This dataset provides a unified, curated collection of experimental results on the hydrothermal conversion of lignocellulosic and lignin-rich biomass, including hydrothermal treatment (HTT), hydrothermal liquefaction (HTL), and related process variants. The database covers diverse biomass feedstocks, reactor configurations, solvents, catalysts, temperatures, residence times, and operating conditions, with all records normalized to consistent units and yield bases.

Data were systematically extracted from peer-reviewed literature, digitized when necessary, and harmonized to a standardized schema. All experimental records include comprehensive provenance tracking. Elemental compositions were normalized to dry or dry-ash-free basis. Missing values for key descriptors (e.g., HHV, lignocellulosic composition, elemental ratios) were imputed using Random Forest models or family-specific median values. Quality-control flags document closure checks for elemental and polymer balances. Derived descriptors include O/C and H/C molar ratios, Lignin Readiness Index (LRI), and energy/carbon recovery metrics.

The dataset comprises one primary CSV file (master_dataset.csv) with 3,693 rows (experimental runs) and 145 columns organized into nine groups:

• Provenance: source publication metadata (DOI, year, author, reference)
• Feedstock Identity: biomass type and family classification
• Feedstock Composition: elemental analysis (C, H, O, N, S, Ash), lignocellulosic components (Lignin, Cellulose, Hemicellulose), HHV, elemental ratios
• Process Conditions: temperature (typically 200-400°C), residence time, pressure, solvent, catalyst, reactor type, heating rate, biomass loading
• Yields: bio-oil, char, gas, and aqueous phase yields (wt%)
• Bio-oil Properties: elemental composition, HHV, O/C and H/C ratios, carbon yield
• Char Properties: elemental composition, HHV, O/C and H/C ratios, carbon yield
• Tracking: flags and methods documenting imputation sources, measurement methods, and quality-control notes
• Other: derived descriptors, closure flags, normalized indices

Format: CSV (UTF-8 encoding)
Size: 3,693 rows × 145 columns
Temporal Coverage: 1982-2026 (40+ years of published research)
Temperature Range: 200-400°C (typical hydrothermal conditions)
Feedstock Families: 15+ standardized categories (e.g., Softwood, Hardwood, Agricultural Residue, Herbaceous, Kraft Lignin)
Data Completeness: Core features (C, O, temperature, time) >98%; product yields 70-80%; detailed product composition 20-40%
Schema: Standardized column names, units, and yield bases; metadata provided in JSON, XML, and CSV formats

• All yields normalized to dry or dry-ash-free feedstock basis as specified in yield_basis column
• Oxygen content calculated by difference when not directly measured (documented in O_method column)
• HHV estimated using Channiwala-Parikh correlation when not experimentally measured (flagged in HHV_feedstock_method)
• Lignocellulosic composition imputed for ~4% of records using ML models trained on family-specific patterns
• Closure flags (LCH_closure_flag) indicate quality of lignocellulosic balance (sum of Lignin+Cellulose+Hemicellulose)
• Some studies report combined gas+water yields; separated when possible, otherwise flagged
• Catalyst = 'none' explicitly indicates non-catalytic runs
• FAIR-compliant: machine-readable metadata, standardized schemas, comprehensive provenance, CC-BY-4.0 license

========================================
TECHNICAL METADATA
Biomass HTT/HTL Dataset

DATASET IDENTIFICATION

Title: Biomass HTT/HTL Dataset
Version: 1.0.0
Created: 2026-01-06
License: CC-BY-4.0

DATASET DIMENSIONS

Total Rows (Experimental Runs): 3,693
Total Columns (Features): 145
Temporal Coverage: 1982-2026 (40+ years)

KEYWORDS

• biomass
• hydrothermal liquefaction
• hydrothermal treatment
• HTL
• HTC
• bio-oil
• biochar
• lignocellulosic biomass
• lignin
• machine learning
• LCA

DATA STRUCTURE

The dataset is organized into 9 column groups with 145 total features:

1. PROVENANCE (6 columns)

   • paper_title, DOI, year, Provenance, Ref, process_type
   • Purpose: Source publication tracking and reference metadata

2. FEEDSTOCK IDENTITY (2 columns)

   • Feedstock, Family_std
   • Purpose: Biomass type identification and standardized classification

3. FEEDSTOCK COMPOSITION (16 columns)

   • Elemental: C_feed_wt_pct, H_feed_wt_pct, O_feed_wt_pct, N_feed_wt_pct, S_feed_wt_pct, Ash_feed_wt_pct
   • Polymeric: Lignin_feed_wt_pct, Cellulose_feed_wt_pct, Hemicellulose_feed_wt_pct, Extractives_feed_wt_pct
   • Ratios: O_C_feed_molar, H_C_feed_molar
   • Energy: HHV_feed_MJ_per_kg
   • Indices: LRI (Lignin Readiness Index)
   • Moisture: Moisture_min_wt_pct_ar, Moisture_max_wt_pct_ar

