Bayer Material Science LLC (BMS) encountered unacceptable levels of dead and slow moving inventory (DSMI) across many of its business units. Furthermore, BMS's compounding business wanted to significantly improve its response time to customers to achieve competitive advantage. The project was commissioned to address dead and slow moving inventory and additional issues such as write downs, inventory space utilisation, selling partial containers, planning & scheduling issues and more. With that baseline established, BMS's Business Excellence Group assigned an experienced Six Sigma team to test several hypotheses and tackle these related problems.
Bayer Material Science LLC, who represents one portion of the worldwide Bayer Material Science business portfolio, is among the world's largest polymer companies. Business activities are focused on the manufacturing of high-tech polymer materials and the development of innovative solutions to problems important to its customers. Bayer Material Science LLC has 30 production sites around the globe.
SherTrack, a leading software-as-a-service provider of Predictive Manufacturing solutions, working with Six Sigma teams, continues to provide the expertise, tools, and proven methodologies to execute complex operating process improvement initiatives in challenging, real-life environments to achieve significant performance improvements.
Rick Baxendell, a BMS Black belt and Bayer Material Science LLC senior Six Sigma manager, realised that BMS's manufacturing facilities with their complex interaction of constraints and operating processes would be very difficult to analyze using traditional approaches. Running trials in live operations to evaluate changes is prohibitively expensive, time consuming and poses unacceptable risks to operations.
The Six Sigma team was chartered to leverage SherTrack's innovative predictive manufacturing and digital modeling technology in order to develop new operating policies and processes to:
The idea to leverage digital modelling and discrete-event simulation with structured design-of-experiment (DOE) methodologies came mid-project, during the Analyze phase of the DMAIC (Define, Measure, Analyze, Improve & Control) project. This idea created a more powerful evaluation tool and allowed many diverse and complex manufacturing hypotheses to be tested in a "safe" modelling environment. The DOE enabled the selection of the optimal policies and set points to maximise business performance.
Multivariate digital models are recognised as the methodology of choice for analyzing complex, non-linear systems. The Analyze and Improve phases of DMAIC are particularly challenged by the multivariate and non-linear relationships in plants with multiple production lines and tens or hundreds of product items, vying for limited capacity. SherTrack's digital models and discrete-event simulation provided sophisticated analytical tools for the BMS Six Sigma team during the DMAIC Analyze and Improve phases to effectively qualify and quantify the impacts of changing set points, policies and procedures in a complex manufacturing environment (See figure 1).
Figure 1
BMS Subject Matter Experts (SMEs) teamed with the SherTrack Six Sigma Services group to build a realistic digital model of the complete order-to-fulfillment process which included sales demand, scheduling and production along with inventory outcomes. The combined team built a working model of the complex compounding facility using historical demand streams and forecasts, schedule and unscheduled downtime, planned and accounted for production transitions, production rates, simulated random Quality Assurance (QA) failures and strict adherence to Make-to-Order (MTO), and Make-to-Forecast (MTF), policies. The SNAPPS* digital model & simulator produced daily operating records that were virtually identical to those generated by the Enterprise Resource Planning (ERP) and Manufacturing Execution System (MES) systems in actual operations. This model validation method is called Post Analysis. For each new scenario/hypothesis, specific performance metrics were developed for each Key Performance Indicator (KPI).
With this capability in hand, the DOE was constructed to extract statistically significant cause and effect relationships between selected discretionary policies and the resulting business performance. The team also conducted a regression-like analysis, early in the process, and was able to assess the strength of the correlation of each factor and combination of factors on the output variables. Thus, using DOE, the team not only assessed the strength of the correlation between "X" variables, such as forecast accuracy and service levels, but was also able to examine the effects of simultaneous changes in other "X" variables such as lead times, production lot sizes and policies of the manufacturing facility. A parametric model was then developed to enable further study. A parametric model is a set of related mathematical equations in which alternative scenarios are defined by changing the assumed values of a set of fixed coefficients (parameters). Parametric statistical methods are mathematical procedures for statistical hypothesis testing which assume that the distributions of the variables being assessed belong to known parametrized families of probability distributions. The cross-functional team of BMS SMEs, Black Belts and SherTrack Services consultants leveraged the strengths of the Six Sigma DMAIC process, digital modelling & discrete-event simulation, multiple-regression analysis and DOE techniques to test the interaction of over 44 different manufacturing hypotheses without inhibiting actual production (See Figure 2).
Figure 2
In the Analyze phase, the team:
In this DMAIC Continuous Improvement project, the combination of DOE structured inputs for SherTrack's business process model demonstrated its value to the business with rapid, quantitative scenario simulation results. Hypotheses were tested and complex cause and effect relationships between project inputs (I), business and production processes (X) and expected outcomes (Y) were explored.
“This methodology provided the BMS project team with a robust decision making tool. The Project team worked with the Business Leaders to find conditions that would improve service, delivery and other criteria. This methodology provided a basis for our supply chain improvements,” said Baxendell.
Lessons learned from this novel approach to evaluate manufacturing responses includes
Figure 3
Figure 4
Figure 5
BMS has already made incremental changes to one supply chain and is evaluating further changes. Real world changes involve time and put money and customer relations at risk. This DOE / Business Process Modeling approach, which took only 10 weeks to complete, coupled with the structured Six Sigma DMAIC process and enabled by SherTrack's digital modelling and discrete-event simulation, has provided a good risk mitigation tool that allows BMS the opportunity to test complex manufacturing hypotheses within a very "safe" environment. This benefits both BMS and its customers because the analysis executed within this statistically-oriented and very structured environment provided the team with ample cause and effect impacts on the plant, BMS and its customers prior to committing funds, resources and time. It also minimizes the risk to BMS customers and operations.
Richard Baxendell
