Multi-Criteria Evaluation and Scenario-Driven Selection of Grounding Connectors Across Materials and Joining Processes

Abstract

Grounding connectors critically influence the safety and long-term reliability of earthing systems through coupled electro-thermal, mechanical, and corrosion behaviors, yet no standardized quantitative framework exists for jointly evaluating these performance dimensions across diverse deployment scenarios. This study introduces a unified multi-criteria evaluation framework applied to six grounding connector configurations spanning four alloy families and three joining technologies. Electro-thermal response was characterized by coupled finite element simulations (0–100 A), mechanical reliability by quasi-static tensile testing (n = 10 per configuration), and corrosion durability by accelerated salt-spray exposure with image-based corroded area fraction quantification. Performance metrics were normalized and aggregated using equal-weight, Analytic Hierarchy Process, and Shannon entropy weighting schemes, with the Technique for Order of Preference by Similarity to Ideal Solution applied for multi-scenario ranking. One-way analysis of variance confirmed statistically significant effects of connector type on tensile performance (F(5, 54) = 3154.90, p < 0.001). The exothermic welded joint achieved the highest mean ultimate tensile load (61.5 ± 1.5 kN), while copper mechanical connectors exhibited the lowest steady-state temperature rise (~2 K above ambient at 100 A). Compression-crimped connectors ranked first under both equal and Analytic Hierarchy Process weighting (closeness coefficients 0.737 and 0.807, respectively), while stainless steel connectors ranked first under corrosion-critical deployment scenarios. Scenario-weighted analyses demonstrate that the optimal material–process combination shifts with environmental severity, current duty, and mechanical demand, providing a reproducible, evidence-based basis for context-dependent connector specification.

1. Introduction

Grounding connectors are critical nodes in utility and substation earthing systems since they directly influence touch/step-potential safety and are the system’s most vulnerable point to environment and electrical payload. Their electro-thermal behavior, mechanical integrity, and corrosion resistance jointly determine long-term reliability [1,2]. In practice, engineering specifications are often framed with reference to established industry standards (e.g., IEEE 837for qualifying permanent substation grounding connections [3], IEEE 80 for substation grounding safety [4], and IEC 62561-1 for connection components [5]), which emphasize current-carrying capacity, mechanical robustness, and environmental durability [6]. However, real-world selection still depends on how specific materials and joining processes behave under combined stresses.

In corrosive or moisture-laden environments, corrosion products and surface films (e.g., oxides, sulfides) can increase electrical contact resistance, promote local heating under load, and reduce clamp force via interface degradation or fretting, accelerating failure [7]. Accelerated salt-spray testing is widely used to screen materials and coatings, yet multiple previous works caution that salt-spray outcomes have limited or context-dependent correlation to field performance; careful protocol control and replication are required for meaningful interpretation [8,9].

The electro-thermal response of joints is tightly coupled to electrical contact resistance (ECR) and Joule heating: increases in ECR—due to insufficient tightening, surface films, or post-installation degradation—raise local temperature rise for a given load profile [10]. Analytical and numerical studies show how contact force, roughness, and wear/corrosion shift resistance and the resulting temperature fields [11,12]. In the connector technologies typically used for grounding (mechanical/bolted, compression/crimped, and exothermic/welded), performance under load is therefore strongly dependent on installation quality, conductor condition, and the severity of the applied thermal–mechanical environment [13,14]. For instance, thermal cycling has been shown to raise crimped connection resistance above a critical threshold of 24 μΩ (corresponding to approximately 4 W of dissipated power at 400 A), at which point the thermal integrity of the joint is compromised [13].

Contact behavior is sensitive to tightening torque and contact pressure: controlled experiments and classic measurements show that inadequate torque or poor surface preparation elevates contact resistance and temperature rise, whereas sufficient clamp force reduces constriction resistance and stabilizes thermal behavior [15,16]. Materials and coatings (e.g., stainless steels, nickel-plated brass, and engineered carbon/ceramic surface treatments) exhibit markedly different responses in accelerated exposures and electrochemical tests, with concomitant effects on interfacial contact resistance [17,18]. Meanwhile, fretting and temperature cycling drive ECR drift and intermittency; the literature documents how corrosion films and mechanical wear increase contact resistance and alter lifetime, particularly under vibration or thermal stress [19,20], and how mechanistic/numerical models connect Joule heating and sliding/wear to evolving ECR and thermal fields, providing a theoretical basis to interpret experimental temperature rise curves and degradation trajectories [21,22]. In controlled large-current temperature rise tests on ultra-high voltage terminal connectors, reducing the tightening torque from the standard 190 N·m to 40% of this value increased contact resistance from approximately 2 μΩ to 14 μΩ, with junction temperatures exceeding 70 °C under six 250 A loadings [14]. For example, untreated 316L stainless steel exhibits contact resistivity in the range of 100–800 mΩ·cm², which can be reduced to below 10 mΩ·cm² through passive film modification—a two-order-of-magnitude improvement that substantially alters interfacial conductance in connector assemblies [17].

Despite this body of work, there remains limited guidance on integrating heterogeneous metrics into a transparent, reproducible multi-criteria score that enables scenario-based engineering selection (e.g., prioritizing high-current duty vs. highly corrosive sites vs. high-vibration installations). Mature multi-criteria decision methods such as the Analytic Hierarchy Process (AHP) and entropy-based objective weighting offer principled ways to normalize, weight, and synthesize cross-domain metrics; however, they are rarely operationalized end-to-end for grounding connector selection with explicit sensitivity analyses [23]. To address this gap, the present study establishes a multi-metric indicator system (electro-thermal, mechanical, corrosion) with explicit normalization and implements transparent weighting schemes (equal, AHP, entropy-based) with scenario-weighted analyses, validating the framework on a structured material × process sample matrix to clarify which domains primarily drive rank differences under practical operating conditions.

2. Materials and Method

Six grounding connector configurations were selected to span the three principal joining technologies and four alloy families commonly encountered in field practice. The three joining geometries are illustrated schematically in Figure 1. For the mechanical bolted category shown in Figure 1a, four connector body materials were evaluated: copper, aluminum, galvanized steel and stainless steel, all of which were commerical purchased from Henan Sida Testing Technology Co., Ltd. (Xuchang, China). Each mechanical connector shares the same assembly architecture—a tapered sleeve body engaging the conductor rod through thread contact, clamped by hexagonal locknuts—with material composition as the sole variable. Two process-bonded configurations were included for comparison: Figure 1b shows crimping and Figure 1c shows welding. All six configurations were joined to standardized specimens serving as the grounding electrode substrate.

