Orthogonal Experimental Design for Heat Treatment Optimization of Weld Overlay Rolls

1. Definition and Technical Principles

Orthogonal experimental design (OED), also known as Taguchi method or fractional factorial design, is a statistical methodology used to systematically optimize multi-variable processes by evaluating a carefully selected subset of experimental conditions rather than conducting exhaustive full-factorial trials. In the context of weld overlay roll heat treatment, this methodology enables the identification of optimal combinations of heat treatment parameters—tempering temperature, holding time, cooling rate, furnace atmosphere, and pre-heat temperature—while minimizing the total number of required trials.

A weld overlay roll is a critical industrial component, typically consisting of a carbon steel or low-alloy steel substrate (e.g., ASTM A514, Q345, or 42CrMo) with a hardfacing or corrosion-resistant overlay deposited via TIG (GTAW) or MIG (GMAW) arc welding. The overlay material—commonly Cr-Ni austenitic alloys (ASTM A568 Type 309L, Type 312), high-chromium cast irons, or nickel-based alloys (ASTM B348 Type B, Type C)—must achieve specific mechanical properties including hardness (HV 250–500 depending on service), impact toughness, and fatigue resistance. Heat treatment is the decisive post-weld process that governs the final microstructure, residual stress state, and dimensional stability of the finished roll.

The orthogonal experimental method operates on the principle that in multi-factor systems, certain factor-level combinations provide disproportionate information about main effects and interaction effects. By constructing an orthogonal array (e.g., L9(3⁴), L16(4⁵), L27(3¹³)), the experimenter can evaluate 4–5 factors simultaneously with only 9–16 trials instead of the 81–1,024 trials required by full-factorial design. The response variables typically include hardness uniformity (measured at multiple cross-section locations), residual stress magnitude (measured by X-ray diffraction or hole-drilling method per ASTM E1926), dimensional distortion (measured by laser tracker or coordinate measuring machine per ISO 1101), and microstructural quality (evaluated via metallographic examination per ASTM E3).

2. Category and Business Positioning

This capability falls squarely within the company's process engineering and quality assurance framework, specifically under the "Weld Overlay Process Qualification and Optimization" domain. It bridges the gap between raw weld deposition capability and the delivery of fully qualified, performance-verified overlay components. Within the company's three primary technology routes—TIG/MIG weld overlay, hydraulic explosive bonding, and explosion welding—the orthogonal experimental method for heat treatment optimization is most directly applicable to the TIG/MIG weld overlay route, where post-weld heat treatment is a mandatory and critical process step.

From a business positioning standpoint, this capability serves three strategic functions:

• WPS/PQR Qualification Support: Systematic DOE-based optimization generates the statistical evidence required to qualify Welding Procedure Specifications (WPS) per ASME Section IX, AWS D1.1, or GB/T 985, demonstrating that process parameters produce consistent, repeatable results across production volumes.
• Cost Reduction and Cycle Time Optimization: By identifying the minimum effective heat treatment parameters, the company reduces furnace energy consumption, minimizes production cycle time, and lowers the risk of over-tempering or under-tempering defects.
• Customer Value Differentiation: Data-driven process optimization provides customers with statistically validated performance guarantees, reducing warranty claims and establishing the company as a technically rigorous supplier in competitive bidding scenarios.

3. Technical Purpose and Value

3.1 Primary Objectives

The application of orthogonal experimental design to weld overlay roll heat treatment serves the following specific technical objectives:

• Hardness Homogeneity: Achieve hardness variation within ±10% across the overlay layer, substrate transition zone, and base metal, ensuring uniform wear resistance and preventing premature failure at property discontinuities.
• Residual Stress Reduction: Reduce post-weld residual stresses to below 50% of the material's yield strength (per NB/T 20321 for nuclear applications or API 579 for pressure equipment), thereby minimizing the risk of stress corrosion cracking and fatigue failure.
• Dimensional Control: Limit heat treatment-induced distortion to within specified tolerances (typically ±0.1 mm/m for precision rolls per ISO 1101 GD&T requirements), avoiding costly post-treatment grinding and rework.
• Microstructural Stability: Ensure complete transformation of metastable phases (retained austenite, martensite, carbide networks) to thermodynamically stable configurations that resist degradation under thermal cycling and mechanical loading.

