# Linearising compound pendulum equation

1. Mar 9, 2012

### seboastien

1. The problem statement, all variables and given/known data
Linearise T=2pi√(K^2 + h^2)/gh K is known constant

This is a compound pendulum equation, I want to plot some kind of formula with variable T against some kind of formula with variable H in order to find g from the gradient.

2. Relevant equations

3. The attempt at a solution

so i've got T/2pi all squared times g all substituted to x, h subbed to y and k^2 subbed to constant C and I've got the equation y^2 -yx + C=0 and tried to solve for y=x+β

I've tried implicit differentiation and it's gotten me nowhere

2. Mar 9, 2012

### tiny-tim

hi seboastien!

(try using the X2 button just above the Reply box )
if K is a known constant, can't you make one of the axes √(h2 + K2) ?

3. Mar 9, 2012

### seboastien

I would have to make the axis √((h^2 + K^2)/gh ) but that is a good point.

However, I would still like to know how I could linearise it further. I know that a taylor approximation is needed but I don't know how to, or what a value to choose

4. Mar 9, 2012

### tiny-tim

√(1 + (h2/K2) = 1 + (h2/K2)/2 + …

5. Mar 9, 2012

### seboastien

????????????????

6. Mar 9, 2012

### tiny-tim

if h/K is small, then √(1 + (h2/K2)) = 1 + (h2/K2)/2 + …

7. Mar 9, 2012

### seboastien

hmmm, my only issue is that its the sqrt of K^2 + h^2 divided by gh

it also turns out that k is the radius of gyration and I have no scales to measure the pendulum's mass. I believe I need a y=mx + c where the y intercept will be determined by k, g by m, x by T and h by y.

is there any way of achieving this?

8. Mar 9, 2012

### tiny-tim

i'm confused

you said that K was known

9. Mar 9, 2012

### seboastien

That's because I thought I was allowed to measure the pendulums mass.

Don't worry I've worked it out...finally, turns out I've been overcomplicating things.

I'll just plot a graph of h^2 against h*T^2 the y intercept will be -k^2 and the gradient will be g/4pi^2.

Thanks anyway.