All roadway lamps in the world must be designed according to specific standards. The most popular standards are a US standard called "IESNA RP 8-00 Roadway lightning" and a European standard called "CIE 140-2000 Road lighting calculations".
In the case of CIE 140-2000, the standard defines only parameters definitions, calculations and measuring technique, the actual parameters values usually regulated by national standards. Some countries like Russia or Brazil use their own standards which do not refer to CIE 140-2000, but use definitions that are the same as CIE 140-2000.
RP 8-00 defines 5 types of streetlight pattern, each with three variations short, medium or long. The general definition of each pattern is shown below
From the perspective of an optical engineer, easiest pattern is Type V, because it has a symmetrical intensity distribution. Usually round bulb lenses are used for that pattern, and require only a simple design. But Type V is not common pattern for roadway. It is best for parking areas, walkway intersections. Type VS is similar as Type V, but has a more square shape, and require bulb lens, but little more bit complex shape.
Most popular streetlight pattern is Type III Medium. It is used for general roads, for wide road and city streets. It is asymmetrical, because lamp placed at the road side or near the road side, so that designer needs to redirect light away from the sidewalk or roadside onto the roadway without losses. This asymmetry makes the optical design more difficult. Furthermore the pattern size across road is narrower than along road. Usually lens design for Type III takes a lot of time. Lens shape is also often bulb type, but has unusual shape. The shape of a Type III lens is unique for the selected LED, which means a lighting company that switches from a CREE to a Nichia or a Lumileds to an Osram chip, needs to design new optics.
The Type II is same complexity as Type III. But it usually used for small roads, walkways. One special consideration is that a long pattern, for Type II (and Type III) increase glares and makes it very difficult to design a proper lamp.
The Type I pattern is easier than Type II or Type III, because lamp placed at the center of road. This makes lens easier, because there is not a requirement for backlight removal. But such lamps require very long pole overhang, and do not work well for for wide roads, and wide city streets. The Type I streetlight pattern is best for lighting walkways, paths and sidewalks.
A Type IV pattern is little bit complex than Type II or Type III It require asymmetrical pattern, but wider across the street. It can be used for wide roads, or for illuminate parking areas near the building walls.
In Europe the main parameters of roadway light and streetlight in Europe are following: longitudinal luminance uniformity Ul, overall luminance uniformity Uo, glare index called threshold increment Ti, average luminance Lavg. And placement of lamp and poles relative to road. Parameter definitions described in CIE 140 2000. A very important parameter is Ti defined as "the measure of disability glare expressed as the percentage increase in contrast required between an object and its background for it to be seen equally well with a source of glare present. Note: Higher values of TI correspond to greater disability glare." Better value for Ti is smaller, for Uo,Ul,Lavg better value is higher value. The long distances between poles makes pattern and lens too complex or impossible, because Ti increases. Also increasing Uo and Ul we can increase Ti. Best value of Ti is below 10%, for most complex patterns it can be increased to 15%.
Note that IESNA RP 8-00 also define glare index and uniformity, but uses other definitions.
Zemax ray tracing design of a streetlight LED optics requires a set of steps and manual control of pattern and lens mechanical properties. For detailed analysis of efficiency, light losses and uniformity, many rays need to be traced from LED model. The best method is to use Radiant Imaging LED models which contain measured data. All these factor above allow obtain simulation results very close to the real lens performance and minimize prototyping and manufacturing costs.