When is geopolymer concrete greener? A harmonized-boundary meta-LCA

Abstract

TITLE When is geopolymer concrete greener? A harmonized-boundary meta-analysis of life-cycle assessments of alkali-activated (geopolymer) concrete versus ordinary Portland cement and LC3. AUTHORS Nguyễn Phú Hoàng (sole author, University of Technology and Education – HCMC). REGISTRATION DATE / SEARCH FREEZE DATE 2026-10-03 RESEARCH QUESTION (PICO) With life-cycle system boundaries harmonized to a common basis, under which conditions does geopolymer concrete (GPC) have lower greenhouse-gas emissions than OPC and LC3 — with respect to precursor type, activator source and dosage, grid carbon intensity, transport distance, and replacement ratio? P Product system: alkali-activated (geopolymer) binder/concrete/mortar. I Intervention: geopolymer binder = precursor + alkaline activator + curing. C Comparator: at least one system computed within the same study (OPC, blended cement, or LC3). O Primary: GHG emissions per harmonized functional unit (kg CO2-eq/m3 at a common strength class; also reported per MPa and per kg of binder). Secondary: CED, AP, EP, HTP, cost, composite sustainability indices where reported. S Study design: quantitative LCA reporting a functional unit and system boundary (ISO 14040/14044 or equivalent description). T 2015-01-01 to 2026-10-03. INFORMATION SOURCES OpenAlex (primary) and Crossref (verification), plus forward and backward citation chaining from seed papers. Scopus and Web of Science were NOT available to the author; this is a stated limitation — studies not indexed in OpenAlex/Crossref may be missed. ELIGIBILITY CRITERIA Inclusion: I1 quantitative LCA of geopolymer system (inventory + functional unit); I2 >=1 comparator computed in the same study; I3 results convertible to a defined functional unit or sufficient inventory provided; I4 system boundary, cut-off rules and allocation method stated; I5 journal article or review, English, peer-reviewed, within the time window (reviews used only for citation chaining, not for quantitative synthesis). Exclusion: E1 mechanical/durability only, no LCA; E2 LCA of unrelated material (unless it is the comparator); E3 not peer-reviewed (conference abstract, poster, preprint, editorial, letter); E4 missing parameters that cannot be converted (to be listed openly, not silently dropped); E5 duplicate data source/overlapping study by the same group (keep the most complete report); E6 retracted. SEARCH STRATEGY (executed 2015-01-01 to 2026-10-03; type:article|review; field: title_and_abstract.search) B1 "geopolymer concrete" AND ("life cycle assessment" OR "life cycle analysis" OR "LCA" OR "carbon footprint") B2 "alkali-activated concrete" AND ("global warming potential" OR "CO2 emission" OR "CO2 emissions" OR "carbon footprint") B3 "fly ash" AND geopolymer AND ("life cycle assessment" OR "LCA" OR sustainability) AND (comparative OR comparison) B4 geopolymer AND (slag OR GGBFS OR "ground granulated blast furnace slag") AND ("environmental impact" OR "life cycle assessment") AND ("ordinary portland cement" OR OPC OR comparison) B5 ("limestone calcined clay" OR LC3 OR "calcined clay") AND ("life cycle assessment" OR "carbon footprint" OR LCA OR CO2) AND (geopolymer OR comparison OR blended) B6 geopolymer AND ("sodium silicate" OR activator OR "alkali activator") AND ("environmental burden" OR hotspot OR "environmental impact" OR "global warming") Every record must carry a DOI read from the API field and must pass a reverse Crossref lookup (title match); records failing verification are excluded. SCREENING PROCESS Records are de-duplicated by DOI. Two screening stages: (1) title/abstract, (2) full text. Honest disclosure: screening and data extraction are performed by a SINGLE author — there is no independent second extractor. Compensating controls: (a) an automated completeness check of the extraction table (script flags every NR cell); (b) a self-re-extraction of a random 20% of rows at least 4 weeks later, logged; (c) raw data and code are released openly. No claim of "two independent extractors" will be made. DATA EXTRACTION AND HARMONIZATION Study-level extraction (one row per study). Three harmonization axes: (A) functional unit — all converted to kg CO2-eq/m3 at a common strength class (30/40/50 MPa); (B) system boundary — labelled per study; analyses run separately for groups that include vs exclude the activator; (C) allocation/cut-off — two scenarios, zero-burden vs economic allocation; a sign change in the conclusion is itself a reported finding. Records requiring an assumed density or binder content are flagged and moved from the main analysis to the sensitivity analysis. QUALITY APPRAISAL Seven criteria (functional unit; explicit system boundary; named + versioned background database; activator included); allocation method justified; sensitivity/scenario analysis; extractable tabulated data), each scored 0-2 (0-14 total). Scores are used for soft exclusion in sensitivity analysis and to contrast "high-quality" (>=10) subgroups. SYNTHESIS Qualitative: contradiction matrix across studies. Quantitative: harmonized kg CO2-eq/m3 by strength class, with four sensitivity analyses — grid carbon intensity (0.02-0.9 kg CO2-eq/kWh), transport distance (0-1000 km), activator dosage (3-10% Na2O), and waste allocation (zero-burden vs economic). Meta-regression will NOT be used if fewer than 10 studies or if variances are missing; a descriptive quantitative synthesis will be used instead. PRE-SPECIFIED STOPPING RULE If fewer than 60 eligible LCA studies are found, the paper will be downgraded to a "critical review with a harmonization framework" and will NOT be called a meta-analysis. CHANGES Because these criteria are registered, any later change is recorded as a dated annex; criteria are never silently altered. CODE AND DATA Search scripts, screening scripts, harmonization code and the harmonized dataset will be released openly (CC-BY).