Elsevier

Computers in Industry

Volume 131, October 2021, 103481
Computers in Industry

Analytics-enabled escalation management: System development and business value assessment

https://doi.org/10.1016/j.compind.2021.103481Get rights and content

Highlights

  • Digitalization facilitates automation of the disruption-handling process in the manufacturing industry.

  • Predictive analytics algorithms (classification and regression) are used to select disruption responders.

  • Business value is evaluated using a real world implementation example and clustered according to the required level of analytics sophistication.

  • Long-term transformational business value exceeds initial automational business value.

Abstract

Industry 4.0 initiatives can help traditional manufacturing industry cope with increasing global competition. Such solutions facilitate transparency, automation as well as business process transformation. This paper elaborates on a collaboration with a medium-sized manufacturing company. We highlight the design, evaluation and roll-out of an escalation management system with integrated data-driven decision support. We do so by applying an action design research process. Thereby, our study focuses on the system design concerning the creation of business value.

The system leverages state-of-the-art machine learning algorithms for disruption type classification and escalation handling duration prediction. These predictions can be embedded in an integrated planning procedure leveraging diverse organizational data sources (e.g., personnel availability, production plans) to instantiate a prescriptive analytics solution. Combined with informative analytics insights, this allows the proposed system to generate significant business value by reducing escalation durations. In the long run, the transformational business value enabled by the system is likely to exceed the automational business value. This highlights the special importance of tight integration of industrial analytics applications within business processes.

Introduction

In recent years increasing global competition (Matschewsky et al., 2018) as well as disruptive supply and demand shocks (e.g., COVID-19 pandemic, US-China trade war) have increased cost pressure across all industries. At the same time, there is a trend towards highly customized products running against the traditional efficiency lever of scaling up production. To better cope with these challenging situations, manufacturing companies seek to increase productivity by adopting lean manufacturing practices or process design and facility layout improvements (Kovács, 2020). More recently, firms started to push forward industrial internet initiatives (Rüßmann et al, 2015, Gilchrist, 2016, Kagermann et al, 2013) in order to support or automate labor-intensive processes and increase their productivity (Ghobakhloo and Fathi, 2020). To achieve this, manufacturing systems are sensorized and connected to IT systems (Monostori et al, 2016, Müller et al, 2017) allowing the automated and continuous collection of information. Wang et al. (2015) notes, that such Industry 4.0 initiatives “emphasize[s] the extension of traditional manufacturing systems to full integration of physical, embedded and IT systems including the Internet.”. We adopt the framework proposed by Mooney et al. (1996), to assess the business value created by Industry 4.0 (I4.0) from a theoretical lens. To account for the recent developments in business analytics, we map the respective success value tiers to the corresponding levels of analytics sophistication (Camm et al., 2020) and database interaction to establish potentials and requirements (Fig. 1).

The automation of processes, such as data collection, creates automational value. This is the basis for informational value, which emerges from information collection and subsequent dissemination, e.g., through process monitoring and dashboards. Leveraging automational and informational business value is supported by existing commercial software solutions for automated (disruption) processing, information visualization or descriptive analytics.1 To go beyond automational and informational value companies have to transform business processes to become data-driven and thereby create transformational business value. This necessitates integration of live production data with other organizational information systems such as manufacturing execution systems (MES) or production planning systems (PPS). I4.0 applications can better justify the required investments by creating business value along multiple dimensions. Furthermore, information processing and system integration for a wide range of applications is necessary.

This paper is concerned with efficient handling of production process disruptions via an analytics-enabled escalation management system. In manufacturing settings, disruptions2 result in a situation where a worker cannot continue the current task. Hence, production at single workstations or even across the complete line is interrupted resulting in sizeable disruption costs. However, due to today's manufacturing processes’ complexity, such disruptions cannot entirely be avoided. To reduce the cost of downtime, companies rely on escalation management systems to efficiently handle disruptions and in turn improve productivity (Lopez-Leyva et al., 2020). Such systems usually prompt a responder3 in order to assist in solving a disruption (Macdonald and Corsi, 2013).

Existing escalation management systems automate the notification process as well as the collection of disruption data and increase shop-floor transparency by means of monitoring tools such as dashboards. These systems typically inform any available response person (comparable to pushing the service button in an airplane). This approach is well-suited for simple settings where tasks can be handled by any responder. However, disruptions in more complex production environments often require specific skills and the notification of any available responder is inadequate. In such settings, predefined responders with a broad skill-set are deployed to analyze the cause of a disruption and subsequently notify a suited expert. This approach avoids sending wrong responders at the cost of an expensive two-step approach.

