The Formula That Governs Every Production System on Earth
In 1961, John D.C. Little, a professor at MIT, published a mathematical proof of a relationship that seems almost too simple to be important:
📐 Little’s Law
CT = WIP ÷ TH
Cycle Time equals Work-in-Process divided by Throughput. The average time a unit spends in a production system equals the average number of units in the system divided by the average rate at which units exit the system.
What makes this formula remarkable is not its complexity — it is its universality. Little’s proof showed that this relationship holds for any stable system, regardless of the arrival distribution, the service time distribution, the number of servers, the routing complexity, or the product mix. It does not matter whether you are running a single CNC machine, a 30-station assembly line, or an entire aerospace manufacturing facility with 2,000 employees. If the system is in steady state (average input rate ≈ average output rate over time), Little’s Law is true.
This is not an approximation. It is not a rule of thumb. It is a mathematical law — as reliable in production management as F = ma is in physics. And yet most production leaders have never applied it to their own operation. They make scheduling decisions, release work orders, and promise delivery dates without ever checking whether those promises are consistent with the WIP already in their system.
That changes today.
The Three Variables: Defined Precisely for Aerospace
Little’s Law has only three variables, but each one is routinely mismeasured in aerospace manufacturing. Getting the definitions right is not pedantic — it is the difference between a formula that gives you actionable intelligence and one that gives you garbage.
| Variable | Symbol | Precise Definition | Common Measurement Error |
|---|---|---|---|
| Cycle Time | CT | Total elapsed time from the moment raw material is released into the production system to the moment the finished unit exits. Includes all queue time, processing time, inspection time, rework time, and storage time. | Confusing CT with “touch time” or “processing time.” Touch time is typically 5–15% of CT in aerospace. CT includes the other 85–95% — the time spent waiting in queues. |
| Work-in-Process | WIP | The total number of units currently inside the production system in any state: being machined, in queue, in inspection, in rework, in staging, in transport, on hold, in storage between operations. If it has been released and has not shipped, it is WIP. | Counting only units “on machines” or “being worked.” The units sitting in queue, in rework, on inspection hold, and in inter-operation storage are all WIP — and they typically outnumber the units being actively processed by 5:1 or more. |
| Throughput | TH | The average rate at which completed units exit the production system over a sustained period. Measured in units per time period (units/day, units/week). | Using theoretical capacity instead of actual observed throughput. If your machine can process 30 units/day but quality holds, rework, and downtime reduce actual output to 22 units/day, your TH is 22. |
Why WIP Is Almost Always Larger Than Management Thinks
In every aerospace facility I have worked in, the initial WIP count surprises management. The number is always higher — often dramatically higher — than what they believed. The reason is that most WIP is invisible to the people who make decisions about it.
Here is how to actually count WIP in an aerospace facility. Walk the floor with a clipboard (or tablet) and count every unit in every state:
| WIP Category | Where to Find It | Why It’s Usually Missed |
|---|---|---|
| Active processing | On machines, at assembly stations | This is the only WIP most managers count |
| Queue WIP | Racks, carts, staging areas in front of machines | Treated as “ready to run” rather than as system congestion |
| Inspection hold | QA cage, inspection benches, awaiting buy-off | Categorized as “quality” problem, not counted as production WIP |
| Rework loop | MRB area, rework benches, awaiting disposition | Falls into a different tracking system; often invisible to production planning |
| Inter-operation storage | Shelves, floor storage, overflow areas between operations | Accumulates gradually; becomes “furniture” that nobody questions |
| Transport WIP | On carts, in transit between buildings, at loading docks | In motion, so it feels like progress rather than WIP |
| Hold/shortage WIP | Parts waiting for missing components, engineering dispositions, or tooling | Parked and forgotten; may sit for weeks without appearing on any active report |
In a typical aerospace machine shop, active processing accounts for 10–20% of total WIP. The other 80–90% is sitting in queues, holds, and storage. This means that when management looks at their “machine loading” report and sees 50 units “in process,” the actual WIP in the system is likely 250–500 units. And every one of those units is adding to every other unit’s cycle time through Little’s Law.
