2

I have a circular ring of unit mass and fixed radius R which lies in the XY plane at point $O$ with coordinates $O:(0,0)$.

I wish to find a formula for the gravitational force at a point $P: (D,0)$ which lies in the same plane as the ring and is at some variable distance D from the ring centre O.

(Note: There are many treatments of the case for a target lying on the axis of the ring. The reference so far found nearest to this co-planar case is Problems 5-12, 5-13, (no solutions given) p.127 in Classical Dynamics of Particles and Systems by Jerry B. Marion.

I expect that the formula should be of the form $F = GM*f(D)$ where $G$ is the gravitational constant, $M$ is the mass and $f$ is some function similar to the Newtonian spherical divergence function $f(D) = \frac{1}{ D^2}$ (where the factor $\frac{1}{4.\pi}$ is absorbed in the value of the constant $G$ ).

So far I have obtained an integral formula by initially modelling the ring as a series of $N$ small point masses of mass $\frac{1}{N}$ separated by angle $\delta\theta$, whose distance from target is $L$ where:

$$L^2 = (D-a)^2+b^2 = D^2-2aD+R^2 = D^2\left(1 -\frac{2a}{D} +\frac{R^2}{D^2}\right)$$

where $a (= R\cos\theta)$ and $b(=R\sin\theta)$ are the $x$ and $y$ coordinates of the point.

Due to symmetry and vector addition of forces there is no net force in the y-direction and so the effective force contribution (along $x$) for a point is given by multiplying by the cosine factor $(D-a)/L$ thus:-

$$ F = \frac{-GM}{N}\frac{1}{4\pi.L^2}\frac{D-a}{L} = \frac{-GM}{ N} \frac{D-a}{L^3} $$

$$ F = \frac{-GM}{ N} \frac{D-R\cos\theta}{\left(D^2\left(1 -\frac{2a}{D} +\frac{R^2}{D^2}\right)\right)^{\frac{3}{2}}} $$

$$ F = \frac{-GM}{ N} \frac{D-R\cos\theta}{D^3 \left(1 -\frac{2a}{D} +\frac{R^2}{D^2} \right)^{\frac{3}{2}}} $$

$$ F = \frac{-GM}{ N} \frac{1-(R/D)\cos\theta}{D^2 \left(1 -\frac{2a}{D} +\frac{R^2}{D^2} \right)^{\frac{3}{2}}} $$

I then obtained the following integral formula for the force exerted on the target point by the ring:-

$$ F = \frac{-GM}{ D^2} \frac{1}{2\pi}\int_0^{2\pi}\frac{1-Q\cos\theta}{\left(1-2Q\cos\theta+Q^2\right)^{\frac{3}{2}}} \text{d}\theta$$ where $Q = R/D$.

$$ F = \frac{-GM}{ D^2} \frac{1}{2\pi} \frac{1}{(2Q)^{3/2}}\int_0^{2\pi}\frac{1-Q\cos\theta} {\left(\frac{Q^2+ 1}{2Q} - \cos\theta\right)^{\frac{3}{2}}} \text{d}\theta$$

Defining $A = \frac{Q^2+ 1}{2Q}$, Wolfram Alpha gives... $$ \int_0^{2\pi}\frac{ 1 - Q \cos x}{(A -\cos x)^{3/2}} dx $$

$$=\left[\frac{2}{(A^2-1)\sqrt{A - \cos x}}\left(A^2-1\right)Q\sqrt{\frac{A - \cos x}{A-1}} \operatorname{F}\left(\frac{x}{2}~\big|~\frac{-2}{A-1}\right)-AQ\sin x- (A-1)(AQ-1)\sqrt{\frac{A-\cos x}{A-1}}\operatorname{E}\left(\frac{x}{2}~\big|~\frac{2}{1-A}\right) +\sin x\right]_0^{2\pi}$$

Where $E(x|m)$ is an elliptic integral of the 2nd kind with parameter $m=k^2$, and $F(x|m)$ is an elliptic integral of the 1st kind with parameter $m=k^2$.

