Technical Summary of Research into
Unit Power Density and Unit Uplight Density:
Results

Table of Contents

Results: Optimization, UPD and UUD

Results from the preliminary phase [3] demonstrate that the optimization routine identifies the system with the greatest spacing and therefore the lowest UPD for that configuration of lighting system and combination of roadway, photometric file and design criteria. Figure 1 shows the results for three different 250W HPS photometric files during optimization for a three lane major (medium) roadway.

The set of values for the first photometric file is on the top, and the second is in the middle, the third on the bottom. Each photometric file produces the best performance - defined as maximum spacing - among systems using that IESNA cutoff classification for that combination of wattage and roadway. Looking at the first one on top, there are three different metrics shown over the horizontal axis which is measured in UPD (Watts per square meter of roadway). The three values scaled together on the vertical axis are - from the bottom up - Lavg (average luminance in cd/m2), STV and UUD (in uplight lumens per square meter of roadway). The values of these three metrics are independent of each other, but each is related to UPD and - as shown - varies as the optimization procedure progresses. The series of values shown are typical over the progress of the optimization routine. The optimization starts with low mounting heights, which correspond to high average luminance and short spacing - equal to high UPD values. So for each photometric files, the results from the start of the optimization are on the far right (or even off right hand side of the figure) and proceed to the left with each increase in mounting height - for a while. When the mounting height is relatively low, the limiting criterion will almost always be the uniformity, keeping the spacing from increasing any further for that mounting height, even though the average is more than its own criterion. As the mounting height increases, a particular combination of mounting height and spacing will appear that just meets the criterion for the average as it also just meets the criterion for uniformity. This is typically the system with the lowest UPD, shown in each of the three parts of Figure 1. As the mounting height increases further, the spacing becomes constrained by the average criterion instead of the uniformity criterion, shown by the succession of Lavg values of 1.0 in Figure 1. Clearly for each photometric distribution, the optimization routine can identify the system with the minimum UPD - the one with the maximum spacing.

The results also show that for a specific photometric file, lower UPD values correspond to lower UUD values. The relationship is direct for each of the three files.

Furthermore the results show that full-cutoff distributions did not provide the lowest UPD or UUD values among those evaluated. In Figure 1, the photometric distribution on the bottom is the best of the full cutoff (FC) for that roadway, and its best UPD value is 0.72 W/m2. The best cutoff (CO) distribution - shown in the middle - has a best UPD of 0.65 W/m2, while the best semi-cutoff (SC) distribution's value is 0.59 W/m2 - over 20% less than the best FC. The same trend appears across the UUD values as well. Figure 2 shows the results for the preliminary phase [3] for 250W HPS luminaires on three lane collector roadways.

It is obvious from this figure that among the best systems and for every specific system, increases in UPD always correspond to increases in UUD, but that two systems can have equal UPD while one has significantly higher UUD than the other. In general the lowest UPD and UUD values are for semi-cutoff (SC) distributions, while the best UPD values for systems with cutoff (CO) distributions are lower than the best UPD values for systems using full-cutoff (FC) distributions.

In the intermediate phase [4], each group was evaluated using presentations like Figures 3 and 4, showing a scatter plot of the average roadway luminance and UPD data pairs for each system in the group. In Figure 3, the group is C07s250H - collector two lane roadway meeting the base case using 250W HPS lamps - and in Figure 4 the group is C07s250M - the same but with metal halide lamps.

Each system has been identified by cutoff classification and is shown on these charts by the corresponding symbol. The distribution of the symbols shows the criterion for the average luminance at 0.6 cd/m2, and that some systems provide higher luminance than the criterion, and some of them do so with lower UPD values, providing more illumination at lower "cost" in terms of UPD.

Inspection of the these figures shows supports the discussions in this document about cutoff classifications, and comparison between the figures supports the discussion about the differences between sources.

