– Excessive working capital
– Suboptimal lot sizes led to:
- Excessive cycle times due to large lot sizes for many SKUs
- Too frequent changeovers due to small lot sizes for some SKUs
- Capacity concerns due to high utilizations
Bayer Material Science’s Baytown facility was challenged to drastically lower working capital. Bayer initially tried using simple inventory analysis (include a link to the Inventory Advisor webpage), but quickly realized that this approach would fall short due to its lack of understanding of their critical capacity issues. One factor contributing to the high levels of required working capital were suboptimal lot sizes throughout the operation. Making this particularly challenging was the fact that some workcenter utilizations were at or near 100%- requiring excessive overtime to meet customer orders.
The team at Bayer realized that this excessive working capital could be controlled if they could take advantage of their knowledge of how lot sizes related to overall costs. As the familiar figure below shows, there are at least two dueling factors to consider when determining an optimal lot size:
1) A fixed component that increases linearly with the lot size. This is due to warehouse and logistic costs as well as holding costs. If only this component were considered, we would always try and have the smallest lot size possible to minimize overall costs.
2) An component that decreases exponentially with lot size. This component is due to yield loss, quality loss and unit reliability costs. If only this component were considered, we would always try and have the largest lot in order to minimize our overall costs.
In the simplified example above, the trick is to find the optimal point that balances both of these components and finds the lot size that minimizes our overall costs.
The challenge is that while these tradeoffs are easily understood, they are exceedingly difficult to actually sit down and calculate for a system of materials running through shared workcenters. In other words, like many manufacturing challenges, once a manufacturing process becomes ‘high mix’, simple techniques and approaches become significantly more challenging to implement. To be specific, the cost functions that describe the curves shown in the Figure above lack any reference to capacity, flow congestion, starvation, customer service level, or any of the needed factors required to make a fully informed planning decision. Clearly, a more sophisticated approach was needed to find the ‘sweet spot’ for Bayer’s lot sizes.
Step 1: Model the flow using Invistics’ Lot Sizer
The team turned to Invistics to assist with the calculation of these optimal lot sizes. Invistics’ ‘Lot Sizer’ uses a unique and patented approach to find this sweet spot between inventory, service, and changeover costs while taking the effect of increasing utilization into account.
Gather Required Data
The first step in the process is to identify the needed set of inputs for Lot Sizer. The diagram below provides a summary:
Using these inputs, Lot Sizer was able to model the flow of materials (including into and out of various storage tanks) through Bayer’s facility.
Step 2: Perform Analysis in Lot Sizer
One the data model was built, Bayer was able to analyze the results in Lot Sizer.
Figure 1: Capacity Analysis from Lot Sizer
The first recommendation highlighted by Lot Sizer was for optimal lot sizes for the Packaging Lines. As is often the case, the current lot sizes in SAP were created years ago using a ‘rule of thumb’ – the specifics of which were now unknown to the team. As the table below shows, Lot Sizer’s optimal values found that some SKUs currently had lot sizes that were smaller than ideal, while others had lot sizes that were too large. The result was an inefficient use of drum line capacity, which resulted in excessively high overall costs.
Figure 2: Partial list of output from Lot Sizer
In fact, even though the average lot size was recommended to decrease, as the figure below shows, there were so many examples of SKUS that had smaller than optimal lot sizes, that the small increase in inventory holding costs from these increased lot sizes was dramatically offset by the cost savings from the expected reduction in setup costs.
The figure below is a partial, but indicative representation of the recommendations from this project. As we expanded to the full product line, similar opportunities were seen for large inventory and overall cost reductions.
Step 3: Confirm Key Inputs as Needed Using Sensitivity Analysis
While most of the needed inputs were straightforward, at Bayer, some required a bit of additional analysis. In particular, the minimum allowable batch size (aka ‘lot size’) was an input that was up for debate. By running Lot Sizer multiple times, allowing varying ‘Minimum Batch Sizes’, the team was able to analyze the sensitivity of the results to the changing input (See Figure below). Doing so helped them realize that the range of their discussion (min batch size somewhere between 5,000 and 10,000 kg, with a best guess of 8,800) wasn’t particularly sensitive to the input, so it was determined that their best guess would work satisfactorily.
Step 4: Implement Recommendations
- At this point, the team decided to move ahead with the recommendations provided by Invistics’ Lot Sizer. Somewhere between 80-90% of the recommended lot sizes were implemented (The other 10% required some small tweaks by the team before putting into place). To implement the lot sizes, the Bayer team simply took the lot size values and entered them into SAP, which then used these improved values in its usual MRP and APO runs.
- For the additional tank capacity recommended during the what-if analysis, the team again decided to move ahead and implement the changes. In this case, this meant dedicating multiple tanks from the tank farm to certain products, rather than having all tanks free to cycle between all products. Here, the change didn’t end up costing anything as the team was able to accomplish the change via a simple management policy rather than through any piping or valving changes.
Within the next year, running with these improved lot sizes, the Baytown facility inventory levels decrease by $4.5 million without any sacrifice to customer service. These improvements were all realized by simply changing the lot size values in SAP to the values calculated by Lot Sizer.
In addition, overall costs were decreased by $500k annually. This improvement was due to a combination of the improved lot sizes and reconfiguring of the tank farm as recommended during the project.
At the time of this writing, Bayer is exploring the expansion of these techniques across their global supply chain.