Here's the situation: a thin concrete wall needs to go up, but it's a very large diameter, say 1 km. The wall thickness is on the order of 20 cm. The wall only needs to be maybe 3 meters high.

I'm assuming a giant single mold shaped like a ring is not practical. Instead, there are multiple molds, shaped as pieces of arc, that the wet cement is poured into.

The thing is, how do you make the wall uniform? I'm thinking the concrete will shrink as it cures, so there will be a gap between each poured piece of wall arc.

Is there some way around this? How do you incrementally extend the wall without gaps? I'm trying to imagine some separater, like a plank of wood, blocking one end of the mold. Once the next mold is set up and poured, you can pull out the plank. But is that practical? How do you seal the plank against the mold so there are no leaks? Besides that, how would you press the sides of the mold together hard enough so there aren't any leaks there either?


I forgot to mention this is non-reinforced concrete.

The environment is kinda cold. I'd say yearly ranges would see -25 to +25 Celsius. However, I would very much love to hear about thermal expansion and how to handle that.

IDK anything about ground conditions. I'm just imagining normal dirt. Someone mentioned a possible need of footers but I can't see that. Remember this wall is very thin. 20 cm thick and 3 meters high. That sounds pretty light to me, but if necessary, I would consider a flaring base for a snowshoe effect. Drilling to the bedrock for footers seems pretty expensive, so some kind of built-in foundation via a flaring base would be my first thought.

  • 4
    $\begingroup$ Why does in need to be uniform without gaps? What's the actual problem that you are trying to solve? $\endgroup$
    – 410 gone
    Apr 19, 2016 at 9:55
  • 1
    $\begingroup$ That's pretty ambitious even by Drumpf standards :-) . You really need to explain what your ultimate design goal is. Among the other things mentioned, what's the local environment? You'll probably need footings every meter or so as a minimum, going down to solid earth or bedrock. $\endgroup$ Apr 19, 2016 at 11:58
  • 3
    $\begingroup$ @DrZ214 That isn't quite so. After all, there will be wind and snow loads, which can be substantial. You'd also have to be careful with how you do the connection of the roof with the walls, otherwise there will be a rotational stiffness which will cause moment and therefore possible tension in the wall. To make that connection you will also need an engineer. Trust me: you need an engineer. $\endgroup$
    – Wasabi
    Apr 19, 2016 at 12:35
  • 2
    $\begingroup$ @DrZ214 I'm not trying to scare you away. I'm trying to give you an understanding of your situation. You won't get adequate answers here. Engineers that don't study your situation carefully, that don't look at the soil conditions in your region, who don't know the precise materials and other constraints in your project (perhaps some which you aren't even aware of) cannot give you adequate advice. Whatever answers you get here will not serve as replacements for an actual engineer. Let me reiterate: You still need an engineer. $\endgroup$
    – Wasabi
    Apr 19, 2016 at 12:47
  • 6
    $\begingroup$ You're trying to build a structure that contains more than 1885 cubic meters (~200 truckloads) of concrete. Yet your question is worded as though you've never worked with concrete before, or even built structures. As such, your question is far too broad to answer in the Q&A format used on SE. The comments are simply trying to point out all of the VERY significant engineering issues that you seem to be ignoring. The details of actually pouring the wall will be handled by your contractor anyway. $\endgroup$
    – Dave Tweed
    Apr 19, 2016 at 13:17

1 Answer 1


You can build this wall as you would any other. Given its size, the fact that it's a ring becomes irrelevant, which means that for all intents and purposes, it can be designed and constructed with the same concepts as that of a linear wall.

This can be demonstrated thusly: the wall is 1000 meters in diameter, which gives it a perimeter of approximately 3141 meters ($p = d\pi$). Now, a common maximum for small angle approximations (where angles can be ignored) is 10 or 14 degrees (source). So, conservatively, the maximum length of an arc of wall which could be designed as a straight wall is 87 meters ($\ell = \dfrac{p}{360}*\theta$).

However, walls can't be built with such dimensions. Not only would they be impractical to build, they would also be very prone to severe cracking due to drying shrinkage and thermal expansion during their service lifetimes. Walls must be designed with control joints which effectively turn a wall into a series of wall segments. Each of these segments will be much shorter than 87 meters and can therefore be designed as if they were straight instead of curved.

These control joints will need to properly treated for waterproofing of course.

  • $\begingroup$ I'm not sure I understand this answer. IDK what the small angle thing is used for, but I'd rather stick with actual curves in the arc. I'm assuming we already have the molds in the right shape. What I'm concerned about is constructing segmentally without leaving and holes/gaps/cracks. However, I will modify the OP because you made good points about thermal expansion. $\endgroup$
    – DrZ214
    Apr 19, 2016 at 12:01
  • $\begingroup$ @DrZ214 The small angle issue simply demonstrates that given the huge diameter of the wall, a curved segment of less than 87 meters will behave almost identically to a straight segment. You can build it curved, but you can use the same concepts as you'd use for a straight one. $\endgroup$
    – Wasabi
    Apr 19, 2016 at 12:04
  • $\begingroup$ This points out that building a circular wall of that diameter in pieces is no different from extending a straight wall. However, it doesn't answer the OP's question of how exactly to extend such a wall, straight or not. $\endgroup$ Apr 19, 2016 at 12:07
  • 3
    $\begingroup$ @DrZ214 "snug, gapless, and crackless" --> You will have to use unobtanium-based concrete. No other formulation will do. $\endgroup$ Apr 19, 2016 at 15:23
  • 1
    $\begingroup$ @CarlWitthoft, well, I took DrZ's comments as relative. Snug is possible. Gapless and crackless are however mutually exclusive. You can either use a construction joint, which implies in a gap (which can later be closed with waterproof filler) or a control joint, which implies in a visible crack (which can later be covered up since it shouldn't be excessively active once the concrete has mostly dried up). $\endgroup$
    – Wasabi
    Apr 19, 2016 at 15:31

Not the answer you're looking for? Browse other questions tagged or ask your own question.