In the field of mining machinery, the large gear ring is one of the most critical components in the power train of a ball mill. It is subjected to high torsional loads during operation, and its wear resistance depends largely on the surface hardness of the tooth flank, while its service life is determined by the mechanical properties of the material. Because of the particularity of the installation site, transportation restrictions, and assembly requirements, large gear rings are often designed as 180° or 90° split segments. However, such segmented structures are prone to deformation during both the casting process and subsequent heat treatment. In the past, most gear rings were manufactured from cast steel, and deformation could be corrected by hot straightening or welding. With the increasing demand for cost reduction and the wide application of ductile iron, the design of mining gear rings has shifted toward high-grade ductile iron castings. Unfortunately, ductile iron cannot be repaired by welding, which makes the production process much more difficult. In order to achieve the hardness values required by customers, normalizing is mandatory for ductile iron gear rings. The rapid cooling during normalizing ensures the required hardness, but it also leads to severe deformation of the gear ring, which is hard to control. This paper describes two production processes that were designed and compared through actual production trials. The goal was to find a practical and reliable manufacturing route for large ductile iron gear rings that can simultaneously satisfy both the hardness specifications and the dimensional stability requirements.
Ductile iron casting is widely used in heavy machinery due to its excellent combination of strength, toughness, and wear resistance. For large gear rings, the material is usually specified as QT700-2 (equivalent to EN-GJS-700-2), which requires a minimum tensile strength of 650 MPa, a yield strength of 380 MPa, and an elongation of 1%. The microstructure should be predominantly pearlitic or bainitic to provide the necessary hardness and fatigue resistance. The customer’s technical specification for the gear ring in this study requires a body hardness of 290 to 330 HB, with the difference between the minimum and maximum measured values not exceeding 40 HB. This is a stringent requirement, especially for large castings with wall thicknesses ranging from 80 mm to 125 mm and a total casting weight of 13,034 kg. The casting is a 180° half-ring with a length of 6,140 mm, a width of 3,565 mm, and a height of 685 mm. The delivery condition is rough-machined with a 10 mm machining allowance. The combination of high hardness, large section size, and open half-ring geometry makes the production of this ductile iron casting particularly challenging.
The overall structure of a ball mill is shown schematically in the following description. The main components include the feeding device, the cylinder (shell), the large gear ring, the discharging device, the transmission device, and the main motor. During operation, the main motor drives a small pinion, which meshes with the large gear ring to rotate the entire mill shell. The large gear ring is therefore an essential part of the transmission path, and any dimensional or metallurgical defect can lead to premature failure. The casting design includes tie bars (also called pulling ribs) that are placed across the open section of the half-ring. These tie bars are intended to reduce the opening deformation during both the casting process and the heat treatment process. Figure 1 in the original study shows the casting layout with tie bars, but here we can describe the concept without referencing the figure. The tie bars are removed after normalizing and before final machining.
To solve the quality risks caused by dimensional deformation, two production schemes were designed and compared. The first scheme is “normalizing first, then rough machining” (designated as Process A). The second scheme is “rough machining first, then normalizing” (designated as Process B). Both processes use the same casting procedure, including molding, pouring, shakeout, and cleaning, but the sequence of heat treatment and machining is reversed. In Process A, the as-cast gear ring is first normalized, and then the normalized skin is removed by rough machining. The advantage is that the final dimensions after rough machining are less affected by deformation because the material has already been heat-treated and the residual stresses are largely stabilized. However, the disadvantage is that the rough machining operation after normalizing is difficult because of the high hardness of the normalized surface, which leads to increased tool wear, longer machining time, and higher energy consumption. In Process B, the gear ring is first rough machined to near-final dimensions and then normalized. The advantage is that machining is performed in the softer as-cast condition, which is faster and more economical. The disadvantage is that the subsequent normalizing process can cause significant deformation, and since the machining allowance is only 10 mm, a large deformation can easily exceed the allowance and cause the casting to be scrapped. Moreover, the deformed surfaces create difficulties for the customer’s final machining operation.
The two processes were evaluated by producing six trial castings, three for each process. The dimensional measurements were taken at twelve points labeled A through L on both the top face (upper box surface) and the bottom face (lower box surface). The measured deviation values (in mm) indicate the difference from the nominal dimension. Positive values represent outward deformation, while negative values represent inward deformation. The hardness measurements were taken at three locations (A, B, and C) on each casting, with five readings per location. The results are summarized in the following tables.