Each configuration was evaluated across three independent performance domains: (i) electro-thermal response, assessed by coupled finite element simulation of steady-state temperature distribution and validated by power-frequency resistance measurement before and after high-current conditioning; (ii) mechanical reliability, assessed by quasi-static tensile testing to determine ultimate load and slip-onset load, supplemented by finite element stress simulation; and (iii) corrosion durability, assessed by accelerated salt-spray exposure with optical image analysis and scanning electron microscopy with energy-dispersive spectroscopy (SEM-EDS). Ten specimens were fabricated and tested for each connector configuration (n = 10 per configuration, 60 specimens in total) to enable statistical assessment of result repeatability. The resulting multi-domain dataset forms the basis for the multi-criteria evaluation framework described in Section 4.

3. Results and Discussion

3.1. Electro-Thermal Response Curves

Coupled electro-thermal finite element simulations were performed to obtain the steady-state temperature distribution across grounding connector assemblies under applied currents from 0 to 100 A at an ambient temperature of 20 °C (293.15 K). The electrical boundary conditions consisted of a uniform current density applied to one end face of the conductor rod, with the opposite end face set to zero electric potential. A convective heat transfer boundary condition with a coefficient of h = 20 W/(m²·K) and an ambient temperature of T∞ = 293.15 K was applied to all external surfaces, representing natural convection in quiescent air. The steady state temperature field T and electric potential V in each connector assembly were governed by the coupled electro-thermal system:

$$\mathit{\nabla }\cdot ({\sigma }_e\mathit{\nabla }V)=0$$

(1)

$$\rho c_p{\displaystyle \frac{\partial T}{\partial t}}=\mathit{\nabla }\cdot (k\mathit{\nabla }T)+Q_J,Q_J={\sigma }_e{|\mathit{\nabla }V|}^2$$

(2)

Material properties for each connector alloy were assigned from standard reference values at 298 K. Equation (1) enforces current continuity within each conductor domain, while Equation (2) couples the resulting Joule dissipation $Q_J$ to the transient heat balance; steady-state solutions were obtained by setting the time derivative to zero. The temperature dependence of electrical resistivity was incorporated to account for the feedback between Joule heating and conduction resistance:

$${\rho }_e(T)={\rho }_0[1+\alpha (T-T_0)]$$

(3)

This linear resistivity model captures the non-linear growth of Joule dissipation with rising temperature and directly explains the progressively widening separation of the temperature–current curves above approximately 40 A, as observed in Figure 2a. Configurations with high bulk resistivity are most strongly affected by this feedback, amplifying their thermal response relative to low-resistivity materials, which can be concluded from the comparison between galvanized steel and copper.

Among the four mechanical connector materials, the copper connector exhibited the lowest temperature rise, remaining within approximately 2 K of ambient at 100 A. Aluminum followed closely with a comparably modest rise of roughly 2–3 K. The stainless steel mechanical connector showed a more pronounced increase, reaching approximately 305 K at 100 A ($\Delta T \approx 7 K$), reflecting its substantially higher bulk resistivity (~70 μΩ·cm) compared to copper (~1.7 μΩ·cm) and aluminum (~2.8 μΩ·cm). The galvanized steel connector recorded the highest temperature of all configurations at approximately 325 K ($\Delta T \approx 27 K$ at 100 A), attributable to the combination of moderate bulk resistivity (~12 μΩ·cm) and less efficient heat dissipation at the connector–rod interface. The process-bonded configurations—welding and crimping—produced intermediate thermal responses, reaching approximately 312 K and 318 K at 100 A, respectively. The higher temperature in the crimped joint relative to the welded joint is consistent with the residual mechanical contact resistance at the crimp interface, whereas the exothermic weld provides a more continuous metallic bond with lower interfacial resistance.

Thermal distribution maps at representative current levels are presented in Figure 2b–g, rendered on a common temperature scale (298–300 K). The copper and aluminum mechanical connectors, shown in Figure 2b,c respectively, exhibit nearly uniform blue/purple coloration, confirming minimal thermal perturbation from the applied current. In Figure 2d, the galvanized steel connector shows a distinctly warmer zone at the connector–rod interface, visible as a yellow-to-orange hotspot. The welded joint depicted by Figure 2f displays localized heating concentrated at the fusion zone, while the crimped configuration in Figure 2g shows warming distributed along the full sleeve length. The use of a common temperature scale across all six panels enables direct visual comparison and confirms the quantitative ranking established by the temperature–current curves. It should be noted that the absolute temperature predictions carry inherent uncertainty arising from reference value material properties (typical tolerance ±2~5%), the uniform convective coefficient assumption (h = 20 W/(m²·K)), and geometric idealization of connector features. However, because all six configurations were simulated under identical boundary conditions and material assignment strategy, the comparative ranking of configurations by steady-state temperature rise is substantially more robust to these uncertainty sources than the absolute values, and it is this comparative ranking that feeds into the multi-criteria evaluation framework.

To quantify the effect of high-current conditioning on contact quality, the power-frequency resistance change ratio was defined as:

$${\displaystyle \frac{∆R}{R_0}}={\displaystyle \frac{R_{final}-R_{initial}}{R_{initial}}}$$

(4)

A negative ratio indicates resistance reduction, interpreted as disruption of interfacial oxide films and exposure of fresh metallic asperities under the combined effect of contact pressure and Joule heating; a positive value indicates net resistance increase due to progressive surface oxidation or thermal degradation of the contact interface.

Power-frequency resistance measurements before and after high-current conditioning are summarized for all six configurations in Figure 2h. Five of the six configurations—aluminum, stainless steel, galvanized steel, welding, and crimping—exhibited resistance decreases after conditioning, ranging from approximately 29% (aluminum) to 56% (galvanized steel). This behavior is consistent with disruption of interfacial oxide films under the combined effect of contact pressure and Joule heating, which exposes fresh metallic asperities and reduces constriction resistance [2,6]. The copper connector, by contrast, shows an anomalous resistance increase of approximately 60% (from ~6.5 to ~9.4 mΩ), indicating that the elevated interface temperature promoted progressive surface oxidation rather than film breakdown—a self-reinforcing degradation mechanism with direct implications for long-term contact stability in service.

3.2. Mechanical Tensile Strength

Tensile performance was evaluated through quasi-static mechanical testing (n = 10 per configuration) and finite element stress simulation to characterize the load-bearing capacity and failure behavior of each connector configuration. The mechanical FEM boundary conditions comprised a fixed constraint applied to one end face of the galvanized steel rod—constraining all translational and rotational degrees of freedom—and a monotonically increasing axial tensile load applied to the opposite end face, consistent with the quasi-static testing protocol. Force–displacement curves and maximum tensile force results are presented in Figure 3a,b, while simulated von Mises stress distributions for representative connector geometries are shown in Figure 3c–h.

