We have a 5% grade on the layout here, and it causes problems with how many cars smaller steam engines can pull upgrade, just like it did on the prototype (Silverton RR).
We’re lucky to get one boxcar and one combo car up the hill with an MDC 2-8-0, just like the real locomotives pulling real cars were. A Spectrum 4-6-0 can pull two 34’ boxcars, or two 34’ flatcars and a shorty bobber caboose upgrade, and a Spectrum Shay can pull two tankers, a gondola, a rail and tie car, a derrick car, and a work caboose up without too much wheelspin. A Spectrum 2-8-0 can haul 4 34’ boxcars, but add a caboose and you’re gambling.
Sometimes it’s a fight. Sometimes we have to back up and try again to make it up grade, sometimes we have to wait for a helper engine, and sometimes we just dedicate a double header to pull the grade, again, all just like the real railroads did. I enjoy winning a long struggle to make it upgrade, whether it’s driving a model train, or my jeep on the 4x4 roads that most of these old grades have become today. Hillclimbs are fun. More fun if you beat the hill.
All that said…friction is a function of downforce per unit area, times total area, times the coefficient of friction. In plain english, increasing friction means increasing the number of pounds per square inch, increasing the number of square inches of contact at that pressure level, or increasing the coefficient of friction, (using a “stickier” surface).
Just about anything you might use to increase the friction coefficient is liable to decrease electrical contact. Sand on the rails ain’t gonna work, here.
Filing the drivers flat, leaving the flanges alone, might increase the area of contact, but the downforce per unit area will decrease proportionally, you are just spreading out the weight of the loco, and not changing the overall equation.
You can lighten the load, fewer cars, lighter weight cars, but then you run into derailing pro