Sensitivity analysis of lamp model parameters in energy and yield simulations of greenhouse

Greenhouses offer a means to extend the growing season and increase local food production. However, greenhouses tend to be energy-intensive buildings. The choice of lighting systems strongly impacts energy consumption and crop yield. Simulations can help determine the optimal lighting conditions. However, lamp models typically include many parameters that are either unknown or hard to obtain. This study conducts a sensitivity analysis of lamp parameters on energy and yield simulations of greenhouses to identify the most influential lamp parameters. For a reference greenhouse, these lamp parameters were varied over a predetermined range through a Latin hypercube sampling strategy, and simulations were run for each combination of these parameters. Then, linear regressions were used to determine the relation between model outputs (energy consumption, yield) and each lamp parameter. The analysis of the standardised regression coefficients demonstrated that the reflection of sun's NIR through the lamps layer and the maximum intensity of the lamps were identified as the most influential factors for energy consumption. For the yield, the maximum intensity of the lamps, the fraction of lamp input converted to PAR, the transmissivity of sun's PAR through the lamps layer, the reflection of sun's PAR through the lamps layer, and the number of photons per joule within the PAR output of the lamps were the most influential parameters. These conclusions can be used to simplify or adjust lamp models and can help manufacturers by identifying the relevant lamp features to provide in their datasheets for simulation purposes.