The force–displacement curves in Figure 3a reveal a pronounced mechanical bifurcation between the mechanical connectors and the process-bonded configurations. The four material-based connectors, copper, aluminum, stainless steel, and galvanized steel, all exhibit a sharp load drop at relatively small displacements (2–5 mm), characteristic of sudden slip onset at the bolt–rod thread or rod–body contact interface, followed by rapid force decay as the conductor rod pulls through the connector body. This failure mode reflects the inherent sensitivity of mechanically bolted joints to interface condition, clamping force, and thread engagement quality, all of which limit the effective load transfer area. In contrast, the crimped and welded configurations sustain elevated loads over substantially larger displacement ranges (up to approximately 45 mm), exhibiting gradual ductile failure initiated in the galvanized steel rod outside the bonded zone rather than at the joint interface itself, which is a clear indication that joint strength exceeded the rod’s material capacity in both process-bonded cases.

The maximum tensile forces summarized in Figure 3b quantify this performance disparity. The welded joint achieved the highest mean ultimate load at 61.5 ± 1.48 kN, followed by the crimped joint at 53.5 ± 2.01 kN. This 7.0 kN difference is statistically significant (independent-samples t-test: t(18) = 9.10, p < 0.001). Among the base-material mechanical connectors, copper recorded the highest ultimate load at 20.0 ± 1.22 kN, followed by galvanized steel (14.5 ± 0.89 kN), stainless steel (13.7 ± 0.80 kN), and aluminum (7.7 ± 0.64 kN). One-way ANOVA confirmed a highly significant effect of connector configuration across all six groups (F(5, 54) = 3154.90, p < 0.001), with within-group coefficients of variation below 8%, confirming good inter-specimen repeatability. The process-bonded joints therefore outperform the best-performing mechanical connector by a factor of approximately 2.5–3.0, with the welded joint exceeding the weakest configuration (aluminum) by more than seven times. The performance ranking among mechanical connectors broadly follows the stiffness and surface hardness of the connector body material, both of which govern thread engagement quality and clamp force retention under applied tensile loading.

Failure location was consistent across all specimens of each configuration type. For mechanical bolted connectors (a, b, c, d), slip onset occurred at the conductor–barrel contact interface, as indicated by the onset of progressive resistance increase under load. For the compression-crimped configuration (f), fracture occurred within the conductor strands immediately adjacent to the crimp barrel exit, consistent with the localized stress concentration at the barrel edge predicted by the von Mises FEM distribution. For the exothermic welded configuration (e), fracture occurred within the weld zone rather than at the weld–rod interface. No interface debonding or sleeve–rod separation was observed in any specimen of the crimped or welded groups; accordingly, the reported maximum loads represent conductor or weld zone failure rather than joint interface failure, confirming that bond strengths for both joining technologies exceed the tensile capacity of the galvanized steel rod substrate under the tested quasi-static loading conditions.

The finite element stress simulations employed the von Mises equivalent stress criterion to identify regions susceptible to yielding or fracture. The von Mises stress is defined as:

$${\sigma }_{VM}={\displaystyle \frac{1}{\sqrt{2}}}\sqrt{{({\sigma }_1-{\sigma }_2)}^2+{({\sigma }_2-{\sigma }_3)}^2+{({\sigma }_3-{\sigma }_1)}^2}$$

(5)

where ${\sigma }_i$ represents principal stresses (Pa). Failure is predicted when ${\sigma }_{VM}\ge {\sigma }_y$. Here, ${\sigma }_y$ denotes a volumetrically averaged yield strength computed over the full multi-material connector assembly, obtained by integrating the element-wise yield strength over the simulation domain and normalizing by the total volume. This scalar threshold is consistent with the comparative use of the von Mises criterion across geometrically and materially heterogeneous configurations. This criterion is well-suited to the ductile metallic alloys examined here and provides a single scalar field for directly comparing stress severity across the geometrically dissimilar connector types shown in Figure 3c–h. The cylindrical sleeve geometries representative of crimped joints in Figure 3c,d exhibit distributed stress fields along the bonded sleeve length, indicative of continuous load transfer across the rod–sleeve interface with no pronounced stress singularity. The mechanical clamp configurations in Figure 3e,f display stress concentrations localized at the nut contact faces and rod–body bearing zones, consistent with the experimentally observed slip-dominated failure at low displacement. The welded connector in Figure 3g exhibits a markedly different response: intense stress concentrations approaching 3 × 10⁸ Pa are observed at the weld toe region, accompanied by visible geometric deformation of the rod, confirming that failure initiates in the rod material adjacent to the fusion boundary rather than within the weld itself. This is in close agreement with the experimentally observed failure locus of rod fracture outside the weld zone. Figure 3h reveals a localized stress hotspot at one end of the crimped sleeve, reaching peak values in the orange–red range of the colorbar scale, which corresponds to the initiation zone of conductor pullout and is consistent with the abrupt post-peak load drop observed experimentally for the crimped configuration.

Taken together, these results demonstrate that the joining process is the dominant factor controlling the mechanical performance of grounding connectors: both crimping and welding provide substantially superior tensile capacity relative to mechanically bolted connections, irrespective of connector body material, and the FEM stress distributions confirm that failure in process-bonded joints is governed by rod material capacity rather than joint interface integrity.

The fracture surfaces of all six configurations after tensile testing were examined by scanning electron microscopy at ×1920 magnification, as presented in Figure 4. The four mechanical connectors, shown in Figure 4a–d, share a common failure mode—interfacial slip at the bolt–rod thread or rod–body contact zone—but exhibit distinct surface morphologies that reflect the mechanical properties of each material. The copper fracture surface shown in Figure 4a displays well-defined parallel striations oriented along the rod withdrawal direction, indicative of progressive thread stripping through adhesive sliding. The aluminum surface in Figure 4b shows similar linear grooves but with a more smeared, plastically deformed texture, consistent with the lower hardness and higher ductility of aluminum alloy, which promotes greater material smearing under thread contact pressure. The stainless steel surface in Figure 4c presents a markedly rougher and more irregular morphology with evident micro-cracking features, reflecting the higher hardness and work-hardening tendency of the stainless steel that resists plastic flow but accumulates interfacial damage during slip. The galvanized steel surface in Figure 4d similarly shows parallel linear marks, with the relatively smooth underlying texture modified by the presence of the zinc coating layer at the contact interface.

In stark contrast, the welded connector shown in Figure 4e exhibits a coarse dimpled surface with abundant microvoid coalescence features, the hallmark morphology of ductile overload fracture in the bulk rod material. This observation directly confirms that failure did not occur at the weld interface itself but in the galvanized steel rod adjacent to the fusion zone, consistent with the sustained high-load ductile behavior observed in the force–displacement curves. The crimped connector in Figure 4f shows a mixed-mode surface combining shallow sliding marks at the sleeve–rod interface with localized deformation features, reflecting the partial pullout mechanism at the initiation zone of the crimp sleeve identified in the stress simulation. Taken together, the SEM evidence corroborates the macroscopic failure classification: mechanical connectors fail by interfacial slip with surface morphologies governed by material hardness, while process-bonded joints transfer failure to the rod substrate, achieving substantially higher energy absorption before fracture.

