All my publications focus on transforming manufacturing operations using data, IIoT, AI and insights


  1. Artificial intelligence for throughput bottleneck analysis – State-of-the-art and future directions

Mukund Subramaniyan, Anders Skoogh, Jon Bokrantz, Muhammad Azam Sheikh, Matthias Thurer, Qing Chang

Journal of Manufacturing Systems, 60, 743-751

Publication Year: 2021

Many engineers spend a significant amount of time searching for bottlenecks on the shop floor. We can now change this situation using digital solutions and artificial intelligence (AI). By analyzing the digital data of the factory, we can quickly identify and predict the bottlenecks. It can also prescribe the necessary elimination actions. That means not only a smoother, faster, and more efficient production but also creating a more predictable, better factory environment without surprises. Curious to know how AI can help eliminate throughput bottlenecks in your factory? Check out the article.

We present the AI architecture, explain with examples how engineers can consume the AI insights, and use them for decision-making in day-to-day operations. We also propose several future research directions and a range of practical recommendations for manufacturers wanting to implement AI in their factories.

  1. A data-driven approach to diagnosing throughput bottlenecks from a maintenance perspective

Mukund Subramaniyan, Anders Skoogh, Muhammad Azam Sheikh, Jon Bokrantz, Björn Johansson, Christoph Roser

Computers & Industrial Engineering, 106851

Publication Year: 2020

Augmenting maintenance engineers’ judgment with data-driven algorithmic recommendations produces high-quality, impactful decisions to increase shop floor productivity. For this to happen, data scientists' and maintenance engineers’ skills need to be pooled when developing data-driven algorithms. Want to know how this can be realized in practice? Then read the article. The article shows how maintenance engineers and data scientists can co-work to develop data-driven algorithms to diagnose bottlenecks in manufacturing systems. We collaborated with a Swedish auto manufacturer to prove the criticality of domain expertise in developing data-driven algorithms.

  1. Data-driven machine criticality assessment–maintenance decision support for increased productivity

Maheshwaran Gopalakrishnan, Mukund Subramaniyan, Anders Skoogh

Production Planning & Control

Publication Year: 2020

The paper presents a data-driven machine criticality tool. The impact is that it will transform the current maintenance practices leading to increased productivity. The tool will help maintenance engineers to quickly identify the very critical machines on the shop-floor.

  1. A generic hierarchical clustering approach for detecting bottlenecks in manufacturing

Mukund Subramaniyan, Anders Skoogh, Muhammad Azam Sheikh, Jon Bokrantz, Björn Johansson, Christoph Roser

Journal of Manufacturing Systems, 55, 143-158.

Publication Year: 2020

Can machine learning (ML) algorithms use to analyze the stock prices of companies be used to analyze bottlenecks in manufacturing? Instead of finding companies that show a unique stock price over time compared to other stocks, the target is to identify machines that exhibit unique behavior as compared to other machines on the shop floor. The unique behavior of those machines is a key indicator that they are bottlenecks. Curious to know how? Then read the article! We explain the methodology based on unsupervised ML techniques, demonstrate it on two real production systems and explain how the algorithmic insights can be consumed by engineers to augment their decisions on bottlenecks. Identifying the right bottlenecks help engineers make correct and more confident decisions to improve shop-floor productivity!

  1. A prognostic algorithm to prescribe improvement measures on throughput bottlenecks

Mukund Subramaniyan, Anders Skoogh, Muhammad Azam Sheikh, Jon Bokrantz, Ebru Turanoğlu Bekar

Journal of Manufacturing Systems, 53, 271-281

Publication Year: 2019

The production and maintenance teams have many difficult decisions to make regarding bottleneck management on the shop floor. Having advance notice of a bottleneck’s location and possible root causes can help them make better-informed and more confident decisions. Want to know how? The read our article. The article exploring how can shop floor data be used for prescribing actions towards bottleneck management using machine learning techniques.