In contrast, we envision a data-driven escalation management system that automatically identifies the underlying disruption cause and directly dispatches the best-suited expert. Clearly, the performance of such a system is driven by the ability to correctly predict the disruption type. While good predictions significantly reduce down-times and costs by sending the correct expert directly, bad predictions lead to wrong dispatches—and therefore unresolved disruptions requiring re-dispatches of other specialists—and incur additional costs.

Such a system is currently not offered by commercial available systems. We collaborated with WITTENSTEIN SE, a German medium-sized manufacturing company with multiple distributed production and assembly lines for highly customized mechatronic products. As part of a large scale company-wide digitalization strategy, an escalation management system has been developed. In the short run the objective was to create automational and informational business value. Beyond these initial benefits the system should generate transformational business value in the future.

This paper discusses how to design integrated escalation management systems facilitating transformational business value creation. In particular, we illustrate how to integrate analytics-enabled decision support and highlight corresponding use-cases.

Our research starts with a review of the current escalation process. We consider a production process spanning the process steps supply of parts, component production, assembly, testing and shipment, each with respective work stations. At each workplace disruptions (e.g., component damage or missing materials) occur frequently. Responders are dispatched to resolve these disruptions. However, dispatched responders are oftentimes not close by, may lack the necessary skills to resolve a given disruption or may be unavailable. This means that the current disruption handling process (Fig. 2) requires workers to search for an appropriate and available responder. After a disruption is communicated and additional information, e.g., on a certain machine or product is obtained, the responder assists in solving the disruption. This evolved process has some obvious shortcomings:

  • Searching for an appropriate and available response person is time-consuming. Most time is taken by the search itself, especially if the first contacted response person cannot assist in solving the disruption.

  • Interruptions of colleagues are common during the status quo process, as possible response persons are interrupted throughout the search process.4

  • The disruption is only communicated once a response person is found, which adds additional time in which the response person has to think about a possible solution before being able to assist the worker in problem solving.

  • The lack of disruption information results in the necessity to interact with organizational databases to find additional information on a machine, workplace or product.

The shortcomings highlight significant improvement potential. Adopting a resource-based view to identify supply chain productivity potentials (Chae et al., 2014), we seek to improve the employees (e.g., worker or response person) utilization through an IT system. To achieve this, the new system must automate communication between a worker and the appropriate response person. In order to notify an appropriate and available person, we need to know what is happening and if the worker has the skills to assist in solving (appropriate) as well as how long the disruption's solution will likely take, to ensure availability and avoid overbooking of responders. To this end, the triggered disruption type has to be classified, the duration predicted, and a response person dispatched. Using advanced machine learning models we establish a data-driven decision support system which assists disruption handling during the production of highly customized products. Data-driven decision support facilitates business value creation through advanced analytics (Davenport and Harris, 2017, Brynjolfsson et al, 2011). Yet, in operational processes such systems are underrepresented in research. Our research sheds light on how analytics-enabled I4.0 applications generate business value along three dimensions—automational, informational, and transformational. In the context of escalation management, we contribute methodologically by integrating suitable analytics approaches.

Section snippets

Related work

We first provide an overview of current information systems research with focus on analytics-enabled business value facilitation. Taking into account the production environment, its digitalization, and possible use-cases of analytics-enabled IT systems, we review recent advances in industrial internet applications and advanced analytics. Subsequently, we highlight recent advances in disruption handling with a special focus on escalation management systems.

Advanced escalation management

To address the shortcomings of the existing escalation process, we design, implement and evaluate an advanced escalation management system (EMS 4.0). Doing so we adopt an action design research process in collaboration with the industrial partner.

Integrated analytics

To tap into the system's transformational benefits, an analytics foundation for responder scheduling is required. We follow Wuest et al. (2016) by relying on supervised machine learning algorithms to deploy the proposed data-driven decision support system. Instead of just collecting data and providing minimal information, we expand the system to include integrated analytics. In particular, we train classification models to predict the disruption type which caused a given escalation.

Evaluation

To evaluate the system, we analyze the collected escalation data of the first eight months after roll-out. We evaluate the escalation handling process as well as the process improvement potential, where we highlight the special importance of additional data sources.