3.3. Corrosion Resistance Performance

To enable quantitative comparison of corrosion resistance across configurations, an automated image analysis procedure was applied to the optical photographs in Figure 5. Pixels satisfying the hue and saturation thresholds characteristic of rust and corrosion products were identified by HSV color-space thresholding, and the corroded area fraction was computed as:

$$f_{corr}={\displaystyle \frac{A_{corr}}{A_{total}}}$$

(6)

where A represents the pixel areas of corroded and total pixels in the frame. This image-based metric provides an objective, repeatable measure of surface degradation that is independent of subjective visual assessment, and directly feeds into the corrosion indicator $S_{corr}$ of the multi-criteria framework. Having established the electro-thermal and mechanical performance characteristics of the six connector configurations in Section 3.1 and Section 3.2, the present section evaluates their corrosion resistance under accelerated salt-spray exposure. As a precursor to quantitative corrosion analysis, Figure 5 presents optical photographs of the as-fabricated connector specimens, documenting their baseline surface condition, joint geometry, and material appearance prior to testing. These visual records provide an essential reference for interpreting subsequent post-exposure surface changes and are directly informative of each configuration’s inherent susceptibility to corrosive degradation.

Corroded surface area fraction was quantified using HSV (hue–saturation–value) color-space image thresholding applied to standardized post-exposure photographs. This approach was selected because it provides an objective, operator-independent measurement that eliminates the subjectivity inherent in visual scoring scales, produces a continuous ratio-scale metric directly meaningful as a surface integrity indicator, and is consistent with optical area fraction methods reported in the corrosion testing literature for accelerated-exposure studies. Hue and saturation thresholds for rust pixel segmentation were calibrated on reference specimens exhibiting known corrosion states and applied uniformly across all six configurations and all replicate specimens. The corroded area fraction S_corr is defined as the ratio of segmented rust pixels to total connector surface pixels visible in each standardized photograph. It is acknowledged that HSV thresholding measures surface projection area and does not resolve corrosion depth, pit density, or subsurface corrosion product distribution; these constitute recognized limitations of the present assessment methodology.

Figure 5a–d show the four mechanically bolted configurations, with the sequence of copper, aluminum, stainless steel and galvanized steel. They share a common assembly architecture consisting of a threaded cylindrical barrel body flanked by two hexagonal clamping nuts, through which the conductor rod is engaged by thread contact and compressive clamping force. Despite their geometric similarity, the four specimens exhibit visually distinct baseline surface characteristics that are directly relevant to their expected corrosion behavior. The copper connector in Figure 5a displays the characteristic warm golden-yellow coloration of the base material, with visible surface tarnishing and localized oxidation patina already present at the barrel and nut faces prior to exposure, suggesting a relatively reactive surface susceptible to further atmospheric degradation. The aluminum connector in Figure 5b presents a brighter, cleaner silver appearance with a noticeably smoother surface finish and minimal pre-existing oxidation, consistent with the protective native alumina passivation layer that is known to limit further surface degradation under moderate corrosive conditions. The stainless steel connector in Figure 5c exhibits the brightest and most uniform metallic luster among the four mechanical configurations, reflecting its passive chromium oxide surface layer and the lowest corroded area fraction under the accelerated salt-spray tests, with no protective coating required. The galvanized steel connector in Figure 5d displays a matte gray surface with slight coating irregularities and a rougher texture characteristic of the hot-dip zinc galvanizing process; the zinc coating is expected to provide sacrificial cathodic protection to the underlying steel substrate during salt-spray exposure.

The welded connector in Figure 5e represents a fundamentally distinct joint architecture in which the conductor rod is fused directly to a flat steel plate through exothermic welding. The weld zone is clearly identifiable as a discolored, rough-textured region along the rod–plate interface, exhibiting the characteristic brownish heat-affected coloration and surface irregularity associated with the weld bead and surrounding heat-affected zone (HAZ). The thermally altered microstructure and surface chemistry of the HAZ are well recognized as preferential sites for corrosion initiation, as local depletion of alloying elements and residual tensile stresses can accelerate oxidation and pitting relative to the unaffected base material.

The crimped connector in Figure 5f is distinguished by its elongated cylindrical sleeve with clearly visible longitudinal crimp indentations at regular intervals along its length. The mechanically deformed sleeve surface and the rod–sleeve crevice geometry formed during crimping represent potential sites for crevice corrosion and moisture retention under salt-spray conditions, which may compromise the interfacial contact integrity that underpins its superior mechanical performance.

Figure 6 presents post-exposure optical photographs of all six configurations, with magnified insets highlighting the representative surface condition of the connector barrel region after salt-spray testing. The corrosion severity differs markedly across configurations. The copper connector in Figure 6a shows the development of a greenish-brown oxidation patina across the barrel surface, consistent with the formation of Cu–Zn oxide and carbonate corrosion products on the brass substrate. The aluminum connector in Figure 6b exhibits localized whitish deposits attributable to aluminum oxide and hydroxide corrosion products, with visible pitting sites on the barrel surface. The stainless steel connector in Figure 6c retains a relatively clean and uniform surface with only minor discoloration, confirming the effectiveness of the passive chromium oxide film in suppressing corrosive attack under the test conditions. The galvanized steel connector in Figure 6d shows moderate surface degradation with white rust products characteristic of zinc hydroxide formation, indicating progressive consumption of the sacrificial zinc coating. The welded connector in Figure 6e presents the most severe surface deterioration among all six configurations, with extensive red–brown rust coverage across the weld zone and adjacent rod surface, reflecting the absence of any passive film or protective coating on the unalloyed carbon steel weld material. The crimped connector in Figure 6f shows comparatively limited surface degradation on the stainless steel sleeve, though some corrosion product accumulation is visible at the rod–sleeve interface and crimp indentation zones, consistent with the expected crevice corrosion susceptibility of those geometrical features.

