*2.1. Light Sources Replacement*

Replacing HID-based luminaires (in particular, the HPS sources) with LEDs is a common retrofit pattern. What is interesting, due to the significant financial benefits, the municipalities decide to install LEDs, even if the end of an HPS fixtures' life cycle is not reached.

Typically, during an HID to LED migration, conversion charts provided by lamp vendors are used (see Table 1). Such a chart allows finding, for a given HID lamp's power, a rough estimation of an equivalent (in terms of a desired luminosity) LED's wattage to be installed.


**Table 1.** HID to LED conversion chart [14].

Although the unit savings shown in Table 1 are, as mentioned, only the rough estimations supplied by a specific manufacturer, they express the order of magnitude of available power reduction ratios. For the considered example, those ratios vary between 32% and 59%. It is obvious that total power savings for any real-life case will depend on both the lighting system structure, i.e., the ratios of particular fixture types being retrofitted, and the installation's size.

### *2.2. Lighting Design Tunning*

Power savings can be obtained thanks to a well suited lighting design. It is achieved by adjusting lamp parameters such as the fixture's photometric solid, dimming, pole height, arm length, fixture's mounting angles, etc. The above installation tuning may be done according to the either standard or customized design approach. In the first case, one assumes regularity and uniformity of a lighting situation, namely, a constant road width and evenly spaced luminaires (Figure 1a). In the second approach, an actual roadway layout is assumed (Figure 1b).

**Figure 1.** The typical (**a**) and customized (**b**) approaches to photometric computations. In the former case, a roadway layout is uniform (averaged). In the latter, an actual roadway layout used in the computations is decomposed into multiple segments which are processed consecutively.

The former method makes an installation meet performance requirements imposed by a lighting standard and minimize power usage. This approach, as a multivariate optimization, requires advanced software tools capable of solving the problem in an acceptable time. It has to be noted that imposing uniformity of a lighting situation leads to over-lighting due to the conservative assumptions on the road width and lamp spacing.

The latter design scheme, relying on the different design paradigm, was introduced in the work [15]. In this approach, the lighting situation is regarded strictly as is, i.e., with its actual geometry recovered on the basis of GIS coordinates (Figure 1b). In addition, the coordinates of luminaires are taken from a GIS data repository. In such an approach, the lighting situation is no longer uniform, which means that neither road width is constant nor luminaires are distributed evenly along a road. Although this method implies a significant computational overhead (one has to analyze multiple road segments rather than a single roadway, as shown in Figure 1), the resultant installation's configuration is designed in such a way that over-lighting is minimized. As shown in [15,16], the customized design method reduces a power usage by 15% compared to the standard approach described above.