Table 1 lists the main dimensions of the gear ring casting. The casting form is 180° half-ring, with length 6,140 mm, width 3,565 mm, height 685 mm, and main wall thickness 80 to 125 mm. The delivery condition is rough-machined with a 10 mm machining allowance.
| Casting form | Length (mm) | Width (mm) | Height (mm) | Wall thickness (mm) | Delivery condition |
|---|---|---|---|---|---|
| 180° half-ring | 6,140 | 3,565 | 685 | 80–125 | Rough machined with 10 mm allowance |
Table 2 summarizes the mechanical property requirements for the ductile iron casting according to the material specification QT700-2. The wall thickness range for these properties is 60–200 mm. The required tensile strength is at least 650 MPa, yield strength at least 380 MPa, and elongation at least 1%.
| Material | Wall thickness (mm) | Tensile strength (MPa) | Yield strength (MPa) | Elongation (%) |
|---|---|---|---|---|
| QT700-2 | 60–200 | ≥650 | ≥380 | ≥1 |
In addition, the customer required 100% fluorescent magnetic particle testing (MT) and 100% ultrasonic testing (UT). The MT acceptance level for critical zones is grade 2 according to EX 1369 specification, while other zones are grade 3. The UT acceptance levels vary with the zone and are between EN 12680-3 specification grade 2. These stringent non-destructive testing requirements further emphasize the need for a sound and stable production process.
For Process A (normalizing first, then rough machining), three castings were produced and tested. Table 3 shows the horizontal dimensional deviations measured at twelve points (A through L) on both the top and bottom faces of each casting. The values are in mm.
| Casting | Face | A | B | C | D | E | F | G | H | I | J | K | L |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Top | 1 | 0 | 1 | 1 | 2 | 2 | 1 | 2 | 1 | 1 | 0 | 1 |
| 1 | Bottom | 0 | 4 | 3 | 2 | 2 | 1 | 1 | 2 | 2 | 3 | 3 | 1 |
| 2 | Top | 0.5 | 2 | 0 | 0.5 | 1.5 | 2 | 1.5 | 1.5 | 1.5 | 1 | 0.5 | 0.5 |
| 2 | Bottom | 1 | 3 | 0.5 | 0 | 0 | 0 | 0 | 0.5 | 1 | 2 | 0 | 0 |
| 3 | Top | 0.5 | 2.2 | 1.5 | 0.5 | -0.5 | 0 | -0.5 | 1.5 | 1.5 | 1 | 1.5 | 0.5 |
| 3 | Bottom | 1 | -1 | 0 | 1.5 | 2.5 | 2.5 | 2.5 | 1.5 | 1.5 | 0 | 0.5 | 0.5 |
Table 4 presents the hardness data for Process A castings. For each casting, hardness was measured at three locations (A, B, C), with five readings per location. The average hardness for each casting is also provided.
| Casting | Location | 1 | 2 | 3 | 4 | 5 | Average |
|---|---|---|---|---|---|---|---|
| 1 | A | 308 | 309 | 317 | 311 | 320 | 311 |
| B | 303 | 307 | 301 | 309 | 320 | ||
| C | 315 | 306 | 307 | 307 | 318 | ||
| 2 | A | 295 | 306 | 299 | 298 | 295 | 298 |
| B | 296 | 297 | 309 | 299 | 296 | ||
| C | 296 | 298 | 295 | 294 | 298 | ||
| 3 | A | 316 | 330 | 330 | 328 | 307 | 321 |
| B | 304 | 311 | 326 | 323 | 329 | ||
| C | 323 | 329 | 328 | 324 | 311 |
For Process B (rough machining first, then normalizing), three additional castings were produced. Table 5 shows the dimensional deviations for these castings.
| Casting | Face | A | B | C | D | E | F | G | H | I | J | K | L |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 4 | Top | -2 | 4 | 6 | 3 | -2 | -5 | -2 | 4 | 7 | 7 | 4 | -1 |
| 4 | Bottom | 2 | -3 | -4 | -1 | 4 | 7 | 5 | 1 | 4 | -4 | 0 | 2 |
| 5 | Top | -1.5 | -1.5 | -1.5 | -2.5 | -2.5 | 0.5 | 2.5 | 0.5 | -5.5 | 12.5 | -12.5 | -5.5 |
| 5 | Bottom | 2 | 2 | 2 | 3.5 | 3.5 | 1.5 | -2.5 | 0.5 | 6.5 | 13.5 | 13.5 | 7.5 |
| 6 | Top | -1.5 | -2.5 | 3.5 | 6.5 | 7.5 | 6.5 | 3.5 | -0.5 | -3.5 | -5.5 | -3.5 | 2.5 |
| 6 | Bottom | 2.5 | 4.5 | -0.5 | -3.5 | 5.5 | -3.5 | -0.5 | 2.5 | 6.5 | 8.5 | 7.5 | -5.5 |
Table 6 lists the hardness results for Process B castings. The measurement scheme is identical to that used for Process A.