Energy dispersive X-ray spectroscopy (EDS) was performed on the surface of each as-fabricated connector specimen to confirm the elemental composition and characterize the baseline surface chemistry prior to salt-spray exposure. The resulting spectra are presented in Figure 7. The copper connector spectrum in Figure 7a is dominated by Cu (21.0 wt%) and Zn (19.0 wt%) peaks, with the Zn Kα line at approximately 7.6 keV registering as the most intense feature. The near-equal Cu and Zn contents reported indicate that the connector body is fabricated from a brass alloy (Cu–Zn) rather than pure copper, which is consistent with common industry practice for mechanical grounding fittings due to the superior machinability and corrosion resistance of brass relative to unalloyed copper. A substantial oxygen signal (12.5 wt%) is also detected, consistent with the surface tarnishing and oxidation patina observed visually in Figure 5a. The aluminum connector in Figure 7b exhibits a single dominant Al peak (67.4 wt%) with a minor oxygen signal (11.1 wt%), consistent with the presence of a thin native alumina (Al₂O₃) passivation layer. As shown in Figure 7, the near-absence of other metallic elements confirms that this connector is fabricated from a commercially pure or lightly alloyed aluminum grade. The passivation layer is expected to provide a degree of inherent corrosion protection under the subsequent salt-spray exposure conditions. The galvanized steel connector spectrum in Figure 7c is dominated by Fe (75.9 wt%), reflecting the underlying carbon steel substrate. A minor Zn signal (1.2 wt%) is detected, confirming the presence of residual galvanized zinc coating, although the relatively low Zn content suggests that the zinc layer is thin or partially depleted in the analyzed region. The low oxygen content (5.0 wt%) indicates a relatively clean metallic surface with minimal pre-existing oxidation. The stainless steel connector in Figure 7c displays the characteristic compositional signature of an austenitic stainless steel, with Fe (46.2 wt%), Cr (11.4 wt%), and Ni (4.7 wt%) as the principal metallic constituents. The Cr and Ni contents listed in

Table 1. Indicator weight vectors for three weighting schemes. AHP consistency ratio CR = 0.0139 (<0.10 threshold).

Indicator	Description	Equal	AHP	Entropy
${\mathit{S}}_{\mathit{t}\mathit{h}}$	Thermal: $\mathrm{\Delta }T$ at 100 A (K)	0.167	0.116	0.083
${\mathit{S}}_{\mathit{res}}$	Resistance stability: $\mathrm{\Delta }\mathit{R}/{\mathit{R}}_0$	0.167	0.065	0.058
${\mathit{S}}_{\mathit{mec}\mathbf{h}}$	Tensile strength: ${\mathit{F}}_{\mathit{max}}$ (kN)	0.167	0.344	0.162
${\mathit{S}}_{\mathit{duct}}$	Ductility: displacement at failure (mm)	0.167	0.068	0.324
${\mathit{S}}_{\mathit{corr}}$	Corrosion resistance: corroded area (%)	0.167	0.181	0.057
${\mathit{S}}_{\mathit{passive}}$	Passive film quality: Cr + Ni content (wt%)	0.167	0.226	0.316

are broadly consistent with a 304-type grade. The moderate oxygen signal (7.1 wt%) is attributable to the passive chromium oxide film that confers the inherent corrosion resistance of this material class. The welded connector spectrum in Figure 7e is overwhelmingly dominated by Fe (67.1 wt%), with a notably elevated oxygen signal (10.6 wt%) compared to the galvanized steel and stainless steel configurations. The higher oxygen content reflects the thermal oxidation of the weld zone and heat-affected zone (HAZ) surfaces during the exothermic welding process. As EDS measurement confirms, the virtual absence of Cr, Ni, and Zn indicates that the weld material is an unalloyed carbon steel, which lacks the passive film protection of stainless steel and may therefore be more susceptible to corrosive degradation. The crimped connector in Figure 7f exhibits a composition closely resembling that of the stainless steel mechanical connector, with Fe (47.9 wt%), Cr (12.1 wt%), and Ni (5.0 wt%) as the major constituents, confirming that the crimping sleeve is fabricated from a stainless steel of comparable grade. The lower oxygen content (5.3 wt%) relative to the welded specimen indicates a less oxidized surface, consistent with the absence of thermal processing in the crimping operation. To provide a single quantitative index of passive film quality for the multi-criteria framework, a composite passive film indicator was derived from the EDS semi-quantitative weight-percent data in Figure 7:

$$S_{passive,raw}=w_{Cr}+w_{Ni}$$

(7)

where $w_{Cr}$ and $w_{Ni}$ represent chromium and nickel weight percentage from EDS analysis, respectively. Cr and Ni are the principal passive-film-forming elements governing long-term corrosion resistance in aqueous and chloride-rich environments. The resulting $S_{passive,raw}$ values span more than two orders of magnitude across the six configurations—from 0.1 wt% for the welded connector to 17.1 wt% for the crimped sleeve—providing strong inter-configuration discrimination that is subsequently exploited by the entropy weighting scheme in Section 4. Collectively, the EDS results presented in Figure 7 establish the baseline elemental composition of each connector configuration and confirm the material identities inferred from visual inspection. These compositional differences, particularly the presence or absence of passive film-forming elements such as Cr and Ni, are expected to govern the relative corrosion performance of each configuration under accelerated salt-spray exposure.

Figure 7. EDS spectra of six grounding connector configurations, from (a–f): copper, aluminum, stainless steel, galvanized steel, welding and crimping.

Figure 7. EDS spectra of six grounding connector configurations, from (a–f): copper, aluminum, stainless steel, galvanized steel, welding and crimping.

4. Multi-Factor-Based Grounding Connector Selection Framework

To systematically compare the six connector configurations across their heterogeneous performance domains, a three-tier multi-criteria decision-making (MCDM) framework was developed, as illustrated in Figure 8, integrating electro-thermal, mechanical, and corrosion metrics into a single quantitative selection tool. The framework proceeds through sequential stages of metric extraction and normalization (Tier 1), multi-criteria weight assignment (Tier 2), and TOPSIS-based scenario ranking (Tier 3), enabling transparent and reproducible scoring under varying deployment conditions.

Six measurable performance indicators were extracted from the experimental and simulation data reported in Section 3.1, Section 3.2 and Section 3.3. The electro-thermal domain contributes two indicators: the steady-state temperature rise at 100 A ($S_{th}$), drawn from the FEM-derived temperature–current curves, and the post-conditioning resistance change ratio $\Delta R/R_0$ ($S_{res}$), which captures the degradation or recovery behavior of the contact interface under high-current load. The mechanical domain contributes two further indicators: the maximum tensile force ($S_{mech}$) and the displacement at failure ($S_{duct}$), both derived from the quasi-static tensile test results. The corrosion domain is represented by the corroded area fraction ($S_{corr}$), quantified via HSV color-space thresholding applied to the pre-exposure optical photographs in Figure 5, and the Cr + Ni passive film content ($S_{passive}$), derived from the EDS surface composition data in Figure 7. The passive film indicator $S_{passive}$ is derived from the EDS-measured Cr + Ni surface content (wt%), which serves as a first-order compositional proxy for passive film formation tendency based on the established relationship between chromium content above approximately 9.5 wt% and the spontaneous formation of a protective Cr₂O₃-rich oxide layer in stainless-grade alloys. It is acknowledged that this proxy does not capture the full complexity of passive film stability in chloride environments, where additional factors—including molybdenum content, microstructure (grain boundary sensitization, carbide precipitate distribution), pitting potential, and galvanic coupling between dissimilar metals at the connector–rod interface—each contribute to actual corrosion susceptibility. $S_{passive}$ should therefore be interpreted as a comparative compositional indicator within the present dataset, not as a comprehensive corrosion resistance descriptor. Electrochemical characterization via potentiodynamic polarization and electrochemical impedance spectroscopy would provide more rigorous passivation metrics and is identified as a direction for future work.

