gregc
you wouldn't consider that case with a continuous 50' curve as 2 sections of 32" radius curve and therefore 4% = 2 + 2 * (32/32)? what if you considered the continuous curve as 4 adjacent 32" curves, would it be 6% = 2 + 4 * (32/32)?
No, because it's still only one curve, while the first two curves on my layout are separated by straight track and curve in opposite directions, and likewise with the third curve, which is separated from the second by a short length of straight track and curves in the opposite direction.
The final two curves are separated from the third curve by a length of straight track over a bridge, again curving in the opposite direction of the previous one, and then segueing directly into the final curve, with no straight track, again in the opposite direction of the previous curve, a true "S" bend.
The direction of each curve counters that of the previous one, and all but the last two are separated by various lengths of straight track.
A locomotive running light would be in only one curve at a time, the radius of that curve only affecting the compensation of the grade in that area.
Where it goes through the two final curves, which are not separated by straight track, I can't say if the transition from a left curve directly into a right curve both add to the compensated grade or if the second one cancels out the previous one...by the time that happens, the loco will already be in the final curve, and onto level track again.
My original tests never even took curves into consideration, as I was unaware, at the time, of the formula to determine the effect of curves on grades, although I guessed that it must have had some effect on performance.
In some instances, the loco on a curve can pull better than on straight track, especially if the engine is on the curve, but the trailing train, at the same time, is mostly on straight track...less friction on the straight, more on the curve.
Wayne