A yard is not slow or fast in the abstract. It is either keeping up with the line or it is not. On a final assembly plant building 1,000 vehicles a day across two shifts, takt is roughly 57 seconds per unit — which means the yard has to deliver a correct trailer to a correct door on a clock that never pauses and never negotiates. Most plant yards were never designed against that clock. They were designed against space, then grew, then absorbed more volume, and the discipline that holds them together is a shunter driver who has been there eleven years and knows where everything is. This guide covers the three things that decide whether a yard runs at production rate: knowing the state of every trailer, staging against the sequence window rather than against arrival order, and stopping congestion before it reaches the gate — because once the queue is on the access road, you have already lost the shift. Request an enterprise demo to see the model running against your own yard layout.
2026 OPERATIONS GUIDE · PLANT YARD
Running a Vehicle Plant Yard at Full Production Rate
Trailer state as a live model, staging discipline enforced against the frozen sequence window, and congestion intercepted before the gate — with the KPI targets that separate a yard keeping pace from one quietly setting the line's ceiling.
Start With the Rate, Not the Yard
Every yard decision is downstream of one number. Before discussing dwell targets or spotter headcount, work out what the line is actually demanding per hour — because that is the demand signal the yard is either meeting or failing, and it is the only honest basis for sizing anything.
The number everything hangs off
Available shift secondsUnits required
≈ 57 sec takt at 1,000 units/day, two shifts
At that rate a station consumes a part roughly every minute. A 90-minute JIT buffer is 90 units of protection — and nothing more.
Frozen sequence window60–120 min
Typical JIT line buffer90–120 min
Frozen zone to line-side, full chain4–8 hrs
OEM delivery window granularity30 min
Gate delay that can reach the line15 min
Premium OEMs now score supplier on-time delivery in 30-minute bands — early, on-window, and late are three different results, not two.
Trailer State: The Model the Yard Actually Runs On
Ask a plant how many trailers are on site and you get a number. Ask what state each one is in and you get a walk around the yard. Every functioning yard operates an implicit state model — it is just held in a shunter's head rather than in a system. Making it explicit is the single highest-leverage change available, because a state that is written down can be measured, alerted on, and handed over at shift change.
Announced
Carrier has confirmed an ETA against a booked window. Not yet on site.Exit condition: arrival at gate queue, or window breach triggering a reslot.
Queued
On the access road or in the holding lot, not yet through the barrier.Exit condition: gate call-forward. This is the state most plants cannot see at all.
Checked in
Through the barrier, identity and load confirmed, awaiting a yard position.Exit condition: parking assignment issued to the shunt driver.
Staged
Parked in a known bin, contents known, waiting on a demand signal.Exit condition: called to a door against the sequence window — not against arrival order.
At door
Spotted at an assigned dock, unload or load in progress.Exit condition: unload complete and confirmed, door released.
Empty pool
Unloaded, awaiting collection, returnable containers reconciled or not.Exit condition: carrier collection, or conversion to an outbound load.
Unknown
On site, no confirmed position, no confirmed contents. Every yard has these.Exit condition: someone walks the yard. Count these weekly — the number is your real maturity score.
The one metric nobody publishes
Count trailers in the Unknown state at the end of each shift. Not dwell, not turnaround — just how many assets are physically on your site whose position or contents cannot be confirmed from a screen. It is the cleanest single indicator of whether a yard is managed or merely populated, and it is the number that improves first when a state model goes live.
How a Fifteen-Minute Gate Delay Becomes a Line Stop
The reason yard problems get under-invested is that they present as small. A truck waits. That is all anyone sees. What follows is the escalation path that turns a queue into a stoppage, and the reason a fifteen-minute delay at a JIS plant gate is not a fifteen-minute problem.
Gate queueTrailer arrives inside its window but waits behind unscheduled arrivals
Staging missTrailer is not in position when the sequence window opens
Door contentionPriority unload competes for a door already committed to a lower-priority load
Sequence breakKit cannot be built in order; the choice is a manual sort or a gap
Line stopStation starves, and everything behind it in weld, paint and final backs up
Each step in that chain is individually cheap to fix and collectively expensive to ignore. Note that four of the five happen in the yard, and only the last one is visible on a production report — which is exactly why yards get diagnosed as a production problem for years before anyone instruments the gate.
