This article explains the design principles of a reflector for street light applications. The OpticStudio CAD Dynamic Link is used to model and optimize this part, in accordance with the mechanical parameters defined by the suppliers of such parts. Optimization and tolerancing are performed directly on the road lighting regulations levels.

Although the sample file couldn't be shared due to confidential, important aspects of the whole design process are highlighted.

Authored By Francesco Aldegheri – Consultant, Updated by Christophe Weisse

Introduction

In the lighting market, aluminum reflectors are frequently used because of their strength and their high optical efficiency. However, aluminum reflector manufacturing uses processes (bending, calendering, deep drawing, ...) with lower mechanical precision compared to other traditional manufacturing processes for optical parts, such injection plastic optics for instance. It is then essential to properly tolerance the metal reflector to give the correct mechanical dimensions range to the supplier, to ensure the needed performance level. This article is a guideline on how to set up, optimize and tolerance this part.

Setting up the system with the Dynamic Link

Road lighting luminaires are required to meet specific regulations – in this example ME3B class with R3 asphalt table. The optical system consists of lamps (LEDs), reflectors and a glass cover. The reflector is made of two central and two side reflectors that were all designed in CAD and imported in OpticStudio using the Dynamic Link.

This approach allows us to align the parameters that need to be shared with the manufacturer (bending angles, length of every bend etc.) It is possible to expose and update these parameters directly in OpticStudio using the CAD modeler engine.

Optimization details

The optimization is achieved using the roadway lighting merit function. Please note that the corresponding wizard only considers weights for the constraints on the NSRW operand: only the OPLT/OPGT operands have a weight. This allows to “drop” the contribution of a given operand to 0 if the condition is met (and then focus on what is still off-target).

In this example, at first, the real luminous flux from the sources is unknown. As a consequence, it's not possible to check the road lighting average luminance parameter and the threshold increment (TI). The weights of the luminance (line 13) and the TI operands (line 19) are then set to zero in the merit function.

The desired light distribution is then reached:

Tolerance analysis

In the following steps we are going to focus on the reflector shape parameters, since this is the data to share with the suppliers. The proper tolerance range needs to be defined for each mechanical dimension used in the manufacturing process.

Note: a tolerance analysis on the entire optical system would require other parameters such as source, reflector, glass cover relative positions (using the TNPS operand for Tolerances on Non-sequential PoSition data) to be taken into consideration. But for the reflector itself, the operand you are looking for is the TNPA operand (TNPA defines Tolerances on Non-Sequential PArameter data). It allows to tolerance any parameter of a given object, including exposed CAD parameters.

The first step is to populate the tolerance data editor.

• The “nominal” column displays the current value of the current mechanical dimension.
• The columns “min” and “max” represent the tolerance range to share with the supplier.

For example, in the current analysis we choose a range of ± 0.5 mm for the bending lengths and a range of ± 1 degree for the bending angles.

The second step is to choose the appropriate tolerance criterion. It is obvious here to use the Merit Function previously customized in merit function editor. This way we will directly check that the "imperfect" reflectors continue to give the correct illumination uniformity on the road.

The third and last step is to configure the Monte Carlo statistical analysis. All parameters from the Tolerance Data Editor simultaneously take a random value within the specified ranges. This simulation generates a series of random reflectors which meet the specified tolerances, then evaluates the merit function.

For this statistical approach, a minimum number of virtual systems need to be tested. A good rule of thumb is to use at least the square of the number of tolerance operands. Therefore, this is about 400 Monte-Carlo runs in our case. After the analysis, a complete report shows the results. In the “Monte Carlo Tab”, we can see the mechanical values of every parameter used in the specific Monte Carlo file, as well as the Merit function value.

Given how the merit function is set-up (using the weights only on OPGT/OPLT constraints), it is then very easy to find out if the tolerances are tight enough. Any value different from 0 points out an off-target system. In this example, it is clear that tighter tolerances should be used.

Conclusion

The manufacturer specifications are keys to decide how the system should be set up. The CAD Dynamic Link provided the needed flexibility and makes it easy to optimize, and tolerance.