Figures (9)  Tables (6)

    • Figure 1. 

      Processing scheme of Reax-Lump for generating lumped mechanisms of polymer plastics based on ReaxFF MD simulations.

    • Figure 2. 

      Confusion matrices of the reaction classification model: (a) the model training set consists of the original dataset (n-dodecane, RP-1, RP-3); (b) the model training set is augmented with an additional 924 data points from the PP pyrolysis dataset.

    • Figure 3. 

      Schematic diagram of the collaborative training framework of Reax-Lump.

    • Figure 4. 

      (a) Number of reactions after different steps for PP heating at 4 K/ps. (b) The optimized loss function and mechanism scale curve for PP heating at 4 K/ps.

    • Figure 5. 

      (a)−(j) Comparison of the predicted number of fragments between ReaxFF MD and lumped mechanisms under 1, 2, 4, and 8 K/ps heating rates. (k) heatmap of R2 for fitting between predicted and experimental concentrations of various species at different heating rates.

    • Figure 6. 

      The yields of small-molecule gases under different temperatures and heating rates.

    • Figure 7. 

      The distribution of the total 29 reaction classes obtained from ReaxFF MD simulation by using SRG-Reax at different heating rates.

    • Figure 8. 

      Evolution trends of (a) C2H4 yields, and (b) RxC2/RxC1 ratios with temperature.

    • Figure 9. 

      Comparison of the proportion of reaction categories related to [C11~C20] at various temperature stages under the heating rate of 1 K/ps. (a) 1,750–2,100 K; (b) 2,100–2,500 K; (c) 2,500–3,000 K.

    • Evaluation indicator The entire
      validation set      		 Part of the validation set
      (reactant carbon numbers > 16)      		 Part of the validation set
      (reactant carbon numbers ≤ 16)      		 Prediction on validation set after
      incorporating PP dataset
      Accuracy 0.5069 0.4192 0.8001 0.9770
      Precision 0.5692 0.4312 0.9125 0.9831
      Recall 0.5069 0.4192 0.8000 0.9770
      F1-score 0.5158 0.4212 0.8184 0.9790

      Table 1. 

      Comparison of evaluation metrics for predictions using different validation sets and classification models.

    • NumberLumped nameLumped ruleRepresentative industrial products
      1[C200+]> 200PP and its oligomers
      2[C41~C200]200 ≥ n > 40Oligomers and wax
      3[C21~C40]40 ≥ n > 20Wax
      4[C11~C20]20 ≥ n > 10Heavy oil
      5[C6~C10]10 ≥ n > 5Light oil
      6[C3~C5]5 ≥ n ≥ 3Gas
      7C2H6EthaneGas
      8C2H4EthyleneGas
      9CH4MethaneGas
      10H2HydrogenGas

      Table 2. 

      Lumping rules for the species in the PP pyrolysis process.

    • Heating rate (K/ps) Detailed reaction number Lumped reaction number Vertical combination reaction number Optimized reaction number
      1 37,931 2,911 1,653 18
      2 19,831 1,881 1,080 19
      4 11,539 1,424 801 26
      8 5,986 1,269 706 27

      Table 3. 

      Reaction scale after different reduction steps.

    • Heating rate (K/ps) Pre-lumping average reaction frequency Post-lumping average reaction frequency Loss value before optimization Loss value after optimization
      1 1.005 13.08 1,804.8 18.6
      2 1.004 7.94 1,184.4 14.87
      4 1.003 7.43 1,746.0 1.28
      8 1.003 4.73 472.0 0.61

      Table 4. 

      Changes in the average reaction frequency and loss function before and after Reax-Lump optimization under different heating rates.

    • Reaction ID Reaction $\ln A $ Ea (kJ/mol)
      1 [C200+] → [C21~C40] + 3[C41~C200] 42.433 326.289
      2 2[C11~C20] → [C21~C40] 41.282 164.841
      3 [C41~C200] + [C6~C10] → 3[C3~C5] + [C21~C40] 42.068 215.119
      4 [C21~C40] + CH4 → [C3~C5] + [C11~C20] 46.783 285.703
      5 [C41~C200] + CH4 → [C11~C20] + 2[C21~C40] 37.222 112.062
      6 [C3~C5] + [C6~C10] → [C11~C20] 34.095 60.606
      7 2CH4 → C2H6 + H2 39.102 158.663
      8 [C21~C40] → 2[C11~C20] + 2[C3~C5] 35.762 229.277
      9 [C3~C5] + C2H6 → [C6~C10] 39.742 123.188
      10 [C3~C5] → C2H6 + C2H4 32.002 200.656
      11 C2H6 → C2H4 + H2 37.813 257.440
      12 [C3~C5] + H2 → C2H6 + CH4 34.647 111.807
      13 C2H4 + H2 → C2H6 34.698 221.893
      14 C2H6 + [C6~C10] → 2[C3~C5] 39.552 108.848
      15 [C6~C10] → [C3~C5] + C2H6 + H2 38.843 256.380
      16 [C3~C5] + C2H6 → C2H4 + H2 + [C3~C5] 38.625 245.262
      17 [C11~C20] + [C3~C5] + C2H6 → 2[C6~C10] + H2 53.427 290.191
      18 [C3~C5] + C2H4 → C2H6 + H2 + [C3~C5] 34.486 203.009
      19 [C11~C20] + C2H6 + H2→ 2[C3~C5] + [C6~C10] 52.412 188.332
      20 [C21~C40] → 2[C11~C20] + [C3~C5] + [C6~C10] 36.157 248.102
      21 [C41~C200] → 2[C11~C20] + [C3~C5] + 3[C21~C40] + CH4 34.970 217.058
      22 [C11~C20] → 4[C3~C5] + [C6~C10] 34.870 175.052
      23 2[C3~C5] → 2C2H6 + 2CH4 38.653 221.624
      24 2C2H6 + 2H2 → 4CH4 50.683 219.295
      25 2[C6~C10] + H2 → 4[C3~C5] 45.566 220.359
      26 2[C21~C40] → [C41~C200] + [C6~C10] 39.542 161.946

      Table 5. 

      Optimized lumped reaction mechanism list for PP pyrolysis at 4 K/ps.

    • Heating rate
      (K/ps)      		RxC4
      (intra-molecular H-shift)      		RxC9 (inter-molecular
      H abstraction by C)      		RxC11 (H radical
      addition to C)      		RxC12 (recombination
      of C radicals)      		RxC31 (chain
      cyclization)      		Ea (kJ/mol)lnA
      122 (2.37%)31 (3.34%)83 (8.93%)778 (83.7%)15 (1.61%)150.0342.03
      27 (1.64%)22 (5.15%)24 (5.62%)372 (87.1%)2 (0.468%)138.0440.63
      43 (1.69%)10 (5.65%)15 (8.47%)145 (81.9%)4 (2.25%)123.1939.74

      Table 6. 

      Numbers and ratios of the top five detailed reaction types for the lumped reaction: [C3~C5] + C2H6 → [C6~C10].