  1. A data-driven algorithm to predict throughput bottlenecks in a production system based on the active period of the machines

Mukund Subramaniyan, Anders Skoogh, Hans Salomonsson, Pramod Bangalore, Jon Bokrantz

Computers & Industrial Engineering, 125, 533-544

Publication Year: 2018

In this paper, we propose a prognostic algorithm based on machine learning techniques that can predict future dynamics of the production system based on the historical data and also predict the future throughput bottleneck locations. The insights from the algorithm can be used to forewarn manufacturing practitioners about the upcoming dynamics of the production system and possible throughput bottleneck locations.

  1. Data-driven algorithms for throughput bottleneck analysis of production systems

Mukund Subramaniyan, Anders Skoogh, Hans Salomonsson, Pramod Bangalore, Maheshwaran Gopalakrishnan, Muhammad Azam Shiekh

Production and Manufacturing Research, 6(1), 225-246

Publication Year: 2017

In this paper, we propose a long-term throughput bottleneck identification algorithm based on statistical methods. Manufacturing practitioners can use this algorithm to identify a set of long term throughput bottlenecks and plan for long term improvement activities on them to get sustainable throughput over time.

  1. An algorithm for data-driven shifting bottleneck detection

Mukund subramaniyan, Anders Skoogh, Maheshwaran Gopalakrishnan, Hans Salomonsson, Atieh Hanna, Dan Lämkull

Cogent Engineering, 3(1), 1239516

Publication Year: 2016

In this paper, we propose an algorithm that can identify the throughput bottlenecks in real-time. Practitioners can use this algorithm to continuously track throughput bottlenecks during a production shift and take actions in real-time to achieve the shift throughput target.

  1. Applications of Big Data analytics and Related Technologies in Maintenance — Literature-Based Research

Jens Baum, Christoph Laroque , Benjamin Oeser, Anders Skoogh, Mukund Subramaniyan

Machines, 6(4), 54

Publication Year: 2018

In this paper, we conduct a systematic literature review on the available data-analytics methods and technologies for effective maintenance decision making in shop floor.


  1. An Enhanced Data-Driven Algorithm for Shifting Bottleneck Detection

Christoph Roser, Mukund Subramaniyan, Anders Skoogh, Björn Johansson

IFIP International Conference on Advances in Production Management Systems, Vol. 630,

Publication Year: 2021

Bottleneck detection is vital for improving production capacity or reducing production time. Many different methods exist, although only a few of them can detect shifting bottlenecks. The active period method is based on the longest uninterrupted active time of a process, but the analytical algorithm is difficult to program requiring different self-iterating loops. Hence a simpler matrix-based algorithm was developed. This paper presents an improvement over the original algorithm with respect to accuracy.

  1. Adaptive Stream-based Shifting Bottleneck Detection in IoT-based computing architecture

Hannaneh Najdataei, Mukund Subramaniyan, Vincenzo Gulisano, Anders Skoogh, Marina Papatriantafilou

24th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA), pp. 993-1000. IEEE

Publication Year: 2019

We propose a processing framework, STRATUM, and an algorithm, AMBLE, for continuous, data stream processing for identifying throughput bottlenecks in production systems. STRATUM seamlessly distributes and parallelizes the processing across the tiers and AMBLE guarantees consistent analysis in spite of timing fluctuations, which are commonly introduced due to e.g. the communication system; it also achieves efficiency through appropriate data structures for in-memory processing.

  1. Stream-IT: Continuous and dynamic processing of production systems data-throughput bottlenecks as a case-study

Hannaneh Najdataei, Mukund Subramaniyan, Vincenzo Gulisano, Anders Skoogh, Marina Papatriantafilou

28th International Symposium on Industrial Electronics (ISIE),pp. 1328-1333. IEEE

Publication Year: 2019

In this paper, we propose an automated configurable streaming-based method for data validation to sanitize and organize manufacturing execution system (MES) data for throughput bottleneck detection. We also show insights on continuous, low-latency data processing as a methodology for throughput bottleneck analysis in production systems.

  1. Analysis of Critical Factors for Automatic Measurement of OEE

Richard Hedman, Mukund Subramaniyan, Peter Almström

Procedia CIRP, 57, 128-133

Publication Year: 2016

In this study, we identify critical factors and potential pitfalls when operating an automatic measurement of OEE. It is accomplished by analyzing raw data used for OEE calculation acquired from a large data set; 23 different companies and 884 machines.


