Somebody proposes renting a small warehouse up-country. Everything goes north on full trailers, gets broken down there, and goes out to customers on small trucks. The transport manager likes it. Finance asks why we would pay to unload and reload our own goods in the middle of a journey. Both are right, and neither has a number.
The question has an answer, and it is not a matter of taste. It is one subtraction. This is how to do it, and what three Thai operations found when they did.
A hub is one thing: an extra unload and reload
Strip away the word. Whether it is called a distribution centre, a cross-dock, a trans-shipment point or just a yard with a forklift, a hub does exactly one thing to a shipment. It stops it, takes it off one vehicle, and puts it on another.
That is a real cost. Labour, equipment, floor space, time, and one more opportunity to damage a pallet. Nobody does it for fun.
So the whole question is what the extra handling buys, and there is only one thing it can buy: it lets each part of the journey use a vehicle that suits it.
Freight moves in two shapes. The long leg wants to be one big vehicle, full, going to one place, and coming back with something on it. The short leg wants to be a small vehicle going to a lot of places. These are opposite requirements, and a single truck driving from your factory to a customer four hundred kilometres away has to be both at once. It is usually neither.
What the handling buys, in one number
The clearest measurement of this comes from a Thai company supplying its own retail branches in Cambodia. Goods were bought in Thailand and sent straight to each branch through the Poipet crossing, with every branch ordering separately.
One outbound run cost 13,700 baht and carried 2,500 kg. That is 5.48 baht a kilogramme. The truck then came home empty.
With a distribution centre in the middle of the network, the return leg had something to carry. The carrier priced the backhaul at 60% of the outbound rate, so 8,220 baht. A round trip loaded both ways therefore cost 21,920 baht and moved 5,000 kg:
| Cost | Load | Per kilogramme | |
|---|---|---|---|
| Direct, empty return | 13,700 | 2,500 kg | 5.48 baht |
| Round trip, loaded both ways | 21,920 | 5,000 kg | 4.38 baht |
A fall of 1.10 baht a kilogramme, or 20%, on a route where not one kilometre changed.
Look at what actually happened there. The hub did not shorten anything. It did not make the driver faster. It created a reason for the vehicle to be full on a leg where it had previously been empty, and the carrier passed back most of the value of that. That is the mechanism, and it is the same mechanism at every scale.
The beverage operator described below runs it the same way at the other end of the size range. Its inbound legs, from fifteen factories, move on 18-wheel and 22-wheel trucks only. Its outbound legs use 4-wheel, 6-wheel, 10-wheel and 22-wheel, chosen against the drop. One network, two completely different vehicle policies, separated by a handling operation.
The two legs move in opposite directions, and only the net matters
This is the part that gets people into trouble, because whoever is arguing tends to quote the leg that suits them.
A beverage logistics operator in Thailand had regional distribution centres at Lampang, Khon Kaen, Nakhon Ratchasima, Surat Thani and Chonburi, plus a modern trade distribution centre, plus three provincial warehouses at Ayutthaya, Saraburi and Lopburi whose delivery areas overlapped that centre. The proposal was to close the three provincial warehouses and run everything through the one main centre: fewer stopping points, longer final deliveries.
Four months of its own data gave this:
| Now | One main centre | Change | |
|---|---|---|---|
| Inbound, factories to the centres | 96,156,821 | 76,688,018 | β19,468,803 |
| Outbound, centres to customers | 140,005,175 | 148,243,406 | +8,238,231 |
| Total transport | 236,161,996 | 224,931,425 | β11,230,571 |
All in baht a year.
Read the two legs against each other. Taking a layer out of the network cut the trunk leg by 20% and put the delivery leg up 5.9%. Anyone quoting only the first number would have claimed a nineteen million baht saving. Anyone quoting only the second would have killed the project. The answer was eleven million, and it was neither person's number.
Add a hub and the same two arrows point the other way: the long leg gets cheaper, the short leg gets more expensive, and you pay for handling on top. The hub is worth it when the fall on the long leg is bigger than the rise on the short one plus the handling. That is the entire decision, and everything below is a way of estimating those three quantities before you sign a lease.