Conclusions and implications

Our research explores the potentials of an I4.0 enabled escalation management system in the traditional manufacturing industry. Our study sheds light on how analytics-enabled industrial applications help create business value. To explore the interplay between analytics and IT business value we relate descriptive, predictive and prescriptive analytics to automational, informational and transformational IT business value. We posit the central importance of analytics for the generation of

Authors’ contributions

Conception and design of study: C.M. Flath, F. Oberdorf, N. Stein.

Acquisition of data: F. Oberdorf.

Analysis and/or interpretation of data: C.M. Flath, F. Oberdorf, N. Stein.

Drafting the manuscript: C.M. Flath, F. Oberdorf, N. Stein.

Revising the manuscript critically for important intellectual content: C.M. Flath, F. Oberdorf, N. Stein.

Approval of the version of the manuscript to be published (the names of all authors must be listed): C.M. Flath, F. Oberdorf, N. Stein.

Funding

None declared.

Conflict of interest

The authors declare no conflict of interest.

Declaration of Competing Interest

The authors report no declarations of interest.

References (70)

  • L. Monostori et al.

    Cyber-physical systems in manufacturing

    CIRP Ann.

    (2016)
  • R. Müller et al.

    Lean information and communication tool to connect shop and top floor in small and medium-sized enterprises

    Proc. Manuf.

    (2017)
  • O. Penas et al.

    Multi-scale approach from mechatronic to cyber-physical systems for the design of manufacturing systems

    Comput. Ind.

    (2017)
  • J. Qin et al.

    A categorical framework of manufacturing for industry 4.0 and beyond

    Proc. CIRP

    (2016)
  • S. Răileanu et al.

    A cloud-based manufacturing control system with data integration from multiple autonomous agents

    Comput. Ind.

    (2018)
  • R. Sanchez-Marquez et al.

    Diagnosis of quality management systems using data analytics-a case study in the manufacturing sector

    Comput. Ind.

    (2020)
  • L. Wang et al.

    Current status and advancement of cyber-physical systems in manufacturing

    J. Manuf. Syst.

    (2015)
  • S. Wang et al.

    Harnessing business analytics value through organizational absorptive capacity

    Inf. Manag.

    (2019)
  • B. Zhang et al.

    Bearing performance degradation assessment using long short-term memory recurrent network

    Comput. Ind.

    (2019)
  • V. Alcácer et al.

    Scanning the industry 4.0: a literature review on technologies for manufacturing systems

    Eng. Sci. Technol. Int. J.

    (2019)
  • G. Almeida Marodin et al.

    Managing barriers to lean production implementation: context matters

    Int. J. Prod. Res.

    (2015)
  • A. Barua et al.

    An empirical investigation of net-enabled business value

    MIS Q.

    (2004)
  • A. Barua et al.

    Information technologies and business value: an analytic and empirical investigation

    Inf. Syst. Res.

    (1995)
  • K.H. Brodersen et al.

    The balanced accuracy and its posterior distribution

  • J. Brownlee

    Better Deep Learning: Train Faster, Reduce Overfitting, and Make Better Predictions Machine Learning Mastery

    (2018)
  • E. Brynjolfsson et al.

    Strength in Numbers: How Does Data-Driven Decisionmaking Affect Firm Performance?

    (2011)
  • J.D. Camm et al.

    Business Analytics

    (2020)
  • B. Chae et al.

    The impact of supply chain analytics on operational performance: a resource-based view

    Int. J. Prod. Res.

    (2014)
  • H. Chen et al.

    Business intelligence and analytics: from big data to big impact

    MIS Q. Manag. Inf. Syst.

    (2012)
  • H. Chen et al.

    Business intelligence and analytics: from big data to big impact

    MIS Q.

    (2012)
  • A.M. Da Silva et al.

    The Andon system: designing a CSCW environment in a lean organization

    Proceedings – 6th International Workshop on Groupware, CRIWG, 2000

    (2000)
  • M. Daneshvar Kakhki et al.

    Information systems for supply chain management: a systematic literature analysis

    Int. J. Prod. Res.

    (2019)
  • T. Davenport et al.

    Competing on Analytics: Updated, With a New Introduction: The New Science of Winning

    (2017)
  • M. Ghobakhloo et al.

    Corporate survival in Industry 4.0 era: the enabling role of lean-digitized manufacturing

    J. Manuf. Technol. Manag.

    (2020)
  • A. Gilchrist

    Industry 4.0: The Industrial Internet of Things

    (2016)
  • Cited by (0)

    View full text