2 excellent laymans books on the subject are The New Science of strong Materials (or why you don't fall through the floor) and Structures (or why things don't fall down). Complete with formulae for what you want.
You can either start with the loads and engineer your beam accordingly, plus safety factors or reverse engineer the steel beam.
The top and the sides of a box beam loaded vertically are different. The top and bottom are loaded in compression/tension, the sides in sheer.
So unis for the top and bottom, +/-45 for the sides.
The top and bottom need +/-45 inside and out to keep the fibres in column.
The sides don't need much, so a core is usually added to give it panel stiffness.
Deeper is better. Twice as deep, 8 times as stiff.
Easiest is probably to make the top and bottom the same theoretical weight as steel , assuming the dimensions are roughly the same. Steel is 7.8, glass 1.9 (you won't get this unless you vac bag it, but we use this for the calculation), so 4 times the thickness of glass to equal the steel. Assume 1mm of glass is 1,000 gsm of cloth (again, only under vacuum), if your steel beams are 2mm top and bottom, the glass will be 8mm, or 8,000 gsm of uni. Every 2mm of thickness, put in a layer of +/-45, plus an internal and external. So 5 layers of +/-45. Use these for the sides (3 outside the core, 2 inside) and you will be pretty close.
Whole lot of disclaimers, but this will get you started. To be more accurate, the load directions, stiffness and point loads need to be considered. As does the result of a failure (although it is most likely to crack first, so you will have plenty of warning, if you look) I'm happy to help with these, but be aware that i am not an engineer.