For that operator the transport saving was not even the main prize. Closing the three provincial warehouses also took out 9,845,435 baht a year of rent on 6,967 square metres and 45 staff worth not less than 8,100,000 baht a year, bringing the total to 29,776,006 baht a year. The transport number was 38% of it.
The break-even is a drop size, not a distance
Here is the practical version, and it is worth doing on paper before anybody visits a site.
Everybody argues about distance. Distance is not what decides it. What decides it is how big each individual delivery is, because that is what determines whether a direct vehicle goes out full or half empty.
Take a factory in the central region and customers clustered around a north-eastern city about 450 km away, 36 tonnes a week between them. Put in your own quoted rates for that lane; these are round numbers for illustration only.
| Vehicle | Payload | One-way rate for the lane |
|---|---|---|
| 4-wheel | 2 t | 9,000 |
| 6-wheel | 6 t | 12,000 |
| 10-wheel | 15 t | 18,000 |
| Trailer | 25 t | 24,000 |
Through a hub, the week costs the same however the 36 tonnes are split up: two trailer loads north at 24,000 (48,000), handling in and out at 300 baht a tonne (10,800), and six local vehicle-days at 4,500 (27,000). 85,800 baht a week, or 2,383 baht a tonne.
Direct, one dedicated vehicle per customer, using the smallest vehicle that carries the drop:
| Customers | Each drop | Direct vehicle | Direct, per week | Direct per tonne | Hub per tonne |
|---|---|---|---|---|---|
| 3 | 12 t | 10-wheel | 54,000 | 1,500 | 2,383 |
| 6 | 6 t | 6-wheel | 72,000 | 2,000 | 2,383 |
| 9 | 4 t | 6-wheel | 108,000 | 3,000 | 2,383 |
| 18 | 2 t | 4-wheel | 162,000 | 4,500 | 2,383 |
| 36 | 1 t | 4-wheel | 324,000 | 9,000 | 2,383 |
Same province, same distance, same 36 tonnes, and the answer flips completely. The crossover sits between seven and eight customers, which is an average drop of somewhere between 4.5 and 5.1 tonnes.
That number is not a coincidence, and it gives you the rule:
The hub starts to win at the drop size where your direct vehicle stops being full.
On this lane a six-wheeler is the smallest vehicle worth sending 450 km, so the crossover lands just under its payload. On a shorter lane, where a direct run is cheap, the crossover moves down and direct wins further. On a longer lane it moves up, and drops have to be bigger still before direct is worth it.
Two honest caveats on the table. The local leg is priced as six-tonne vehicle-days; for twelve-tonne drops you would use a bigger local vehicle and the hub column improves, but not nearly enough to catch a direct ten-wheeler that is already full. And the trailers north are only running at 72% of payload, which is what happens when your volume does not divide neatly, and is a real cost of the hub rather than a flaw in the sum.
There is also a version of this that catches new shippers. If your drop is too small to fill any truck, "direct" was never on the menu. A one-tonne consignment sent 450 km is not going on its own vehicle; it goes on somebody's consolidated service, which is a hub. The only question left is whether it is your hub, chosen and located by you, or a carrier's, chosen and located by them.
What the table leaves out
Three things, all of which have made a hub look better on paper than it turned out to be.
The handling is time, and time is the constraint
At the beverage operator's main centre, an arriving vehicle took an average of 38 minutes from reaching the site to the goods being put away, with seven staff on the job. Dispatch took 34 minutes from a vehicle arriving to it leaving, again with seven staff.
Just over an hour, twice-staffed, per vehicle passing through. That is the honest price of the word "cross-dock". It also has to be inside somebody's working day at both ends, and it is why a hub with one dock and a shared forklift stops being a hub the moment two trucks arrive together. If your own sites already release vehicles slowly, assume the new one will too until you have measured it.
The hub has its own bill, and it eats a chunk of the saving
Thailand Post Distribution's Chumphon centre serves the southern provinces, much of it on 4-wheel trucks and some of it with no route planning at all. Simulating a second centre at Kok Kloi in Phang Nga moved transport and distribution cost from 983,955.82 to 494,346.84 baht a month, a fall of just under half, with 1,868.38 fewer kilometres a day.
Then the study did the thing most proposals skip: it added the new centre's own cost. Model 2 came out at 527,728.98 baht a month all in.