💡 Little’s Law Is Always True
Little’s Law is always true — the only question is whether you are using it deliberately or suffering it accidentally. Every production system on earth obeys CT = WIP / TH. If you have not calculated your CONWIP limit, you are still operating under Little’s Law — you are just operating at whatever random WIP level your MRP system and management decisions have created, and getting whatever random lead time that WIP level produces.
The WIP Trap: Why Releasing More Work Orders Makes Things Worse
This is the single most counterintuitive — and most important — insight from Little’s Law. When a production facility falls behind schedule, management’s instinctive response is to release more work orders. “We need to get more started so we can get more finished.” This feels logical. It is mathematically catastrophic.
Here is why. Little’s Law says CT = WIP / TH. Throughput (TH) is bounded by the constraint — the slowest operation in your value stream. Releasing more work orders does not increase the constraint’s capacity. The constraint can only process what it can process. So when you release more WIP into the system while TH remains constant, the only variable that can change is CT. And it changes in exactly the wrong direction: it goes up.
⚠️ Starting More Jobs Is the Most Counterproductive Response to a Schedule Emergency
When you release additional work orders into a system that is already behind, you are not accelerating production. You are adding to the queue in front of the constraint, which increases the wait time for every job in the system — including the critical ones you are trying to expedite. The math is unforgiving: more WIP at constant throughput always means longer lead time. Always.
The mechanism is queue congestion. Every unit in the system must wait its turn at every operation. When you add more units, every queue gets longer. The constraint — which was already the bottleneck — now has an even longer queue in front of it. Jobs that were going to take 20 days now take 30 days. The schedule miss gets worse, not better. And if management responds to the worsening miss by releasing even more work orders, the death spiral accelerates.
Scenario: A 5-axis CNC machining cell processes structural aerospace components. Current state data:
| Variable | Current Value |
|---|---|
| WIP (total units in the cell, all states) | 400 units |
| Throughput (average daily output) | 20 units/day |
Step 1: Calculate current cycle time.
CT = WIP ÷ TH = 400 ÷ 20 = 20 days
Every unit released into this cell will take an average of 20 days to exit. This is the lead time you should be quoting to your internal customers.
Step 2: Management releases 200 additional units to “get ahead.”
New WIP = 400 + 200 = 600 units. Throughput has not changed — the constraint machines can still only process 20 units/day.
New CT = 600 ÷ 20 = 30 days
Step 3: Interpret the result.
By releasing 200 additional units, management has increased the lead time for every job in the system — including the ones that were already late — by 10 days. The jobs they were trying to accelerate are now even later. The “critical” part that was 5 days behind schedule is now 15 days behind schedule.
What management usually does next: Expedites the critical jobs by pulling them to the front of the queue, which pushes every other job back further, triggering more expedites, more re-sequencing, and eventually a complete breakdown of scheduling integrity. This is the WIP death spiral.
What a Process Architect does instead: Stops releasing new work orders until WIP drops to the target CONWIP level, then implements a release discipline tied to the constraint’s consumption rate.
CONWIP: Controlling Lead Time by Controlling WIP
CONWIP (Constant Work-in-Process) is the direct application of Little’s Law as a production control method. The concept is simple: set a cap on total WIP in the system. When a unit exits (ships, moves to the next value stream), release a new unit at the entrance. Never exceed the cap.
The CONWIP limit is calculated directly from Little’s Law by rearranging the formula:
📐 The CONWIP Formula
WIPmax = Target CT × TH
The maximum WIP you should allow in the system equals your target cycle time multiplied by your throughput rate. This is Little’s Law solved for WIP.
Scenario: You have committed to your program that machined components will have a 10-day lead time through the machine shop. Your constraint throughput is 25 units/day.
Step 1: Calculate the CONWIP limit.
WIPmax = Target CT × TH = 10 days × 25 units/day = 250 units
To guarantee a 10-day lead time at 25 units/day throughput, you must never have more than 250 units in the system.