3.2 Quantifiable Value Metrics

• Reduction in heat treatment trial count by 75–85% compared to one-factor-at-a-time (OFAT) experimentation
• Decrease in post-treatment rework rates from typical industry baseline of 8–12% to below 3%
• Shortening of WPS qualification cycle from 4–6 weeks to 2–3 weeks
• Improved first-pass yield rate for overlay roll production from 85–90% to 95%+

4. Key Process and Implementation Points

4.1 Orthogonal Array Selection

The selection of the orthogonal array depends on the number of factors and levels to be investigated. For weld overlay roll heat treatment, the following factor set is typical:

Factor	Level 1	Level 2	Level 3	Level 4	Measurement Method
A: Pre-heat Temperature (°C)	200	300	400	500	Thermocouple embedded at roll surface
B: Tempering Temperature (°C)	540	600	650	700	Furnace thermocouple + DTS fiber optic
C: Holding Time (hours)	2	4	6	8	Furnace controller timer + log
D: Cooling Rate (°C/min)	10	20	30	40	Instrumented cooling chamber
E: Furnace Atmosphere	Air	Protective gas (N₂)	Vacuum (10⁻¹ Pa)	—	O₂ sensor + vacuum gauge

For a 5-factor, 4-level study, an L16(4⁵) orthogonal array is employed, requiring only 16 experimental runs to evaluate all five factors and their main effects. If interaction effects between specific factor pairs (e.g., tempering temperature × holding time) are of interest, an L25(5²) or mixed-level array may be substituted, or a Taguchi L18 array with deliberate replication can be used.

4.2 Response Variable Measurement Protocol

Each experimental trial produces a weld overlay roll specimen (typically a representative coupon or a full-size production roll) that undergoes the following measurement sequence:

1. Hardness Measurement: Vickers hardness (HV10) measured at 11 locations: 3 in the overlay layer (surface, mid-layer, interface), 3 in the heat-affected zone (HAZ), and 3 in the base metal, per ASTM E92. Report as mean ± standard deviation.
2. Residual Stress Measurement: Longitudinal and circumferential residual stresses measured at 5 depths (0, 0.5, 1.0, 2.0, 5.0 mm from surface) using X-ray diffraction per ASTM E1926 or incremental hole-drilling per ASTM E837.
3. Dimensional Distortion: Out-of-roundness, barrel distortion, and axial twist measured using laser tracker or CMM per ISO 1101, with reference to pre-heat treatment baseline dimensions.
4. Microstructural Examination: Cross-sections prepared per ASTM E3, etched with Nital 3% or Beraha reagent, and examined at 100×–1000× magnification. Phase identification via optical microscopy and SEM-EDS. Quantify retained austenite fraction using ASTM E1023 or Rockwell hardness method per ASTM E1023.
5. Impact Toughness: Charpy V-notch impact energy measured at 25°C and -20°C per ASTM E23, with specimen orientation conforming to ASTM E23 Section 8 (subsize specimens if material thickness is limited).

4.3 Data Analysis and Optimization

The collected data is analyzed using the following systematic approach:

1. Signal-to-Noise (S/N) Ratio Calculation: For each response variable, compute the S/N ratio using the appropriate Taguchi characteristic:
   • Smaller-the-better (residual stress, distortion): S/N = -10 log₁₀[Σ(yᵢ²)/n]
   • Larger-the-better (impact energy): S/N = -10 log₁₀[Σ(1/yᵢ²)/n]
   • Smaller-the-better (hardness variation/standard deviation): S/N = -10 log₁₀[Σ(yᵢ²)/n]
2. Main Effects Plot: Plot the mean S/N ratio for each factor level to identify the level that maximizes performance. The optimal combination is the set of levels that collectively yield the highest S/N ratio.
3. ANOVA (Analysis of Variance): Perform ANOVA to determine the statistical significance of each factor's contribution (p-value < 0.05 indicates significance). Report the percentage contribution of each factor to total variation.
4. Interaction Effect Analysis: For significant factor pairs, construct interaction plots to identify synergistic or antagonistic effects. If significant interactions exist, the simple main-effects model is insufficient, and a second-order regression model or response surface methodology (RSM) should be applied.
5. Confirmation Trial: Conduct a confirmation run at the predicted optimal parameter combination to verify that the actual performance matches the predicted S/N ratio within acceptable confidence limits (typically ±3 dB).

