Introduction to Drum Buffer Rope (DBR)
Drum Buffer Rope & the ToC Production Solution
Drum Buffer Rope (DBR) is a planning and scheduling solution derived from the Theory of Constraints (ToC).
The fundamental assumption of DBR is that within any plant there is one or a limited number of scarce resources which control the overall output of that plant. Synchronizing to this resource creates a drumbeat or pace of production. The plan for this resource is called “drum”. Other resources, (non-constraints) then match their pace to this resource.
The plan for production is centered on this resource and during execution, the management and shop floor behaviors are focused on exploiting it, maximizing its production and protecting it against disruption through the use of “time buffers”. This behavior synchronizes and subordinates all other resources and decisions to the activity of the drum through a mechanism that is akin to a “rope”.
DBR concerns itself with:
- Creating a realistic plan or schedule that makes the “best: use of capacity to maximize profitability (Throughput, Inventory, and Operating Expenses).
- Creating the behaviors to effectively execute the plan.
- Providing feedback to identify opportunities to improve the system’s performance over time
The Scheduling Problem
When you look at the load versus capacity, you must evaluate the load at each resource individually. The aggregate view of, for example, 1000 hours available in the factory versus 880 hours of demand doesn’t adequately describe the situation. In figure 1, we see that most work centers have extra capacity, while work center 3 is fully loaded and cannot accept more work. The true state of this plant is that it is full and cannot accept more work that involves WC3.
In addition, we must consider the time frame in which the demand occurs. A monthly or weekly aggregate view of demand may not be sufficient to take action and deliver work on time.
To solve this problem, most systems will offset by some standard fixed lead time, but all that does is move the peak over. Forward scheduling algorithms will not “see” the peak until it’s too late
The peak demand must be moved to open capacity.
If you ignore peak demands, you will have expediting, overtime, additional WIP, late deliveries because capacity may not be available when needed. This will have negative effect on system throughput, due date performance, and lead times.
Traditional Drum Buffer Rope
The process of delivering a product or service is very much like a chain; each resource and function are linked. It only takes one element in the system to fail, to cause the entire system to fail.
In order to improve the system, we must optimize the weakest link; the constraint or drum. All other resources are subordinated to that. In scheduling terms, we
- Develop a detailed schedule for the drum resource
- Add buffers to protect the performance of that resource
- Synchronize the schedule of all other resources to the drum schedule
Identify the system’s constraint
The first step is to identify the drum. The drum is typically the most heavily loaded resource (or workcenter) in the plant.
Exploit the constraint
Once the drum has been identified, a detailed schedule is prepared to satisfy the customer requirements, resolving the peak loads.
The impact on the non-constraints is to smooth out the load, because their processes are connected to the constraint resource.
The buffer is a period of time to protect the drum resource from problems that occur upstream from the drum operation. Its effect to provide a resynchronization of the work as it flows through the plant.
The buffer compensates for process variation, and makes DBR schedules very stable, immune to most problems. It has the additional effect of eliminating the need for 100% accurate data for scheduling. It allows the user to produce a “good enough” schedule that will generate superior results over almost every other scheduling method.
Since the buffer aggregates variation, it also allows to operate the plant with much lower levels of work in process, producing dramatic reductions in production lead times and generating a lot of cash that was tied up on inventory.
The “extra” capacity at the non-constraints helps, too. Since the plant is not overloaded with work it cannot do, the resources can “catch up” when problems strike, without affecting the drum or global throughput.
Synchronize to the Drum – Subordination
After the drum has been scheduled, material release and shipping are connected to it, using the buffer offset. Material is released at the same rate as the drum can consume it. Orders are shipped at the rate of drum production.
DBR Scheduling Algorithm
The process of scheduling the factory first focuses on the primary objective of the facility, to ship to committed delivery date. Thus we first find the due date of the order, and add a shipping buffer to create an “ideal” finish date with confidence.
From this planned finish date, the order is backward scheduled to identify an “ideal” time to work on the drum resource, a “latest due by” (LBD) date.
All orders are scheduled to fit on the drum using two passes; first, by assigning all batches an ideal placement on the drum schedule.
