The picture below shows an image of a bed from a 3d printer, which I intend to equip with an array of 15 x 3 mm Neodym magnets to hold down a steel plate. The steel plate itself is not and can not be equipped with magnets. Theres a heated aluminium bed between magnets and steel plate and the magnets are specified to withstand 150°C.

The left side shows all 12 magnet arranged with the same orientation, on the right side every second magnet is flipped. The distance between the magnet is wide enough so they don't noticeably attract/distract/disturb each other.

Question: The goal is that the steel plate lays flat (Z direction) and does not accidentally move in XY direction. Is either magnet orientation better than the other in achieving this?

My gut feeling is that the second option is better. It wouldn't be a particularly difficult experiment to do.

You might find this useful

https://en.wikipedia.org/wiki/Halbach_array

Got the some opinion on another forum, where someone wrote: „The second orientation (right side) provides return lines for the magnetic flux and leads to stronger induced magnetic fields in the plate.“

Thank you!

The steel would concentrate the magnetic fields, in the second orientation the resultant field would be much weaker than the first and thus would allow for easier movement along any axis.

2 minutes ago, Butch said:

The steel would concentrate the magnetic fields, in the second orientation the resultant field would be much weaker than the first and thus would allow for easier movement along any axis.

No, that's not it.

The magnetic field is a vector, and with all of the same pole present the field from one magnet will oppose the field from the adjacent one.

 Wookbert's finding about flux return is correct.

4 minutes ago, swansont said:

No, that's not it.

The magnetic field is a vector, and with all of the same pole present the field from one magnet will oppose the field from the adjacent one.

 Wookbert's finding about flux return is correct.

You are correct! 

It is counter intuitive.