Still a large win, and the point is not that the hub failed. The point is that the transport saving and the answer are two different numbers, and the gap between them is the rent, the staff, the equipment and the racking. A proposal that shows you only the first is not finished.
Stock is the cost that does not appear until later
A hub that holds inventory rather than passing it straight through adds a stocking point, and stocking points multiply safety stock. That is a slower and larger effect than anything in the tables above, and it is the reason network decisions are not really transport decisions. It is covered in full in where to put a warehouse so trucking costs stay low.
Cross-docking avoids most of it, which is exactly why people reach for the word. But a genuine cross-dock only works if everything arrives to a schedule tight enough to be sorted and gone the same day. It also has a habit of pushing the inventory upstream rather than removing it, because suppliers hold more to feed it. If your inbound is unreliable, what you have designed is not a cross-dock. It is a warehouse that has not admitted it yet.
Two Thai constraints that decide it before the money does
The driving day
Under Section 103 bis of the Land Transport Act B.E. 2522, in a twenty-four hour cycle a licensed driver may not drive more than four consecutive hours from the moment of starting. After a break of at least half an hour, up to four more consecutive hours are allowed. The section is expressly subject to the labour protection law, which governs the working day on top of that.
So the Act's own ceiling is eight hours of driving in a day, arranged in two blocks with a break between them.
Now put a 450 km lane against it. That is roughly six to seven hours of driving one way in normal conditions, which fits, and which leaves almost nothing. There is no version of that day that also includes a multi-drop delivery round at the far end. Two legal driver-days are needed whatever the cost sheet says.
A hub does not merely make this cheaper. It makes it possible, by turning one impossible day into a trunk day and a delivery day that can run in parallel, with different drivers, on different vehicles. That is worth checking before the money argument starts, because if the direct version does not fit inside a legal day, the money argument is about an option you do not have.
Whether the far end has freight coming back
The Cambodian case above saved 20% a kilogramme purely because the return leg found a load. That is not a general law. It is a fact about a particular corridor.
Thai regional flows are not balanced. Some lanes run heavy outbound and thin inbound, and on those a hub gives you a place to consolidate a return load that would otherwise not exist. On others there is nothing to bring back regardless, and the trunk leg is priced as a round trip whatever you do. Ask your carriers what they currently carry back from the region you are considering, before you assume the saving. The answer changes the trunk rate more than the building does.
The test, in five lines
- Work out your average drop size into the region, in tonnes or pallets. Not your total volume. The average individual delivery.
- Ask what the smallest sensible direct vehicle for that lane is, and whether your average drop fills it. If it does, direct probably wins and no amount of handling will change that. If it does not, the hub is live.
- Price both legs separately and subtract. The fall on the trunk leg has to beat the rise on the delivery leg plus handling. One leg on its own is half an answer.
- Add the hub's own annual bill to the losing side: rent, staff, equipment, and stock if it holds any. The transport saving is not the saving.
- Check the direct version fits inside a legal driving day. If it does not, you were never comparing two live options.
What to change on Monday
- Work out the average size of one delivery into your furthest region, and put it next to the payload of the smallest vehicle you send there. That single comparison decides most of this
- Ask each of your carriers what they carry back from that region today. A lane with a return load prices completely differently from one without
- If somebody brings you a hub proposal, ask for the delivery leg number as well as the trunk leg number. Refuse to read one without the other
- Ask for the hub's own annual cost on the same page as the transport saving, not in a separate appendix
- Measure how long a vehicle takes from arriving to leaving at your existing sites, and use that number for the new one rather than an assumption
- Time your longest direct run against eight hours of driving in two four-hour blocks. If it does not fit today, it is already costing you two driver-days without a hub
- Before designing a cross-dock, check whether your inbound arrives to a schedule. If it does not, you are designing a warehouse
- If your drops into a region are under a tonne, stop calling the current arrangement direct. You are already using somebody's hub, and you should find out whose and where
None of this says a hub is good or bad. It says the decision is a subtraction between two legs that always move in opposite directions, and that the number people quote first is almost never the one that settles it. The same cost per tonne can be true and useless at the same time if it is measured on one leg of a two-leg network, and the same load can be cheaper combined or cheaper split depending on nothing more exotic than how big each delivery is.