ED Sciences Physiques et de l'Ingénieur
Study of an ecological geopolymer concrete with ultra high performance based on industrial by-products and recycled materials.
by Youssef SLEIMAN (I2M - Institut de Mécanique et d'Ingénierie de Bordeaux)
The defense will take place at 9h00 - Salle A9 Université de Bordeaux, 351 cours de la Libération 33405 Talence CEDEX
in front of the jury composed of
- Nadia SAIYOURI - Professeure des universités - Université de Bordeaux - Directeur de these
- Zoubir-Mehdi SBARTAÏ - Professeur des universités - Université de Bordeaux - CoDirecteur de these
- Mokhfi TAKARLI - Maître de conférences - Université de Limoges - Examinateur
- Karim AÏT-MOKHTAR - Professeur des universités - Université de La Rochelle - Examinateur
- Yun JIA - Maîtresse de conférences - Université de Lille - Rapporteur
- Ali-Nordine LEKLOU - Professeur des universités - Université de Nantes - Rapporteur
Offshore floating photovoltaic substructures are currently built in Portland-cement ultra-high-performance concrete (UHPC), whose embodied carbon sits near the upper bound of the construction sector. This thesis develops, characterizes, models, and assesses a 100% cement-free, ambient-cured Ultra-High Performance Geopolymer Concrete (GEO) as a replacement candidate, within the Horizon Europe project NATURSEA-PV. A six-stage, bottleneck-driven optimization of a GGBS and silica fume binder activated by sodium silicate and 10 M KOH, reinforced with 108 kg/m3 of tire-derived recycled and virgin steel fiber, satisfies the five project acceptance criteria simultaneously under ambient curing: 65 MPa at 24 hours, 152 ± 5 MPa at 28 days on EN 196-1 prism halves, 208 mm slump flow, 23 min open time, and 60 wt % recycled fiber. The 28-day value is 92% of the Portland-cement reference REF, at 165 ± 4 MPa on the matched geometry. Electrochemical impedance spectroscopy from 1 to 28 days shows the optimal strength indicator to be binder-specific: the Warburg coefficient tracks GEO strength (r = 0.983 on log axes, p < 0.001), statistically indistinguishable from the gel pore resistance R2 adopted for REF (r = 0.987), and neither parameter serves the other binder. Acoustic emission under three-point bending identifies a cohesive, debonding-dominated fracture pathway in GEO against friction-dominated pull-out in REF; a continuum-damage model calibrated on the measured flexural response reproduces the peak load to within 2.3% but not the post-peak branch. Twelve months of sulfate and acid immersion, chloride migration and thermal programs, and a 10-month offshore campaign at HarshLab establish superior sulfate and acid resistance, no chloride or carbonation front after real immersion, and full retention of the UHPC mechanical class in all three exposure zones; the single deficit, 36% compressive retention at 600 °C against 90% for REF, lies outside the marine service envelope. A cradle-to-grave carbon footprint assessment under EN 15804+A2 returns a 37% reduction per m3 and 32.1% per tonne against a verified third-party EPD for the reference UHPC, meeting the project carbon target, while the cost-to-strength target is not met, at 10.9% against a 35% requirement. GEO is therefore mechanically compliant, monitorable, marine-durable, and substantially lower in carbon than the UHPC it replaces.