Building on my previous thread and taking into account suggestions Long distance travel by non-motorised rolling vehicles

I wish to consider two-way rivers that flow in opposite directions at the same time.


There is a high fertile plateau surrounded by flat scrubland. Rain is plentiful on top of the plateau and waterfalls occur around its edges. At the moment, these falls create rivers that flow away from the plateau in various directions, ending at the sea. This is great for transport by boat when exporting crops from the plateau, however it is hard work going upriver when the boats are bringing imports to the plateau.

The proposed system

A trial is to be performed with the biggest of the rivers. It will flow to the sea as usual but, before it gets there, it will be sent on a long diversion. Canals are to be built that carry the river to outlying cities and back.

Boats do not need to be powered, they all travel with the flow of the river.


I know about canals that currently exist on Earth but as far as I know, the flow is prevented by locks and any flow is one-directional. Has anyone used this proposed system before? If so where and when. If not why not?

enter image description here

  • $\begingroup$ If your city is on the top of the plateau, but the canals only go to the bottom of the plateau, then you've still left yourself a lot of work. $\endgroup$
    – Cadence
    Mar 10, 2023 at 22:43
  • 2
    $\begingroup$ A boat which moves with the current cannot be steered and will eventually run aground. In order to be able to steer the boat it must maintain a certain minimum speed relative to the water. (And a river needs a mimum slope in order to flow. There would have to be at least a difference of 100 meters in elevation between the outgoing and incoming 50 mile canals at the point where they touch the green area. That is a lot of dirt to be moved.) $\endgroup$
    – AlexP
    Mar 10, 2023 at 22:44
  • 1
    $\begingroup$ When you say "bidirectional river", do you mean both directions within the same channel, or do you mean two separate channels separated by some distance of land? If the latter, why are you playing a sort of semantics game? The picture just looks like one whopping long canal, and that is perfectly feasible! $\endgroup$
    – elemtilas
    Mar 11, 2023 at 1:35
  • 2
    $\begingroup$ @chasly-supportsMonica: Water is not liquid helium; it does have a certain viscosity. As the slope becomes closer and closer to zero, the speed of the flow reduces so that other effects become more important. For example, the average local sea level at Suez is about 1.5 meters higher than at Port Said -- yes, the eastern Mediterranean is a little below the global mean sea level. But over the 200 km of the Suez Canal this difference is height is negligible, and water flow in the canal is determined more by the rate of evaporation in the Bitter Lakes. $\endgroup$
    – AlexP
    Mar 11, 2023 at 6:48
  • 1
    $\begingroup$ Regarding the steering of such "passive" boats, you can check how belyanas (en.wikipedia.org/wiki/Belyana) used to be steered. They were (very) large boats carrying wood down the large russian rivers, steered mostly with anchors $\endgroup$
    – Axel B
    Jul 10, 2023 at 7:49

6 Answers 6


No, no one has used this system before on a scale larger than that of the Lazy River at a water park.

Why? Because you need one of two situations to make this work:

  1. A perfectly level surface along the entire length of your canal system, with pumps to keep the water moving in the correct direction rather than flooding out near the inlet from the natural river.
  2. A consistent grade along the entire length of your canal system, which will result in a significant altitude difference between the outgoing and incoming sides, and which would require a ridiculous amount of civil engineering to construct, because that will not happen naturally.

In real-world canal systems, locks do not exist primarily to prevent flow, although they do do that to some extent. Locks exist to handle elevation changes, precisely so that you don't have to do the unreasonable amount of grading and earthmoving needed to get a navigable slope along their entire length.

The solution to your problem is to build canals with locks to each of your cities, and use pack animals to pull barges along said canals. You also save labor by building a single double-wide channel rather than two separate channels.