4. PROCESS CONDITIONS (16 columns)

   • Temperature: T_reaction_C (typically 200-400°C)
   • Time: t_residence_min, t_ramp_min
   • Reactor: process_subtype, reactor, atmosphere
   • Medium: solvent_or_medium, IC_feed_wt_pct_slurry, water_biomass_ratio_kg_kg
   • Pressure: pressure_reaction_MPa
   • Catalyst: catalyst, cat_biomass_ratio_kg_kg
   • Other: heating_rate_C_per_min, stirring_rpm, yield_basis, separation_method

5. YIELDS (7 columns)

   • Mass yields: Yield_biooil_wt_pct, Yield_char_wt_pct, Yield_aqueous_wt_pct, Yield_gas_wt_pct, Yield_gas_water_wt_pct
   • Energy yields: Energy_yield_biooil_pct, Energy_yield_char_pct

6. BIO-OIL PROPERTIES (9 columns)

   • Elemental: C_biooil_wt_pct, H_biooil_wt_pct, O_biooil_wt_pct, N_biooil_wt_pct, S_biooil_wt_pct
   • Ratios: O_C_biooil_molar, H_C_biooil_molar
   • Energy: HHV_biooil_MJ_per_kg
   • Carbon recovery: Carbon_yield_biooil_pct

7. CHAR PROPERTIES (9 columns)

   • Elemental: C_char_wt_pct, H_char_wt_pct, O_char_wt_pct, N_char_wt_pct, S_char_wt_pct
   • Ratios: O_C_char_molar, H_C_char_molar
   • Energy: HHV_char_MJ_per_kg
   • Carbon recovery: Carbon_yield_char_pct

8. TRACKING (24 columns)

   • Method documentation: C_method, O_method, OC_method, HC_method, S_method, etc.
   • Imputation flags: LRI_imputed, HHV_feedstock_imputed, Lignin_imputed, cellulose_imputed, hemicellulose_imputed, Ash_imputed
   • Quality notes: C_Note, O/C_Note, H/C_Note, t_note
   • Source tracking: LRI_imputed_source

9. OTHER / DERIVED DESCRIPTORS (76 columns)

   • Compositional: LCH_total_wt_pct, Lignin_share_pct, Holo_share_pct, sum_LCH_wt_pct, sum_CHONSAsh_wt_pct
   • Quality flags: LCH_closure_flag, Ash_adjusted
   • Imputed values: LCH_total_imputed_wt_pct, Lignin_imputed_from_LCH, cellulose_imputed_from_LCH, hemicellulose_imputed_from_LCH
   • Readiness indices: LRI_dd, CeRI_dd, HeRI_dd, HRI_dd, CRI_dd, HeRI_comp_dd
   • Fractions: CeFrac_dd, L_share_idx, CeFrac_idx, CeFrac_mb
   • Estimates: Lignin_HHV_est, Lignin_C_est, Lignin_est, Holocellulose
   • Normalized features: inv_OC, inv_HC, ash_inv, C_norm, HHV_norm
   • Performance indices: LPI, HoPI
   • Source tracking for all imputed/derived features

DATA COMPLETENESS BY GROUP

Core Features (>98% complete):

• C_feed_wt_pct: 100.0%
• O_feed_wt_pct: 100.0%
• H_feed_wt_pct: 98.7%
• T_reaction_C: 99.97%
• t_residence_min: 99.65%
• O_C_feed_molar: 100.0%
• H_C_feed_molar: 100.0%
• HHV_feed_MJ_per_kg: 100.0%
• Family_std: 100.0%

Compositional Features:

• Lignin_feed_wt_pct: 96.18%
• Cellulose_feed_wt_pct: 95.61%
• Hemicellulose_feed_wt_pct: 95.61%
• Ash_feed_wt_pct: 99.68%
• N_feed_wt_pct: 78.53%
• S_feed_wt_pct: 54.78%

Process Conditions:

• reactor: 92.5%
• solvent_or_medium: 99.86%
• IC_feed_wt_pct_slurry: 98.08%
• water_biomass_ratio_kg_kg: 90.74%
• atmosphere: 65.88%
• heating_rate_C_per_min: 24.51%
• stirring_rpm: 21.18%

Product Yields:

• Yield_biooil_wt_pct: 80.67%
• Yield_char_wt_pct: 71.41%
• Yield_gas_wt_pct: 30.73%
• Yield_aqueous_wt_pct: 26.1%

Bio-oil Properties:

