Featured Story September - October 2019
Towards integration of simulation and optimization for manufacturing problems

By Gregory Kasapidis, Phd Candidate.

Modern industries that adapt to the Industry 4.0 paradigm [1] tend to incorporate decision support systems into their production planning processes. Decision support systems comprise from a set of intelligent tools that help production engineers to understand and improve the performance of complicated production processes and the efficient utilization of resources. Two very important types of elements that can be included in decision support systems are optimization and simulation tools [2].

Automobile Manufacturing Environments

A typical example of a manufacturing industry that has already started to adapt to this paradigm is automobile production. Automobile production involves thousands of components that come from a large network of suppliers. Components are delivered using trucks from the suppliers to the plant. Usually, a make-to-order policy is followed, while safety stock levels for critical components are kept as low as possible. The shop-floor typically consists of a paced assembly line with many process steps, where multiple car models are sequenced for production. A car model enters the assembly line once all necessary components have arrived at the production floor, together with the main body that has exited the paint shop. Given a master production schedule, the material requirements are determined and the necessary replenishment orders for all components are issued. A pre-generated dock schedule provides details regarding the quantities, the arrival times and assigned dock station for each truck. To that end, a production plan is generated that describes the exact sequence of production orders per day for a specific planning horizon (typical several weeks of production). In practice, this detailed production plan is subject to change on a weekly basis due to the occurrence of unexpected events (e.g. material unavailabilities, machine break-downs and quality problems). In terms of optimization, such manufacturing environments can be modeled using the job-shop scheduling framework. This combinatorial optimization problem is one of the most well-known and hardest to solve problems since the 1950s [3, 4]. Its basic formulation can be extended to handle a wide range of extra features and constraints such as: stochastic processing times, setup times, sequence dependent setup times, resource constraints, unavailability windows and many more.

Real-time disruption events

In such environments, any disruption in the scheduled arrivals and availability of components may cause production delays with significant financial impact. Real-time monitoring of the supply chain and the production processes via cyber-physical systems (CPS) and complex event processing (CEP) tools can be used to generate events that may trigger an alert to handle this disruption. This is a case where the utilization of simulation and optimization tools can become critical. Simulation tools can evaluate the impact of events on various KPIs with a great amount of precision, while optimization tools can also evaluate and re-schedule the production plan accordingly to restore feasibility or to achieve the desired optimization objective. More specifically in re-scheduling cases, the optimization objective is usually hierarchical and can include other than the common metrics, the minimization of the number of changes from the base production plan.

Real-time Co-simulation Optimization

Real-time optimization and co-simulation frameworks [5] try to combine the best of the two worlds and also try to provide a much more realistic handling of day-to-day operations of manufacturing environments. In such a framework, simulation is used as the schedule evaluator that helps optimization take better decisions in terms of minimizing an objective function for example, but also to provide more specific details regarding the status of the production floor during the execution of the production plan. An example of such a use-case could be the maximization of the "robustness" of the solution in the aforementioned automobile manufacturing environment. Schedule robustness can be seen as time-buffer and it is calculated as the time difference between the start time of a production order and the last replenishment time of critical components that are prone to stock-outs. The idea behind this metric is to anticipate and absorb potential late deliveries. For the calculation of the scheduling robustness, the optimization is assisted by simulation in order to identify these critical components. In this optimization and co-simulation scheme whenever a new production schedule is generated, simulation is triggered to evaluate the schedule. Among other KPIs, simulation can identify the components that are more prone to cause production delays. On return, the optimization algorithm is using this information (as described above) so as to introduce time-buffers and improve the overall robustness of the schedule.


Decision support systems play a very important role on the day-to-day operations on modern manufacturing environments such as an automobile manufacturing facility. We introduce the notion of co-simulation optimization where in particular, simulation can be used in conjunction with optimization so as to provide a much lower level of detail on the status of the shop-floor, which enables a much more thorough optimization where new environment features and objectives can be considered.

This article is based on a presentation "Rescheduling and co-simulation of a multi-period multi-model assembly line with material availability restrictions" by Kasapidis G. A. et al. at the 9th Manufacturing Modeling Management and Control Conference (MIM) in Berlin (Aug. 2019).



Power Daniel J. (2008), Decision Support Systems: A Historical Overview, Daniel J.Power, Handbook on Decision Support Systems

Zhang J, Ding G, Zou Y, Qin S, Fu J (2019) Review of job shop scheduling research and its new perspectives under Industry 4.0. Journal of Intelligent Manufacturing 30(4):1809-1830

Lenstra J.K. (1992) Job Shop Scheduling, Combinatorial Optimization: 199-207

Osorio C, Selvam Krishna Kumar (2017) Simulation-Based Optimization: Achieving Computational Efficiency Through the Use of Multiple Simulators. Transportation Science 51(2): 395-411

M A N A G E M E N T    S C I E N C E    L A B O R A T O R Y   








Copyright 2013 Website Management Science Laboratory. All rights reserved.