Throughput bottleneck detection in manufacturing: a systematic review of the literature on methods and operationalization modes

Figures & data

Figure 1. Illustration of throughput bottlenecks in a serial production line.

Figure 2. Systematic sampling process.

Table 1. Distribution of publications across journals/conferences.

Journal/Conference Name	Count
Proceeding of the Winter Simulation Conference	5
International Journal of Production Research	4
Journal of Manufacturing Systems	2
Computers and Industrial Engineering	2
Production and Operations Management	2
Journal of Manufacturing Systems and Engineering	2
Other journals (8 different journals)	8
Other conferences (2 different conferences)	2

Figure 3. Process states of the station during a production run (adapted from (Roser et al., 2001)).

Table 2. Summary of different bottleneck detection methods identified from the literature review.

Focus	Name	Description	Capability
Queue State	Queue Length Method (Lawrence & Buss, 1994)	The station whose queue has the largest number of jobs is the bottleneck	Momentary bottleneck, average bottleneck
	Waiting Time Method (Roser et al., 2001)	The station for which jobs in the queue have the longest waiting times is the bottleneck	Average bottleneck
Process state	Utilization Method (Hopp et al., 2007)	The station with the highest utilization is the bottleneck	Average bottleneck
	Active Period Percentage Method (Roser et al., 2001)	The station with the highest active state time duration in a time period is the bottleneck	Average bottleneck
	Average Active Period Method (Roser et al., 2001)	The station with the longest average active duration is the bottleneck	Average bottleneck
	Shifting Bottleneck Detection Method (Roser et al., 2002a)	The station being the longest time momentary of shifting bottleneck during a time period is the bottleneck	Average bottleneck
	Longest Current Active Period Method (Roser et al., 2002a, Subramaniyan et al., 2016)	At any given time, the process with the longest active duration is the bottleneck	Momentary bottleneck
	Inactive Period Method (Sengupta et al., 2008)	The station with the minimum sum of the blocked state time and the starved state time is the bottleneck.	Average bottleneck
	Inter-departure Time Variance Method (Betterton & Silver, 2012)	The station with the smallest work-in-process inter-departure time variance is the bottleneck	Average bottleneck
	Arrow Method (Kuo et al., 1996)	Uses blocking and starvation information to indicate whether the bottleneck is upstream or downstream	Average bottleneck
	Turning Point Method (L. Li et al., 2009) (Li, 2009)	In addition to the condition that a station’s blockage and starvation are smaller than its neighboring stations, the bottleneck station should be a ‘turning point’.	Average bottleneck
Queue and Station (or System) State	Bottleneck Walk Method (Roser et al., 2015)	Uses blocking and starvation information together with information on buffer inventory levels to indicate whether the bottleneck is upstream or downstream	Momentary bottleneck, average bottleneck
	Bottleneck Index Method (Huang et al., 2019)	Uses a composite measure of utilization and the number of jobs in the queue to identify the bottleneck.	Average bottleneck
	Sensitivity of the System Method (Chang et al., 2007)	The station with the largest ratio of change in the system state is the bottleneck	Average bottleneck

Table 3. Throughput bottleneck detection methods and operationalization modes.

Focus	Method	Operationalization modes
		Gemba Walk	DES	Data science
Queue State	Queue Length		(Lima et al., 2008)
	Waiting Time		(Lima et al., 2008)
Process state	Utilization		(Lima et al., 2008)	Cao et al., 2012) (Yu & Matta, 2016) (Subramaniyan et al., 2020)
	Active Period Percentage		(Roser et al., 2001Lima et al., 2008)	(Subramaniyan, Skoogh, Salomonsson, Bangalore, Gopalakrishnan, et al., 2018) (Subramaniyan, Skoogh, Salomonsson, Bangalore, & Bokrantz, 2018) (Subramaniyan et al., 2019) (Yu & Matta, 2016) (Subramaniyan et al., 2020)
	Average Active Duration		(Roser et al., 2001)	(Subramaniyan, Skoogh, Gopalakrishnan, & Hanna, 2016) (Yu & Matta, 2016) (Subramaniyan et al., 2020)
	Longest Current Active Period		(Roser et al., 2002a)	(Subramaniyan, Skoogh, Gopalakrishnan, Salomonsson, et al., 2016) (Yu & Matta, 2016) (Subramaniyan et al., 2020
	Shifting Bottleneck Detection		(Roser et al., 2002a, Roser & Nakano, 2015)	(Subramaniyan, Skoogh, Gopalakrishnan, Salomonsson, et al., 2016) (Yu & Matta, 2016) (Subramaniyan et al.,2020)
	Inactive Period		(Sengupta et al., 2008)	(Yu & Matta, 2016, Subramaniyan et al., 2020)
	Inter-departure Time Variance		(Betterton & Silver, 2012)	(Yu & Matta, 2016, Subramaniyan et al., 2020)
	Arrow		(Kuo et al., 1996Chiang et al., 2000)	(Yu & Matta, 2016, Subramaniyan et al., 2020)
	Turning Point		(L. Li et al., 2009) (Li, 2009Li, 2018)	L. Li et al., 2011) (Lai et al., 2018) (Yu & Matta, 2016) (Subramaniyan et al., 2020
Queue and Station (or System) State	Bottleneck Walk	(Roser et al., 2015)
	Bottleneck Index			(Huang et al., 2019)
	Sensitivity of the System		(Chang et al., 2007)

Table 4. Operationalization modes: suitability, advantages and challenges.

