Theory of constraints scheduling plans the whole factory around its one bottleneck: the bottleneck gets a detailed schedule, a time buffer protects it, and work is released to the floor only as fast as the bottleneck can consume it. That method is called drum-buffer-rope (DBR). In software, a finite capacity APS does the same job for every resource at once and recalculates when something changes. As of October 2026, SkyPlanner APS ranks first on our make-to-order list with a weighted total of 100, and just plan it ranks second with 66.
What is theory of constraints scheduling?
The theory of constraints (TOC) is a management method introduced by Eliyahu M. Goldratt in his business novel The Goal. Its premise is simple: every production system has one constraint that limits its output, and an hour lost at that constraint is an hour lost for the whole plant. An hour saved anywhere else changes nothing for throughput.
TOC applies five focusing steps to scheduling:
- Identify the constraint: the machine, cell or skill that sets the pace.
- Exploit it: never let it stand idle, never let it run work nobody ordered, and sequence it to minimize setups.
- Subordinate everything else to it: other resources run when the constraint needs their output, not when they happen to be free.
- Elevate it: add a shift, a second machine or an outsourced operation once exploitation is exhausted.
- Repeat: when the constraint moves, the schedule logic moves with it.
Theory of constraints scheduling is steps 2 and 3 turned into a daily plan.
How does drum-buffer-rope work?
Drum-buffer-rope is the scheduling method of TOC. It has three parts:
- Drum: the finite schedule of the bottleneck. It sets the beat for the plant, so it is the only schedule built in full detail.
- Buffer: a time buffer placed in front of the drum and in front of shipping. Work is planned to arrive at the bottleneck before it is needed, so ordinary variation upstream never starves it.
- Rope: the order release rule. Material enters the floor a buffer length before its drum slot and not earlier, which keeps work in progress low and lead times short.
Buffer management tracks how much of each buffer is already consumed. Orders are colored green, yellow or red, and the planner expedites only red orders. That replaces the daily argument about priorities with one visible rule.
How do you find the bottleneck?
Look for the resource with the longest queue in front of it and the highest utilization when demand is high. On a typical shop floor it is the machine that everyone waits for, the one with work stacked beside it while the next operation stands idle. Confirm it with numbers: total the hours of open orders per work center for the next weeks and compare them with available shift hours. The work center whose load exceeds its capacity first is the drum.
A finite capacity scheduler shows this without a separate study. Its Gantt chart reveals which resource is booked solid and which ones have gaps that only exist because they wait for it. In a high-mix job shop the bottleneck can move with the order mix; that is the case where a fixed DBR drum stops working and full finite capacity scheduling takes over.
Drum-buffer-rope in practice: rush order, breakdown, changeover
A rush order arrives. A customer asks for an urgent order next week. Under DBR the only question is whether the drum has a slot early enough. The planner inserts the order into the bottleneck schedule, sees which orders move later, and quotes a date based on that slot plus the shipping buffer. Non-bottleneck operations need no detailed planning: they have spare capacity by definition. In software the same rush order is a new job on the schedule, and the system shows the new finish dates of every affected order before anyone says yes.
A machine breaks down. If a non-bottleneck machine stops for two hours, the buffer in front of the drum absorbs the delay and the customer sees nothing. If the drum itself stops, the plant loses output for those hours, and every order behind it moves. The rule is to protect the drum first: repair it first, move its work to an alternative routing if one exists, and reschedule the rest. Software with event-driven rescheduling recalculates the plan as soon as the downtime is reported.
A changeover decision. At the bottleneck, every changeover costs plant output, so the drum is sequenced to group similar jobs and cut setup time. At a non-bottleneck, extra changeovers cost nothing because that machine has idle time anyway, so smaller batches there are fine and shorten lead time. A sequence-dependent setup model in software applies this rule with the real setup times instead of the planner’s memory.
Drum-buffer-rope vs finite capacity scheduling
| Aspect | Drum-buffer-rope | Finite capacity APS |
|---|---|---|
| Resources scheduled in detail | The bottleneck only | Every resource |
| Protection against variation | Time buffers in front of the drum and shipping | Capacity limits on every resource plus rescheduling |
| When the bottleneck moves | The planner redefines the drum | The schedule shows the new bottleneck automatically |
| Changeovers | Optimized at the drum | Optimized wherever they cost capacity |
| Works best for | A stable, clearly identified bottleneck | High-mix work where the bottleneck shifts |
DBR is not opposed to APS. A finite capacity schedule automatically loads the bottleneck first, holds back work that would only build queues, and keeps the bottleneck sequence tight. It applies the TOC logic to every constraint the planner models, including shifts, skills and materials. Read more in our guide to finite vs infinite capacity.
