Assessment of Weld Metal Compositional Prediction Models Geared Towards Submerged Arc Welding: Case Studies Involving CaF₂-SiO₂-MnO and CaO-SiO₂-MnO Fluxes

Abstract

Submerged arc welding has been performed by utilizing CaF₂-SiO₂-MnO and CaO-SiO₂-MnO fluxes over a wide range of compositions and basicity index values. Contents of essential elements, including O, Si, and Mn, in the weld metal, are predicted by employing the basicity index model, slag–metal equilibrium model, and gas–slag–metal equilibrium model. Capabilities of each model to predict weld metal compositions have been evaluated from thermodynamic perspectives. The results show that the basicity index model is capable of predicting the variation trend of O content with MnO addition, but fails to differentiate O levels when fluxes with same basicity index are applied. The slag–metal model overestimates the contents of Si and Mn due to underestimated O level or overestimated oxide activity. The gas–slag–metal equilibrium model, on the other hand, offers better prediction accuracy for O content than the basicity index model, and is able to differentiate the O content of the weld metals produced by fluxes with varying formulas but same basicity index. Furthermore, when the gas–slag–metal equilibrium model is applied, the prediction error for Si and Mn contents is significantly reduced as compared to the slag–metal equilibrium model. Thermodynamic calculation data indicates that the consideration of gas formation, which essentially controls the predicted flux O potential and oxide activity, is necessary to improve the overall prediction accuracy.

Appendices

Appendix A: Prediction of O Content Using Flux BI Model

O contents are predicted from Figure A1 using both Tuliani and Eagar BIs illustrated in Table I. The predicted O levels from Tuliani and Eagar BIs are summarized in Table AI.

Fig. A1

Predicted O content of WM as a function of flux BI

Table AI Predicted O Content (Parts Per Million) From Flux BI

Appendix B: Prediction of Si and Mn Contents Using Slag–Metal Model

Mn and Si contents are predicted from the slag–metal equilibrium model proposed by Chai et al., as follows:

The equilibrium constant of Reaction [3] is referenced as Eq. [B1].

$$ \log K_{1} = - \frac{28360}{T}{ + 10} . 6 1 $$

(B1)

The equilibrium temperature of 2273 K is set, and all interaction terms are neglected. Then, Eq. [B2], proposed by Chai et al., is applied to predict Si content ([pct Si]);[7,18,24] in Eq. [B2], [pct O] is the predicted O level from the Eagar BI (see Table AI). The data of SiO₂ activity $\left( {\alpha_{{{\text{SiO}}_{ 2} }} } \right)$ is obtained from FactSage 7.3 (Equilib Module) using the FToxid database (ASlag-liq all oxides and S (FToxid-SLAGA) solution phase are selected) with flux formulas in Table I as input.[45] The activities of SiO₂ $\left( {\alpha_{{{\text{SiO}}_{ 2} }} } \right)$ used in slag–metal equilibrium model are given in Table IV.

$$ \left[ {{\text{pct}}\,{\text{Si}}} \right] = \frac{{\alpha_{{{\text{SiO}}_{ 2} }} }}{{ 7 3. 6\cdot \left[ {{\text{pct}}\,{\text{O}}} \right]^{2} }} $$

(B2)

Similarly, the equilibrium constant of Reaction [4] is referenced as Eq. [B3]; ignoring all interaction terms, Eq. [B4] is used in the models of Chai et al.[7,24] to predict Mn content ([pct Mn]). The activity of MnO activity (α_MnO) used in Eq. [B4] are calculated from FactSage 7.3 (Equilib Module) using the FToxid database (ASlag-liq all oxides and S (FToxid-SLAGA) solution phase are selected), during which flux formulas in Table I were set as input. The activities of MnO (α_MnO) used in slag–metal equilibrium model are summarized in Table V.[45]

$$ \log K_{2} { = } - \frac{12760}{T}{ + 5} . 6 8 $$

(B3)

$$ \left[ {{\text{pct}}\,{\text{Mn}}} \right] = \frac{{\alpha_{\text{MnO}} }}{{ 0. 8 6\cdot \left[ {{\text{pct}}\,{\text{O}}} \right]}} $$

(B4)

Appendix C: Prediction Using Gas–Slag–Metal Equilibrium Calculation

Similar to previous studies, nominal compositions, which refer to the contents considering only the dilution effects of the BM and electrode,[15,30] are used as the input metal chemistries.[8,9,18,27] Nominal compositions are calculated from Eq. [C1], in which d_BM represents the dilution value of BM.[8,46]

$$ {\text{Nominal}}\,{\text{comp}} .\,{ = }\,d_{\text{BM}} \, \times \,{\text{base}}\,{\text{metal}}\,{\text{comp}} .\,{ + }\,\left( { 1\, - \,d_{\text{BM}} } \right)\, \times \,{\text{electrode}}\,{\text{comp}} . $$

(C1)

d_BM equals to the ratio of the area below the plate surface to the total WM area. To determine these areas, WMs were cross-sectioned, polished, and etched using 4 wt pct nital solution.[8,9,18,22,29,38] Calculated nominal compositions and d_BM values are given in Table CI (‘N’ subscript means nominal composition).

Table CI Nominal Compositions (Weight Percent) and d_BM

The flux formulas given in Table I are set as input flux compositions.

Equilib Module of FactSage 7.3 is employed to perform gas–slag–metal equilibrium following the settings in our previous study:[27]

• 1. FToxid, Fstel, and FactPS databases are selected. Solution phases of ASlag-liq all oxides, S (FToxid-SLAGA), and LIQUID (FStel-Liqu) were selected to model the molten slag and steel phases.

• 2. The equilibrium temperature in SAW of 2273 K is set.

• 3. The mass ratio of flux to the electrode is set as unity.

Parts of gas compositions calculated from gas–slag–metal equilibrium calculations are summarized in Table CII. Activities of SiO₂ and MnO calculated from gas–slag–metal equilibrium models are given in Tables IV and V.

Table CII Gas–Slag–Metal Equilibrium Gas Components