  1. How do you select the shop floor problems for the AI solution? (Upcoming article)

Mukund Subramaniyan

Advectas AB (A part of Capgemini)

Publication Year: 2021

  1. AI on the shop floor: A game-changer to achieve economies of speed

Mukund Subramaniyan

Advectas AB (A part of Capgemini)

Publication Year: 2021

Shop floor engineers have many questions about AI. Will AI make a difference in shop floor management? What are the use cases? How to prioritize the use cases for AI implementation? Is the impact significant? How should the AI journey start in a factory? In my recent article, I have answered these questions using a combination of my academic research and real-world manufacturing experience. Take a look at the white paper!

  1. Balancing the role of domain experts and data scientists in an AI-powered factory

Mukund Subramaniyan

World Manufacturing Foundation (WMF) Report (Page 86)

Publication Year: 2020

Manufacturing companies have concerning questions on the role of shop floor and AI engineers in AI-powered factories. E.g., Will the manufacturing expertise of the shop floor engineers gained through years of education and experience be relevant in the future? Will AI experts replace the shop floor engineers? I have shared my perspectives in the white paper.

This essay was has been selected by the WMF editorial board as the best one on the topic “The ethical questions surrounding AI in manufacturing: What needs to be done?”.


  1. Data-driven throughput bottleneck analysis in production systems

Mukund Subramaniyan

Ph.D. (Doctoral) Thesis, Chalmers University of Technology

Publication Year: 2021

This thesis offers a series of data-driven solutions that can provide a multifold increase to a company's bottom line by eliminating throughput bottlenecks in the factory. Today manufacturing companies' factory floor productivity is alarmingly low at 50%. Practitioners are exploring new ways to increase factory floor productivity. One way to increase productivity is to get higher factory throughput. Some machines constraints the throughput on the factory floor. These machines are called throughput bottlenecks. When practitioners eliminate throughput bottlenecks, they can get higher throughput. But how can practitioners find, analyze, and eliminate throughput bottlenecks? Currently, practitioners spend a lot of time (sometimes hours or days) on the shop floor to search for bottlenecks and make ambiguous experience-based decisions. But this can be changed using digital solutions. How? My research answers this question. In this thesis, I build data-driven approaches to analyze throughput bottlenecks in less than seconds. The input to a data-driven approach is digital machine data. Then, the data-driven approach quickly analyses the digital data using artificial intelligence techniques. The outputs of a data-driven approach are insights on throughput bottlenecks.

Within this thesis, I propose four data-driven approaches for different types of throughput bottleneck analysis. First, I present different data-driven approaches to identify historical throughput bottlenecks. With these, practitioners can quickly identify the bottleneck location in a production system. Second, I propose a data-driven approach to diagnosing historical throughput bottlenecks. It will help to understand the possible root-causes of the throughput bottlenecks. Third, I offer a data-driven approach to predict throughput bottlenecks for the next production day. It will help to take proactive actions on throughput bottlenecks. Fourth, I propose a data-driven approach to prescribe actions on predicted throughput bottlenecks. It will give information on specific measures one can proactively perform on predicted throughput bottlenecks. In sum, these data-driven approaches will help practitioners to make faster, confident, and informed decisions on throughput bottlenecks, which will help to maximize the throughput from production systems.

Overall, data-driven approaches are similar to GPS. People use GPS to find the best way. The GPS eliminates blind alleys. Similarly, practitioners can use data-driven approaches to eliminate throughput bottlenecks and create a more predictable, safe, and better factory environment without surprises.

  1. Production data analytics - To identify productivity potentials

Mukund Subramaniyan

Master Thesis, Chalmers University of Technology

Publication Year: 2015

In this thesis, manufacturing system data from different real-world Swedish manufacturing companies were used to identify the productivity potentials. The analysis results show that the average OEE of Swedish manufacturing industries in 2015 was 50% approx. This indicates there is an enormous potential to increase the OEE of Swedish manufacturing industries. Also, this thesis shows concrete examples of how Swedish manufacturing companies can increase the OEE.