The presence of residual stress within a gray iron casting is an inevitable consequence of the complex thermal history and phase transformations experienced during solidification and cooling. These locked-in stresses, existing in a state of equilibrium at room temperature, are primary culprits behind dimensional instability, distortion, and in severe cases, catastrophic cracking of cast components. While post-casting treatments like natural aging, low-temperature annealing, or vibration stress relief can mitigate these stresses, the most cost-effective and production-efficient strategy is to minimize their generation during the casting process itself. This study, therefore, employs an integrated approach combining Computer-Aided Engineering (CAE) simulation and Design of Experiments (DOE) methodology to systematically investigate and rank the influence of key process parameters on the residual stress development in a standard gray iron casting. The objective is to move beyond traditional trial-and-error methods and provide a data-driven foundation for process optimization in gray iron foundries.
Residual stress in a gray iron casting originates from three fundamental sources: thermal stress, phase transformation stress, and mechanical constraint stress. Thermal stress arises due to non-uniform cooling, where different sections of the casting contract at different rates. Phase transformation stress, particularly significant in ferrous alloys, results from volumetric changes associated with metallurgical transformations, such as the austenite-to-pearlite/ferrite transformation in gray iron. Mechanical constraint stress develops from the resistance offered by the mold and cores to the natural shrinkage of the casting. The final residual stress state is a complex superposition of these contributions. For a hypoeutectic gray iron casting, the sequence involves the precipitation of primary austenite dendrites followed by the eutectic solidification of austenite-graphite. Upon further cooling, the austenite decomposes. The temperature ranges and magnitudes of these transformations are highly sensitive to composition and cooling rates, directly influencing the resultant stress.
The core of this investigation utilizes a CAE-based simulation approach to predict residual stresses. The governing equations for a coupled thermo-mechanical-metallurgical analysis involve solving the heat transfer equation for temperature field evolution and the equilibrium equations for stress-strain development, considering temperature-dependent material properties and phase transformation kinetics. The heat conduction is governed by:
$$ \rho c_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + Q_{latent} $$
where $\rho$ is density, $c_p$ is specific heat, $T$ is temperature, $t$ is time, $k$ is thermal conductivity, and $Q_{latent}$ is the latent heat source term from solidification and phase changes. The mechanical response is often modeled using a thermo-elasto-plastic constitutive model. The total strain increment $d\varepsilon$ is decomposed into elastic $d\varepsilon^e$, plastic $d\varepsilon^p$, thermal $d\varepsilon^{th}$, and transformation-induced $d\varepsilon^{tr}$ components:
$$ d\varepsilon = d\varepsilon^e + d\varepsilon^p + d\varepsilon^{th} + d\varepsilon^{tr} $$
The transformation strain is related to the volumetric change $\Delta V/V$ associated with a given phase transformation. For the decomposition of austenite ($\gamma$) to a mixture like pearlite (P), it can be expressed as:
$$ \varepsilon^{tr} = \frac{1}{3} \ln\left(1 + \frac{\Delta V}{V}\right) f_P $$ where $f_P$ is the volume fraction of pearlite formed. The von Mises stress ($\sigma_{vm}$), a scalar measure of the distortional energy, is used as the key indicator for residual stress and potential yielding. It is calculated from the principal stresses ($\sigma_1, \sigma_2, \sigma_3$) as:
$$ \sigma_{vm} = \sqrt{\frac{(\sigma_1 – \sigma_2)^2 + (\sigma_2 – \sigma_3)^2 + (\sigma_3 – \sigma_1)^2}{2}} $$
A high value of maximum $\sigma_{vm}$ in the casting pinpoints the most critical location for stress concentration and potential failure.
To efficiently evaluate the effect of multiple process parameters, a Design of Experiments (DOE) methodology was adopted. This statistical approach allows for the simultaneous variation of multiple factors and the quantification of their main effects and interactions with a minimal number of simulation runs. The standard “stress frame” or “stress bar” geometry was selected as the test casting. This geometry, comprising sections of different thicknesses (a central thick section and two outer thinner sections connected by crossbars), is designed to inherently generate thermal gradients and thus residual stresses, making it an ideal benchmark. The geometry and dimensions of the stress frame used in this study are provided below.
| Section | Dimensions (mm) | Description |
|---|---|---|
| Central Thick Beam | 54 x 54 (cross-section) | Slowest cooling section |
| Outer Thin Beams | 34 x 34 (cross-section) | Faster cooling sections |
| Crossbars | 16 x 20 (cross-section) | Connecting elements |
| Overall Length | 375 mm | – |
Four key factors were identified for investigation, each at three levels, as shown in the following table. The material was a standard HT250 grade gray iron. Adjusting the Si/C ratio within the software automatically updates critical material properties like eutectic temperature, elastic modulus, and thermal expansion coefficients.