Replacing $\cos x$ by $1$ and $\sin x$ by $0$... $$=\frac{2}{(A^2-1)\sqrt{A -1}}*\left[(A^2-1)Q \operatorname{F}\left(\frac{x}{2}~\big|~\frac{-2}{A-1}\right)-(A-1)(AQ-1) \operatorname{E}\left(\frac{x}{2}~\big|~\frac{-2}{A-1}\right)\right]_0^{2\pi}$$

Cancelling $(A^2-1)$... $$=\frac{2}{\sqrt{A -1}}\left[Q\operatorname{F}\left(\frac{x}{2}~\big|~\frac{-2}{A-1}\right) - \frac{(AQ-1)}{A+1} \operatorname{E}\left(\frac{x}{2}~\big|~\frac{-2}{A-1}\right) \right]_0^{2\pi}$$

Being unfamiliar with Elliptic Integrals, this is as far as I can comfortably go at present.


After reading the wikipedia article Elliptic Integral, proceeding tentatively, from the definitions of elliptic integrals I think that $E(x|k^2)$ and $F(x|k^2)$ both go to zero when $x$ is zero, thus...

$$=\frac{2Q}{\sqrt{A -1}}\left[\operatorname{F}\left(\pi~\big|~\frac{-2}{A-1}\right)-\frac{(AQ-1)}{AQ+Q} \operatorname{E}\left(\pi~\big|~\frac{-2}{A-1}\right)\right]$$

Next perhaps it would be helpful to reformulate the problem so that the amplitude(?) term in the elliptic integrals changes from $\pi$ to $\pi/2$, thereby making the elliptic integrals "complete" and permitting them to be expressed as power series. This reformulation could be done by modelling the gravitational effect ($Fx$ component only) of two half rings (positive $y$ and negative $y$), independently, and using respectively the angles $\theta_1$ and $\theta_2$ which both range from $0$ to $\pi/2$ but in different directions.

steveOw
  • 941
  • 2
    See, these elliptic integrals are not going to vanish by some smart trick. So the only realistic option is to incorporate them into your experience. – Ivan Neretin May 28 '19 at 19:30
  • @Ivan Neretin I appreciate your tip and have stumbled ahead a bit. – steveOw May 29 '19 at 10:09

3 Answers3

2

Considering similar problems it is usually simpler to consider the potential rather than force. The latter can be found later as the negative of the potential gradient. Assuming the masses of the test point-like body and the ring be $m$ and $M$, respectively, we have in spherical coordinates with the1 origin at the center of the ring and the polar axis directed perpendicular to the plane of the ring: $$ U({\bf r})=-\frac{GmM}{2\pi}\int_0^{2\pi}\frac{d\theta}{\sqrt{r^2+R^2+2rR\sin\phi\cos\theta}},\tag1 $$ where (following the "maths" convention referred to in the Spherical Coordinates link and for consistency with the Question) $r,\phi,\theta $ are radial distance, polar angle, and azimuthal angle of the point ${\bf r}$, and $R$ is the radius of the circle.

The integral $(1)$ can be dealt in the following way:

$$\begin{align} \int_0^{2\pi}\frac{d\theta}{\sqrt{r^2+R^2+2rR\sin\phi\cos\theta}} &=2\int_0^{\pi}\frac{d\theta}{\sqrt{r^2+R^2+2rR\sin\phi\cos\theta}}\\ &=2\int_0^{\pi}\frac{d\theta}{\sqrt{(r^2+R^2+2rR\sin\phi)-4rR\sin\phi\sin^2\frac\theta2}}\\ &=\frac{4}{\sqrt{r^2+R^2+2rR\sin\phi}} \operatorname{K}\left(\frac{4rR\sin\phi}{r^2+R^2+2rR\sin\phi}\right), \end{align} $$ where we used the convention $$ \operatorname{K}(m)=\int_0^{\pi/2}\frac{d\theta}{\sqrt{1-m\sin^2\theta}} $$ for the complete elliptic integral of the first kind.

