Bottleneck Migration: What Happens to Your Schedule After You Solve the First Constraint

You spent three months justifying, purchasing, and installing a second VMC. Output went up 15%. Six months later you're late on deliveries again, and now everyone's pointing at the grinding department. You solved the bottleneck. The bottleneck moved. Welcome to constraint migration — the phenomenon every Theory of Constraints practitioner knows and most production planners discover the hard way.

Bottleneck migration isn't a failure. It's proof the system is working correctly. The constraint will always find the weakest link, and when you strengthen one link, it migrates to the next. The shops that handle it well are the ones that plan for the migration before the first machine ships. This post covers the mechanics of why constraints migrate, how to predict where they'll land, and what your scheduling system needs to do differently at each step.

For broader context on managing capacity through a finite lens, see our guide to finite capacity planning.

Why the Constraint Always Moves After Improvement

Goldratt's fundamental insight is that every system has exactly one constraint at any point in time. Throughput is limited by that one resource — all other resources have excess capacity by definition. When you add capacity to the constraint, one of two things happens: output rises until the next weakest resource hits 100% utilization, or demand absorbs the gain and you're back to the same delivery performance with a higher cost base.

In a 12-work-center job shop, the math is simple. If your turning center runs at 94% utilization and everything else runs below 80%, the turning center is your constraint. Add 20% turning capacity. Turning drops to 78%. Now your milling cell at 79% becomes the constraint. The migration happened the moment the first chip cut on the new lathe.

What makes this operationally painful is the lag. The new bottleneck doesn't announce itself. Queue builds slowly over 4-8 weeks as the downstream cell absorbs the extra output from the newly liberated constraint. By the time the queue is visible, you've been expediting for a month and the connection to the original investment is obscured.

The 5 TOC Focusing Steps — Repeated, Not Completed

Most production planners treat the TOC focusing steps as a one-time project:

1. Identify the system constraint
2. Exploit the constraint (maximize its throughput without additional investment)
3. Subordinate everything else to the constraint's rhythm
4. Elevate the constraint (add capacity if exploitation isn't enough)
5. Repeat — go back to step 1

The critical word is step 5: repeat. The five steps are a continuous loop, not a project plan with a completion date. Every successful elevation triggers a new constraint. Every new constraint requires a new exploitation analysis. Shops that treat step 4 as the finish line spend the next year confused about why their new equipment didn't deliver the ROI the spreadsheet promised.

At User Solutions, over 35 years of working with manufacturers from GE to small job shops, the pattern is consistent: the first constraint is usually a single machine or work center. The second constraint is usually labor — specifically a skilled operator class (machinists, welders, CMM programmers) that can't be purchased on a 6-week lead time. The third constraint is frequently floor space or material flow between work centers. Each migration type requires a different response.

How to Predict Where the Constraint Will Migrate

You don't have to wait for the queue to build. Before you elevate the current constraint, calculate projected utilization at every work center under the anticipated throughput increase.

The formula is straightforward:

Projected Utilization (%) = (Demand Units × Standard Hours per Unit) / Available Capacity Hours × 100

Run this for every work center at the demand level you expect after elevation. The work center with the second-highest projected utilization is your next constraint. If two work centers are within 3-5 percentage points of each other, plan to address both — you'll hit them in quick succession.

A practical example: a 35-person precision machining shop running 4-axis CNC work. Before adding a second 4-axis machine:

The second 4-axis drops CNC utilization to approximately 52%. At the same throughput increase, deburring hits 94% within 8 weeks. The shop should have hired one deburring operator before the machine arrived, not after deburring became the crisis.

The Scheduling Re-Optimization Required After Capacity Changes

Adding capacity changes more than utilization. It changes the optimal sequencing logic at every work center upstream and downstream of the constraint.

Before elevation, the scheduling priority was simple: protect the constraint. Every upstream work center was subordinated to feed the constraint at exactly the rate the constraint could absorb. Batching rules, setup sequences, and priority rules were all oriented around one goal — keep the constraint running.

After elevation, the former constraint is no longer the gating resource. Scheduling priorities need to reset. Three specific changes are required:

1. Recalculate protective capacity at the new constraint. The new bottleneck needs a time buffer — typically 10-20% of its available capacity held in reserve to absorb variation. This buffer doesn't exist yet because the new constraint was previously a non-bottleneck running with casual excess capacity.

2. Resequence the drum-buffer-rope logic. In TOC scheduling, the constraint sets the "drum beat" — the pace of the entire system. Upstream release of work-in-process is tied to the constraint's pace. When the constraint moves, the rope anchor moves with it. Releasing work at the old constraint's rate will starve the new constraint or flood it with WIP.

3. Audit setup sequencing rules at the new constraint. The most efficient setup sequence at the old constraint (minimizing changeover time to maximize constraint throughput) may not apply at the new one. If the new constraint is a grinding cell with a family-based changeover structure, your sequencing logic needs to be rebuilt around grinding wheel changes, not the spindle-type groupings that made sense at CNC.

RMDB handles this reconfiguration with constraint-aware scheduling that allows planners to designate the active constraint work center and rebuild buffer and rope logic in a single planning run. See how RMDB handles finite capacity scheduling for details on constraint designation.

