Simulation of the energy, carbon, and water balance of a fluid-roof greenhouse

Abstract

The energy, water, and CO(,2) balance of a greenhouse was analyzed with two simulation models: an existing one that simulated the flows of energy and mass in a greenhouse, based on a single leaf crop representation, and an expanded version that included a more detailed crop submodel. Each model was tested with data obtained on a test greenhouse fitted either with a fluid roof or a double plastic roof. Data describing the ambient weather and greenhouse conditions were collected over nine periods of 1-4 days. The weather data included solar irradiance, air and dew point temperatures, wind speed, and, in three tests, CO(,2) concentration. These data were used to drive each of the two models, for calculating greenhouse conditions. Greenhouse data included air, crop, individual leaf, roof, and fluid storage tank temperatures, interior air humidity, and in four tests, CO(,2) concentration. The exchange rates of water vapor and CO(,2) were also calculated from the measurements. Direct measurements of crop water use were available for the first five tests. Results from a comparison of calculated and measured greenhouse variables were mixed. The simulations of the energy and water vapor balances were adequate, but the simulation of the CO(,2) balance was not. Temperatures were generally simulated within 2.5(DEGREES)C (root mean square deviation), and the humidity within 2.0 g/m('3), during those tests in which reliable psychrometric data were obtained. The net radiation above and below the roof was simulated within 15-65 W/m('2). The simulated CO(,2) concentrations and exchange rates differed significantly from the measurements, but the latter showed considerable random error. There was no clear distinction in the performance of the two models. The simpler model fitted the data better in 67 of 162 variable-days, the expanded model in 70, and 25 were even. The expanded model required 6 times the computing time as the original one, hence one should use the simple one unless details of the crop action are of interest. In future uses of the simple model, transfer rates of the crop could be made dependent on leaf area. Also, the ventilation rate should be measured more accurately.