Most plants can name the line stop. Very few can name which of the four upstream steps caused it.
An enterprise demo runs your yard layout, gate volumes and sequence windows through the same cascade, and shows where your site's chain actually breaks.
Staging Against the Sequence Window
This is the discipline that separates a plant yard from a distribution yard. In a DC, staging order is a throughput optimisation. In a vehicle plant, staging order is a contract with the build sequence — the plant publishes a frozen window covering the next 60 to 120 minutes of takt, and everything upstream is obliged to serve that window in order. A trailer that arrives early and parks in the wrong place is not neutral. It is occupying capacity that belongs to the window.
One sequence window, four hard checkpoints
T-120Window freezes. Build order published. No further sequence changes accepted.
T-90Required trailers must be in Staged state with position and contents confirmed.
T-45Call-forward to door in window order. Door assignment scored on demand, not arrival.
T-0Window opens at the line. Anything not present is now a sort or a stop.
The checkpoints matter more than the exact timings, which vary by plant and by commodity. What does not vary: if staging is confirmed only at T-0, you have no recovery time. Every hour of visibility you move earlier converts a stoppage into an expedite, and an expedite into a non-event.
Congestion Prevention Starts Outside the Fence
Once trucks are queuing on the access road, every available option is bad. The gate cannot process faster than its physical throughput, the yard cannot absorb arrivals it has no space for, and the plant is now negotiating with a queue rather than managing a plan. Congestion control is therefore an arrival-shaping problem, and it is solved in four layers — three of which happen before a vehicle reaches your barrier. Our dispatch and load scheduling capability covers the carrier-side half of this.
Layer 1
Arrival shaping at booking
Slots issued against demonstrated dock and yard capacity per hour, not against a static timetable. Slot density flexes with the build plan, so a high-variant shift does not get the same arrival profile as a steady one. Appointment adherence in the 75–85% range within a ±15-minute window is a realistic target; anything materially below it means the slots themselves are wrong.
Layer 2
In-transit ETA and reslotting
A carrier running forty minutes late should be reslotted while still on the motorway, not discovered at the barrier. This is the layer that most directly prevents the cascade, because it converts a late arrival from a gate event into a planning event — and planning events are cheap.
Layer 3
Off-site holding and call-forward
A holding lot outside the fence with a call-forward instruction to the driver's handset. Early arrivals wait somewhere that costs nothing rather than somewhere that consumes yard capacity, and the yard only ever holds trailers it has an active reason to hold. This is the single most effective structural fix for a yard running above 80% space utilisation.
Layer 4
Gate throughput itself
Only now does barrier speed matter. Pre-matched documentation, automated identity capture and dedicated in and out lanes should hold a gate transaction to three to five minutes. Optimising this layer while ignoring the three above is the classic mistake — a fast gate feeding an overloaded yard just moves the queue inside the fence.
Benchmarks Worth Holding a Yard To
Published ranges, with the caveat that any plant yard running JIS should be targeting the tight end of every band. Baseline these before changing anything, because the argument for investment is only ever as good as the number you started from. Our analytics and reporting module is where these get tracked over time.
← Swipe to see all columns →
Set targets by percentile, not average
"Average dwell is two hours" hides the trailers that sat for nine. Those outliers are what generate detention invoices, carrier complaints and sequence misses — averages are structurally blind to exactly the events you are trying to eliminate. Write the target as "90% of trailers dwell under four hours" and the conversation immediately moves to the worst cases, which is where all the recoverable cost lives.
Yard Maturity, Honestly Assessed
Five levels. Most plant yards sit at Level 1 or 2 and describe themselves as Level 3. Read down and stop at the first description that sounds uncomfortably familiar — that is your actual level, and the level above it is your next twelve weeks of work.
L0ReactivePaper gate log, radio dispatch, trailer position held in memory. Shift handover is a conversation. Nobody can answer a location question without walking.
L1RecordedA spreadsheet exists. It is accurate at the moment it is updated and degrades from there. Dwell is calculable after the fact but not visible during the shift.
