Panel saw beam adjustment

I am trying to help a friend who has a vertical panel saw. Somebody before me did some adjustments and result is bent upper beam. It is just a round bar of steel around 5 meters long supported by adjustment bolts for levelling and carriage is running on two rollers.
As all of subsequent saw geometry is based on correct straightness of the beam, this has to be fixed first.
Problem is if the beam is even very little tilted in some areas, crosscut result is not 90 degrees.

I have Tesa Niveltronic electronic level, but this seems too sensitive and little imperfections on beam surface give me false interpretations (noise is killing trend).

If somebody has experience or idea please let me know.

You should ask this question in the Woodworking forum

A taught wire might get you the reference you need to adjust the bar. There is reference information for wire size, tension, and the amount of sag to expect in the middle. That is, it won't be dead straight but you can know how much un-straight it is at any distance from the end. You could then use a dial or digital caliper to see where the position needs to be adjusted. I realize the carriage is there and in the way, but the wire could be positioned above it. If you had, hypothetically, a wire over the top of 1-2-3 blocks at each end you could measure anywhere in between and subtract sag to find the correct bar position. A spreadsheet would be your friend here.

An optical level is another option, as used in aircraft assembly, by machine builders etc.

How are you checking the squareness of the item being cut after the fact? In my experience, the truth is in the cut, so you don't need to use fancy devices to measure the upper beam, you merely need to fiddle with it until the cut piece is true, and that's what you need to focus your measuring on. Tweek the bar until the end product is perfect. Make sense?

Stuart

OP should probably expand a little on where the problems are. He says that "beam is even very little tilted in some areas" If it's tilted in some areas and not others then it's not straight so the crosscut angle would be different depending on where along the beam he cuts. If it's the same angle everywhere (even if not 90 degrees) then it's straight at least, even if not parallel with the bottom edge.

If rip cuts are done by turning the saw and pushing the stock through, then it doesn't even have to be parallel within reasonable bounds. Straight and square with the bottom edge are sufficient. If rip cuts are made by pushing the carriage, then he has to have both squareness and parallelism.

If he's assured of straightness of the bar the carriage rides on, then a cross cut and flipping the sheet over the vertical will show a double angle of the error and he can work on correcting that.