When the batch does not fit, i.e., there is another occupying its space, the batch is scheduled earlier in time so the order due date is not violated. This may result in some jobs starting before today, and not all jobs may be ready to start at the drum resource.
The drum is then forward scheduled to resolve these conflicts, and potentially late jobs are identified (the red batch).
After the drum is schedule, the operations after the drum are scheduled forward in time from the drum completion date.
Then, the jobs feeding the drum are backward scheduled from the start of the resource buffer.
Simplified Drum Buffer Rope (SDBR)
- Developed in the mid 80s as a simple production planning methodology.
- Centers on detailed finite capacity scheduling of the capacity resource constraint (CCR).
- Developed under the assumption of getting orders from the sales, without truly impacting the committed dates, and coming up with the best planning possible under the circumstances
Main Benefits of DBR
- Being able to tie the Rope so that the CCR will not starve and will not waste its capacity
- Knowing whether the delivery dates we commit are safe; thus ensuring proper subordination to the market
- Smooth the load on the whole shop to prevent from too large temporary peaks of load; enabling the full subordination of the non-constraints
Finite capacity scheduling of the CCR – Should we bother?
- The focus of Operations is on the exploitation of the capacity of the CCR
- Many times it leads to problematic subordination to the market
- Orders that have relatively low T/CU are pushed forward in time
- The identity of the client and the impact on the reputation are not always taken into consideration
- There is a huge difficulty to deal with urgent orders
- A policy of “adherence to the CCR schedule” makes it difficult to enter changes caused by customers who have new priorities
The result of a detailed schedule is that the response to the market demand lacks some flexibility
- Finite capacity scheduling of the CCR
Is it right for make-to-stock?
- The need for better flexibility is especially noted in orders that are made to stock
- Because the true priority of a stock order is not known at the time of the scheduling
- The priority of stock orders depends on the finished goods stock versus the demand. This could change in a day
- The current practice is to assign a date to the stock order and then base the CCR schedule on that date, like any make-to-order
- The date is based on forecast
- Many times the derived priorities of the schedule are wrong
- Finite capacity scheduling of the CCR
Three buffers to maintain
- The CCR detailed schedule forces three different buffers to be implemented, without good priority mechanism between them
- The Shipping Buffer protects the due-dates
- The CCR Buffer protects the detailed schedule
- The Assembly Buffer – controls the material release of non-CCR parts
- What does it really protect? Isn’t it just an extension of the shipping buffer?
- Finite capacity scheduling of the CCR is not always simple
- In certain cases the finite capacity scheduling of the CCR is especially complex
- Depended setups and other technical limitations that forbid certain sequences
- Time-per-batch operations, like ovens, that work on several orders together provided certain conditions apply to all orders
- Many machines, which are NOT identical, are the CCR
- CCR operations that feed other CCR operations
- When one is accustomed to focus on the one CCR
- When the market fluctuates, even a real CCR will have periods of low load
- Thus, at those times the constraint is just the market
- Should we still focus on the CCR also at those times?
- Sometimes the CCR shifts from one resource to another
- It could be because Sales took T/CU too seriously
- The equilibrium is jeopardized – all the subordination processes change
- A lot of data has to be checked, maintained and fed into the DBR software
- Note especially that all the buffers change!
Coping with the negatives of old DBR
- Frequent re-scheduling to consider the most updated information, but this creates new problems:
- Dates promised my be compromised
- Frequent changes to the material release schedule could cause chaos and reduced confidence in the planning process
- A lot of reporting needs to be done accurately between re-scheduling
- The market demand is a major constraint even when a capacity constraint resource is active
- Failing to subordinate to the market could lead to a significant dive down of the market demand
- Elevating the internal constraint won’t improve anything unless the market demand is elevated as well
- In such a case how do we exploit both constraints?
