Using inference from user attribution of models to support high resolution modelling

Abstract

Air flow networks are for many practitioners a here-bedragons territory and despite providing useful information short of CFD assessments it has remained a niche activity. The paper discusses how we might transition from user-imposed air flows to computed air flows for long term high frequency assessments within simulation practice. It explores the historical, developmental, domain knowledge and model quality barriers to this transition. It describes the implementation of an inference facility to assist in the creation of flow networks of hundreds of entities and applies this in case studies to show what a new normal might look like.