  • $\begingroup$ Why is a grade or pump needed at all? Place a length of guttering flat on the ground; stop one end and leave the other end open. Pour water in the stopped end and it will flow towards and out of the open end. The water creates its own grade Spillage can easily be dealt with - it can be used for crops at the base of the plateau. Or surplus water can be diverted down the old dried up river bed ay source. $\endgroup$ Mar 10, 2023 at 23:07
  • 4
    $\begingroup$ @chasly-supportsMonica: To calculate the steady-state flow depending on the slope and cross section of the canal, use the Manning equation. At what slope does it become faster to walk to the destination? At what slope does the speed of the flow become insufficient for the canal cannot absorb the incoming flux of water so that the water will prefer spilling over the banks instead of flowing down the canal? $\endgroup$
    – AlexP
    Mar 11, 2023 at 7:02
  • $\begingroup$ I wrote an answer only to realize that it was no different from this answer. Spillage isn't ignorable over hundreds or thousands of kilometers unless you want to ground your ships mid-route. Your world would, presumably, experience drought and have water flow and availability limitations. But more to the point, the complexity of the lock system combined with artificial water ways would make this the slowest transportation system on your world. Walking the route with pack mules would be faster and maintenance would be debilitating. So, Logan... +1. $\endgroup$
    – JBH
    Mar 11, 2023 at 7:58
  • 1
    $\begingroup$ @Logan R. Kearsley a consistent gradient across multi-kilometer long structures , designed to transport large volumes of water with a near-insignificant altitude difference at each end was achieved over two thousand years ago by the Romans. Aqueducts! Scaling this up to a 'river' would be far more difficult, but not insurmountable by any means. $\endgroup$ Mar 12, 2023 at 2:28
  • 1
    $\begingroup$ @JeffereyDawson: Roman aqueducts aimed to maintain a uniform slope, usually between 1 in 300 and 1 in 600; the lowest slope ever measured is an exceptional 1 in 1500. That's not "insignificant". The goal was to have enough flow to fulfill the water requirements at the destination and at the same time keep the flow low enough to prevent turbulence because that would quickly erode the walls of the water channel. (And if you think about it, it would be quite impossible to have "a near-insignificant altitude difference" -- the sources are in the mountains, the cities are in the plains.) $\endgroup$
    – AlexP
    Mar 12, 2023 at 4:07

Unfortunately, this project will be done in by its very poor economics.

Flowing water is a problem. It erodes the canal banks in some places, and carries the resulting sediment downstream. Eventually it deposits the sediments, which is even worse because it causes the river to meander and can make the waterway too shallow for safe navigation.

This can be avoided by additional engineering; a lot of additional engineering. Essentially, the river would need to be a concrete-walled culvert for at least much of its length to prevent erosion, and continuous dredging will be required to keep it navigational in the lower reaches.

Also, a river has tributaries, through which it receives water from the whole river basin. If the plateau has irregular rainfall (i.e. either a wet and dry season, or occasional heavy showers), the river will flood at some times, and the floods will introduce sediment into the channel, with all the resulting problems. At other times, the river may not carry enough water for navigation. This will be amplified by the river being a concrete culvert, as it must be to avoid erosion: all the rain will flow downstream unimpeded in one big pulse. So the river would require major flood defences all along its length to cope with these pulses, and also some form of storing potentially several months' worth of water.

Meanwhile a vanilla one-way canal avoids all these problems by not flowing, requires much less engineering in its construction as a result, and has a system of locks to control the water it contains. It would just be so much cheaper.

And on top of all that, the resulting circuitous route will make exports from the plateau more difficult, i.e. more expensive. At least a significant part of the plateau society will oppose such scheme on that basis; in particular the major landowners, who are both most likely to produce cash crops for export, and also to hold a lot of influence on account of being rich. Since the canal would need to cut through a lot of the plateau, they may well be able to block its construction, or to bankrupt the project by stalling and making things difficult in various ways.

  • 1
    $\begingroup$ I should have made it clear that most of the plain is currently uninhabited. The idea would be to build walled canals, or rather a single canal divided down the middle with a central wall plus side walls. Overflowing would not be a problem - just let the excess water flow over the sides onto the plain. What do you think? $\endgroup$ Mar 11, 2023 at 23:29
  • 1
    $\begingroup$ I don't understand why the problem of flooding caused by irregular waterfalls and the problem of sedimentation is a problem for this type of canal and not a problem for vanilla canals? My understanding is that the sediment problem depends of the nature of the ground in which the canal is built? $\endgroup$ Mar 12, 2023 at 10:16
  • $\begingroup$ Chasly - supports Monica, thank you. Unfortunately, letting the canal overflow is not quite a solution. The water is still lost from a system that requires water to work. The area being uninhabited is a new problem though, because this means it also lacks roads, and you would need some big roads to transport construction materials for a project of that size. $\endgroup$
    – ihaveideas
    Mar 12, 2023 at 10:20
  • $\begingroup$ TotalMongot, thank you. Flooding due to irregular rainfall is a problem for every watercourse which has a catchment area. The proposed canal, being a redirected river, will at the minimum have the catchment area of that river above the point of being captured; we know that this river is already big enough to be navigable, so necessarily has a catchment area to match. The sediment becomes a problem whenever the flow slows down, as it will when the river reaches the plains; this does not depend on the ground through which it flows, only the amount and type of sediment does. $\endgroup$
    – ihaveideas
    Mar 12, 2023 at 10:30
  • 1
    $\begingroup$ @ihaveideas - Transport for building materials is easy. Just use barges on what has been built of the canal so far. This was exactly how the railroads were built across the USA - use the existing train tracks. $\endgroup$ Jul 10, 2023 at 10:58