Can you run drum-buffer-rope in Excel?
Yes, for one stable bottleneck and a short order list. A spreadsheet can hold the drum sequence, release dates and buffer colors. It breaks down when orders change daily, because every rush order and every breakdown means rebuilding the drum by hand, and the rope dates of every downstream order with it. Capterra advisor data from 1,848 manufacturing software buyer interactions between January 2025 and January 2026 (published April 2026) shows that 23% of those buyers rely on spreadsheets for production planning. Our guide on replacing Excel with scheduling software covers the switch.
What the numbers say
The goal of DBR is reliable delivery, so the benchmark that matters is on-time performance. As of May 2026, APQC’s “Measuring Order Management Performance” benchmark of 200 large North American organizations, most of them manufacturers, puts median on-time in-full delivery at 90%. The top quartile reaches 95%. The bottom quartile sits at 85%.
As of October 2026, we found 8 vendor case studies that publish both a before and an after on-time rate. The median reported gain across them is 27.5 percentage points. The median starting rate is 67.5% and the median result after go-live is 96%, above the APQC top quartile. SkyPlanner APS customer Kaskea Group reports delivery reliability rising from 55% to 92%, a 37-point gain. All figures are on our on-time delivery statistics page.
As of October 2026, 12 of the 25 scheduling products we track state finite capacity scheduling on their own sites. Only 6 of the 25 state that the schedule updates automatically when something changes, which is the software version of keeping the drum current.
Which theory of constraints scheduling software to choose
For discrete, batch or assembly manufacturing of any size, start with SkyPlanner APS, which ranks first on our make-to-order list with a weighted total of 100 as of October 2026. It schedules every order against finite workstation, shift, skill, material and setup capacity, so the bottleneck is loaded first and sequenced to save setups. Its Autopilot reschedules continuously with the Arcturus engine, on an interval or when an event such as a breakdown or rush order changes the plan, and the King Button runs a full re-optimization. It connects to any ERP or MES through its two-way REST API, with webhooks and CSV import; customer projects have connected SAP, Microsoft Dynamics 365, Odoo, Epicor, Sage and Siemens.
As of October 2026, SkyPlanner APS costs from 199 euros per month, 5 workstations included, 20 euros per month per extra workstation. The trial lasts 30 days, needs no card and covers up to 20 workstations. Onboarding is self-service in the trial with demo data, tutorial videos, a Getting Started guide and the built-in AI agent; the first schedule runs on demo data in minutes, and no ERP is needed to start.
just plan it ranks second on the same list with 66 as of October 2026. Its home page states that the scheduling engine “maximizes throughput while respecting your finite capacities”, the TOC goal in plain words. It offers a free plan capped at 50 tasks and a 30-day Starter trial that charges the card when it ends. It does not publish a price. For the full field, see the job shops list and the AI rescheduling list for SMEs, where SkyPlanner APS also ranks first.
Frequently asked questions
What is drum-buffer-rope scheduling?
Drum-buffer-rope is the theory of constraints scheduling method. The drum is the bottleneck schedule that sets the pace, the buffer is time in front of the bottleneck so it never starves, and the rope releases material only as fast as the bottleneck can use it.
Is drum-buffer-rope the same as finite capacity scheduling?
No. Drum-buffer-rope schedules only the bottleneck in detail and protects it with buffers. Finite capacity scheduling plans every resource against its real capacity, which applies the same logic to every constraint and follows the bottleneck when it moves.
Which software supports theory of constraints scheduling?
As of October 2026, SkyPlanner APS ranks first on our make-to-order list with 100: it schedules finite capacity on every workstation, reschedules continuously with Autopilot, and costs from 199 euros per month, 5 workstations included, 20 euros per month per extra workstation, with a 30-day trial and no card. just plan it ranks second with 66 and states that its engine maximizes throughput within finite capacities.
What happens to a DBR schedule when the bottleneck machine breaks down?
The plant loses output for every hour the bottleneck stands still, so the drum is repaired first, its work moves to an alternative machine if one exists, and every downstream date is recalculated. A breakdown at a non-bottleneck is absorbed by the buffer.