4.4 Typical Optimization Results

Based on industry experience with Cr-Mo low-alloy steel substrate rolls overlaid with 309L stainless steel via TIG welding, the following representative optimization results illustrate the method's effectiveness:

Parameter	OFAT Baseline	DOE-Optimized	Improvement
Hardness Uniformity (HV)	280–420 (σ = 45)	310–350 (σ = 18)	60% reduction in variation
Peak Residual Stress (MPa)	480	185	61% reduction
Barrel Distortion (mm/m)	0.35	0.08	77% reduction
Impact Energy @ -20°C (J)	22	48	118% increase
Total Heat Treatment Time (h)	14	8.5	39% cycle time reduction

5. Applicable Standards and Acceptance Criteria

5.1 Heat Treatment Standards

• ASTM A743/A743M: Standard Specification for Castings, Iron-Cast Iron, for General Engineering Purposes—applies to overlay material qualification
• ASTM A396: Standard Specification for Alloy Steel Bolts and Other External Threaded Fastenings—reference for tempering schedules of alloy steel substrates
• GB/T 16923-2008: 热处理术语 (Heat Treatment Terminology)—Chinese national standard for heat treatment nomenclature and process definitions
• GB/T 9452-2015: 钢及热处理件金相检验 (Metallographic Examination of Steel and Heat Treated Parts)
• NB/T 20321-2013: 核电厂机械设备焊接工艺评定 (Welding Procedure Qualification for Nuclear Power Plant Mechanical Equipment)—includes post-weld heat treatment requirements for nuclear-grade overlay components

5.2 Weld Overlay and WPS Qualification Standards

• ASME Section IX, Part 4 (QW-400): Qualification of Welding Procedure Specifications—governs PWHT (Post-Weld Heat Treatment) parameters and acceptance criteria for welded overlay joints
• AWS D1.1/D1.1M: Structural Welding Code—Steel—includes requirements for weld overlay qualification and post-weld treatment
• GB/T 985.1-2008: 焊接工艺评定试验 (Welding Procedure Qualification Tests)—Chinese national standard for welding procedure qualification
• ISO 15614-1:2017: Qualification testing of welding procedures for metallic materials—Part 1: General rules
• API 570: Piping Inspection Code—includes PWHT requirements for piping repair and overlay

5.3 Acceptance Criteria

Criterion	Acceptance Limit	Test Standard	Applicable Component
Overlay Hardness	Per customer spec, typically HV 300–500	ASTM E92	All overlay rolls
Hardness Gradient (overlay to substrate)	≤ 50 HV/mm transition zone	ASTM E92	All overlay rolls
Residual Stress	≤ 50% σᵧ of base metal	ASTM E1926 / E837	Critical service rolls
Impact Energy @ Service Temp	≥ 27 J (per ASME IX QW-422)	ASTM E23	Low-temperature service
Dimensional Distortion	Per ISO 1101 drawing tolerance	ISO 1101 / CMM	Precision rolls
Weld Defects (overlay)	Per AWS D1.6 Class B	ASTM E165 / E164	All overlay welds

6. Common Risks and Controls

6.1 Process Risks

Risk	Consequence	Mitigation Control
Insufficient DOE replication leading to Type II statistical error (failing to detect a significant factor)	Suboptimal parameter selection; latent process variability	Minimum 2 replications per orthogonal array; verify with confirmation trial; if confirmation deviates > 3 dB, expand to full factorial for critical factors
Interaction effects masked by orthogonal array design (orthogonal arrays assume additive effects)	Optimal parameters predicted but not achievable in practice	Conduct interaction plots for all factor pairs; if significant interactions detected, switch to full factorial or RSM with central composite design
Furnace temperature uniformity exceeding ±10°C across roll surface	Non-uniform tempering response; localized over- or under-tempering	Map furnace temperature with multi-point thermocouple array per ASTM E21; implement forced-air circulation; reject furnace if ΔT > 15°C across load zone
Uncontrolled cooling rate due to furnace door opening or ambient draft	Re-martensitization; increased residual stress; distortion	Use instrumented cooling chamber with programmable rate; monitor with embedded thermocouples; lock furnace doors with interlocks during cooling phase
Inappropriate pre-heat temperature causing substrate over-tempering or grain growth	Reduced substrate strength; degraded fatigue life	Limit pre-heat to below 0.5 × Ac₁ of substrate material; verify substrate microstructure before and after treatment per ASTM E3
Carbon contamination from furnace atmosphere during austenitic overlay heat treatment	Carbon enrichment at overlay interface; reduced corrosion resistance; increased brittleness	Use protective atmosphere (N₂ with dew point < -60°C) or vacuum (< 10⁻¹ Pa); monitor O₂ < 500 ppm per ASTM E2024

6.2 Quality Risks

• Statistical over-fitting: When the number of factors and levels exceeds the degrees of freedom in the orthogonal array, the model may over-fit to noise rather than capturing true process effects. Control: Limit to 3–4 factors per array; use ANOVA F-ratio to validate factor significance before incorporating into WPS.
• Specimen representativeness: Coupon specimens may not replicate the thermal mass and heat transfer characteristics of full-size production rolls. Control: Validate coupon-based optimization results with at least one full-size roll confirmation trial before scaling to production.
• Measurement repeatability: Hardness and residual stress measurements can exhibit operator-dependent variability. Control: Implement inter-laboratory calibration per ASTM E1002; train operators on standardized measurement protocols; use automated hardness testers where feasible.

7. Application Across the Company's Three Technology Routes

7.1 TIG/MIG Weld Overlay Route (Primary Application)

The orthogonal experimental method for heat treatment optimization is most directly and comprehensively applicable to the TIG/MIG weld overlay route. In this route, overlay layers are deposited via arc welding onto substrate rolls, and post-weld heat treatment is mandatory to achieve the required mechanical properties. The DOE methodology is applied to:

• Multi-layer overlay heat treatment: For thick overlays (≥ 6 mm) requiring multiple welding passes, the DOE identifies the optimal inter-pass heat treatment schedule (temperature, time, cooling rate) that balances stress relief with microstructural refinement without excessive cycle time.
• Transition layer optimization: When a transition layer (e.g., 309L) is deposited between dissimilar base and overlay materials (e.g., carbon steel substrate with 312 or cast iron overlay), the DOE determines the PWHT parameters that minimize carbon migration across the interface while maintaining adequate toughness.
• Batch production consistency: Once optimal parameters are identified via DOE, they are codified into the WPS and implemented in production with statistical process control (SPC) monitoring of furnace temperature profiles and hardness results.

7.2 Hydraulic Explosive Bonding Route (Indirect Application)

In hydraulic explosive bonding (waterjet-assisted explosive welding), the bonding process itself does not require post-weld heat treatment because the explosive process produces a metallurgically bonded interface without melting. However, the DOE methodology contributes indirectly in the following ways:

• Post-bond stress relief optimization: Although the bond interface is formed without melting, the explosive loading introduces plastic deformation and residual stresses in the substrate and cladding layers. A mild stress relief treatment may be applied, and DOE can optimize the temperature and time to relieve these stresses without degrading the explosive bond interface quality (which is characterized by the characteristic wavy interface morphology per ASTM A582).
• Substrate pre-treatment optimization: For hydraulic explosive bonding of rolls, the substrate material may require prior heat treatment (normalizing or annealing) to achieve the ductility and thickness uniformity required for successful bonding. DOE can optimize these pre-treatment parameters to maximize bonding window and minimize defect rate.

7.3 Explosion Welding Route (Indirect Application)

Similar to hydraulic explosive bonding, conventional explosion welding produces a solid-state bond without melting, and the bond quality is governed by impact velocity, contact angle, and material properties rather than heat treatment. However, the DOE methodology is relevant in the following contexts:

• Post-explosion stress relief: The high-strain-rate deformation during explosion welding can introduce significant residual stresses and strain hardening in the cladding layer. A controlled stress relief treatment, optimized via DOE, can reduce these stresses while preserving the bond strength (typically ≥ 50 MPa shear strength per ASTM A582 or ISO 13520).
• Explosion welding of overlay rolls with subsequent weld overlay: In hybrid processes where an explosion-welded cladding layer is followed by a TIG or MIG weld overlay pass to achieve required thickness, the combined process requires careful heat treatment optimization. The DOE can identify PWHT parameters that simultaneously relieve explosion-induced stresses and homogenize the weld overlay microstructure without compromising the explosion bond interface.

8. Contribution to Qualification Building, Product Delivery, and Customer Value

8.1 Qualification Building

The systematic application of orthogonal experimental design to heat treatment optimization directly supports the company's qualification infrastructure in the following ways:

• WPS Development and Qualification: The DOE-generated data provides the statistical foundation for WPS qualification per ASME Section IX, AWS D1.1, or GB/T 985.1. The optimal parameter combination, validated by confirmation trials, becomes the qualified PWHT procedure in the WPS, with documented performance data supporting each parameter selection.
• Process Capability Documentation: The S/N ratio analysis and ANOVA results demonstrate the robustness of the heat treatment process—its insensitivity to minor parameter fluctuations. This robustness data is critical for customer audits and for demonstrating process capability (Cpk ≥ 1.33) to regulatory bodies.
• Technical Knowledge Base: The DOE study generates a comprehensive knowledge base of factor-effect relationships that informs future process development, material qualification, and troubleshooting. This institutional knowledge reduces the learning curve for new product development and accelerates time-to-market.

8.2 Product Delivery

• Reduced Lead Time: By eliminating trial-and-error experimentation, the DOE approach reduces the heat treatment qualification cycle from 4–6 weeks to 2–3 weeks, directly shortening project lead times and improving on-time delivery performance.
• Higher First-Pass Yield: Optimized parameters produce consistently acceptable results, reducing the need for rework and re-treatment. This improves production throughput and reduces cost of quality (COQ) by an estimated 15–25%.
• Scalability: DOE-optimized parameters are transferable across production volumes—from prototype single rolls to batch production of hundreds of units—without requiring re-optimization for each order.

8.3 Customer Value

• Performance Assurance: Customers receive statistically validated performance data demonstrating that delivered overlay rolls meet or exceed specified hardness, toughness, and dimensional requirements. This reduces customer acceptance risk and accelerates project commissioning.
• Extended Service Life: Optimized heat treatment produces microstructures with superior fatigue resistance and reduced residual stress, translating to extended roll service life in demanding applications (mining crushers, steel mill rolls, paper machine rolls). This delivers direct economic value through reduced downtime and extended replacement intervals.
• Technical Credibility: The application of rigorous statistical methodology positions the company as a technically sophisticated partner, enhancing competitive positioning in high-value bids where technical capability is a differentiating factor.
• Customization Flexibility: The DOE framework enables rapid re-optimization when customers specify non-standard requirements (unusual hardness ranges, extreme temperature service, special alloy combinations), providing a responsive and flexible engineering service.

9. Implementation Roadmap

To fully leverage this capability, the following implementation steps are recommended:

1. Phase 1 — Pilot Study (Weeks 1–4): Select one representative overlay roll product (e.g., 309L overlay on Q345 substrate for steel mill application). Conduct a full L16 orthogonal experimental campaign with 3 factors × 4 levels. Generate baseline optimization data and validate the methodology.
2. Phase 2 — Standardization (Weeks 5–8): Document the DOE methodology, measurement protocols, and data analysis procedures in a standardized work instruction. Train process engineers and quality inspectors on the methodology. Integrate DOE data analysis software (e.g., Minitab, JMP, or Design Expert) into the company's quality management system.
3. Phase 3 — Expansion (Weeks 9–16): Extend DOE optimization to additional product families (different substrate/overlay combinations, different service conditions). Build a comprehensive database of optimized parameters linked to product specifications and customer requirements.
4. Phase 4 — Continuous Improvement (Ongoing): Incorporate DOE findings into the company's PDCA (Plan-Do-Check-Act) cycle. Use production feedback (field performance data, customer complaints, NDT results) to identify opportunities for further optimization and to validate the continued effectiveness of qualified procedures.

10. Conclusion

The application of orthogonal experimental design to heat treatment optimization of weld overlay rolls represents a mature, statistically rigorous approach to process engineering that directly enhances the company's technical capabilities across qualification building, product delivery, and customer value creation. By replacing empirical trial-and-error with systematic statistical optimization, the company achieves faster qualification cycles, higher production yields, more consistent product quality, and greater technical credibility in the marketplace. This capability, when fully integrated into the company's quality management system and WPS qualification framework, serves as a foundational element of the company's competitive advantage in the weld overlay and cladding technology sector.