Systematic audit of uncertainty treatment in high-impact biorefinery techno-economic assessment and life cycle assessment case studies (2020–2025)

This dataset was compiled to test the hypothesis that a structural decoupling exists in how uncertainty is treated in high-impact biorefinery techno-economic assessment (TEA) and life cycle assessment (LCA). Specifically, we hypothesized that while probabilistic methods are routinely applied to downstream financial and environmental indicators, intrinsic feedstock variability is rarely propagated through the non-linear technical core of process models—creating a systematic blind spot that deterministic averaging cannot reveal. Data collection: Scopus was searched using TITLE-ABS-KEY ((TEA OR LCA) AND (Biorefinery OR Biomass) AND (Sensitivity OR Monte Carlo)). From 757 initial records (2020–2025), studies with Field-Weighted Citation Impact > 2.0 were retained. Phase 1 screening required primary case studies with explicit process modeling and reported feedstock composition (at minimum mean values). After exclusions, 159 studies underwent full methodological coding in Phase 2. Classification framework: Studies were categorized by uncertainty approach: Hybrid stochasticity (HYB; probabilistic methods applied to financial/LCA parameters only), Local sensitivity (OAT; one-at-a-time perturbations), Scenario-based (SCN; discrete scenarios), and Deterministic (DET; point estimates only). Feedstock variability reporting was coded as Yes (dispersion metrics provided), Partial (incomplete), or No (point estimates only). What the data show: The dataset reveals a persistent methodological asymmetry. Probabilistic methods are widely adopted for market and environmental indicators, yet upstream technical transformation stages remain predominantly mean-based. Only a minority of studies propagate feedstock variability through non-linear process cores where kinetic constraints, transport limitations, and mass–energy coupling govern system behavior. When feedstock variability is reported, it is often incomplete or collapsed into point estimates. Key findings: Of the 159 audited studies, 47% employed OAT, 31% HYB, 15% SCN, and 7% DET. Full propagation of feedstock variability occurred in only 12% of cases; 23% partially reported variability, and 65% collapsed inputs into point estimates. This layered methodological decoupling means that curvature-driven effects originating from intrinsic variability remain unresolved, even when sophisticated financial risk analysis is applied. How to interpret and use these data: Each row represents one peer-reviewed study with full methodological classification. The "Observed modeling approach" column directly maps to the categories defined above. "Feedstock variability explicitly reported?" indicates data quality. Researchers can use this dataset to: (1) benchmark their own uncertainty practices against the field, (2) identify literature gaps for meta-analyses, or (3) extract case studies illustrating specific uncertainty approaches.本数据集旨在验证下述假说：在高影响力生物炼制技术经济评估（techno-economic assessment, TEA）与生命周期评估（life cycle assessment, LCA）中，不确定性的处理方式存在结构性脱钩。具体而言，我们提出的假说为：尽管概率方法已常规应用于下游财务与环境指标的分析，但原料本征变异性极少通过过程模型的非线性技术核心进行传递，由此产生系统性盲区——而确定性平均法无法揭示此类盲区。 数据采集：通过Scopus数据库，以标题-摘要-关键词（TITLE-ABS-KEY）检索式((TEA OR LCA) AND (Biorefinery OR Biomass) AND (Sensitivity OR Monte Carlo))进行检索。在2020至2025年间的757条初始记录中，保留领域加权引文影响分值（Field-Weighted Citation Impact）大于2.0的研究。第一阶段筛选要求纳入包含明确过程建模且至少报告了原料组成均值的核心案例研究。经排除后，共159项研究进入第二阶段的完整方法学编码流程。 分类框架：研究按不确定性处理方法划分为以下类别：混合随机性（Hybrid stochasticity, HYB）：仅针对财务与LCA参数应用概率方法；局部敏感性分析（Local sensitivity, one-at-a-time perturbations, OAT）：单次单变量扰动法；基于情景分析（Scenario-based, SCN）：离散情景法；确定性分析（Deterministic, DET）：仅采用点估计法。原料变异性的报告情况被编码为「是」（提供离散程度指标）、「部分」（报告不完整）或「否」（仅报告点估计值）。 数据集所揭示的结论：本数据集展现出持续存在的方法学不对称性。概率方法已被广泛应用于市场与环境指标的分析，但上游技术转化环节仍主要采用均值基分析。仅有极少数研究将原料变异性传递至非线性过程核心——该核心的系统行为由动力学约束、传输限制以及质能耦合关系所决定。当报告原料变异性时，其结果往往不完整，或被简化为点估计值。 关键发现：在159项经审核的研究中，47%采用单次单变量扰动法（OAT），31%采用混合随机性方法（HYB），15%采用基于情景分析法（SCN），7%采用确定性分析法（DET）。仅12%的研究实现了原料变异性的完整传递；23%的研究仅部分报告了变异性，65%的研究将输入变量简化为点估计值。这种分层的方法学脱钩意味着，即便采用了成熟的财务风险分析方法，由本征变异性引发的曲率驱动效应仍未得到解决。 数据解读与使用方法：每一行代表一项经过同行评审的研究，并附带完整的方法学分类信息。"Observed modeling approach"列直接对应上述定义的类别。"Feedstock variability explicitly reported?"项用于表征数据质量。研究人员可通过本数据集实现以下用途：(1) 对照当前领域基准评估自身的不确定性处理实践；(2) 识别元分析（meta-analyses）所需的文献缺口；(3) 提取用于说明特定不确定性处理方法的案例研究。