To place these dimensionally disparate metrics on a common [0, 1] scale, min–max normalization was applied column-wise. Benefit-type indicators ($S_{mech}$, $S_{\mathit{duct}}$, $S_{\mathit{passive}}$) are normalized as $S={\displaystyle \frac{x-x_{\mathit{min}}}{x_{\mathit{max}}-x_{\mathit{min}}}}$, mapping the highest-performing configuration to 1.0. Cost-type indicators ($S_{th}$, $S_{\mathit{res}}$, $S_{\mathit{corr}}$) are normalized as $S={\displaystyle \frac{x_{max}-x}{x_{max}-x_{min}}}$, mapping the lowest-cost configuration likewise to 1.0. The individual indicator scores for each configuration are visualized in Figure 9. Inspection of Figure 9 reveals that crimping and welding lead on the mechanical indicators, with $S_{\mathit{mech}}$ and $S_{\mathit{duct}}$ values approaching 1.0, consistent with their superior tensile capacity and ductility reported in Section 3.2. Stainless steel and crimping achieve near-maximum passive film scores (0.944 and 1.0, respectively) owing to their Cr + Ni content of 16.1 and 17.1 wt%—an order of magnitude above the remaining four configurations. Copper scores 0.0 on $S_{\mathit{res}}$, reflecting its anomalous +60% post-conditioning resistance increase. Galvanized steel and aluminum achieve the highest $S_{\mathit{corr}}$ scores of 1.000 and 0.988, respectively, since both exhibited near-zero corroded area fractions in the baseline photographs.

Three complementary weighting schemes were applied to assess the robustness of the rankings to different expert and data-driven perspectives, with the resulting weight vectors presented in Table 1. Under equal weighting, each indicator receives a weight of 1/6, establishing an unbiased baseline. For the Analytic Hierarchy Process (AHP) scheme, expert pairwise judgments were encoded in a 6 × 6 reciprocal comparison matrix following the Saaty scale [13,14], assigning highest relative importance to tensile performance, reflecting the structural load requirements of earthing joints in service. The consistency of the expert judgment matrix was verified using the Saaty consistency index (CI) and consistency ratio (CR):

$$\mathit{CI} = {\displaystyle \frac{{\lambda }_{\mathit{max}}-n}{n-1}}$$

(8)

$$\mathrm{CR}=\mathrm{CI}/\mathrm{RI}$$

(9)

where ${\lambda }_{\mathit{max}}$ is the principal eigenvalue of the 6 × 6 pairwise comparison matrix, n = 6 is the number of criteria and RI = 1.24 is the random consistency index for n = 6 [13,14]. Second-highest relative to passive film chemistry, and reflecting long-term durability, the derived priority vector satisfies the Saaty consistency criterion, with a consistency ratio CR = 0.0139, well below the accepted threshold of 0.10, confirming acceptable coherence of the expert judgments. The resulting AHP weight vector ranks indicators as S_mech = 0.344 > S_passive = 0.226 > S_corr = 0.181 > S_th = 0.116 > S_duct = 0.068 > S_res = 0.065 in descending order. The entropy-based objective weighting scheme derives weights from the Shannon information divergence of each indicator across the six configurations [15]: indicators with greater discriminating power receive higher weights, independent of expert preference. For each indicator j, the normalized score proportion, Shannon entropy and objective weight were computed sequentially as:

$$p_{\mathrm{ij}}={\displaystyle \frac{s_{\mathrm{ij}}}{{\displaystyle {\sum }_{i=1}^m}{s}_{\mathrm{ij}}}}$$

(10)

$$E_j=-{\displaystyle \frac{1}{\mathit{ln}(m)}}{\displaystyle \sum _{i=1}^m}{p}_{\mathrm{ij}}\mathit{ln}(p_{\mathrm{ij}})$$

(11)

$$w_j={\displaystyle \frac{1{-E}_j}{{\displaystyle {\sum }_{k=1}^n}(1-E_k)}}$$

(12)

where $s_{\mathit{ij}}$ is normalized score of configuration i on indicator j, m = 6 is the number of configurations and n = 6 is the number of indicators. $E_i\in [0,1]$ with $E_j\to 1$ indicating a non-discriminating indicator. Indicators with broader spread across configurations yield lower $E_j$ and therefore receive higher weights under Equation (12), ensuring that the weighting reflects the actual information content of the data rather than subjective judgment. Under this scheme, S_duct (0.324) and S_passive (0.316) emerge as the dominant contributors, since the configurations span a factor of 30 in displacement-to-failure (1–30 mm) and more than two orders of magnitude in Cr + Ni content (0.1–17.1 wt%), providing the greatest inter-configuration discrimination. S_corr receives the lowest entropy weight (0.057) because the pre-exposure corroded area fractions are clustered near zero for five of the six configurations, limiting discriminating power at this baseline stage.

The AHP pairwise comparison matrices were constructed following Saaty’s eigenvalue method [22,23]. Relative importance ratios between the six performance indicators were assigned on the basis of expert pairwise judgment reflecting the structural and durability requirements of earthing joints in service, with highest relative importance assigned to tensile performance (S_mech) and second-highest to passive film chemistry (S_passive). The consistency ratio CR = 0.0139 confirms logical consistency of the comparisons (CR < 0.10 required; Saaty [22]). The resulting AHP weight vector, together with the equal-weight and Shannon entropy weight vectors derived independently from indicator data dispersion, is reported in Table 1 to enable full reproducibility of the ranking calculations.

TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) was applied to the normalized score matrix under each of the three global weighting schemes, computing the relative closeness C_i of each configuration to the positive ideal solution. For each configuration i, weighted Euclidean distances to the positive ideal solution A+ and the negative ideal solution A− were computed as:

$$d_i^{+}=\sqrt{{\displaystyle \sum _{j=1}^n}{w}_j{(s_{\mathit{ij}}-s_j^{+})}^2}$$

(13)

$$d_i^{-}=\sqrt{{\displaystyle \sum _{j=1}^n}{w}_j{(s_{\mathit{ij}}-s_j^{-})}^2}$$

(14)

where $s_j^{+} = {\max }_i(s_{\mathit{ij}})$ is for benefit indicators and ${\min }_i(s_{\mathit{ij}})$ is for cost indicators; $s_j^{-}$ is defined conversely; is indicator weight under the chosen weighting scheme. The relative closeness of each configuration to the ideal solution is then:

$$C_i={\displaystyle \frac{d_i^{-}}{(d_i^{+}+d_i^{-})}}$$

(15)

with $C_i\in [0,1]$; $C_{i }= 1$ corresponding to a configuration coinciding with the positive ideal on all indicators simultaneously and $C_i = 0$ to one coinciding with the negative ideal. Configurations are ranked in descending order of $C_i$. The composite scores and rankings are assembled in

Table 2. TOPSIS relative closeness scores (C_i) and rankings under three weighting schemes. Rank 1 denotes the most preferred configuration.