Has anyone used this proposed system before? If so where and when. If not why not?

Because digging and supporting long canals is prohibitively expensive.

And quite pointless too. Towns form in places where they have enough food supply from surrounding lands. Food surplus is what makes them possible. If there is not enough food to feed 10k extra people there will be no 10k town.

In your example there seem to be no reason for cities in the scrubland to exist. Settling these people along old river bed or near the sea just makes much more sense.

High fertile plateau surrounded by flat scrubland is generally very strange. "Rain is plentiful on top of the plateau" - hmmm. Why exactly? It smells with magic. Is it still science-based?


What you need is a series of locks. Its a bit of a 'long' form video about someone going up and down the Bingley Five Rise Locks but the system here is pretty much what you're thinking of.

Locks ARE bidirectional - just one direction at a time

Well, unless there's a ship stuck in it

While boats in real life are powered, You might have something that might allow for the water flow to pull/push the boats along. Having paired locks would allow for you do do it simultaneously and if one route is blocked, you can use the other. In real life, a common way to deal with having a long route 'one way at a time' is to have lakes in between to act as harbours for waiting ships, mainly cause one line is cheaper than two.

For steering I wonder if having a pair of rails for each loch and having some short of connection would work...

enter image description here

The big problem here though is well ... you don't actually need the water to go back to the sea.

What you kind of need is a lake on the plateau to act as a catchment, and 3 separate canals leading to each endpoint - the 2 cities and the sea. Have a few lakes in between to act as harbours in between. Needs a bit of hydroengineering but other than that its not implausible, just fairly complex

  • $\begingroup$ Interesting. Although there is no up or down in my system, unlike Bingley, I can see how a series of locks could work. I'll give it some thought. $\endgroup$ Mar 12, 2023 at 11:25

Some water in the Netherlands is used as/like a river most of the time but in times of draught the water can flow the other way around.
That is possible because of the lay of the land is very flat, no more than one meter height difference over several km length (3 feet over several mile) and the source of water into the canal either higher or forced by pumps.

To replicate this you need a completely level plane, a good source of water on either end which can be regulated to stream on request or a double canal system as in your drawing where the two canals on each stretch have to have a water proof divider between them, (like the dikes along the Dutch canals, which are often between a canal and a wet ditch or other canal on a different level.)

To make it work round the year you might want a big lake at the plateau, with locks to regulate the inflow into the canal. Water that streams downhill out of a lake does flow a lot faster than you would like. And canals with too much streaming will try to meander, breaking your dike/land between the canals and spreading out into the land on the other side or you need very strong dikes on both sides of the water.
On the other hand, the faster the flow of the water, the faster the boats will travel on it. With a very low flow rate you may still want some sort of power in the boats, like rowing to get started.

This quote from the Wikipedia page on one of the canal systems explains it, in Dutch.

Het Amsterdam-Rijnkanaal speelt een belangrijke rol in de aanvoer van water voor de regio Utrecht en Zuid-Holland in perioden van droogte. Het water van het kanaal wordt dan getransporteerd naar de Hoogheemraadschappen De Stichtse Rijnlanden, Rijnland en Delfland. Hiermee worden deze gebieden van zoet water voorzien, terwijl ze anders in droge perioden afhankelijk zouden zijn van verzilt water in de directe omgeving.