• C_biooil_wt_pct: 36.15%
• HHV_biooil_MJ_per_kg: 36.61%
• H_biooil_wt_pct: 35.07%
• O_biooil_wt_pct: 34.09%
• Carbon_yield_biooil_pct: 32.39%
• Energy_yield_biooil_pct: 31.76%

Char Properties:

• HHV_char_MJ_per_kg: 25.24%
• H_char_wt_pct: 21.31%
• C_char_wt_pct: 20.82%
• O_char_wt_pct: 20.8%
• Energy_yield_char_pct: 22.2%
• Carbon_yield_char_pct: 19.04%

DATA TYPES

• String (categorical): 44 columns
  Examples: Feedstock, Family_std, reactor, catalyst, process_type, DOI

• Float (continuous): 91 columns
  Examples: T_reaction_C, t_residence_min, Yield_biooil_wt_pct, C_feed_wt_pct

• Integer: 1 column
  Examples: year

• Boolean (flags): 9 columns
  Examples: Ash_adjusted, HHV_feedstock_imputed, Lignin_imputed, N_imputed

UNITS OF MEASUREMENT

Temperature: °C (degrees Celsius)
Time: min (minutes)
Pressure: MPa (megapascals)
Energy: MJ/kg (megajoules per kilogram)
Composition: wt% (weight percent, dry basis unless specified)
Yields: wt% (weight percent of initial dry feedstock)
Ratios: dimensionless or molar ratios
Speed: rpm (revolutions per minute)
Heating rate: °C/min (degrees Celsius per minute)

KEY INDICES AND DERIVED FEATURES

1. Lignin Readiness Index (LRI):
   Formula: (Lignin/100 + C/60 + HHV/22 + (1/(O/C))/2 + (1/(H/C))/2) / 6
   Purpose: Composite indicator of feedstock suitability for lignin-focused conversion

2. O/C and H/C Ratios:
   Van Krevelen diagram coordinates for characterizing biomass and products

3. LCH Closure:
   Sum of Lignin + Cellulose + Hemicellulose with quality flags

4. Carbon and Energy Recovery:
   Mass balance tracking for carbon and energy distribution

5. Readiness Indices (LRI_dd, CeRI_dd, HeRI_dd):
   Data-driven indices for component-specific conversion prediction

QUALITY CONTROL FEATURES

• LCH_closure_flag: Polymer mass balance validation
• sum_CHONSAsh_wt_pct: Elemental mass balance check
• Imputation tracking: Boolean flags for all imputed values
• Method documentation: Source/calculation method for derived values
• Provenance: Complete source publication tracking

DATA NORMALIZATION

• All yields normalized to specified basis (dry, daf, as-received)
• Elemental compositions on dry or dry-ash-free basis
• Consistent unit conversions applied across all sources
• Standardized feedstock family classification
• Harmonized column naming convention

IMPUTATION METHODOLOGY

• Random Forest models for continuous features
• Family-specific median values as fallback
• Formula-based calculation for stoichiometric features
• All imputations documented with method flags and source columns
• ~4% of lignocellulosic composition imputed

SPECIAL NOTES

• Oxygen calculated by difference when not measured (O_method column)
• HHV estimated via Channiwala-Parikh correlation when needed
• Catalyst = 'none' explicitly indicates non-catalytic runs
• Some gas+water yields reported together (flagged accordingly)
• pressure_reaction_MPa stored as string to accommodate "autogenous"

FILE FORMAT

Primary file: master_dataset.csv
Encoding: UTF-8
Delimiter: comma (,)
Quote character: double quote (")
Header: Yes (first row contains column names)

RELATED FILES

• metadata.json: Complete metadata in JSON format
• metadata.xml: Complete metadata in XML format
• metadata_radar.xml: RADAR repository-compliant metadata
• column_metadata.csv: Column-level documentation with descriptions and units
• RADAR_DESCRIPTION.txt: Structured description for repository upload
• ABSTRACT.txt: Dataset abstract

CITATION

When using this dataset, please cite:
[Citation information to be added after DOI assignment]

FAIR COMPLIANCE

• Findable: Unique DOI, comprehensive metadata, standardized keywords
• Accessible: Open license (CC-BY-4.0), standard formats (CSV, JSON, XML)
• Interoperable: Standardized schemas, documented units, machine-readable formats
• Reusable: Complete provenance, quality flags, imputation documentation

APPLICATIONS

• Machine learning model development for process optimization
• Comparative analysis of conversion technologies
• Life cycle assessment (LCA) studies
• Feedstock-to-product relationship modeling
• Process condition optimization
• Digital chemistry and data-driven discovery

========================================
Last Updated: 2026-01

The related GitHub repository containing advanced data analysis, sanity checks, metadata , curation examples... is provided through this link https://github.com/SFETNI/BiomassLignocellulose-HTL-HTT-HTC-Conversion-Data