Step 2: Compare to current state.
Your current WIP is 600 units. Your current CT = 600 ÷ 25 = 24 days. You are promising 10 days and delivering 24 days.
Step 3: Calculate the required WIP reduction.
You must reduce WIP from 600 to 250 — a reduction of 350 units (58%). This cannot happen overnight without disrupting production. See Guide 07: CONWIP for the controlled WIP reduction protocol.
Step 4: Understand the transition.
During WIP reduction, you stop releasing new work orders while the system drains. Throughput continues at 25 units/day (the constraint doesn’t slow down — it still has 600 units to work through). After 350 ÷ 25 = 14 days of reduced releases, WIP reaches 250 and the CONWIP limit takes effect. From that point forward, one in = one out.
The counterintuitive result: By releasing fewer work orders, you did not reduce output. Throughput stayed at 25 units/day throughout. What changed was lead time — it dropped from 24 days to 10 days. You delivered the same number of parts, faster, with less WIP, less congestion, and better schedule predictability.
📐 The Math Proof: WIP Reduction Always Reduces Cycle Time
CT = WIP ÷ TH. If TH is held constant (the constraint’s capacity hasn’t changed), then CT is directly proportional to WIP. Reduce WIP by 30%, and CT drops by 30%. Reduce WIP by 50%, and CT drops by 50%. There is no scenario in which reducing WIP at constant throughput increases cycle time. This is not a theory — it is arithmetic.
Little’s Law in the Make Shop vs. the Assembly Shop
Little’s Law applies universally, but the way you count the variables differs between the two production environments described in Guide 01.
| Variable | Make Shop (Asset-Bound) | Assembly Shop (Labor-Bound) |
|---|---|---|
| WIP unit | Individual part or batch (e.g., one bracket, one batch of 20 fittings) | One aircraft position, one major assembly, one ship set |
| WIP count method | Count every part in every state: on machine, in queue, in inspection, in rework, in storage | Count every position occupied on the assembly line, including positions with incomplete work or holds |
| CT definition | Time from raw material release to finished part delivery (days) | Time from assembly start (position entry) to assembly complete (position exit) — often measured in “flow days” |
| TH measurement | Parts per day exiting the last operation | Aircraft (or major assemblies) per month moving to the next value stream |
| Typical CT range | 5–60 days for machined aerospace parts | 20–180 flow days for major airframe assemblies |
Scenario: A wing assembly value stream has three major stations: Station 1 (structural assembly), Station 2 (systems installation), Station 3 (functional test and closeout). Each station is designed for a 10-day Takt — one wing assembly advances every 10 working days.
| Station | Design Takt | Current WIP at Station | Actual CT at Station |
|---|---|---|---|
| Station 1: Structural Assembly | 10 days | 2 units (1 active + 1 in queue) | 2 ÷ 0.1 = 20 days |
| Station 2: Systems Installation | 10 days | 3 units (1 active + 2 in queue/hold) | 3 ÷ 0.1 = 30 days |
| Station 3: Test & Closeout | 10 days | 2 units (1 active + 1 awaiting buy-off) | 2 ÷ 0.1 = 20 days |
Note: Throughput is 0.1 units/day (one unit every 10 days) at each station, since the line is designed for a 10-day Takt.
Full value stream:
Total WIP = 2 + 3 + 2 = 7 units
System TH = 0.1 units/day (the line produces one wing every 10 days)
System CT = 7 ÷ 0.1 = 70 days
Interpretation: The design intent was a 30-day flow through (10 days × 3 stations). The actual flow-through is 70 days — more than double — because of queue WIP at each station. Station 2 is the worst offender with 3 units of WIP, suggesting it is the constraint or has a systemic hold/quality issue.
The fix: Reduce WIP at Station 2 by resolving the root cause of the queue buildup (likely material shortages or engineering holds). If Station 2’s WIP drops from 3 to 1, total system WIP drops to 5, and system CT drops to 5 ÷ 0.1 = 50 days — a 20-day improvement without changing throughput, headcount, or shift structure.