Configuration	Ci (Equal)	Rank (Equal)	Ci (AHP)	Rank (AHP)	Ci (Entropy)	Rank (Entropy)
Copper	0.326	6	0.260	6	0.342	6
Aluminum	0.475	4	0.345	4	0.493	4
Galvanized Steel	0.421	5	0.329	5	0.471	5
Stainless steel	0.568	3	0.486	3	0.551	3
Welding	0.594	2	0.622	2	0.580	2
Crimping	0.737	1	0.807	1	0.674	1

. Under both equal and AHP weighting, crimping ranks first (C_i = 0.737 and 0.807, respectively), followed by welding (C_i = 0.594 and 0.622). Stainless steel occupies third position under both schemes (C_i = 0.568 and 0.486), benefiting from its superior passive film chemistry despite intermediate tensile performance. Copper consistently ranks last (C_i = 0.326 and 0.260), penalized primarily by its anomalous post-conditioning resistance increase, which imposes the maximum cost on S_res and reflects a self-reinforcing oxidation mechanism with adverse implications for long-term contact stability. Under entropy weighting, crimping retains first place by a wider margin (C_i = 0.674) because both S_duct and S_passive—the two highest-weight indicators in this scheme—are simultaneously maximized by crimping, while welding retains second (C_i = 0.580), with stainless steel in third (C_i = 0.551) owing to its passive film advantage.

TOPSIS was selected for its computational tractability, mathematical transparency, and established validation record in engineering multi-criteria decision-making, including material selection, infrastructure assessment, and power system equipment evaluation. The method requires only the normalized decision matrix and weight vector as inputs and executes via straightforward matrix arithmetic implementable in any standard numerical computing environment or spreadsheet application without specialized optimization software. Extension of the framework to additional connector configurations requires only the addition of new rows to the normalized matrix; inclusion of new performance indicators requires updating the weighting vector accordingly. Computational cost scales linearly with the number of alternatives and indicators, rendering the approach accessible for routine use in engineering specification and procurement workflows. Furthermore, TOPSIS simultaneously measures each alternative’s distance from the positive ideal solution and from the negative ideal solution, thereby satisfying the axiom of rational choice and avoiding the single-reference bias inherent in simpler scoring approaches. TOPSIS also produces continuous ratio-scale closeness coefficients C_i rather than ordinal-only rankings, which allows the magnitude of performance differences across configurations and scenarios to be directly compared. The domain-level sensitivity analysis presented in Figure 10 further supports this choice: the leading rank positions of crimping and welding under mechanically oriented weighting, and of stainless steel and aluminum under thermally oriented weighting, remain stable across a broad range of weight perturbations, indicating that the principal conclusions of this study are not strongly dependent on the choice of aggregation method. It is acknowledged that indicator-level measurement uncertainty propagation through the TOPSIS computation was not explicitly quantified in this study; this represents a methodological refinement recommended for future work.

To assess performance under differentiated engineering requirements, TOPSIS was re-run with four scenario-specific weight profiles that redistribute emphasis across the three performance domains. The resulting scenario rankings are summarized in

Table 3. Scenario-specific TOPSIS rankings. SC1 = high-current substation; SC2 = coastal/corrosive; SC3 = mechanically stressed; SC4 = balanced general-purpose.

Configuration	SC1: High-Current Substation	SC2: Coastal/Corrosive	SC3: Mechanically Stressed	SC4: Balanced General
Copper	5	6	5	6
Aluminum	2	5	6	4
Galvanized Steel	4	4	4	5
Stainless steel	1	2	3	3
Welding	3	3	2	2
Crimping	6	1	1	1

and visualized as a heatmap in Figure 10b. In the high-current substation scenario, where thermal and resistance stability indicators carry fourfold excess weight relative to mechanical and corrosion metrics, stainless steel rises to the top rank owing to its combination of low steady-state temperature rise (7 K at 100 A) and substantial post-conditioning resistance reduction (−44%), while aluminum advances to second place by virtue of its minimal thermal perturbation (2.5 K). This rank inversion from the global result underscores that connector selection under high-current duty must weight thermal efficiency preferentially over structural considerations. In the coastal/corrosive scenario—in which S_corr and S_passive together account for two-thirds of the total weight—crimping regains first rank due to its stainless steel sleeve composition (Cr + Ni = 17.1 wt%), with stainless steel in second. The mechanically stressed scenario restores crimping to first and welding to second, reflecting the co-dominance of S_mech (53.5 and 61.5 kN) and S_duct (30 and 18 mm) under high mechanical weighting. The balanced general-purpose scenario reproduces the equal-weighting result, confirming crimping and welding as the default recommendations in the absence of application-specific constraints.

The normalized performance profiles of all six configurations are rendered in the radar chart in Figure 10a, in which each configuration is represented as a closed polygon spanning the six indicator axes. The chart graphically confirms that no configuration dominates on all axes simultaneously: welding achieves the largest projection on S_mech, crimping spans both S_duct and S_passive jointly, and copper achieves near-maximum reach on Sth alone—an inversion from its last-place TOPSIS rank that illustrates how thermal advantage is insufficient to compensate for resistance degradation in multi-criteria scoring. The near-complete overlap between the stainless steel and crimping polygons on the S_passive axis visually confirms that passive film protection is the shared advantage of these two stainless-grade configurations and the primary driver of their convergence under corrosion-heavy scenarios. The contraction of all mechanical connector polygons toward the origin on the S_mech and S_duct axes further illustrates the fundamental performance gap between process-bonded and mechanically bolted assemblies established in Section 3.2.

The quantitative TOPSIS rankings should be interpreted alongside the practical installation and operational characteristics of each joining technology. Mechanical bolted connectors (configurations a, b, c, d) are field-adjustable and reusable, requiring no specialized tooling beyond a calibrated torque wrench; however, they are sensitive to installation torque variability and require periodic re-torquing to compensate for contact relaxation induced by thermal cycling under service loads. Compression-crimped connectors (f) require a calibrated hydraulic crimp tool for field assembly; once formed, the crimp produces a permanent, mechanically consistent bond with minimal ongoing maintenance requirements and no susceptibility to torque relaxation. Exothermic welded joints (e) require a trained operator, proprietary thermite charges, and graphite molds; the resulting metallurgical bond is irreversible, maintenance-free under service conditions, and insensitive to fretting corrosion at faying surfaces. These installation and lifecycle characteristics are qualitative factors that practitioners should weight alongside the quantitative TOPSIS scores when applying the framework to specific procurement or specification decisions.

