STRATEGIES FOR A CONSTANT DAILY LIGHT INTEGRAL IN GREENHOUSES

At Cornell University a continuous-mode pilot-plant is producing lettuce at a constant rate throughout the year. Plants of all ages, floating on nutrient solution, are present in the greenhouse at all times. They move from the starting point to the harvest point in 24 days, all under the same environmental conditions, repeated day after day throughout the year. The key to the success of such an operation is a constant daily light integral, which can be achieved only through shading and supplemental lighting, as required. Since the costs of supplemental lighting (in winter) and the damage from leaf tip burn (LTB) from excessive light (in summer) are high, a rather tight light control is required. Two control strategies are considered: (A) a rule-based strategy which utilizes a projection of solar radiation over the next few hours, and (B) a strategy with feedback from the difference between actual and reference light trajectories, without a view into the future. These two are compared with a reference strategy (R), which is based on perfect single-day totals light-forecasts. The results show that the success of all strategies depends on the goal function, which is particularly sensitive to surplus in the 3-day light integral (which triggers LTB). The reference strategy R produces the best results. Strategy A, achieving about 10 $/m² less per year than Strategy R (for Ithaca, NY), is much improved when a 3-day light target, rather than a single-day target, is considered. Strategy B, calibrated by means of three parameters, results in a performance similar to that of Strategy A. The conclusions from this initial study are: (1) It is important to take into account the economic goal function when developing a control strategy. (2) There is an advantage in a good (perfect) total daily light forecast, but whether this is possible is doubtful. (3) Strategy B, relatively simple and devoid of any projection into the future, seems as successful as the more elaborate rule-based strategy A, which involves light forecasting. (4) In the examined system, shading control is more critical than lighting control.