HOPCA: Hospital Layout Design Optimization using Computational Architecture

Abstract

Hospital facilities are known to be functionally complex buildings in various ways namely due to their non-trivial spatial connectivity requirements. There are several typical problems in hospitals that can be attributed to the configuration of the building namely the inefficient circulation of medical staff difficult wayfinding for visitors lengthy and complex procedures long walking times and so on. This Ph.D. research aims to investigate the relation between the performance of hospital buildings focusing on their configurational layout at various levels of abstraction and to develop a computational design methodology for configurational layout optimization of hospital buildings. To this end spatial layout methods were devised and tested by using Computational Design techniques derived from the fields of Graph Theory Operations Research and Computational Intelligence. The presented research was carried out based on the actions performed in three main parts: Literature Review Preliminary Research and Method Development with Case Studies. Firstly a worldwide literature review was conducted on “configurational layout in general” and “hospital layout” studies to highlight the correspondence between configurational layout methods and hospital layout problems in the current literature. Secondly preliminary research was conducted on configurational design requirements to identify a typical program of requirements (PoR) and spatial connectivity adjacency and closeness requirements of hospitals (REL-Chart) for which expertise knowledge on-site observations hospital design codes and standards were utilized. In addition international hospital design guidelines were reviewed and a discrete event simulation method that can model the patient flows was developed and programmed as a design toolkit to define the space requirements in CAD software by considering the patient waiting times. The last part is Method Development & Case Studies in which two parallel tracks of work were reported on “Reconfiguration of existing hospitals” and “Designing hospitals from scratch”. The core concept of the reconfiguration track is assigning hospital units to the existing locations such that the assignment cost (flows of people distances) is minimized. The core concept of designing from scratch is based on a holistic approach consisting of stacking zoning and routing. In stacking a spectral clustering method was proposed for defining the levels of hospital units. In the stage of zoning hospital units were assigned to the faces of a mesh-surface representation of the floorspace such that the closeness of the interrelated spaces is minimized subject to various constraints. In routing the corridors were designed by assigning circulation routes to the edges of a mesh-surface representation of the available floor space considering wayfinding and work-related interruptions. Due to the nature of the problems in the first track for redesigning existing hospitals Quadratic Assignment Problem (QAP) formulation with geodesic distances and a heuristic optimization algorithm were proposed. In the second track for designing new hospitals Mixed Integer Programming (MIP) formulations were suggested. The two models have been programmed as a computational design toolkit that is available in popular CAD software in architecture using C# and Python programming languages. The models are not case-based, but for the test and validation of the layout results each of the models was applied to a real-world case study hospital in Izmir Turkey. The models were created considering Turkey's hospital design codes and standards and specific to mid-size general hospitals however the models can be modified later according to the different regulations of other countries. Effectively this project presents an interdisciplinary methodological framework that can tackle hospital layout design problems by integrating Computational Design workflows Graph Theory techniques Operations Research and Computational Intelligence into the field of Architectural Space Planning. © 2023 Elsevier B.V. All rights reserved.