Modes	When to use	Advantages	Challenges
Gemba Walk	Suitable for non-digitalized production system. Can be used to detect throughput bottlenecks when practitioners are present on the shop floor. Needs relatively stable production system. Can identify momentary and average bottlenecks.	Allows for directly observing the production system dynamics and identifying throughput bottlenecks. Facilitates quick decisions. Checks and aligns the practitioner’s perception on throughput bottlenecks with reality.	Considers limited stochastic events. Needs extensive manual efforts and is time consuming. Limited possibilities to detect shifting bottlenecks in real-time. Lack of predictive capabilities.
DES	Suitable for both non-digitalized and digitalized production systems. Production system data should be available (either in digital format or manual recorded observations). Suitable for back-office analysis of throughput bottlenecks. Suitable to identify average throughput bottlenecks.	Considers all types of stochastic events. Can support “what if” analysis, i.e. test interventions to resolve throughput bottlenecks before implementation.	Time consuming to build simulation models. Requires high input data quality. Need of extensive manual efforts and expertise (e.g. programming) to build and verify simulation models. Difficult to keep the model updated in dynamic contexts.
Data science	Suitable for digitalized production systems. Digital production system data should be available (e.g. event log data). Back-office analysis of the throughput bottlenecks. Suitable to identify momentary, average and shifting bottlenecks.	Considers all stochastic events (even disturbances in seconds time scale). Can identify the throughput bottlenecks in real-time. Enables quick decisions. Can predict future throughput bottlenecks. Has the capability to automatically learn about changing production system dynamics (i.e. without much manual efforts compared to DES).	Requires high input data quality. There may be a bias in the input data. Limited capability to test interventions before implementation in the real-world.

Roser, C., Nakano, M., & Tanaka, M. (2001). A practical bottleneck detection method. In B. A. Peters, J. S. Smith, D. J. Medeiros, & M. W. Rohrer (Eds.), Proceedings of the 2001 Winter Simulation Conference (pp. 949–953). IEEE. https://doi.org/10.1109/WSC.2001.977398

 Google Scholar

Lawrence, R. S., & Buss, H. A. (1994). Shifting production bottlenecks: Causes, cures and conundrums. Production and Operations Management, 3(1), 21–37. https://doi.org/10.1111/j.1937-5956.1994.tb00107.x

 Google Scholar

Hopp, W. J., Iravani, S. M. R., & Shou, B. (2007). A diagnostic tree for improving production line performance. Production and Operations Management, 16(1), 77–92. https://doi.org/10.1111/j.1937-5956.2007.tb00167.x

 Web of Science ®Google Scholar

Roser, C., Nakano, M., & Tanaka, M. (2002a). Shifting bottleneck detection. In E. Yucesan, C.-H. Chen, J. L. Snowdon, & J. M. Charnes (Eds.), Proceedings of the 2002 Winter Simulation Conference (Vol. 2). https://doi.org/10.1109/WSC.2002.1166360

 Google Scholar

Subramaniyan, M., Skoogh, A., Gopalakrishnan, M., Salomonsson, H., Hanna, A., & Lämkull, D. (2016). An algorithm for data-driven shifting bottleneck detection. Cogent Engineering, 3(1), 1–19. https://doi.org/10.1080/23311916.2016.1239516

 Google Scholar

Sengupta, S., Das, K., & VanTil, R. P. (2008). A new method for bottleneck detection. In S. J. Mason, R. R. Hill, L. Monch, O. Rose, T. Jefferson, & J. W. Fowler (Eds.), Proceedings of the 2008 Winter Simulation Conference, Miami, Florida, US, (pp. 1259–1267).

