AASHTO green book equation in wind farms road design

I’ve been recently asked to justify the roadbed thickness for a wind farm I’ve designed.

For several reasons (mainly because the majority of documents are redacted by non-civil engineers) the engineering companies supporting our customer ask for a written demonstration that the road design comply with the requirement of the famous AASTHO 1993 green book.

Unfortunately, it is not possible to use it for wind farms, and I’ll explain you why in this post.

As you will probably know, AASHTO defined an empirical equation  after a series of full scale test done about 50 years ago in the USA, the famous “Road test”.

This equation, very large and complicated indeed, gives as a result the “structural number” (SN) – a number that can be used to define the required roadbed thickness.

The formula looks very complicated, but the idea behind it it’s pretty easy: given the expected number of vehicle using the roads (defined as standard “equivalent single axis loads”) and other physical and project related variables you can define the correct thickness of the various materials selected for the road bed.

This is how the equation looks like:


W18 = Predicted number of 80 kN (18,000 lb.) ESALs (equivalent single axis loads). Basically different type of vehicles (car, trucks, bikes, etc.) will use the road. To simplify the calculation, all this different axes are concerted to “standard axes”.

ZR = Standard normal deviate.

So = Combined standard error of the traffic prediction and performance prediction. Both ZR and So choice depend on the type of the road (for a major highway you will need more confidence in the result, while for a local road you can assume some risk).

SN =Structural Number (an index that is indicative of the total pavement thickness required).

Basically, each layer has a thickness (D) and a “layer coefficient” (a) representing the quality of the material.

In wind farm construction normally only one or two gravel layers are used.

Therefore the equation SN=a1D1 + a2D2m2 + a3D3m3+… will simplify becoming SN=a1D1

a1 = Layer coefficient. Gravel would be around 0.14

D1 = Layer thickness (inches).

ΔPSI = Difference between the initial design serviceability index, p0, and the design terminal serviceability index, pt. This concept is needed to incorporate in the equation the quality of the road at the beginning of the considered timeframe, p0 and the quality of the road at the end of the life span (pt).

MR = sub-grade resilient modulus (in psi). This number indicates the quality of the sub-grade.

Said that, let’s see why this beautiful and highly effective equation is of little (if any) utility for wind farm design.

Basically, a highway or an urban road is damaged by the recurring transit of heavy loads – that is, bus, trucks, etc. This trucks use the road for several years, causing accumulated damage.

What happens in a wind farm is that, when the WTGs are installed and producing, no one will use the internal roads – only a few service cars every now a then. The ESAL number will be almost zero.

What normally damage wind farms internal roads without heavy traffic is poor drainage, incorrect roadbed material selection or poor construction (e.g. incorrect compaction), not cyclical mechanical loads above the elastic limits.

Therefore we normally design the roadbed based on the CBR value: we know that with a very good CBR in dry climates 20 cm are normally enough, while for low to very low CBR (>5) we use 40 to 50 cm.

Below CBR=3% special solutions are normally needed.

  1. William Devabalan’s avatar

    I am facing a similar predicament with an access road to a wind farm I have been asked to look at. The conventional way of designing pavement does not seem to apply here. I guess your approach to designing pavement is explained in the last paragraph. How do we corelate the thickness of the roadbase with the CBR value of the underlying soil? Is there a case study or reference that I can use to justify my design with the client? Your response is appreciated. Thanks.


    1. Francesco Miceli’s avatar

      Hello William,
      sorry for the delay. I assume that it's already too late to answer you - by the way I've just written a post about an alternative method:
      Regards, Francesco



Your email address will not be published. Required fields are marked *