Results: Design Criteria

The preliminary phase included limited calculations done with the same software as the optimization program, but "by hand" (to check the optimization routine and) to investigate the differences that design criteria made to the UPD and UUD results. Following the "base case" runs, the systems with the lowest UPD were identified for each photometric classification of full cutoff (FC), cutoff (CO) and semi-cutoff (SC). Runs were made with those photometric files meeting just the luminance criteria and meeting the Small Target Visibility (STV) criteria. The results of the preliminary phase [3] show that significant reductions in UPD - up to 33% - may be achieved by using these design methods compared to the "base case" of meeting both the illuminance and luminance criteria. The preliminary results also show that full cutoff (FC) distributions do not necessarily provide the lowest UPD or UUD values. For the photometry considered, the full cutoff (FC) distributions consistently use more energy and produce more uplight than the semi-cutoff distribution (SC) (and the cutoff (CO) also outperformed the full cutoff (FC) in most cases.) Some of the results are shown below in Table 9.

Table 9 includes the values for FluxRoad, Flux Off and Flux Up, which are the lumens for each path to uplight in one spacing of that particular system. Therefore the total of all three Flux numbers is equal to the lumens out of the luminaire, and each lumen follows one of three paths. Flux Up is lumens leaving the luminaire above horizontal which contribute directly to uplight. FluxRoad is the lumens onto the roadway surface, which are multiplied by the roadway reflectance (7%). Flux Off is lumens off of the roadway surface (equal to downward lumens less lumens onto the roadway), which are multiplied by the off-roadway reflectance (18%) to establish the amount of uplight due to that path. These three values are summed to make the total uplight, which is divided by the area (spacing times the roadway width of 7 meters in Table 9). The relationship between Flux Up and the total uplight is clearly not direct - the lowest UUD does not correspond to the lowest Flux Up, as seen for the full cutoff distributions.

This relationship between the three different contributions to uplight was investigated further, with example calculations and intermediate values shown. That investigation is reported in the "white paper" at "http://www.resodance.com/mdi/UUDCalc.html" and the photometry used for those calculations is also available through that website.

Results: Illuminance and Luminance

Results from the intermediate phase [4] clearly show that under the "base case" (Both) the optimization is almost always limited by the illuminance criteria, and an example of this - for Collector roadway of two lanes with 250W MH - is shown in Figure 5. The letters (F, C, S, or N) at the bottom of the figure corresponds to the cutoff classification of the luminaire used in that system.

When the illumination requirements increase and the roadway is particularly wide, for example Major roads with four or more lanes, the luminance criteria may be limiting the optimization. An example with Major roadway with 6 lanes using 250W MH is shown in Figure 6.

This shift generally improves the performance of more stringent cutoff classification distributions more compared to other distributions. Comparison of Figures 5 and 6 show this to some extent, as the distributions with the most improvement (i.e. reduction in UPD) using the illuminance method are consistently full cutoff (F in these figures). This is quite evident in Figure 6 where - counting from the left end - the fourth and fifth and the ninth, tenth and eleventh are all type F and have significant improvement for the illuminance method alone compared to the base case (Both). Otherwise almost all the improvements from the base case are for luminance method and tend to correspond to less stringent cutoff classifications, such as non-cutoff (N) and semi-cutoff (S). Throughout this research, this is the dominant trend in the results [3, 4, 5]. Systems using the luminance method consistently have lower UPD values than those using the illuminance method, except for situations with wider roadways and distributions with more stringent cutoff.

Results: Cutoff Classifications

Results from all three phases [3, 4, 5] indicate that - comparing the best from each classification - the more restrictive cutoff classifications correspond to higher UPD values than the less stringent classifications. Figure 1 shows the performance of the systems with the lowest UPD values in that classification, for the full cutoff (bottom), cutoff (middle) and semi-cutoff (top) distributions in the preliminary phase. This trend is also shown in Figures 2, 3, 4, 5 and 6, which show results for a variety of roadways and include some results from the intermediate phase. Part of the investigation in the advanced phase [5] was to look closely at the difference in system performance when full cutoff lighting is required (as is the case now due to legislation or ordinance) compared to when any distribution is allowed. This is not exactly the same as comparing the performance of systems with full cutoff distribution to systems with any other distribution, although in response to a discussion at the 2002 IESNA conference, such data was developed. Instead the comparison between "full cutoff only" and "any distribution allowed" provides clear information on the cost of requiring full cutoff distributions, as is now mandated in some legislation and ordinances.

UPD values for systems with "any distribution allowed" are shown in tabulated form in Table 10 below, and values for systems using "full cutoff only" are shown in Table 11. A graphical presentation of the percentage differences between these two sets of data is shown in Figure 7.