Beyond Point Calibration: Identifiability and Observable Design in Discrete Element Calibration for Near-Nozzle 3D Concrete Printing
Journal of Architectural and Planning Research · 30.06.2023 · 37 (3) · 31-42
DOI: 10.66033/japr2022-303 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Architectural and Planning Research, online erschienen am 30.06.2023. Verfasst von Donald Shoup und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Calibration studies for fresh cementitious materials are still commonly judged by whether one selected parameter set reproduces one or two macroscopic tests. Much less attention is paid to whether the chosen observables actually identify the underlying DEM micro-parameters in a unique and stable way. This issue is especially important for near-nozzle mixing in extrusion-based 3D concrete printing, where calibration must distinguish free-flowing aggregate behavior, fluid-like paste behavior, and cohesive mixed-mortar behavior within one workflow. Using the central composite design (CCD) data of as a secondary-analysis dataset, this study quantifies the identifiability gains obtained by enriched observables in DEM calibration. A total of 46 simulation runs were reanalyzed: 13 for Poraver®, 20 for paste, and 13 for mortar. Second-order response-surface surrogates were re-estimated, and admissible parameter regions were compared under competing observation schemes using feasible-set fraction, parameter-span reduction, and Pareto trade-off analysis. The reanalysis shows that observability is strongly material dependent within the CCD domain. For Poraver®, an angle-of-repose target alone leaves essentially the full restitution search width admissible at a 1% tolerance, whereas adding the pile-shape index reduces the feasible fraction of design space from 0.702% to 0.089% and contracts the admissible restitution width by 86.7%. For paste, slump diameter alone weakly constrains the friction parameters; both friction coefficients remain effectively unresolved over the search range, while low-cracking Pareto-efficient solutions concentrate in a narrow low-adhesion band and therefore motivate a better experimental crack descriptor. For mortar, the value of the standardized Haegermann sequence is pronounced: at a 2% tolerance, using only the fifteenth-stroke diameter leaves 11.755% of the parameter space admissible, whereas using the full d5–d10–d15 sequence reduces this fraction to 0.036% and confines the admissible surface-energy interval to 6.02 J m−2–6.40 J m−2. The main contribution of enriched calibration protocols is therefore not only better fit, but materially improved parameter identifiability. Because the analysis is based on deterministic surrogates fitted within CCD bounds, the claims are comparative rather than universal; within that scope, the study offers a transparent framework for judging which experiments are genuinely informative in DEM calibration for additive manufacturing in construction.
Zitieren
APA
Shoup, D., & Liu, L. (2023). Beyond Point Calibration: Identifiability and Observable Design in Discrete Element Calibration for Near-Nozzle 3D Concrete Printing. Journal of Architectural and Planning Research, 37(3), 31-42. https://doi.org/10.66033/japr2022-303
Vancouver
Shoup D, Liu L. Beyond Point Calibration: Identifiability and Observable Design in Discrete Element Calibration for Near-Nozzle 3D Concrete Printing. Journal of Architectural and Planning Research. 2023;37(3):31-42. doi:10.66033/japr2022-303
BibTeX
@article{Shoup2023,
author = {Shoup, Donald and Liu, Lijun},
title = {Beyond Point Calibration: Identifiability and Observable Design in Discrete Element Calibration for Near-Nozzle 3D Concrete Printing},
journal = {Journal of Architectural and Planning Research},
year = {2023},
volume = {37},
number = {3},
pages = {31--42},
doi = {10.66033/japr2022-303}
}
EUVIMED