 Google Scholar

Betterton, C. E., & Silver, S. J. (2012). Detecting bottlenecks in serial production lines – a focus on interdeparture time variance. International Journal of Production Research, 50(15), 4158–26. https://doi.org/10.1080/00207543.2011.596847

 Web of Science ®Google Scholar

Kuo, C.-T., Lim, J.-T., & Meerkov, S. M. (1996). Bottlenecks in serial production lines: A system-theoretic approach. Mathematical Problems in Engineering, 2(3), 233–276. https://doi.org/10.1155/S1024123X96000348

 Web of Science ®Google Scholar

Li, J., Blumenfeld, D. E., Huang, N., & Alden, J. M. (2009). Throughput analysis of production systems: Recent advances and future topics. International Journal of Production Research, 47(14), 3823–3851. https://doi.org/10.1080/00207540701829752

 Web of Science ®Google Scholar

Li, L. (2009). Bottleneck detection of complex manufacturing systems using a data-driven method. International Journal of Production Research, 47(24), 6929–6940. https://doi.org/10.1080/00207540802427894

 Web of Science ®Google Scholar

Roser, C., Lorentzen, K., & Deuse, J. (2015). Reliable shop floor bottleneck detection for flow lines through process and inventory observations: The bottleneck walk. Logistics Research, 8(1), 1–9. https://doi.org/10.1007/s12159-015-0127-2

 Google Scholar

Huang, B., Wang, W., Ren, S., Zhong, R. Y., & Jiang, J. (2019). A proactive task dispatching method based on future bottleneck prediction for the smart factory. International Journal of Computer Integrated Manufacturing, 32(3), 278–293. https://doi.org/10.1080/0951192X.2019.1571241

 Web of Science ®Google Scholar

Chang, Q., Ni, J., Bandyopadhyay, P., Biller, S., & Xiao, G. (2007). Supervisory factory control based on real-time production feedback. Journal of Manufacturing Science and Engineering, 129(3), 653. https://doi.org/10.1115/1.2673666

 Web of Science ®Google Scholar

Lima, E., Chwif, L., & Barreto, M. R. P. (2008). Metodology for selecting the best suitable bottleneck detection method. In S. J. Mason, R. Hill, L. Mönch, O. Rose, T. Jefferson, & J. W. Fowler (Eds.), Proceedings of the 2008 Winter Simulation Conference, Miami, Florida, US, (pp. 1746–1751).

 Google Scholar

Cao, Z., Deng, J., Liu, M., & Wang, Y. (2012). Bottleneck prediction method based on improved adaptive network-based fuzzy inference system (ANFIS) in semiconductor manufacturing system. Chinese Journal of Chemical Engineering, 20(6), 1081–1088. https://doi.org/10.1016/S1004-9541(12)60590-4

 Web of Science ®Google Scholar

Yu, C., & Matta, A. (2016). Data-driven bottleneck detection in manufacturing systems: A statistical approach. International Journal of Production Research, 54(21), 6317–6322. https://doi.org/10.1080/00207543.2015.1126681

 Web of Science ®Google Scholar

Subramaniyan, M., Skoogh, A., Muhammad, A. S., Bokrantz, J., Johansson, B., & Roser, C. (2020). A generic hierarchical clustering approach for detecting bottlenecks in manufacturing. Journal of Manufacturing Systems, 55, 143–158. https://doi.org/10.1016/j.jmsy.2020.02.011

 Web of Science ®Google Scholar

Subramaniyan, M., Skoogh, A., Salomonsson, H., Bangalore, P., Gopalakrishnan, M., & Sheikh Muhammad, A. (2018). Data-driven algorithm for throughput bottleneck analysis of production systems. Production and Manufacturing Research, 6(1), 225–246. https://doi.org/10.1080/21693277.2018.1496491

 Web of Science ®Google Scholar

Subramaniyan, M., Skoogh, A., Sheikh Muhammad, A., Bokrantz, J., & Turanoğlu Bekar, E. (2019). A prognostic algorithm to prescribe improvement measures on throughput bottlenecks. Journal of Manufacturing Systems, 53, 271–281. https://doi.org/10.1016/j.jmsy.2019.07.004

 Web of Science ®Google Scholar

Subramaniyan, M., Skoogh, A., Gopalakrishnan, M., & Hanna, A. (2016). Real-time data-driven average active period method for bottleneck detection. International Journal of Design & Nature and Ecodynamics, 11(3), 428–437. https://doi.org/10.2495/DNE-V11-N3-428-437

 Google Scholar

Roser, C., & Nakano, M. (2015). A quantitative comparison of bottleneck detection methods in manufacturing systems with particular consideration for shifting bottlenecks. IFIP International Federation for Information Processing 2015, 1, 273–281. https://doi.org/10.1007/978-3-319-22759-7

 Google Scholar

Chiang, S.-Y., Kuo, C.-T., & Meerkov, S. M. (2000). DT-bottlenecks in serial production lines: Theory and application. IEEE Transactions on Robotics and Automation, 16(5), 567–580. https://doi.org/10.1109/70.880806

 Google Scholar

Li, L. (2018). A systematic-theoretic analysis of data-driven throughput bottleneck detection of production systems. Journal of Manufacturing Systems, 47, 43–52. https://doi.org/10.1016/j.jmsy.2018.03.001

 Web of Science ®Google Scholar

Li, L., Qing, C., Xiao, G., & Ambani, S. (2011). Throughput bottleneck prediction of manufacturing systems using time series analysis. Journal of Manufacturing Science and Engineering, 133(2), 1–8. https://doi.org/10.1115/1.4003786

 Web of Science ®Google Scholar