Lean Six-Sigma has brought great changes in the industrial world. The idea of removing wastes and defects using statistical data analysis and embedding the culture of continuous improvement as an organizational framework has become typical practice in the improvement process of manufacturing industries, health care and other service systems. Nevertheless, most Lean Six-Sigma projects still fail in different stages of implementation of the DMAIC steps, and the reasons of failure come in many different flavors, but in this project, Multi-Method Simulation was used as an instrument to mitigate the risks of failure. Simulation modeling is a tool that computationally replicates the manufacture and supply chain operations while forecasting the statistical effect of operational changes performed in a virtual or simulated environment as a quick and cheap way to analyze “what-if” scenarios. The integration of Multi-Method simulations (Agent based, Discrete events, Systems Dynamic) is necessary to establish a robust quality framework where potential root causes can be tested to increase production throughput, improve processes, deal with uncertainties, and reduce manufacturing wastes. Here I present a business case study of a manufacturing facility to test different production parameters of a LED factory as a way to improve the processes by using simulations within the Six-Sigma methodology. The following image shows the process of this factory.
Having the model validated as a close representation of reality, the improvement step focuses on finding ways in which the process can be improved by changing certain parameters, adding strategic policies or finding a solution for the problems that were found in the analysis phase. A large queue indicates a potential bottleneck and an opportunity to increase production rate, without the need of increasing human resources, and according to the bottleneck information, the biggest problem occurs in the process of combining Reflectors with Heat Sinks, which has a high accumulation of raw products and a high utilization of resources. Hence, a solution for that problem must be found and tackled before having a fresh view over the oven and PCB clamping processes that are also candidates with a high product accumulation and a high utilization respectively.
Before coming up with a solution, there are two important points that must be understood: the first point is that the employees of this factory have every day a station assigned to them where they will work for the full day without changing their position, the second point is that the arrival of PCBs, Heat Sinks and Reflectors is not scheduled in advance and no action is taken when raw products arrive to the factory. These two points are important because they highlight the fact that there is no communication between operations (employees who handle the manufacturing dynamics) and deliveries (employees who request new raw products according to inventory levels and demand).
The potential solution and hypothesis to test is very simple. To avoid future bottlenecks, whenever there is inventory of Heat Sinks, PCBs or Reflectors, increased focus should be placed in clamping and combining with the Reflector in order to reduce inventory waste and have as many heat sinks assembled and ready for the potting process as possible. To do this in the simulation, a variable is used to increase the capacity of combining reflectors and clamping PCBs, while reducing the capacity of the potting process. This means that a resource that is usually assigned to the potting process will be moved to the clamping and combining processes when there are raw products available in the inventory. This is not performed as an emergency situation, but instead it’s a strategy to avoid bottlenecks that are predicted to happen in the future. Testing this in the real world would take weeks, but with the help of simulations, the hypothesis can be tested in just a few seconds. It is expected for the production to increase when running this simulation again with this small change, which translates into reducing cost per unit on the short term, a solution that has no financial costs to the factory. The implementation of this policy would require the deliveries department to communicate in advance the time of delivery to the operational manager in order to plan the distribution of resources ahead of time. The reduction of inventory wastes is one of the basic principles of Lean Manufacturing.
The solution requires moving human resources that normally work in potting into clamping and combining when deliveries arrive. This means that the redistribution of the resources available in the factory, in conjunction with a correct coordination between deliveries and operational managers to adjust roles in the factory with the objective of having the clamping and reflector combining processes flowing as fast as possible when deliveries are made, can result in a huge improvement in production numbers and cost. The reduction in cost is directly associated with manufacturing efficiency and lower human resource cost per unit produced. With this solution, production was increased from 126 to 170 products per week in average. Also, cost per unit was reduced from $147 to $135.
The process of building simulations demands a deeper understanding of the system, so additional valuable information can be found and reported to upper management. For instance, in terms of defects, the main root causes that are responsible for the defects were identified. The following figure shows a fishbone diagram explaining the main root causes of defects. It was built with the team of manufacturing specialists during the analyze phase.
The Fishbone Diagram shows 4 areas of interest. First, the clamping stage has two problems that can cause defects. One is the presence of bubbles in the RTV and the other is the occurrence of delamination. The delamination itself has two causes, which are problems with the ejectors and with the silicone not adhering to the RTV. The Heat Sinks cause defects because of warping, the Reflectors cause defects because of their appearance and the removal of the metalized layer. Finally the process itself has three potential reasons for defects: an inefficient layout, a process that is too complicated and a high employee turnover resulting in no retention of knowledge among the staff.
The application of Multi-Method simulations with Six Sigma’s DMAIC methodology can enable operators to develop comprehensive models of their systems and engage in improvement measures based on the results that they record in the simulated environments. In this study, the intent was to implement Six Sigma’s DMAIC methodology along with a Multi-Method simulation to improve the operations of a components manufacturing factory. In the process, case-specific variables such as oven loading times, storage and delivery times, and delays will require the use of AnyLogic’s capacity to simulate discrete events, system agents, and the dynamics of these components’ interactions regarding their influence on the factory’s operations. In the process, we can identify the strategic changes that can result in optimum efficiency for the production-facing aspects of the business as well as the resource utilization necessary for information and materials to flow across the supply chain.
- Industry: Manufacturing
- Model: Discrete-Events, Agent-Based, System Dynamics
- Duration: 6 months