Adapting building stock archetype methodology for greenhouse stock characterisation

The development of archetypes is a well-established approach for modelling the energy consumption of building stocks; however, its application to greenhouse stocks remains largely unexplored. This article proposed a method for greenhouse stock characterisation, using electric greenhouses in Quebec as a case study. The proposed method relied on k-means clustering applied to a sample of 104 greenhouses, categorised into three electric energy use types: electric heating (GHe), electric heating with artificial lighting (GHLe), and electric lighting with non-electric heating (GLe). Clustering was based on heating degree-days and energy use intensity per production days (EUI_pp) for GHe and GHLe , and on latitude and EUI_pp for GLe . The optimal number of clusters (k = 3) was determined using the elbow method and silhouette index, with scores ranging from 0.54 to 0.72, resulting in the identification of nine distinct archetypes. Each archetype was characterised according to its geometry, cover material, thermal transmittance, and crop type. Geometry was assigned in a deterministic manner, resulting in three individual and six multi-span greenhouse archetypes. Cover material exhibited a discrete distribution (68% double polyethylene (Ped), 32% glass), while the associated thermal transmittance was characterised by a uniform probability distribution (UPD) specific to each material. Crop type was defined by a discrete distribution: 85% for warm-season and 15% for cool-season. This approach provided a structured segmentation of a heterogeneous greenhouse stock into distinct clusters and archetypes, thereby establishing the foundation for energy modelling of Quebec’s electric greenhouses and supporting its transferability to other regions with analogous data.