Multi-Constraint Scheduling: When You Have Two Bottlenecks Simultaneously

Pure TOC theory says there is exactly one constraint. Reality in a job shop says otherwise, especially in mixed-product environments where different product families route through different work centers. A shop running aerospace forgings and commercial castings on the same floor may face a constraint in their 5-axis machining cell for aerospace work and a separate constraint in their casting prep area for commercial work — simultaneously, for different demand streams.

This is called a roving bottleneck or product-mix-dependent constraint scenario. It's more common than textbooks acknowledge because most real shops don't run a single product family.

Managing multi-constraint environments requires:

Segmented capacity views by product family. Aggregate utilization numbers hide product-mix-dependent constraints. If 5-axis CNC runs at 75% average utilization, but aerospace orders consume 95% of available 5-axis time while commercial orders barely touch it, the aggregate number is meaningless for scheduling.

Priority rules that resolve conflicts at shared resources. When a single work center serves two product families with different constraint dynamics, you need explicit priority rules. Options include: contractual priority (defense contracts above commercial), revenue per constraint hour (rank jobs by their margin contribution per unit of bottleneck time), or scheduled windows (aerospace on M-W-F, commercial on T-Th). Each approach has trade-offs that need to be documented and agreed upon before the schedule runs.

Constraint calendars, not static designations. In a seasonal business, the bottleneck in Q4 may be a completely different work center than Q2. Finite capacity software should track effective constraint status by time period, not just point-in-time utilization.

Real Examples of Constraint Migration Patterns

Pattern 1: Machine to Labor. A metal fabrication shop adds a second laser cutter. Laser utilization drops from 91% to 52%. Within 6 weeks, the press brake cell hits 97% utilization because more laser output feeds it directly. The press brake can run more parts, but only has 2 qualified operators. The constraint migrated from a capital asset to a human skill. Lead time to resolve: 4-6 months of hiring and training.

Pattern 2: Labor to Material Flow. A furniture manufacturer adds two cabinet makers to their assembly cell. Assembly utilization drops from 88% to 61%. Within 10 weeks, the staging area between CNC routing and assembly is physically full. Parts are being stacked in aisles. The constraint migrated from labor to floor space and material flow. The fix required a kitting supermarket between routing and assembly — a lean solution, not a hiring solution.

Pattern 3: Internal to External. A medical device job shop adds CNC capacity across three work centers. Internal constraints largely disappear. Within one quarter, late deliveries are being caused by a single outsourced anodizing vendor who can't scale with the shop's new throughput. The constraint migrated outside the four walls. Resolution required qualifying a second anodizing vendor — a 6-month supplier qualification process.

Each pattern requires a different response. The mistake is applying the same solution (buy equipment, hire people) regardless of where the constraint actually landed.

How RMDB Tracks Constraint Migration Across Plan Horizons

Knowing that the constraint will migrate is useful. Knowing specifically where it will migrate, and when, before it happens — that's operational leverage.

RMDB's finite capacity engine builds a forward-looking utilization profile across every work center for the full planning horizon. When you simulate a capacity change (add a machine, add a shift, increase a work center's efficiency), RMDB recalculates forward utilization across all downstream and upstream resources under the same demand load. The constraint migration shows up as the next work center to hit capacity ceiling in the simulation.

This means planners can see the cascading effect of a capacity investment before committing to purchase orders. A $180,000 VMC investment that shows RMDB projecting the inspection cell hitting capacity 11 weeks post-installation gives a plant manager the data to hire a CMM operator concurrently — not reactively.

The planning horizon matters here. Constraint migration analysis is only useful if your planning window extends far enough to see the next constraint emerge. For most job shops, a 12-16 week horizon is the minimum. For shops with 20+ week lead times on raw material, a 26-week horizon gives better visibility into where the constraint will be when new material arrives.

What to Do Before Elevation, Not After

The operational playbook for constraint migration should be executed before you elevate, not after:

1. Run utilization projection at +20% throughput across all work centers. Identify the top 3 by projected utilization — one of them is your next constraint.
2. Pre-hire or pre-train for the most likely next constraint type (usually labor). If the next constraint is a skill-intensive work center, start hiring the week the elevation capital is approved, not the week the new machine runs its first part.
3. Pre-plan the scheduling logic change. Document the drum-buffer-rope adjustments needed when the constraint moves. Brief your schedulers so they're not rebuilding scheduling logic under deadline pressure.
4. Set a 90-day post-elevation constraint review. Put it on the calendar before the machine arrives. Utilization numbers at 90 days post-elevation will confirm whether the predicted next constraint actually materialized.
5. Update your finite capacity software's constraint designation the moment queue at the new work center exceeds 3 days of buffer. Don't wait for it to become a crisis.

The shops that execute constraint migration proactively — rather than reactively — typically see 18-25% throughput gains from each elevation cycle instead of the 8-12% realized by shops that scramble after the fact.

Ready to see constraint migration before it happens? Contact User Solutions to see how RMDB projects constraint migration across your full planning horizon. Trusted by GE, Cummins, BAE Systems for 35+ years of finite capacity scheduling.