In de droge zomer van 2003 is extra zoet water vanuit het Markermeer via de zogenaamde Tolhuissluisroute (Singelgracht - Amstel - Amstel-Drechtkanaal - Tolhuissluis) naar de Rijnlandse boezem in het Aarkanaal gevoerd om in het tekort aan zoet water te voorzien en om zout water te kunnen verdrijven. Ook werd in juli van dat jaar het peil van de kanalen met 5 centimeter opgezet om de wateraanvoer naar de boezemwateren van bovengenoemde waterschappen te bespoedigen.1

Shortened translation, as far as I understand the matter, and I am not a specialist: The Amsterdam-Rhine canal plays an important rol in supplying water for the region Utrecht and Zuid-Holland in periods of draught. < - > This way the area get sweet water, while otherwise they would be getting partially silt water out of the direct area

In the summer of 2003 extra sweet water out of the Markermeer is brought in via waters in Amsterdam to the area, also the water in the canals was raised by 5 cm to improve the transfer of water out into the smaller canals/wet ditches.

  • $\begingroup$ This sounds very encouraging. Do you have any links to descriptions/maps where I can get more information. I'm not sure what search terms to use. (Presumably they will be Dutch ones?) $\endgroup$ Mar 12, 2023 at 14:16
  • 1
    $\begingroup$ @chasly-supportsMonica, look at maps of the Netherlands with elevation lines on them, (you may have to search very hard to find the lines, they need to be rather close together to even find them on a map) and you can see how level most canals are. $\endgroup$
    – Willeke
    Mar 12, 2023 at 14:26

Water only ever flows one way, and that is "down". Sometimes it seems to flow a different way, this is usually some sort of illusion.

You could conceivably build these canals, so long as both angle at least a little bit down, but if you want them to be flowing constantly, you risk running out of water.

Canal based infrastructure has existed throughout history, but it typically used the water as a way to make pushing/pulling loads easier, not as a way to avoid pushing/pulling them, Because of how fidlly that would be.

You can certainly engineer around the issues with this. It's definitely doable. But most cultures would consider doing that to be a much bigger pain than just latching a couple oxen to a boat. Most people want to take the energy that would be required engineering this infrastructure, and spend it on things like building temples, palaces and monuments.

That's not to say you shouldn't have these canals, it only means it should say something important about what kind of people it was who built them.

  • $\begingroup$ Water can flow along level ground. That is how floods happen. I'm just providing a very long drainage channel. Here's a small-scale example: Place a long trough on a level surface and fill it with water. Now knock a hole in one end of the perfectly level trough and water will flow through the hole. If you continually fill the trough at the other end, water will continue to flow along the level trough. If you remove the whole of one end of the trough the water will flow faster. $\endgroup$ Mar 18, 2023 at 12:13
  • $\begingroup$ @chasly-supportsMonica still only going down. You still need a reservoir at high ground and a place to drain it at low ground. And making a channel that is level the entire way would be significantly more difficult than one that gently slopes. Water in a flood is still only going down. Liquid water only flows down. It never flows up, unless it is coming from somewhere even higher that before (or special cases like a spring or geyser) $\endgroup$
    – user102593
    Mar 21, 2023 at 15:29
  • $\begingroup$ Never say never! have you heard of a siphon? The liquid has to flow up before it can flow down. You are correct that the net flow of water in my trough example is downward but in order to get down it must also flow along the level trough. In what sense would a level channel on level ground be more difficult? More difficult to construct? I don't understand. $\endgroup$ Mar 22, 2023 at 12:07
  • $\begingroup$ @chasly-supportsMonica oh I will say never. but that is a long topic. too long for the comments. short version, nothing ever goes up, unless something bigger goes down. ever. this is basic entropy. everything goes down, unless you put in energy, and the only way to get enough to make one thing go up, is if something bigger than it goes down. all of created existence works this way. even if you syphon it. that is an example of an illusion. you cannot use a syphon to move water to a high container, the syphon effect breaks as soon as the net effect stops being "water goes down" $\endgroup$
    – user102593
    Mar 24, 2023 at 13:42
  • $\begingroup$ In my proposed system, as I clearly said in my previous comment, the net flow of water is down, it goes from the mountain to the sea. However all rivers have a horizontal component to their flow apart from where there is a waterfall. It is that inevitable horizontal component that interests me. $\endgroup$ Mar 24, 2023 at 22:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .