The BIW inspection problem: 5,000 welds, mostly unverified

A typical automotive car body contains approximately 5,000 spot welds, more than 95% of which are produced by resistance spot welding (RSW), according to a 2023 patent from South China University of Technology. Each joint directly determines vehicle stiffness and crash safety — yet the majority of current inspection methods remain destructive, semi-destructive, or reliant on expensive off-line equipment that cannot verify every weld in a high-volume production line. The scale of the problem makes it structurally different from most NDT challenges: 100% inline coverage is the only commercially meaningful standard.

The research and patent landscape reviewed here encompasses more than 50 sources spanning ultrasonic NDT, eddy current testing, dynamic resistance monitoring, thermography, vision systems, and impedance-based measurement techniques including EIS. The most active assignees in weld quality inspection using electrical and impedance methods include LG Energy Solution, South China University of Technology, Denshi Jiki Kogyo (Japan), and independent inventors in Japan. The dominant NDT modalities in the spot welding field remain ultrasonic and eddy current testing, but electrical resistance and impedance techniques are gaining traction — particularly for detecting weak or under-bonded welds.

The innovation trend visible across the patent and literature dataset is a clear progression: from single-parameter DC resistance measurement, to multi-parameter dynamic resistance monitoring, to frequency-resolved AC impedance (EIS) analysis. The introduction of machine learning — neural networks, support vector machines, convolutional networks — to interpret the resulting signal space is accelerating this trajectory, as demonstrated by the Korea Institute of Industrial Technology’s weld quality prediction algorithm achieving 93–94% accuracy for tensile shear strength and indentation depth.

How EIS detects weld defects: the physics of impedance contrast

Electrochemical impedance spectroscopy detects spot weld defects by exploiting the measurable difference in electrical impedance between a fully fused metal nugget and an incomplete, cold, or voided joint. As established in research from the University of Florence, EIS works by applying an electrical sinusoidal perturbation with a fixed frequency and measuring the electrical impedance Z of the sample — and ISO standards 16773 and 17463 have institutionalised it as a non-destructive method for assessing corrosion resistance of metal surfaces. The technique is inherently non-destructive because the excitation amplitude is kept small enough to avoid perturbing the material under test.

The physical basis of impedance contrast is well-understood. A fully formed weld nugget consists of resolidified metal with a continuous crystal lattice across the joint interface, producing low bulk resistance and negligible interfacial capacitance. A cold weld, partial weld, or weld with internal voids presents an interface layer with higher resistivity, and the interfacial charge-transfer dynamics produce a characteristic semi-circular arc in the Nyquist plot corresponding to a parallel RC element. This frequency dependence is the fundamental advantage of EIS over simple DC resistance measurement: it allows decomposition of bulk, interfacial, and diffusion contributions, providing far greater diagnostic resolution.

“EIS allows decomposition of bulk, interfacial, and diffusion contributions — providing diagnostic resolution not available from single-frequency DC resistance methods.”

The bridge from EIS corrosion science to spot weld inspection is provided by a patent from the China Electronics Product Reliability and Environmental Testing Institute, which explicitly applies impedance measurement to battery tab welding quality evaluation. In this method, multiple external voltages are applied to the weld zone, the resulting currents are measured with a sensitivity of 10⁻⁶ to 10⁻³ A/V, and the slope of the resulting I-V curve (i.e., the admittance) yields the impedance of the weld. The patent explicitly mentions the use of an electrochemical workstation and techniques including AC impedance (EIS), cyclic voltammetry, and linear sweep voltammetry. The standard impedance value from a defect-free weld serves as the reference, and the difference between test impedance and standard impedance determines pass/fail — an architecture that is fully transferable to automotive BIW spot welds on steel and coated steel sheets.

For zinc-coated automotive steels — precisely the materials that cause increased electrode wear and quality variability in BIW production — EIS could monitor both the zinc layer integrity and the underlying weld quality from the same measurement. The University of Florence study on corrosion resistance testing of electroplated metals using fast EIS shows that EIS can differentiate coating quality on metal substrates within minutes, a timescale compatible with industrial inspection.

A practical barrier to BIW deployment — the need to apply EIS without immersing the joint in liquid electrolyte — has been addressed in the corrosion testing literature. The Silesian University of Technology developed a prototypical flexible measuring probe with gel electrolyte for EIS testing on any surface regardless of position, demonstrating that dry-surface impedance measurement of steel structures is feasible. This directly removes the most-cited practical obstacle to applying EIS in an automotive assembly environment.

Impedance measurement architectures: from four-probe DC to full Cole-Cole EIS

Several distinct measurement architectures have been patented for applying impedance and resistance techniques to spot weld quality assessment, forming a progression from simple DC methods to full frequency-swept EIS. Understanding the trade-offs between them is essential for selecting the right approach for a given BIW production environment.

Four-probe DC resistance

The simplest architecture is the DC four-probe resistance method, in which a constant current is injected through a pair of electrode probes and the resulting surface potential is measured by a separate pair of voltage probes. Denshi Jiki Kogyo (Japan) patented spot welding strength evaluation using the constant-current four-terminal probe method, measuring resistance R1 at the weld zone and R2 at an unwelded zone and evaluating weld strength from their ratio. The same assignee extended this to a multi-probe head with sequential probe switching to map the surface potential distribution and thereby compute the effective nugget diameter non-destructively. This is the most mature and commercially deployed architecture for non-destructive weld assessment.

Micro-resistance threshold comparison

LG Energy Solution’s multiple welding defect inspection patents deploy micro-resistance measurement at nanoohm-to-microohm resolution for battery weld inspection, with the threshold resistance comparison method providing a practical inline framework. A reference population of known-good welds establishes the threshold resistance, and welds exceeding this threshold are flagged as defective. The resolution requirements — nanoohm-to-microohm — are achievable with modern instrumentation and are consistent with the resistance changes expected from incomplete nugget formation in steel sheet joints. According to NIST metrology standards, four-terminal resistance measurement at sub-microohm resolution is well within the capability of commercially available precision instruments.

Electrode-integrated impedance measurement

The French patents from LEON PAUL describe a method for assessing the surface condition of a spot welding electrode by measuring the impedance of a volume at the electrode surface. While oriented toward electrode condition monitoring rather than weld inspection, this demonstrates that impedance measurement during or between welds can be integrated into the spot welding machine itself without additional external hardware — a significant advantage for BIW integration.

Full EIS with Cole-Cole analysis

The most diagnostically rich approach — full EIS with Cole-Cole plot analysis — is implemented in the LG Energy Solution battery weld state inspection patent. Alternating current or voltage at a set frequency band is applied to the battery prior to electrolyte injection, and the resulting Cole-Cole (Nyquist) plot is used to determine whether the weld between the current collector and the tab or can is sound. The x-intercept of the Cole-Cole plot — representing the real part of the impedance at high frequency, dominated by contact and bulk resistance — serves as the quality indicator. The frequency-resolved nature of this measurement means that it can, in principle, distinguish between different defect types: an unbonded interface will produce a high-frequency arc dominated by contact resistance, while a porous or partially fused nugget may produce an additional low-frequency arc characteristic of a diffusion-limited interface.

Why ultrasonic and eddy current methods fall short of 100% inline inspection

Ultrasonic C-scan testing is not suitable for inspection of every weld in high-volume production due to the considerable operator skill required and the requirement to take parts off-line — a finding from the Australian National University’s comparison of dynamic resistance and ultrasonic C-scan approaches. This is not a marginal limitation: in a BIW line producing hundreds of car bodies per shift, removing parts for off-line ultrasonic inspection is operationally incompatible with production targets.

Even in-process ultrasonic systems — such as the novel online real-time ultrasonic NDE system presented by Liverpool John Moores University — require coupling media and careful transducer placement in the electrode assembly, as demonstrated by the Tessonics Corporation patent which positions a transducer inside the electrode to transmit through the weld tip. The Poznan University of Technology study confirms strong correlations between 20 MHz longitudinal wave parameters and nugget strength, but the methodology still requires dedicated high-frequency probes and controlled contact conditions. According to WIPO patent data, the volume of ultrasonic weld inspection patents far exceeds that of impedance-based methods — yet the fundamental access and throughput constraints remain unresolved.

Dynamic resistance monitoring during the weld cycle itself is a more competitive approach. Both Nissan Motor Co. and early General Motors patents established that electrode voltage, current flow time, voltage pulse count, and the integral of excess electrode voltage over a base voltage — combined with inter-electrode resistance — can determine weld quality. The Korea Institute of Industrial Technology later demonstrated a weld quality prediction algorithm using dynamic resistance and electrode displacement signals with prediction accuracies of 93–94% for tensile shear strength and indentation depth. However, these in-process methods can only assess the weld at the moment of formation and cannot be applied retrospectively to completed assemblies — a critical limitation for audit and traceability requirements in automotive production, which are increasingly mandated by standards bodies such as ISO and the IATF.

Key assignees, active patents, and the machine-learning frontier

LG Energy Solution is the most prolific filer in resistance and impedance-based weld inspection, with at least eight active or recently active patents across US, EP, KR, JP, and IN jurisdictions. Their technology progresses from simple threshold-resistance comparison to full EIS-based Cole-Cole analysis, and their most recent patents (2024–2025) demonstrate active innovation in this space across multiple patent families. While their primary application is battery manufacturing, their methods are directly applicable to resistance spot welding of thin metal sheets in automotive BIW production.

Denshi Jiki Kogyo (Japan) holds two active Japanese patents (2008 and 2012) specifically on spot welding strength evaluation by the four-probe resistance method for metallic materials, with explicit claims for non-destructive weld nugget assessment. South China University of Technology contributes both a 1D convolutional neural network approach to dynamic resistance analysis for BIW spot weld quality and an ultrasonic metal welding quality assessment method, positioning themselves across multiple sensing modalities. General Motors Corporation established foundational patents in the 1980s on predicting resistance spot weld quality and detecting edge welds, both relying on electrode electrical signals — the conceptual precursor to modern impedance-based approaches. Nissan Motor Co. filed parallel patents in Germany and the UK on resistance-based weld quality determination combining electrode voltage, current time integrals, and inter-electrode resistance for pass/fail determination.

The integration of machine learning with impedance and resistance signal data represents the dominant innovation direction for inline spot weld quality control. The vision-system and fuzzy support vector machine approach from Wuhan University of Technology, the review of advances in RSW quality control from Universidad de Valladolid, and the 1D-CNN dynamic resistance approach from South China University of Technology all point toward the same convergence: rich, multi-dimensional signal datasets — precisely the kind produced by EIS — interpreted by trained models rather than fixed thresholds. As reported by IEEE, the intersection of frequency-domain electrical measurement and machine learning is one of the fastest-growing areas in industrial NDT research, and EIS fits naturally into this trajectory by producing frequency-domain datasets well-suited to model-based interpretation.

“EIS fits naturally into the machine-learning trajectory for inline weld quality control by producing rich, frequency-domain datasets well-suited to model-based and neural-network interpretation.”

The patent landscape reviewed here — spanning more than 50 sources from 1970 to 2025 — reveals that the direct application of full EIS to automotive BIW spot weld inspection remains an emerging and partially patent-protected area. The most significant open technical challenge is adapting the Cole-Cole measurement architecture from battery cells (where the weld is accessible and the geometry is controlled) to the complex, multi-layer steel sheet stacks of a BIW assembly. The flexible gel-electrolyte probe developed by the Silesian University of Technology for in-situ EIS of anti-corrosion coatings on steel structures offers one validated pathway. The integration of EIS measurement into the welding gun itself — as demonstrated in principle by the LEON PAUL electrode impedance patents — offers another. The combination of both approaches, with machine-learning interpretation of the resulting Cole-Cole data, defines the frontier of the field as of the patent and literature landscape reviewed here.