How to solve an iterative equation

[ptc-2923222]

OK, I think I need a solve block, but I can't seem to get the format correct or am doing something else wrong as I know the answer, and can't even get close. I am trying to work out the leeward post embedment calculation from the National Frame Builders Design Manual on page 9-9. The equation is d^2 = [7.02 Va + (7.65 Ma / d)] / S' b where Va = 441 lbf , Ma = 1640 lbf * ft . S' = 532 lbf / ft^3 and b = 0.64 ft so I tried building it in MathCAD 14 using solve and the symbolic arrow to the right, but it gives me some huge number with the third root of a square root . . . . . I read where it will solve using units, but maybe that is the issue here? I know the answer is

d^2 = [7.02 441lbf + (7.65 1640 lbf-ft / d)] / 532 lbf/ft3 * 0.64 ft d = 4.22 ft

But I can't figure out how to ask MathCAD to solve for d . . . any suggestions? Did I miss it in the help section? Solve block syntax?

[jkowalski · Accepted solution]

it is imposible in mathcad 14 or 15

see:

http://communities.ptc.com/message/53285#53285

http://communities.ptc.com/message/180157#180157

but with Prime 2.0v its work fine

[ptc-2923222]

Jan,

Thanks for the help - Unless Mr. Hensley chimes in, I don't see how it will work either. If anyone else wants to take a shot, I included the file where I got it to work without the units and my attempt to incorporate units when it went haywire.

[Werner_E]

Of course you can do it with units in Mathcad 14/15 as well, as it seems, that you are not after a symbolic solution anyway.

The problem is, that the symbolic processor does not know anything about units (and thats true for Prime, too - jan has solved numerically). So the units you use are seen as variables - too much for the symbol processor to find a solution using the solve block.

Concerning the symbolic solve Jan has shown you a trick (assume ft=1) you can use as you know the unit of the outcome.

Here is another one - advantage: you dont't loose the unit ft.

In your file basically I found two errors, apart from trying to solve symbolically with units:

1) If you evaluate the solve block numerially (as you have to do with units), yoe need to provide a guess value for the iteration to start with. Depending on the equations to be solved this guess value may have great influence on which solution is found or if a solution is found anyway. I had luck with d=1ft.

2) You mixed the variable names between your first try (without units) and your second try. That way it could not work.

If you correct that, Mathcad 15 does exactly what you suppose it to do:

[jkowalski]

CHUCK I'm sorry that I introduced in error, I thought it was not possible

[Werner_E]

Nobody's perfect!

[ELSID]

You didn't miss it. This is an empirical formula, not symbolic.

An issue I frequently run into with empirical equations in solve blocks are constants missing the units. The constants have a unit value which you have NOT accounted for. In order for MC to carry the units correctly, the user needs to multiply them in. Example would be Q=500*GPM*dT where 500 has units to make it work.