| Factor | Level 1 | Level 2 | Level 3 |
|---|---|---|---|
| A: Pouring Temperature (°C) | 1380 | 1400 | 1420 |
| B: Si/C Ratio | 0.567 (e.g., 1.9%/3.35%) | 0.606 (e.g., 2.0%/3.3%) | 0.646 (e.g., 2.1%/3.25%) |
| C: Shake-out Time (s) | 6000 | 8000 | 10000 |
| D: Cooling Condition (Chill Size: mm³) | 50x55x50 | 80x55x50 | 100x55x50 |
A Taguchi L9(3^4) orthogonal array was selected to design the simulation matrix. This array requires only 9 simulation runs to study the four 3-level factors, efficiently covering the design space. The assignment of factors to the orthogonal array columns and the resulting simulation matrix are presented below.
| Run No. | A: Pour Temp. | B: Si/C | C: Shake-out | D: Chill Size |
|---|---|---|---|---|
| 1 | 1380 (L1) | 0.567 (L1) | 6000 (L1) | 50x55x50 (L1) |
| 2 | 1380 (L1) | 0.606 (L2) | 8000 (L2) | 80x55x50 (L2) |
| 3 | 1380 (L1) | 0.646 (L3) | 10000 (L3) | 100x55x50 (L3) |
| 4 | 1400 (L2) | 0.567 (L1) | 8000 (L2) | 100x55x50 (L3) |
| 5 | 1400 (L2) | 0.606 (L2) | 10000 (L3) | 50x55x50 (L1) |
| 6 | 1400 (L2) | 0.646 (L3) | 6000 (L1) | 80x55x50 (L2) |
| 7 | 1420 (L3) | 0.567 (L1) | 10000 (L3) | 80x55x50 (L2) |
| 8 | 1420 (L3) | 0.606 (L2) | 6000 (L1) | 100x55x50 (L3) |
| 9 | 1420 (L3) | 0.646 (L3) | 8000 (L2) | 50x55x50 (L1) |
For each of these 9 process configurations, a fully coupled thermal-stress simulation was performed using a commercial CAE software capable of modeling the solidification, cooling, and stress development in gray iron casting, including the effects of phase transformations. The output of primary interest was the maximum von Mises stress ($\sigma_{vm}^{max}$) found anywhere in the stress frame after complete cooling to room temperature. The results for all nine simulation runs are compiled in the table below.
| Run No. | Max. Von Mises Stress, $\sigma_{vm}^{max}$ (MPa) |
|---|---|
| 1 | 276.1 |
| 2 | 265.2 |
| 3 | 255.6 |
| 4 | 255.0 |
| 5 | 275.4 |
| 6 | 261.0 |
| 7 | 268.7 |
| 8 | 259.0 |
| 9 | 279.2 |
The DOE analysis proceeds by calculating the average response (mean $\sigma_{vm}^{max}$) for each factor at each of its three levels. For example, the average stress for Pouring Temperature (Factor A) at Level 1 (1380°C) is the mean of the results from runs 1, 2, and 3. This calculation is repeated for all factors and levels. The main effect of a factor is then determined by the range (R) between its highest and lowest average response. A larger range indicates a stronger influence of that factor on the residual stress in the gray iron casting. The calculated mean responses and ranges are summarized in the following table.
| Factor | Mean $\sigma_{vm}^{max}$ at Level 1 | Mean $\sigma_{vm}^{max}$ at Level 2 | Mean $\sigma_{vm}^{max}$ at Level 3 | Range (R) (MPa) | Rank |
|---|---|---|---|---|---|
| A: Pouring Temp. | 265.63 | 263.80 | 268.96 | 5.16 | 2 |
| B: Si/C Ratio | 266.60 | 266.53 | 265.26 | 1.34 | 3 |
| C: Shake-out Time | 265.36 | 266.46 | 266.56 | 1.20 | 4 |
| D: Chill Size | 276.90 | 264.96 | 256.53 | 20.37 | 1 |
The analysis yields clear and significant insights. The most striking result is the overwhelming influence of Cooling Condition (Factor D, Chill Size) on the residual stress in this gray iron casting. With a range of 20.37 MPa, its effect is nearly four times greater than that of the next most influential factor. This underscores the paramount importance of controlling cooling gradients, often managed through chills, padding, or mold material selection, in determining the final stress state of a casting. The effect is strongly monotonic: larger chills (increased cooling intensity on the thick section) lead to significantly lower residual stresses. This can be explained by a more balanced cooling rate between the thick and thin sections, reducing the thermal gradient and consequently the thermal stress.