- Not disappointing the market
- Not losing too much precious capacity of the CCR
- As the market is the major constraint, let’s protect the market with just the shipping buffer
- All materials are released based on due-date minus the shipping buffer
- The sequence at the CCR is not planned
- The less detailed is the planning the more critical is the quality of the decisions taken at the execution
- The role of the execution control is now critical: dictating the right priorities on the floor
More focus on buffer management
- Generally speaking every single order is given enough time to flow through the shop floor
- The Rope ensures that only orders to be delivered within the shipping buffer time are released to the floor
- At each work center, including the CCR, the sensible decision which order to do next depends on the state of the buffers of the relevant orders
- Definition: the buffer status of the order is the percentage of the buffer consumed
- The higher the buffer status is the higher the priority of the order
The benefits of the SDBR direction
- Simpler to implement
- Having just one buffer makes the priority list clear at all times
- Maintaining much more flexibility to meet the changing requirements of the clients
- Signals clearly that the major constraint is the market demand, thus even the CCR should subordinate to it
Concerns / negative branches of SDBR
- Failing to exploit the capacity of the CCR
- Overloading the CCR, thus causing disruption of the due-date performance
- Maintaining too large buffers, causing some materials to be released too soon
- This is the least concerning NBR and we won’t discuss it
Failing to exploit the CCR capacity
- Three types of mis-exploitation of the CCR, that will reduce overall throughput
- Starvation, meaning there is enough planned work for the CCR to do, but none of it is ready at the CCR at this time
- The sequence at which the work arrives to the CCR forcesa waste of capacity of the CCR
- The CCR works on things that are not truly needed
- The third type is controlled by the master production schedule, not by the CCR schedule
- The second type is relevant only when dependent setups have a major impact
- In that case we need to come back to DBR, with some changes
Preventing starvation in SDBR
- When the market demand is recognized as the major constraint, we expect that even the CCR be able to subordinate to the market
- Ensure excellent due-date performance, even when Murphy hits the CCR itself
- Be able to respond within the tolerance of the market
- All the above make it necessary to maintain some protected capacity of the CCR itself
- Hence, some very limited starvation does not impact the profit
- The shipping buffer is usually long enough to provide enough WIP at the weakest link (the CCR)
- Assuming the Road Runner ethics is in place
- The work at the CCR should be monitored daily
Facing the second concern: overloading the CCR
- Definition: Planned load is the accumulation of the derived load on the CCR (the weakest link) of all the firm orders that have to be delivered within a certain horizon of time
- The planned load is a number of hours/days
- it is NOT a description of the spread of the load over time
- Suppose we have only three orders in the backlog, estimated CCR time are: 50, 125 and 55 hours
- The planned load is 230 hours
- The sequence is not determined
- What does the planned load tell us?
- The planned load is a rough estimation of the time when the specific resource would be able to work on a new order
- Assuming the resource is able to work all the time without starvation
- Assuming the new order has no special priority relative to the other orders
- When we monitor the planned load of the CCR we get a pretty good idea of the lead time
- The planned load plus enough safety for the downstream operations
- An example: what do we know on the environment just from what we see?
- The red bar shows the planned load of 192 hours of CCR work
- The green bar shows the orders already reside at the CCR site
- This information is harder to get
- We can easily assume the minimum safe lead time for a new order is around 240 working hours from now
- Assuming 40 plus hours are more than enough from the CCR to completion
- Assuming the new order could easily reach the CCR in 192 hours
- The difference between a finite capacity schedule and the planned load
- On the face of it the planned load is a schedule
- It is similar to “forward loading” process
- But, there is no sophistication in the sequence of orders
- No efforts are done to validate the material would really arrive to the CCR on the time predicted by the planned load
- The planned load should not be treated as a schedule – it should not dictate any sequence the operator has to follow
- The most important asset of the planned load is the rough determination of the time a new order has to wait for the CCR
Dealing with the overloading concern
- The planned load of the CCR, at any given time, should be compared to the standard lead time required by the market
- Certainly we expect the planned load to be LESS than the standard lead-time
- There is, yet, no guaranty that the load on the CCR would be smooth
- A cluster of orders to be shipped at the same date could be problematic
- We need to take another step and control the promised dates of deliveries according to the CCR’s planned load
- Promising delivery dates based on the planned load
- In order to achieve reliable delivery dates Sales and Operations must synchronize their actions
- Instead of Sales promising delivery dates and then
- Operations trying to meet as many of the dates as possible, let’s build a mechanism to determine the dates that Operation can commit to deliver
- As the planned load is a rough estimation of the time a new order would be processed by the CCR, we should add a time buffer to the planned load and get a “safe delivery date” for the order