022
Accepted 29 June 2022
Industrial Management & Data
Systems
Vol. 122 No. 8, 2022
pp. 1853-1881
© Emerald Publishing Limited
0263-5577
DOI 10.1108/IMDS-09-2021-0584
manufacturers (Flath and Stein, 2018). As a latest addition to this set, prescriptive design of
operations is attracting a lot of research interest (Clausen et al., 2020).
Effective prediction as well as prescription for a manufacturing product or process is
based on three things (Gr€
oger, 2018). It requires a quick processing of enormous amount of
data into useful information, integration of data from a variety of sources (product, process
and system) to decipher the pattern of events and a rapid response system that predicts the
future based on historical as well as prevailing system state. Industry 4.0 puts forward the
creation of advanced simulation models as a means to meet the above requirements (Gr€
oger,
2018;Alc
acer and Cruz-Machado, 2019).
The reported study aims at joining the two crucial aspects of Industry 4.0 vision, i.e.
advanced analytics and simulation, with an objective to facilitate real-time control of
manufacturing functions. The reported work is a graduation from robust prediction
mechanism to prescription scheme. The work puts forward a framework for developing a
simulation model for a multi-machine manufactu ring environment, which facilitates
prediction followed by prescription. As a connecting link between prediction and
prescription mechanism, heuristics for intervention have been proposed. Sequential flow of
activities that begins with system assessment, development of a virtual replica of a multi-
machine and a prediction mechanism along with a scheme of intervention that ultimately
leads to prescription are the highlights of the current work. The proposed plan of prescriptive
control has been demonstrated through a case of a high pressure die casting (HPDC)
manufacturer.
The paper is organized in sections with Section 2 discussing the problem under
consideration. Section 3 gives an account of the literature reviewed with a subsection on the
observations drawn from the literature. Section 4 elaborates the development of the
prescription mechanism with subsections on development of simulation model, the prediction
mechanism and the intervention plans. Section 5 presents a case study of a HPDC
manufacturing system with subsections on system description and assumptions made for the
model demonstration. Sections 6 and 7, respectively, demonstrate the applicability of
intervention plans and mechanism of prescription generation for the observed HPDC shop
floor. Section 8 presents the practical implications and limitations of the current study and the
paper concludes with Section 9.
2. Problem statement
The research problem discussed in this paper is related to a manufacturing shop floor. By
shop floor, the reference is to the area in a manufacturing facility where production work
takes place. A manufacturing shop floor relies on the efficient interaction of its robust
production schedule, maintenance planning and execution and quality monitoring for its
improved performance. Each of these functions has to keep accommodating the uncertainties
associated with one another (Kumari et al., 2017). A prediction mechanism presented by
Kumari and Kulkarni (2016,2019a,b) for a single machine system aims to identify the
environmental variables that require intervention. The developed prediction mechanism aids
in screening out the domain(/s) that require intervention which can be scheduling,
maintenance or quality individually or jointly. Having developed and demonstrated the
usability of the prediction models for manufacturing shop floor, a compelling question that
arises is as follows: “Even if the environmental variables requiring intervention are identified,
what should a decision maker do with them?”
To add to it, other level of intricacy arises as the system configuration graduates from
single to multiple machine system configuration. A multi-machine system configuration, as a
factual representation of a typical manufacturing setup, calls for a hierarchical decision-
making structure. Within this hierarchy, the machine-level performance is of interest to the
IMDS
122,8
1854
To continue reading
Request your trial