The sensitivity of TOPSIS ranking to domain weight was evaluated by sweeping the weight of each performance domain from 0 to 100%, with the residual weight distributed equally between the remaining two domains. The resulting rank trajectories are displayed in Figure 10c (thermal domain), Figure 10d (mechanical domain), and Figure 10e (corrosion domain). Figure 10d shows that crimping and welding hold first and second rank for any mechanical domain weight above approximately 30%, indicating a robust hierarchy under a wide range of mechanically oriented weighting profiles. Figure 10c reveals a sharp rank inversion near a thermal domain weight of 35%, at which point stainless steel and aluminum displace crimping and welding from the leading positions, identifying the crossover threshold beyond which high-current thermal efficiency becomes the decisive criterion for connector selection. Figure 10e demonstrates that crimping and stainless steel converge to the top two ranks for corrosion domain weights above approximately 55%, reflecting the step-change in passive film content between the stainless-grade and non-stainless configurations. Across all three panels, copper consistently occupies the fifth or sixth rank, and galvanized steel remains in the lower half of the table, confirming the robustness of their relative assessments to changes in weighting strategy. Taken together, the framework results demonstrate that crimping is the highest-ranking configuration under the majority of weighting conditions, while connector selection for specialized environments—particularly high-current substations—requires deliberate reweighting toward domain-specific performance metrics.

Several limitations of the proposed multi-criteria evaluation model should be noted. First, all six performance indicators are derived from a single connector geometry and galvanized steel rod diameter tested at a specified current range; the absolute TOPSIS C_i scores and scenario rank positions are specific to these conditions and should not be directly extrapolated to other connector designs, conductor cross-sections, or current ratings without repeating the indicator extraction procedure under the relevant service conditions. Second, S_passive uses EDS-derived Cr + Ni surface content as a first-order compositional proxy and does not capture molybdenum content, microstructural effects, pitting potential, or galvanic coupling between dissimilar metals at the joint interface (as noted in the model limitations discussion above). Third, AHP weights reflect expert judgment under generalized grounding system deployment conditions; while the domain-level sensitivity analysis in Figure 10 demonstrates that the leading rank positions are robust to weight perturbations, domain-specific applications may require recalibration of the pairwise comparison matrix. Fourth, accelerated salt-spray testing (ASTM B117/ISO 9227) [24] provides a reproducible comparative ranking but does not constitute a quantitative prediction of absolute long-term field corrosion rates. Fifth, the sensitivity analysis sweeps domain-level AHP weights but does not propagate indicator-level measurement uncertainty through the TOPSIS computation; indicator-level uncertainty propagation represents a methodological refinement recommended for future work.

5. Conclusions

The present study characterizes static mechanical performance under monotonic quasi-static tensile loading; cyclic mechanical loading arising from thermal expansion–contraction cycles, vibration, or repeated fault-current surges is not addressed. The evolution of contact resistance, mechanical integrity, and surface corrosion under cyclic thermomechanical loading and long-term aging conditions represents an important extension of this baseline comparative assessment and is identified as a priority for future investigation.

This study has presented a systematic multi-criteria comparative evaluation of six grounding connector configurations across three conductor materials and three joining technologies. Three principal scientific findings emerge from the combined experimental and analytical program. First, the electro-thermal finite element modeling results demonstrate that electrical conductivity of the connector body material is the dominant driver of steady-state temperature rise under service current: copper (a) and aluminum (b) configurations maintained the lowest operating temperatures, while galvanized steel (d) exhibited the highest, consistent with material resistivity differences. Second, the tensile mechanical analysis confirms that exothermic welding (e) and compression crimping (f) produce joint strengths that exceed the tensile capacity of the galvanized steel rod substrate, whereas mechanical bolted connectors are governed by slip onset at the conductor–barrel contact interface. The exothermic welded joint achieved the highest maximum load; however, both process-bonded configurations outperformed all mechanical bolted alternatives. Third, the corrosion assessment reveals that the stainless steel connector (c) exhibited the lowest corroded area fraction after accelerated salt-spray exposure, attributable to its high Cr + Ni surface content. The exothermic welded configuration (e) exhibited pronounced localized rust in the iron-rich weld zone, yielding a lower total corroded area fraction than the non-stainless mechanical bolted connectors but substantially higher than the stainless steel configuration. The integrated multi-criteria decision-making framework demonstrates that no single configuration is universally optimal: in thermally critical applications, copper and aluminum configurations score highest; in mechanically demanding or maintenance-free applications, exothermic welding and crimping are preferred; in corrosion-critical environments, the stainless steel mechanical connector ranks first. The scenario-specific ranking results and sensitivity analysis provide a transparent, reproducible basis for context-dependent connector selection in engineering practice.

Several limitations of this study should be noted. The quantitative TOPSIS rankings are specific to the tested connector geometry and galvanized steel round rod substrate; direct extrapolation to other conductor cross-sections or connector designs requires repeating the indicator extraction procedure under the relevant conditions. Corrosion performance is assessed by an accelerated salt-spray protocol (ASTM B117/ISO 9227) [24], which provides a reproducible comparative ranking but does not constitute a prediction of absolute long-term field durability. Under prolonged field exposure, relative rankings may shift due to differences in real atmospheric chloride deposition rates and wetting–drying cycles compared with continuous spray; progressive galvanic coupling effects as corrosion products accumulate at dissimilar-metal interfaces; contact-resistance increases, driven by oxide layer growth under cyclic thermal loading; and localized crevice corrosion develops at the conductor–sleeve interface, which is not captured by short-duration immersion or spray protocols. Cyclic mechanical loading, thermal fatigue, and electrochemical characterization of passivation kinetics lie outside the current scope and are identified as important future directions. Future research should also address: (i) transient electro-thermal analysis under fault-current surge conditions, preceded by a mesh sensitivity study to establish convergence of the steady-state solution and to ensure that absolute temperature predictions meet the accuracy requirements of safety-critical surge assessments; (ii) fatigue and cyclic loading characterization under thermal and vibration excitation; (iii) long-term field monitoring to validate accelerated corrosion outcomes against service exposure data; (iv) electrochemical characterization (electrochemical impedance spectroscopy, potentiodynamic polarization, open-circuit potential evolution) to complement the salt-spray visual assessment; (v) extension of the multi-criteria decision-making framework to additional connector sizes, current ratings, and environmental exposure classes; and (vi) detailed microstructural characterization of joint interfaces using optical metallography, electron backscatter diffraction, or transmission electron microscopy to elucidate bonding mechanisms and microstructure–property relationships.

Finally, formal propagation of material property and boundary condition uncertainty through the electro-thermal finite element modeling predictions was not performed; quantification of absolute temperature prediction uncertainty and its effect on the multi-criteria ranking electro-thermal indicator is identified as a direction for future work.