DBR, Drum-Buffer-Rope, is a TOC planning methodology for manufacturing companies. In this post I focus on DBR/SDBR for make-to-order (MTO), when client orders specify the products, the quantities and the delivery time. I’ll dedicate another post on the boundaries of managing stock. On top of the planning, there is Buffer-Management, which is used to guide the priorities in the execution.
I’m not going to discuss here the details of DBR and the differences between DBR and SDBR. I wish to outline the situations where the rationale of DBR/SDBR is valid. For instance, Goldratt created CCPM because it became evident that DBR would NOT work for projects. Is it clear to YOU, the reader, what basic assumption(s) behind DBR are not valid in multi-project environment and vice-versa?
The basic situation is manufacturing or service organizations getting orders from clients who ask for certain packages of well-defined products or services to be delivered within a certain lead-time. The organization does not usually hold these products in stock, so there is a need to produce them. The organization might need materials purchased from suppliers and then produce the specific orders for those clients.
Key basic assumptions where using DBR/SDBR would yield reliable delivery performance:
- The net touch-time of any production order is very low relative to the lead-time, the time from accepting an order until delivery.
- Goldratt assumed less than 10% of the time-buffer used for such an order.
- The rest of the time the production order waits for resources to become available, which means that the utilization level of many resources is not very low.
- The level of statistical fluctuations within the shop-floor itself is not too large.
- For instance, when the average setup is two hours it is unlikely to take 20 hours.
- Overall such an environment is exposed to much less fluctuations than in projects!
- This assumption also addresses the level of scrap. It is unlikely, or very rare, that a whole production order is scrapped.
- It means that the shop-floor maintains good enough quality in all stages.
- The organization maintains acceptable control on our supply and outsourcing.
- All the operations are at the same location, or close enough, so transportation time between the facilities is short relative to the time-buffer.
- This assumption could be viewed as an extension of assumption #1. We need to understand that transportation, also dry-time, is part of the “touch time”.
One key assumption for Buffer-Management:
- Most orders either finish in Yellow or finish very close to their penetration into Red.
When one assumption, or more, is not valid it does not mean that DBR/SDBR is useless, but it does mean certain changes are mandatory.
I like to explain more the impact of touch-time on the methodology. Actually the assumption about touch-time is the main difference between production and multi-projects. In projects we expect that the time it takes to process the tasks along the critical chain is equal to the time to complete the whole project. This is in sharp contrast to production where orders wait between operations for quite long time. Because of this key difference the term “critical path” or “critical chain” is not relevant for production.
What happens when the touch-time of a certain operation is about 30% of the time-buffer?
In SDBR the time-buffer covers the whole production process, from the release of the materials to completion. When one operation takes 30% of that time we have to ask two critical questions:
- Is the time-buffer enough to protect the due-date from fluctuations along the whole production process? The 30% touch-time is not part of the protection time, so, do the rest 70% provide enough protection?
- In buffer management for DBR/SDBR the exact location of the production order is not reflected in the status of the buffer, because when the touch-time is negligible then it does not truly matter. However, when one specific operation takes 30% of the buffer it matters a lot whether the production order has still to go through it or already passed it. When the order is still behind the long-operation the remaining effective buffer is significantly shorter than the remaining time until the due-date.
A new application of buffer management within SDBR implementation facing high touch-time was developed and presented at the TOCICO conference in 2012 (Lisa Scheinkopf, Yuji Kishira, and Amir Schragenheim). This is an example of understanding the boundaries of the basic assumptions and the level of required changes when a specific assumption is invalid.
It is important to emphasis again that the above assumptions refer to make-to-order production. The critical distinction between MTO and MTA (make-to-availability) was recognized much later than DBR and also SDBR. The assumption that we have a specific quantity to deliver at a specific date was not clearly verbalized in both MRP and DBR. Once the role of that assumption became clear a different methodology for MTA had been developed. The lesson is to try our best to verbalize the underlining assumptions, which together define the boundaries where a specific methodology is valid. Then we can identify the situations that lie beyond those boundaries and look for an appropriate solution, which can be close or very different from the original methodology.
5 thoughts on “The boundaries of DBR/SDBR”
Probably not less important assumptions of DBR/S-DBR for MTO are:
● The use of load control mechanisms, in order to maintain a smooth flow on the floor and consequently excelent production lead times.
● The due-dates for the orders are safe, in order to maintain reliability.
● Mechanisms for availability of raw materials and even completness of “kits” are in place to prevent late releases (penetrations in The Rope) and consequently too much orders into Red.
Gustavo, your first and third assumptions are directed at what is needed in order for DBR/SDBR to truly work.
The boundary assumptions define the environmental conditions where the methodology is supposed to work.
Within the methodology itself there are more assumptions that present the logic of how the methodology work.
Your second assumption is relevant for the boundaries. I tried to include it in the broad description of the make-to-order: there are orders for specific products, well-defines quantities and agreed-upon due-dates. Making the due-dates safe can be added, but then you need also to make the required quantity safe as well. Certainly when the client gives forecasts, but without commitment to the quantities until the very last minute – the solution of DBR/SDBR cannot be applied as is and would need to include stock as well.
LikeLiked by 1 person
Eli, first I want to congratulate you for this exemplar post.
In my humble opinion, this is the type of knowledge that is truly missing in the public domain. The impression is that the community mantain it in private domain to safe-guard markets. But this is exactly what is missing for most of the practictioners. A more clear specification of boundary conditions and principles that apply to each deviation.
Here in Brazil we have almost none public classes in TOC, so there are lots of gaps that only the more resilient students, with strong purposes, came to fill.
Thanks a lot!
Dear Mr. Schragenheim,
Thank you very much for sharing all this great knowledge with all of us . I just have a question.
Let’s suppose that we have one job shop – it is built from only one machine, one CNC lathe machine and we shape different metals. This is an imaginary scenario – I do not know how a CNC lathe machine works.
First Stage – 35 minutes
Let’s suppose that there are 4 tasks before we actually shape a piece of metal, like moving close the parts, clean the CNC machine from the previous work, set up the machine and position the piece of metal in the CNC lathe. Let’s suppose that for these 4 tasks we need 35 minutes.
Second Stage – 30 minutes
Then, we are working on the specific metal and we modify it according to a customer’s specifications. We need different durations depending on the type of the work. Let’s assume that on average is 30 minutes.
Third Stage – 15 minutes
After we have finished with the shaping of the metal, there are 3 tasks to remove the metal. Firstly, we unloose certain parts of the CNC machine, we wait for the metal to cool down and we remove the metal. Let’s suppose that these three tasks need 15 minutes to be completed.
Let’s suppose now that I want to increase the flow through above scenario/layout by implementing DBR. The Drum is the time I spend shaping the metal. This is the longest and the most valuable one, the Buffer is the next order I have on hand, the rope is the signal to my helper to bring parts of the next order near to the CNC machine.
My observations are that although in the traditional DBR only the processes before the constraint can actually affect the Throughput – in this case all the processes including those after the constraint affect the total Throughput since they keep the constraint from working. In this imaginary scenario in order to increase Throughput we need to reduce all the time between the operation of the constraint – this time includes stage three from the previous operation and stage one from the next.
My question – is this a typical application of DBR? Tasks after the constraint are not considered by DBR BUT they affect the overall lead time. How can this be handled?What do you think?
Warm Regards and once more your willingness to share your knowledge with the rest of us is highly appreciated and respected.