L2ScheduledAppointments are booked and mostly honoured. Gate knows who is expected. Yard positions are assigned but not verified, and the Unknown count is real but uncounted.
L3StatefulEvery trailer holds a live state with a timestamp. Shunt tasks dispatch to a handset. Dwell clocks run in real time and alert on threshold. Handover is a screen, not a conversation.
L4SynchronisedStaging and door assignment are scored against the frozen sequence window. Arrivals are shaped in transit. Off-site holding absorbs earliness. The yard has recovery time built in rather than borrowed from the line.
See Level 4 Running Against Your Yard
An enterprise demo is built around your site: your gate volumes, your yard layout, your sequence windows and your current KPI baseline. We show the trailer state model, the staging logic and the pre-gate control layers operating on your numbers rather than a generic dataset.
Live trailer state model
Sequence-window staging
Pre-gate arrival shaping
Runs on existing handsets
Frequently Asked Questions
Why is a plant yard different from a distribution centre yard?
Because the consumer is a moving assembly line rather than a picking wave. A DC can absorb a late trailer by resequencing work later in the day. A plant running just-in-sequence cannot — the build order is frozen, delivery timing is measured in minutes, and the line does not wait. That difference means plant yard staging has to be scored against the production sequence, while DC staging can legitimately be scored against throughput. Applying DC logic to a plant yard is the most common structural error we see.
Our yard has enough space. Is congestion still a problem?
Space and congestion are only loosely related. A yard at 60% utilisation still congests if arrivals cluster, if door assignment is reactive, or if shunt drivers spend their shift searching rather than moving. Conversely, once sustained utilisation passes about 80%, every move gets longer regardless of how well the yard is run, because there is no slack to shuffle into. Treat space as a constraint you should stay under, not a problem you have solved by having some.
What should we measure first?
Trailers in the Unknown state at shift end, then gate-to-gate turnaround at the 90th percentile. The first tells you how much of your yard is genuinely visible; the second tells you how bad your worst days are without letting averages hide them. Both can be baselined in a fortnight with no system change, and both move measurably in the first phase of any competent deployment — which makes them the right pair to build a business case on.
Do we need an off-site holding lot?
Not always, but you need somewhere for earliness to go. If your yard runs comfortably under 75% utilisation and arrivals are well shaped, in-yard buffering is fine. If you are regularly above 80%, or if a meaningful share of arrivals turn up more than an hour early, holding capacity outside the fence pays for itself quickly — it costs less per square metre than yard space and it stops early arrivals consuming positions that the sequence window needs.
How do we handle carriers who will not adopt appointment discipline?
Measure them, publish it, and price it. Adherence in the 75–85% band within a fifteen-minute tolerance is achievable across a mixed carrier base, but only when performance is visible per carrier and fed into commercial reviews. The other half of the fix is your own: carriers disengage from slot systems that issue windows their transit times cannot meet. Before escalating, check whether your slot plan is actually achievable from their origin points.
Does this require new gate hardware?
The first three control layers — arrival shaping, in-transit reslotting and off-site call-forward — need no gate hardware at all. They run on booking data, carrier ETAs and driver handsets. Fixed infrastructure like cameras, RFID portals or automated barriers only becomes worth costing once the arrival profile is under control, because a faster barrier feeding an unshaped arrival curve simply relocates the queue from outside the fence to inside it.
How long before the line feels the difference?
Sequence-related interruptions attributable to yard causes are usually the last metric to move, because they sit at the end of the cascade. Gate queue length and trailer Unknown counts move within weeks; turnaround and dwell percentiles within a quarter; sequence protection once staging is genuinely scored against the frozen window rather than arrival order. Plan the business case around the early metrics and treat line protection as the compounding return. A
demo is the fastest way to put timings against your own site.
Bring Your Yard Layout. We Will Bring the Model.
Trailer state at production rate, staging enforced against the frozen sequence window, and congestion intercepted before it reaches your barrier — demonstrated against your gate volumes, your dock count and your takt, not a generic benchmark set.
No gate hardware required to start · Runs on existing handsets · Site-level configuration
August 11, 2026By Derek Goes
All Blogs
