# Gauss' Law for Gravity

1. Aug 10, 2010

### ayae

I'm a little stuck, how can I go about calculating the gravitational field on the surface of a mass in the shape of a torus using Gauss' Law.

2. Aug 10, 2010

### Naty1

[1/4piG] closed integral (g . dS)

3. Aug 10, 2010

### ayae

I know that but I cannot simplify it to produce a workable result.

4. Aug 10, 2010

### Staff: Mentor

That shape would not have enough symmetry to make using Gauss' law profitable.

5. Aug 10, 2010

### ayae

When you say profitable, do you mean possible or worth while doing?

Because I really had my mind set on using Gauss' law for this example, is there no way of numerically calculating it?

6. Aug 10, 2010

### Staff: Mentor

I mean worth doing. Gauss' law always applies, but you can only use it to find the field in certain cases of high symmetry. (Such as spherical or cylindrical symmetry.)

Step one is to find a gaussian surface with a uniform field. Can you do that for a torus?

You'd need to use superposition, not Gauss' law.

7. Aug 10, 2010

### ayae

This is what I feared, I didn't know whether it was symmetrical enough. Thanks for clearing it up.

You're going to have to forgive my scientific illilteracy, but I don't understand this.

All I need to know is the gravitational acceleration at the very centre edge of the torus. So is it safe for me to make the assumption that g.n = g at this centre strip? And simplify down to g integral dA = 4 Pi G M, (which I can easily calculate).

#### Attached Files:

• ###### show.jpg
File size:
56.9 KB
Views:
127
Last edited: Aug 10, 2010
8. Aug 10, 2010

### Staff: Mentor

Is the field strength constant over that center strip? Sure. But that's not a gaussian surface.
Sounds like you're still insisting on applying Gauss' law. You have to integrate over a closed gaussian surface, not just that center strip.

9. Aug 10, 2010

### ayae

If I can't find a gaussian surface for a which passes through the center strip what can I do? How would I do this using superposition?

Can I not utilize the property of the torus being rotationally symetrical around z? :(
Thanks for the help so far.

10. Aug 10, 2010

### Staff: Mentor

Find the field contribution from each element of mass and add them up. Not trivial, I'm afraid.

I don't see any obvious way. You want the field at the inner edge, not along z.

11. Aug 10, 2010

### ayae

Would it be scientifically sound to conclude with the use of shell theorem that; if the radius is considerablly large compared to the thickness of the torus that the torus can be considered a cross section of a sphere and if we consider that this cross section is the middle of the sphere then the net gravity in the middle area of the torus will be close to zero?

12. Aug 10, 2010

### stevenb

13. Aug 10, 2010

### ayae

14. Aug 11, 2010

Staff Emeritus
Yes, but you will have to work out what the divergence operator is in toroidal coordinates. That may be non-trivial.

15. Aug 11, 2010

### stevenb

Actually, it's not too hard to derive that if you have a couple of hours to kill, but one can also find this worked out. The attached pdf gives the formulas for general curvilinear coordinates and the previous references cited above give the h1, h2 and h3 scale factors.

Personally, the issue I'm having is visualizing in this coordinate system. It's not clear to me that the symmetry works out. At first it seems as if the symmetry is not there, but the coordinate system itself resembles the expected field pattern (EDIT: correction, not field but potential function), so there might be a way to make it work. I am curious if it can be done, so if someone is successful, please post the results.

#### Attached Files:

• ###### GenCurvCoord.pdf
File size:
1.8 MB
Views:
86
Last edited: Aug 12, 2010
16. Aug 12, 2010

### gabbagabbahey

Personally, I wouldn't bother trying to find a Gaussian surface. Instead, just use Cartesian coordinates and find the contribution to the total field (at the field point you are interested in) by an arbitrary infinitesimal piece of mass on the toroid. Integrate over the entire toroid and you're done. You should find that due to symmetry (when you measure the field at a point along the inner middle track of the toroid) two Cartesian components of the field will be zero, and you will only really have to do the integration for one component. The integration itself shouldn't be too bad with a clever substitution.

17. Aug 12, 2010

### Staff: Mentor

Exactly. (Still, non-trivial.)

18. Aug 12, 2010

### stevenb

So, the above advice is definitely good, and as an engineer I'd take that approach to get a practical answer.

Still, the motivation of the OP has not been expressed and it's not clear if an nifty and elegant approach is being sought for some esoteric reason. For this reason, I hope I can take some poetic licence to do some (possibly dubious) brainstorming, making clear that these ideas may be deadends, or at best needlessly complicated.

My previous mention of using Gauss's law in toroidal coordinates does not seem fruitful. At least I can't seem to make it work, as I don't see the needed symmetry. Still, there is some gut level feeling that I have that somehow this coordinate system may be useful, if it turns out that a closed form solution can not be found using the above recommendations in rectangular coordinates. Note, that I haven't tried, so I don't know if that works out to find the field at all points in space.

With that preface, note the following. As pointed out in the following link, Laplace's equation is separable in toroidal coordinates. This means that the exterior solution (for potential) can be found from the Laplace equation and boundary conditions. One can then investigate what happens with Poison's equation inside the torus shaped mass. I expect that the mass density is either constant or only dependent on the v-coordinate, in order to have toroidal symmetry. This mass symmetry could (possibly) allow determination of the interior solution for potential and then gravitational field can be found from the gradient of the potential.

http://mathworld.wolfram.com/LaplacesEquationToroidalCoordinates.html

Last edited: Aug 12, 2010