The failure mode this illustrates: Management often looks at this value stream and says “we need to speed up Station 2.” But the queue at Station 2 may not be caused by Station 2 being slow — it may be caused by upstream quality escapes, missing parts, or engineering holds that park units in the queue. Speeding up Station 2’s processing time does not help if the units in its queue are waiting for disposition, not processing. Diagnose the queue before prescribing the solution.
Using Little’s Law to Commit to Delivery Dates Honestly
Most aerospace delivery commitments are made by looking at the schedule, counting the operations remaining, summing the standard times, and adding a “buffer.” This method is wrong because it ignores queue time — which is typically 80–90% of total lead time.
Little’s Law gives you an honest delivery commitment method:
Count your current WIP
Walk the floor. Count every unit in every state. This is your actual WIP number — not the ERP “open work order” count (which often misses units on hold, in rework, or in inter-operation storage).
Measure your actual throughput
Use the last 4–8 weeks of shipping data. How many units actually exited the system per day/week? Do not use theoretical capacity — use observed output.
Calculate cycle time
CT = WIP ÷ TH. This is the average time a unit currently spends in your system. Any new work order released today will take approximately this long to exit.
Quote the honest number
If CT = 25 days, tell your customer 25 days (or 25 + a small buffer for variability). Do not tell them 12 days because “the standard time is 12 days of processing.” They will not receive it in 12 days. They will receive it in 25 days. Promising 12 and delivering 25 is worse than promising 25 and delivering 25.
The beauty of this method is that it also shows you how to improve. If you want to promise 15 days instead of 25, Little’s Law tells you exactly what must change: WIP must drop to 15 × TH. If TH = 20 units/day, then WIP must drop to 300 (from whatever it currently is). You now have a specific, quantified target for WIP reduction — not a vague aspiration to “improve flow.”
The Financial Cost of WIP: Inventory Carrying Cost
Every unit of WIP in your system is money sitting on the floor doing nothing. In aerospace, where individual parts and assemblies carry significant material and labor value, the inventory carrying cost of WIP is substantial.
The standard formula for annual inventory carrying cost is:
📐 Inventory Carrying Cost
Annual Carrying Cost = Average WIP Value × Carrying Cost Rate
The carrying cost rate in manufacturing is typically 20–30% of inventory value per year. This includes the cost of capital (what you could earn if the money were invested elsewhere), storage space, insurance, obsolescence risk, and handling.
| WIP Scenario | Avg Units | Avg Value/Unit | Total WIP Value | Annual Carrying Cost (25%) |
|---|---|---|---|---|
| Current state (bloated WIP) | 600 | $8,500 | $5,100,000 | $1,275,000/year |
| After CONWIP implementation | 250 | $8,500 | $2,125,000 | $531,250/year |
| Annual savings | $2,975,000 freed | $743,750/year |
The WIP reduction from 600 to 250 units frees nearly $3 million in working capital and saves $744K per year in carrying cost — in addition to the lead time reduction from 24 days to 10 days. This is money that was invisible on most management dashboards because WIP is treated as an asset on the balance sheet, not as the cost that it actually represents.
⚠️ WIP Is Not a Measure of Productivity
WIP is a measure of system congestion. A facility with high WIP is not “busy” — it is congested. Just as a highway with bumper-to-bumper traffic is not moving more cars than a highway with smooth flow at 60 mph, a production system with excessive WIP is not producing more output than one with controlled WIP. It is producing the same output, slower, at higher cost, with worse predictability.
🎯 The Bottom Line
Little’s Law is the foundation of every production control decision you will make as a Process Architect. CT = WIP ÷ TH. Control WIP, and you control lead time. Control lead time, and you control delivery promises. Control delivery promises, and you control customer confidence, program reputation, and contract retention. The next guide — Kingman’s Equation — explains the second half of the physics: why the queues that inflate your WIP form in the first place, and what utilization and variability have to do with it.
Stop reading, start modeling
Model your process flow, run simulations, optimize staffing with TOC math, and test your knowledge with 107 interactive checks — all in one platform.