A truck is loaded in forty minutes. It spends three hours at your site. Nobody on the dock thinks they are slow, and nobody is lying.
Both statements are true because a loading dock is a queue, and queues do not behave the way people expect. The wait does not grow in step with how busy the dock is. It grows slowly, then suddenly, and the last two trucks of the day do most of the damage.
This article is about your own site, where you control every minute: the outbound dock your customers' and carriers' trucks collect from, and the receiving dock your suppliers deliver to. What the hours cost the carrier, and how the charge should be written, is covered in what a waiting time charge is and how much waiting is reasonable. Here the question is how to get the hours back.
First, find out where the three hours actually are
"The trucks wait" is not one problem. A truck's visit is a chain, and it can stall at any link:
| Stage | What the truck is waiting for |
|---|---|
| Gate to door | A free door, or a free crew |
| At the door, before loading | The goods: not picked, not packed, not in the right order |
| Loading | The work itself |
| After loading | Documents, a count, a signature |
| Door to gate | Other vehicles, a single exit, security |
Each stage has a different fix, and the fixes cost very different amounts. A second loading crew does nothing for a truck waiting on paperwork. A faster paperwork process does nothing for a truck waiting on a pallet that has not been packed.
Researchers at Chulalongkorn University showed how easy it is to fix the wrong stage. At the busiest warehouse of a Thai wood substitute manufacturer, the average visit from joining the loading queue to leaving was 139 minutes, and 55% of trucks took more than an hour. The two stages that looked like the bottleneck were the queue at the loading point and the loading itself: 14.9 minutes of queue and 23.9 of loading for a flatbed, 21.7 and 44.3 for a container truck.
Add those up and they come to 39 minutes for a flatbed and 66 for a container truck. The rest of the 139 minutes sat outside the two stages everyone was looking at. Keep that in mind for later in this article, because it decided which fix worked.
Why a queue explodes
Take one loading crew, a 40-minute average load, and a nine-hour loading day. Trucks turn up through the day at no particular time, and some loads take longer than others, as real loads do. Standard queue arithmetic then gives the average time a truck waits before loading even starts:
| Trucks a day | Share of the day the crew is busy | Average wait before loading starts |
|---|---|---|
| 8 | 59% | 58 minutes |
| 9 | 67% | 80 minutes |
| 10 | 74% | 1 hour 54 minutes |
| 11 | 81% | 2 hours 56 minutes |
| 12 | 89% | 5 hours 20 minutes |
Read down the last column. The ninth truck adds 22 minutes to the average wait. The eleventh adds an hour. The twelfth adds almost two and a half hours, and the crew is still idle one minute in nine.
That is the whole explanation for the three-hour dock. A crew that is busy 80% of the day is not 80% of the way to a problem. It is already in one. A dock manager looking at a crew with spare time in it sees no reason for a queue, and the queue is there anyway, because the spare minutes fall in the wrong places. The truck that arrives at 10:15 cannot use the fifteen idle minutes the crew had at 08:40.
At eleven trucks a day, that queue costs about 32 truck-hours between them, every day, on a working day that is capped by law.
The model is a simplification: it assumes arrivals scattered at random and loading times that vary in the usual way. Real docks are often worse, because trucks do not arrive at random. They arrive at 08:00.
The same arithmetic tells you which lever is strongest
Stay at eleven trucks a day and one crew, with the three-hour average wait, and change one thing at a time:
| Change | Average wait before loading |
|---|---|
| Nothing | 2 hours 56 minutes |
| Load 5 minutes faster, 35 minutes | 1 hour 27 minutes |
| Load 10 minutes faster, 30 minutes | 47 minutes |
| Book arrivals evenly through the day, trucks keep their slot | about 1 hour 28 minutes |
| Even arrivals, and loading times made steadier | about 22 minutes |
| Add a second crew | 8 minutes |
The last three rows use a standard approximation rather than an exact formula, and the booked-arrival rows only hold if trucks arrive when they are booked. The shape is what matters.
Three things fall out of it.
Near the top of the table, a small speed-up is worth a lot. Five minutes off each load halves the wait, because it pulls the crew back down the steep part of the curve. That is the opposite of how improvement projects are usually judged, where five minutes on a forty-minute job sounds like 12% and gets a 12% budget.
Evenness is worth as much as speed. Spreading the same eleven trucks across the day halves the wait without anyone working faster. Making loading times less erratic on top of that takes it down to about 22 minutes. Neither costs a forklift.
A second crew is the strongest lever on the queue, and the most expensive. It takes the wait almost to nothing. Whether it takes the visit to nothing is a separate question, and the Thai case below is the reason to ask it.
A second crew killed the queue and saved 3.5 minutes
The Chulalongkorn warehouse is a clean test, because the researchers built a model of the site on a year of its own operating records and tried four fixes against it.
The building was designed with four loading platforms for flatbeds and two for container trucks. It was worked with one loading set per truck type: two forklifts and two loaders. So however many platforms stood empty, one flatbed and one container truck could be loaded at a time. The constraint was not the doors. It was the crews.