Pouring Temperature (Factor A) is the second most influential parameter. Interestingly, its effect is non-monotonic. The lowest average stress (263.80 MPa) occurs at the intermediate level of 1400°C, with both higher and lower temperatures resulting in increased stress. This nonlinear behavior can be attributed to the complex interplay between superheat, solidification morphology, and phase transformation kinetics. Higher superheat can lead to coarser microstructures (larger eutectic cells, wider dendrite arm spacing), which may affect both the thermal contraction behavior and the transformation characteristics. The latent heat release profile also changes with pouring temperature, altering the thermal history. The existence of an optimal pouring temperature for minimizing stress is a critical finding for process engineers working with gray iron casting.
The effects of Si/C Ratio (Factor B) and Shake-out Time (Factor C) are considerably smaller and of comparable magnitude (Ranges of 1.34 MPa and 1.20 MPa, respectively). For the Si/C ratio, a higher value (0.646) tended to produce slightly lower stresses. A higher Si/C ratio promotes graphitization and ferrite formation, which could lead to different transformation strains compared to a lower ratio that might favor pearlite. However, within the typical compositional range studied, its effect on the overall stress magnitude for this geometry was secondary to thermal management factors.
The result for Shake-out Time challenges conventional wisdom. The data indicates a slight increase in average residual stress with longer shake-out times (from 265.36 MPa at 6000s to 266.56 MPa at 10000s). The traditional belief is that allowing the gray iron casting to cool slowly in the mold acts as an in-situ stress relief anneal, reducing stresses. However, for this relatively small, restrained stress frame, the mechanism may be different. During prolonged cooling in the mold, the casting is mechanically constrained by the sand. As it cools through the elastoplastic temperature range, it can accumulate significant plastic strain due to this constraint. When the mold is finally removed (shaken out), the casting experiences a sudden change in boundary condition. The elastic portion of the strain from the mold period may recover, but the accumulated plastic deformation can now manifest differently, potentially leading to a new, and sometimes higher, state of elastic stress upon final cooling to room temperature. This highlights that the relationship between shake-out time and final residual stress is not universal; it depends heavily on the casting geometry, the level of mold restraint, and the cooling path. For some gray iron castings, an earlier shake-out followed by controlled free-air cooling might be more beneficial.
While the main effects provide the primary ranking, potential interactions between factors can also be explored through the DOE data. For instance, the combined effect of Pouring Temperature and Chill Size might be significant. A simple interaction plot can be conceptualized by grouping the data. Let’s consider the average stress for different chill sizes at low (1380°C) and high (1420°C) pouring temperatures from the available runs. Although a full factorial analysis would be needed for definitive conclusions, the strong main effect of chill size likely dominates most interactions in this study on gray iron casting.
The practical implications of this study are substantial for foundries producing gray iron castings, especially those requiring high dimensional stability. The key recommendation is to prioritize the management of cooling conditions above all other parameters studied. Implementing judicious use of chills to balance section cooling is the most powerful lever to reduce residual stress. Secondly, pouring temperature should not be viewed simplistically; an intermediate optimal value likely exists and should be determined for specific casting geometries, potentially through similar CAE-DOE studies. Thirdly, the shake-out practice should be critically evaluated. The blanket advice of “longer shake-out for lower stress” is not always valid. For geometries prone to high mold restraint, a shorter shake-out time might be advantageous, though this must be balanced against the risk of hot tearing or distortion during free-air cooling.
This investigation demonstrates the potent synergy between CAE simulation and DOE statistical design. The CAE tool provides a virtual laboratory to obtain precise, repeatable “measurements” of residual stress—a parameter extremely difficult to measure comprehensively in physical castings. The DOE framework organizes the simulations efficiently, extracts quantitative effect magnitudes, and reveals non-intuitive relationships. This integrated approach enables rapid, cost-effective exploration of the process parameter space for gray iron casting, leading to optimized processes that enhance quality, reduce scrap, and minimize the need for corrective stress-relief operations. Future work could expand this methodology to include other factors like carbon equivalent (CE), alloying elements (e.g., Cu, Sn), mold material properties, and more complex, production-relevant casting geometries to build a comprehensive knowledge base for residual stress control in gray iron foundries.