(EN) Reducing Lead Time 2 – Fluctuations and Utilization

This second post in a series on how to reduce your lead time looks deeper at the effect of fluctuations and utilization. Improving these will reduce your inventory and hence, as per Little’s Law, reduce your lead time.

Reduce Fluctuations

Your inventory helps you to cover fluctuations. The next step would be to reduce fluctuations (in Japanese „mura„). This is also not easy. I recently did a multi-post series on how to reduce fluctuations in the source, make, and deliver side of your value chain. It is a Sisyphean task that never ends, since fluctuations always tend to increase unless they are actively reduced. The Kingman formula which I showed in the last post includes the fluctuations for the arrival and processing times, but this is a simplified model. In reality, you have multiple part types, multiple processes, and multiple queues. On top of that, the arrival of parts is often initiated by the departure, as in a pull system the departure of one part initiates reproduction. Hence, fluctuations in the demand pass through to fluctuations at the arrival.

Overall, it is hard to say how beneficial a reduction in fluctuations will be, but it will be beneficial. Hence, reducing fluctuations is an very important but underrated aspect of lean manufacturing.

One fluctuation that affects the lead time especially is prioritization. If you have a priority „VIP“ queue, the jobs in the VIP queue are prioritized and have a shorter lead time. This, however, comes at a cost of longer lead times of the non-VIP jobs. As long as you prioritize no more than 20%–30% of the work, the negative impact on the non-VIP jobs is negligible. However, the more jobs you prioritize, the worse the effect on the non-VIP jobs. Eventually, the non-VIP jobs may have a lead time approaching infinity.

With make-to-order, reduction in fluctuations help primarily with reducing lead time. With make-to-stock, the primary goal of reducing fluctuations is to reduce inventory, and a reduction in lead time is a nice side effect with additional benefits.

Reduce Utilization

Another major factor in the Kingman equation is the utilization. As the utilization approaches 100%, the waiting time approaches infinity. The utilization is – indirectly – another way to manage fluctuations. There are three ways to decouple fluctuations, inventory, capacity, and time. Having a utilization of less than 100% gives you a capacity buffer to decouple fluctuations. If you want to have high utilization, you need lots of inventory in the system to reduce the chances of running out of material. Lots of inventory – you guessed it – means a long lead time. Hence, 100% utilization is a really bad target for a production system!

For one, your production system is not balanced perfectly, and some processes are needed more than others. It is absolutely normal and okay for non-bottlenecks to have less than 100% utilization. Even bottlenecks have less than 100%, since the bottlenecks usually shift. It also depends on how you measure utilization. Is it the percentage of the time a machine is scheduled for production? Or is it the ratio of the actually produced good parts to the maximum number of parts that could theoretically have been produced in the same time? This is better known as the OEE. Aiming for a 100% OEE is unrealistic (unless the number is fudged, and I have seen plenty of OEE claimed to be above 100%). Just as a reference, Toyota aims for around 90% OEE.

In any case, a utilization of less than 100% gives you a capacity buffer to handle fluctuations. You could also use inventory to decouple fluctuations. However, a capacity buffer does not increase your lead time, whereas inventory does. On the downside, a capacity buffer with workers and machines waiting for parts costs you too. The question is a trade-off between unused capacity and additional inventory. Furthermore, this trade-off is not linear as shown in this graph based on the Kingman equation. The closer you get to 100% utilization, the more inventory you need to cover even the smallest fluctuations. For an utilization of 100% you need – in theory – an infinite inventory. Please note that the graph measures only the waiting time for a simplified system with one queue and one process, but similar behavior can also be found in more complex systems.

Hence, reducing utilization a bit below 100% will save a lot of inventory and hence a lot of lead time. A sweet spot is often around 80%–90% utilization, although it depends on the details of your system. Reducing the utilization further has much less impact on the inventory and hence the lead time. For example, if you reduce your utilization from 60% to 40%, the impact on the inventory and lead time is negligible, but now your workers stand around even more, which costs money.

The Kingman equation gives a nice relation between utilization and lead time. But again, the Kingman equation is a simplified model of the real world. Regarding utilization, it assumes a round-the-clock non-stop working system. Such systems exist in reality. However, most systems do not work twenty-four hours a day seven days a week. Instead, you may have one shift or two shifts four, five, or six days per week. This gives you another way to use capacity for handling fluctuations, namely with overtime. Overtime where you call people in when you need them is cheaper than having them waiting around if there is no work. On the downside, overtime requires a bit of preparation, and hence can cover medium- or longer-term fluctuations like seasonality. Overtime can also handle fluctuations whose short-term effects were buffered by inventory. If you see your buffer inventory going down, you may schedule overtime.

But keep in mind that even with overtime, your utilization will not be 100%. Overtime cannot cover short-term fluctuations unless you have inventory. Trying to reach 100% utilization will again blow up your inventory and lead time. In sum, it is okay to aim for getting the most out of your system. But keep in mind that it will not have a 100% OEE, and that is okay. Also keep in mind that if you run your system around the clock like a paper mill or similar, you need an additional capacity buffer (a lower utilization) if you want a short lead time or a good product availability.

Hence, reducing fluctuations in general and utilization to a reasonable level will reduce your inventory and hence your lead time. Now, go out, reduce your fluctuations, manage your utilization, control your lead time, and organize your industry!

Series Overview

• Reducing Lead Time 1 – Inventory
• Reducing Lead Time 2 – Fluctuations and Utilization
• Reducing Lead Time 3 – Throughput and Lot Size
• Reducing Lead Time 4 – Development

P.S.: This series of posts is based on an inspiration by Rajan Suri, and also chapter 7 of his book Quick Response Manufacturing and chapter 3 from his book It’s About Time.

• Suri, Rajan. It’s About Time: The Competitive Advantage of Quick Response Manufacturing. 1 edition. New York: Productivity Press, 2010. ISBN 978-1-4398-0595-4.
• Suri, Rajan. Quick Response Manufacturing: A Companywide Approach to Reducing Lead Times. Portland, Oregon, USA: Taylor & Francis Inc, 1998. ISBN 978-1-56327-201-1.