The causes the researchers found were three: boards being re-stacked and moved during loading, which was the most frequent problem at 39%, goods not yet produced or packed when the truck was called in, and limits on equipment and crews. The four fixes tested each attacked one of them.
| Fix | Queue, flatbed | Queue, container | Whole visit | Cost |
|---|---|---|---|---|
| None | 14.9 min | 21.7 min | 139.0 min | |
| Lay the warehouse out by product type | 13.3 | 18.2 | no significant change | 66,650 baht once |
| Confirm the goods are ready before confirming the truck | no significant change | 19.6 | no significant change | Coordination only |
| Both of those | 13.1 | 16.9 | no significant change | 66,650 baht once |
| A second loading set per truck type | 1.1 | 1.3 | 135.5 min | 136,000 baht a month |
The second loading set did exactly what the arithmetic above predicts: the queue at the loading point fell by about 14 minutes for a flatbed and 20 for a container truck, to almost nothing. It cost 136,000 baht a month, 100,000 of it for four rented forklifts and 36,000 for four staff.
And the whole visit fell by 3.5 minutes.
The paper does not trace where the freed time went, so do not read more into it than the numbers say. What the numbers do say is that the queue at the platform was never where most of the 139 minutes were, and that buying capacity at the one stage everyone could see moved the wait somewhere nobody was measuring. The same thing happens on the road: a route plan that ignores the receiving gate relocates the queue rather than removing it.
The readiness fix is worth a separate look, because it is the one that costs nothing. When the site trialled checking that goods were produced and packed before telling the carrier to come, over a month from October to November 2019, visits that had been waiting on production or packing got 10% shorter. A truck called in for goods that do not exist yet is not waiting at the dock. It is waiting at the factory, and the dock is just where it happens to be parked.
Waiting for the right pallet, not a free door
A second Thai study, at a distribution centre with 42 delivery vehicles and seven loading doors, found the wait somewhere else entirely.
The researchers timed every step of getting one job from storage onto a delivery truck. The steps that looked like work were small. The step called "waiting to be loaded" was most of the job:
| Job type | Total time | Waiting to be loaded | Share |
|---|---|---|---|
| Pallet | 526 seconds | 312 seconds | 59% |
| Box | 936 seconds | 418 seconds | 45% |
| Long items | 769 seconds | 577 seconds | 75% |
There was a free door, and there was a forklift. The goods were waiting because drivers handed their paperwork to the door staff in whatever order it came, not in the order the truck would deliver. The forklift brought work to the door in the wrong sequence, and the right pallet sat behind the wrong one.
The fixes cost almost nothing:
- Drivers sort their own jobs into delivery order before handing them over
- A coloured laminated card on each pallet, yellow for the first truck queued at a door and red for the second, showing load order and queue position, with a card on the door post showing which queue is being loaded
- A marked place for empty pallets at each door, which cut the time to find one from 44 seconds to 9
- One check instead of two: the paper check against the staging sheet was dropped and the barcode scan kept
The pallet job fell from 526 to 379 seconds, the box job from 936 to 640, and the long-item job from 769 to 519. That last one is the telling result. None of the long-item job's own handling steps were changed, only the queue system at the door it shared with everything else. It still got 32% faster, because the jobs around it stopped blocking the door.
That is the second lever in the table above, evenness, showing up in real life. Nobody loaded faster. The work stopped arriving at the door in a scramble.
Receiving is the same problem pointing the other way
Everything above applies to supplier trucks at your receiving dock, with two additions.
Book inbound loads in advance. The purpose is not tidiness. A booking is what lets you have a crew, a door and a space ready for a load before it arrives, rather than discovering it at the gate. That is standard practice in large warehouses for exactly that reason.
Check against what was sent, not against what was ordered. Checking each delivery line by line against the purchase order is slow, and it happens with the truck at the door. The faster alternative is to check the goods against the supplier's advance shipping notice for that vehicle.
What to do, in order of cost
1. Put six times on every truck for two weeks. Gate in, arrival at the door, loading start, loading end, documents in the driver's hand, gate out. Do it for inbound and outbound separately. Most sites already have three of the six on a gate log or a system and have never lined them up.
2. Split the visit into its stages and find the biggest. Waiting for a door, waiting for goods, loading, waiting for paperwork. Do not start a project until you know which one it is. At the Chulalongkorn site, the stage everyone could see was not the stage that held the time.
3. Work out how busy each crew really is, hour by hour. Trucks in the hour, multiplied by average loading minutes, divided by the crew minutes available. Anything above about three quarters means long waits are not bad luck, they are arithmetic.
4. Do not call a truck for goods that are not ready. Confirm the carrier only when the load is picked, packed and staged. This is the cheapest hour on the list.
5. Put the work in load order before the truck arrives. Sequence by the delivery plan, mark the staging lanes by door and queue position, and keep empty pallets at the door.
6. Remove the second check. One scan or one count. A load checked twice is a load that waits twice.
7. Book the day instead of letting it happen. Spread collections and deliveries across the crew's hours rather than letting everyone arrive at opening. Then hold both sides to it: a slot the site does not keep is not a slot.
8. Only then price a second crew, and judge it on the whole visit. Put its monthly cost against the truck-hours it returns. If the queue at the door falls and the gate-in to gate-out time does not, the money bought a shorter queue and not a shorter visit.
The one line worth keeping
A dock does not start queueing when it is full. It starts queueing when it is busy, and at 80% it is already losing hours. The fastest way back is usually not another forklift. It is making sure the right goods are at the right door, ready, when the truck you called actually arrives.