As I've brought up in many previous posts, the external forces to be overcome by my vehicle at speed are rolling resistance and aerodynamic drag. I've also mentioned that the aerodynamic drag increases with the square of speed, whereas rolling resistance is independent of speed. The latter contention will be, I suspect, debated by experts. I've read extensively and, though several authors contend that rolling resistance increases linearly with speed, I have found none that support that theory with data or analysis.
My admittedly simplistic evaluation revolves around dimensional analysis. While this topic is far too deep to cover in a blog post, I can at least mention the principle involved. In an equation, the units on the left side must be the same as the units on the right side. For example: distance=speed times time. Distance may be in miles, speed in miles per hour, and time in hours. So on the right side, miles per hour times hours is miles, the same as the left side. Physicists will say "length = speed times time" so that they can use miles, centimeters, inches, furlongs, leagues, or parsecs for length, etc. Thus, they deal with the dimension of length rather than the specific unit of miles, for example.
For our problem, we want to know what affects rolling resistance. Resistance on the left side of the equation we're seeking is a force, so we want to know how force is affected by various things that may be on the right side of the equation. Likely candidates for what might affect this force are vehicle weight and speed. So we look for a combination of the dimensions of weight and speed that result in a force. But weight is a force, so if we multiply it by any power of speed, we'll no longer have a force and the dimension on the right side will not result in a force. While dimensional agreement does not assure the correctness of an equation, lack of dimensional agreement assures its incorrectness.
Now, it's true that dimensional analysis cannot, alone, give the entire equation. It cannot account for constants, for dependence on exponential and trigonometric functions, etc. And the method is also highly dependent on the accurate physical intuition of the analyst in determining the factors that may affect the dependent variable. For example, in this case is tire diameter (a length) a possible factor? Inflation pressure? How about bulk modulus of tire rubber? Certainly these could be factors, but a more thorough dimensional analysis indicates that, at least without taking even more arcane factors into account, they are not. For the physicists and automotive engineers reading this, I recognize that this is very simplistic and yet, to the accuracy possible by reading speedometers, odometers, and gas pumps, I believe it represents a valid analysis.
So, we have F[total]=.5*p*C[drag]*A*v^2+C[rolling]*W where F[total]is total external force on my vehicle, p is air density, C[drag] is the coefficient of drag, A is the flat plate area, v is speed, and C[rolling] is the coefficient of rolling resistance. This can be written as a quadratic equation in v, or F[total]=k*v^2+d where k=.5*p*C[drag]*A and d=C[rolling]*(weight). Using a typical value for air density and the other values for my Land Rover LR3 HSE, k=.775 and d=393. So we have F[total]=0.775*v^2+393.
From there, I can produce a graph that shows the fraction of resistive force from rolling resistance and aerodynamic drag at each speed. Below is a plot of each component of resisting force. The aerodynamic drag is the red plot, the blue is rolling resistance. They are equal at about 22.5 meters/second or approximately 50 m.p.h. I took the graph to 40 meters/second, or about 90 m.p.h. (though that speed is irrelevant to me because I never drive that fast).