Finally $$ U({\bf r})=-\frac{2GmM}{\pi\sqrt{r^2+R^2+2Rr\sin\phi}}\operatorname{K}\left(\frac{4rR\sin\phi}{r^2+R^2+2rR\sin\phi}\right).\tag2 $$

In the plane of the circle $\phi=\frac\pi2$ and the above equation simplifies to: $$ U({\bf r})=-\frac{2GmM}{\pi(R+r)}\operatorname{K}\left(\frac{4Rr}{(r+R)^2}\right). $$


To obtain the expression for the acting force recall that: $$ \nabla f={\partial f \over \partial r}\hat{\mathbf r} + {1 \over r}{\partial f \over \partial \phi}\hat{\boldsymbol \phi} + {1 \over r\sin\phi}{\partial f \over \partial \theta}\hat{\boldsymbol \theta}. $$

As the potential $(2)$ does not depend on $\theta$ only two first terms remain.

Tedious but straightforward calculation reveals: $$\begin{align} {\bf F}_r&=\frac{GmM}{\pi}\frac{(R^2-r^2)\operatorname{E}\left(1-\frac {y^2}{x^2}\right)-y^2\operatorname{K}\left(1-\frac {y^2}{x^2}\right)}{rxy^2};\tag3\\ {\bf F}_\phi&=\frac{GmM}{\pi}\frac{(R^2+r^2)\operatorname{E}\left(1-\frac {y^2}{x^2}\right)-y^2\operatorname{K}\left(1-\frac {y^2}{x^2}\right)}{rxy^2}\cot\phi,\tag4\\ \end{align} $$ where $x=\sqrt{R^2+r^2+2Rr\sin\phi},\ y=\sqrt{R^2+r^2-2Rr\sin\phi}$.

user
  • 26,272
  • 1
    For $r>>R$ and $R$--> $0$ this should presumably tend to the Newtonian formula for the grav potential from a point source at distance $r$ namely $U(r) = 1/r$ which (barring the excess factor 4 in the numerator) your formula appears to do. Many thanks. – steveOw May 29 '19 at 13:27
  • Thank you for catching this. Probably the idea to get rid of the physical constants was a bad one, so I have inserted them now. As $\lim_{x\to0}\operatorname{K}(x)=\frac\pi2$, the potential has correct asymptotic $U(r)\sim-\frac{Gm}r$ for $r\gg R$. – user May 29 '19 at 13:48
  • Some notes:- 1. you have used for angle in XY plane where I used ; 2. You are (generously!) providing a general expression for angles (your) between + and ; 3. The first equation involving and comes from geometry; 4. I have editted in an intermediate step, to clarify (for me) how the eqtn in $K$ is derived.. – steveOw Jun 14 '19 at 03:01
  • Also 5. If possible, could you differentiate the potential to give an equation for the force? – steveOw Jun 14 '19 at 12:29
  • I have added the explanation of the notation and gave a link to an article about spherical coordinate system. 2. Yes, my expression is valid for whole space (excluding the points of the ring). 3. Yes 4. I do not understand which edit do you mean. In fact there were no derivation (except for using simple substitution $\frac\phi2\to\phi$ in the integral, then the definition of $\operatorname{K}(x)$ was used).
  • – user Jun 14 '19 at 14:35
  • I will so it as soon as I have enough time. I have noticed that you have accepted the answer and then changed the mind. Is it related to this issue?
  • – user Jun 14 '19 at 14:42
  • Link is where? My point is that I was using $\theta$ for the angle in the XY plane. 4. My edit was awaiting mediation but now I cannot find it anywhere. 5. re acceptance - yes, sorry; obtaining the force (rather than just the potential) is critical and (for me) the differentiation is non-trivial. It seemed unfair to accept your answer while the other answer does give a force. But your answer is very helpful already.
  • – steveOw Jun 14 '19 at 18:34