Ok, now what should I measure?

Many sensors inside a model.

Determining the measurement strategy is an important part of the model design. This should be strongly guided by decisions made when constructing the testing plan mentioned above. Some questions can be answered with a single interior sensor placement and measurement while others require an elaborate network of measurements under multiple model configurations. Knowing exactly what will be measured in advance will greatly influence the construction of the model.

The CERES artificial sky instrumentation package has 10 Li-Cor photometric sensors numbered 1 through 10. Sensor 1 is reserved for measurement of the light available outside of the model. Sensors 2 though 10 may be used for interior model measurements. Users of the artificial sky are not obligated to use all 10 sensors; however, any testing exercise must use at least 2 sensors, one outside the model and one inside the model. The outside measurement is always required to determine the DF value previously discussed.

Good use of wires on the inside of the model.

Sometimes it is sufficient to place the available sensors into fixed positions for testing; other times more readings are required and sensors must be made mobile to accommodate all of the desired measurement locations. When testing a larger number of locations in a model, often it is more convenient to use fewer sensors and move them during testing and take multiple readings. Frequently the desired study requires the measurement of horizontal illumination at a given height in a space. In this case a measurement grid across the floor area of the space could be established. For example, after examination it may be determined that a uniform measurement grid of 6 x 7 across the floor plan would be useful. In this case it would likely be best to use 7 interior sensors in a single row and take 6 sets of measurements by moving the sensors between sets. Some considerations for placing and moving sensors are discussed in the next section.

While horizontal illumination is most frequently measured, sometimes vertical measurements are more valuable. For example, the determination of the DF and potential daylighting illumination on the vertical surface of book stacks in a library would require that the sensors be recessed flush into vertical surfaces of a modeled book stacks.

I hate spaghetti - Sensor/Wire Management 

T-sensor and how sensors fit into this setup.

The Li-Cor photometric sensors are connected to the data collection instrument by wires that are approximately 10' in length. This normally provides more than sufficient wire length for routing the wire from the instrument into the model and around obstacles to the desired measurement location. However wires entering the model must be planned for and accommodated before model construction. The wires are not detachable from the sensors so the sensor must be placed into the model in a manner that allows the wire to be routed back through to the outside of the model. The wires are black and so can absorb and obstruct the entry of light. Therefore it is not recommended that sensors wires be routed through available daylight apertures. Instead, special portals and paths for sensors and wires should be designed into the model. These should make the wires as unobtrusive as possible and care should be taken to cover any openings and portals in a light-tight manner after sensors and wires are in place. The number of sensors used and their location will guide the design of any wire ports or channels.

The cylindrical shape of the sensor, the stiffness of the wire, its length and the attachment point between the wire and sensor all combine to make the sensors extremely susceptible to tipping over. This can be extremely inconvenient if it happens once all the sensors are in place, the model is closed up and testing is in progress. To minimize the occurrence of this, it is desirable to use a small piece of scotch tape to hold the sensors in place and in the upright position. Care should be taken to keep tape from touching the white sensor surface and from obstructing light from reaching the sensor. In some situations it is sufficient to tape the wire down a short distance away from the sensor cylinder rather than placing tape on the sensor itself.

The previous section mentioned a measurement technique in which few sensors are used and are moved from location to location for multiple sets of measurement. To save time when this technique is used, a specialized "sensor carrier" is sometimes constructed. This specially designed component holds the sensors in an upright position and at the proper spacing from one another in one direction. The sensor is then repeatedly moved a defined distance as measurements are collected in "rows". This technique works particularly well when the desired measurement pattern is a rectangular grid. The carrier can be designed to manage the sensor wires and can also be designed to be moved from outside the model. These carriers are sometimes constructed in a "T" configuration so that the handle can be pulled; moving the sensors without disassembling the model and gathering data sets quickly. This can speed the testing process by not requiring the model to be opened as the sensors are moved from row to row. Again, care must be taken to insure that the carrier, handle and sensor wires do not introduce any light leaks into the model.

This one is better than that one - Changeability

Making interchangable parts for your model
shows you different design alternatives.

As has been mentioned, the best use of the Mirror Box Artificial Sky is for comparing design alternatives. This likely means that the model will be tested at least twice and that something about the model will be changed between the tests. Planning for and accommodating these changes in advance can save time during testing. Consider designing and constructing the model to allow easily swappable components, walls, daylight apertures and/or surfaces.