Traduzioni digitali di architetture di carta. Un approccio HBIM per l'interpretazione e la comunicazione degli archivi di architettura del Novecento.

The research addresses the digitisation of twentieth-century architectural archives as a question of cognitive accessibility, interpretation and communication of design knowledge. The online availability of records, inventories and digital surrogates extends the consultation of archival holdings, yet it often leaves implicit the relationships between sources, design sequences, variants, gaps and interpretative decisions. For this reason, the thesis adopts the notion of "paper architectures" in an extended sense: unrealised projects, transformed works, lost buildings, configurations documented through physical models, and hypotheses reconstructed from fragmentary sources. Under these conditions, the archival document is considered part of a relational system in which drawings, models, photographs, reports and administrative materials preserve traces of the design process and require a critical reading grounded in architectural drawing and representation. The work proposes an "as-drawn" HBIM methodology aimed at making explicit the transition from primary source to information model. The methodology integrates survey and virtual reordering, multi-level description, GIS geodatabases, document digitisation, photogrammetric survey of physical models, information modelling and digital communication devices. The HBIM model is understood as a verification environment in which geometry, semantics, sources, decisions and paradata remain queryable. The operational structure is organised into coordinated models: COO, dedicated to documentary and geometric coordination; STR, aimed at the structural hypothesis; ARC, devoted to architectural restitution; and ANL, understood as a federated model for queries, schedules, thematic views, graphic outputs and inference control. The decision matrix records sources, critical issues, DDI, rule type, geometric strategies, operations and paradata notes, making it possible to distinguish direct evidence, deduced data and interpretative hypotheses. The methodological verification is developed through two main categories. The "holding to be reactivated" concerns the Salvatore Rattu fonds preserved at the DICAAR of the University of Cagliari, characterised by heterogeneous documentation, partly ordered and partly unpublished. The research tests its critical recomposition through three archival units: the project for the Casa del Mutilato, as an existing case; the project for a Scuola Professionale Marinara, as an unrealised architecture; and the new parish church of San Paolo, as a transformed architecture structured through operational variants. The "holding to be communicated" concerns the Spanish case studies of Miguel Fisac’s Laboratorios Jorba and the Anteproyecto del edificio de Instituto de Restauraciones Artísticas by Fernando Higueras and Rafael Moneo, in which HBIM translation supports the communication of lost, transformed or fragmentarily documented architectures. The selection of the case studies also highlights the relevance of serialisation and industrialisation in twentieth-century building processes, since repetitions, modules, construction rules, variants and exceptions make the relationship between design structure and information structure particularly verifiable. Parametric modelling enables the formalisation of families, types and dependencies, while the quantification of inference measures the distribution of the interpretative load across models, disciplinary domains and documentary configurations. The overall objective is to define a digital archive-museum capable of connecting conservation, description, interpretation and communication. The thesis demonstrates how digital translation can make fragile documentary heritage newly queryable, while keeping verifiable the relationship between what is modelled, what is inferred and the sources that justify each decision throughout the methodological process and its communicative outcomes.