As I said in my last posting, in this posting we’ll discuss the concept of queuing and some basic laws of variability. Once again, I highly recommend Hopp and Spearman’s book, Factory Physics for a more detailed discussion of this subject.
A queuing system combines the impact of the arrival of parts from other processes and received parts from outside suppliers, the production of the parts, and the inventory or queue waiting to be processed. Hopp and Spearman go into much depth on this subject, and I suggest you read their work, but the important thing to remember is this. Since limiting interstation buffers is logically equivalent to installing a kanban, this property is a key reason that variability reduction (via production smoothing, improved layout and flow control, total preventive maintenance, and enhanced quality assurance) is critical to reducing variability, especially in the constraint.
In my last post, we identified and discussed a number of causes of variability and how they might cause congestion in a manufacturing system. We also said that one way to reduce this congestion is to reduce variability by addressing its causes. Hopp and Spearman point out that another, and more subtle, way to deal with congestion effects is by combining multiple sources of variability known as variability pooling. An everyday example of this concept is found in routine financial planning. Virtually all financial advisers recommend investing in a diversified portfolio of financial investments. The reason, of course, is to hedge against risk and uncertainty. It is highly unlikely that a wide spectrum of investments will perform extremely poorly at the same time. At the same time, it is unlikely that they will perform extremely well at the same time. Hence, we expect less variable returns from a diversified portfolio than from a single asset.
Hopp and Spearman go on to discuss how variability pooling affects batch processing, safety stock aggregation, and queue sharing, but the important point to take away is this. Pooling variability tends to reduce the overall variability just like a diversified portfolio reduces the risk of up and down swings in your earnings. The implications are that safety stocks can be reduced (resulting in less holding costs) or that cycle times at multiple-machine process centers can be reduced simply by sharing a single queue.
There are two basic but fundamental laws of factory physics relevant to variability provided to us by Hopp and Spearman.
1. Law (Variability): Increasing variability always degrades the performance of a production system. This is an extremely powerful concept since it implies that variability in any form will harm some measure of performance. Consequently, variability reduction is absolutely essential to improving the performance of a system.
2. Law (Variability Buffering): Variability in a production system will be buffered by some combination of:
a. Inventory
b. Capacity
c. Time
This law is an important extension of the variability law because it specifies the three ways in which variability impacts a manufacturing process and the choices we have in terms of buffering for it.
The primary focus of this blog posting was to explain the effect of processing time variability (PTV) on the performance of production lines. The primary points, conclusions, or principles are provided, once again, by Hopp and Spearman.
1. Variability always degrades performance. As variability of any kind is increased, either inventory will increase or lead times will increase or throughput will decrease or a combination of the three. Because of the influence of variability, all improvement initiatives must include variability reduction. As presented in earlier posts, and my last book, The Ultimate Improvement Cycle, there are important steps that focus on variability reduction.
2. Variability buffering is a fact of manufacturing life. If you can’t reduce variability, then you must buffer it or you will experience extended cycle times, increased levels of inventory, wasted capacity, reduced throughput and longer lead times all of which result in declining revenues, missed delivery dates and poor customer service.
3. Flexible buffers are more effective than fixed buffers. By having capacity, inventory or time available as buffering devices, you now have a flexible combination of the three to reduce the total amount of buffering needed in a given system. Examples of each type of buffer are included in Table 1.
| Flexible Buffer Type | Buffer Example |
| Flexible Capacity | Cross-trained workforce – by moving flexible workers to operations that need the capacity, flexible workers can cover the same workload with less total capacity than would be required if workers were fixed to specific tasks. |
| Flexible Inventory | Generic WIP held in a system with late product customization. That is, having a product platform that results in potentially different end products. |
| Flexible Time | The practice of quoting variable lead times to customers depending upon the current backlog of work (i.e. the larger the backlog, the longer the quote). A given level of customer service can be achieved with shorter average lead time if variable lead times are quoted individually to customers instead of uniform fixed lead time quoted in advance. This is possible if you significantly reduce your cycle time to the point that the competition can’t match it. |
Table 1
4. Material is conserved. Whatever flows into a workstation must flow out as acceptable product, rework or scrap. It’s either good, bad or re-workable product, but obviously we prefer only good product.
5. Releases are always less than capacity in the long run. Although the intent may be to run a process at 100 percent of capacity, when true capacity, including overtime, outsourcing, etc., is considered, this really will never occur. It is always better to plan to reduce release rates before the system “blows up” simply because they will have to be reduced as a result of the system “blowing up” anyway. This is why Drum Buffer Rope works so well….it prevents WIP explosions.
6. Variability early in a line is more disruptive than variability late in a line. Higher front end process variability of a line using a push system will propagate downstream and cause queuing later on in the process. By contrast, stations with high process variability toward the end of the process will affect only those stations. Remember, variability propagates, so the further up-stream the variability occurs, the more disruptive its effects will be.
7. Cycle time increases nonlinearity in utilization and efficiency. As utilization approaches 100 percent, long-term WIP and cycle time will approach infinity. Companies that attempt to drive the total process utilization and/or efficiency higher and higher will clearly have problems with excessive WIP, long cycle times, missed delivery dates and poor customer satisfaction…..plus huge quality problems. It’s why I loathe efficiency and utilization as a performance metric in any place other than the system constraint.
8. Process batch sizes affect capacity. Increasing batch sizes increases capacity and thereby reduces queuing while increasing batch size also increases wait-to-batch and wait-in-batch times. Because of this, the first focus in serial batching situations should be on setup time reduction, enabling the use of small, efficient batch sizes. If setup times cannot be reduced, cycle time may well be minimized at a batch size greater than one. In addition the most efficient batch size in a parallel process may be in between one and the maximum number that will fit into the process. The bottom line here is, the whole idea of economic batch quantity is riddled with wrong assumptions…..but don’t try and convince your local cost accounting group of this because they’ll accuse you of a sacrilege.
9. Cycle times increase proportionally with transfer batch size. Because waiting to batch and unbatch is typically one of the largest sources of cycle time length, reducing transfer batch sizes is one of the simplest and easiest ways to reduce cycle times. Instead of waiting for the full batch to be produced and moved to the next process step, move product periodically to the next step so it can be worked. But then again, I don’t believe in large batches at all. To me, the key is to simply reduce your process batch size and focus efforts on reducing changeover times through SMED.
10. Matching can be an important source of delay in assembly systems. Lack of synchronization, caused by variability, poor scheduling, or poor shop floor control, will always cause significant buildup of WIP, resulting in component assembly delays. Once again Drum Buffer Rope comes to the rescue by significantly improving synchronization of your process, much better scheduling and excellent shop floor control. In Bruce and my new book, Epiphanized which hopefully will be released this month, we go into much detail about DBR and how it will help your company become world class.
I hope you have enjoyed this 3-part series and I look forward to reading any comments or questions you might have.
Bob Sproull
No comments:
Post a Comment