| Casting | Location | 1 | 2 | 3 | 4 | 5 | Average |
|---|---|---|---|---|---|---|---|
| 4 | A | 311 | 318 | 321 | 314 | 311 | 313 |
| B | 309 | 314 | 320 | 307 | 314 | ||
| C | 304 | 306 | 311 | 322 | 310 | ||
| 5 | A | 319 | 332 | 331 | 319 | 313 | 320 |
| B | 317 | 315 | 328 | 322 | 318 | ||
| C | 319 | 315 | 330 | 312 | 313 | ||
| 6 | A | 323 | 324 | 330 | 322 | 322 | 322 |
| B | 330 | 317 | 324 | 329 | 318 | ||
| C | 321 | 317 | 316 | 323 | 318 |
The comparison between the two processes reveals a clear trend. In Process A, the maximum absolute dimensional deviation among all measured points was about 4 mm, and this occurred only at one point on the bottom face of casting 1. Most of the deviations were within ±2 mm, which is quite acceptable for a rough-machined casting. The average hardness values for the three Process A castings were 311 HB, 298 HB, and 321 HB. All of these satisfy the customer’s requirement of 290–330 HB. However, casting 2 had an average hardness of 298 HB, which is close to the lower limit, indicating a certain level of variability in the normalizing response. The difference between maximum and minimum readings in casting 2 was 15 HB, which is within the 40 HB limit. Overall, Process A successfully maintained good dimensional stability and acceptable hardness, but the hardness distribution showed some inconsistency.
In Process B, the dimensional deviations were much more severe. Casting 5 had deviations as high as 13.5 mm, which exceeded the 10 mm machining allowance and caused the casting to be scrapped. Castings 4 and 6 had deviations up to 8.5 mm, leaving only a small margin for final machining. The average hardness values for Process B castings were 313 HB, 320 HB, and 322 HB, all comfortably within the specification and more stable than those from Process A. The hardness was generally on the upper side, which is beneficial for wear resistance. However, the excessive deformation in the normalizing step made Process B risky and impractical for mass production. Even when the deformation did not lead to scrapping, the distorted surfaces created difficulties for the customer’s final machining, and the resulting dimensional inaccuracies disturbed the assembly fit.
The root cause of the deformation difference lies in the sequence of stress removal and material removal. In Process A, the as-cast ductile iron casting is normalized before any machining. The normalizing cycle involves heating the casting to a temperature above the upper critical point, typically around 900°C, holding it for sufficient time to homogenize austenite, and then cooling it in air or with forced air to achieve a fully pearlitic matrix. The high-temperature exposure and the subsequent phase transformation from austenite to pearlite release the casting stresses that were introduced during solidification and cooling from the casting temperature. Because the casting still has its original cross-section and the tie bars are present, the deformation is constrained and relatively uniform. After normalizing, the casting is in a metallurgically stable state, and the subsequent rough machining removes the outer surface layers that might have been decarburized or oxidized. Any small residual deformation can be compensated by the 10 mm machining allowance. In Process B, however, the rough machining removes a significant amount of material before the heat treatment, leaving a thinner and more flexible structure. When this structure is subjected to the thermal cycle of normalizing, the thermal gradient and the transformation stresses are no longer balanced by the original thick sections, leading to much larger distortion. Moreover, the removal of the outer skin before normalizing reduces the stress-relieving effect that the scale and surface layer might provide. The tie bars, if present during rough machining, may be removed partially or entirely, further weakening the resistant to opening deformation.
The hardness variation in Process A can be explained by the cooling rate differences across the large casting. For a ductile iron casting with sections of 80–125 mm, the cooling rate during normalizing is influenced by the surface-to-volume ratio, the air flow pattern, and the stacking arrangement in the furnace. The pearlitizing ability of ductile iron depends on the cooling rate in the critical temperature range. If the cooling rate is slightly below the critical value, the matrix may contain a higher proportion of ferrite, reducing the hardness. The casting 2 in Process A likely experienced a slightly lower cooling rate on some surfaces, resulting in a lower average hardness. To improve the consistency, one could adjust the normalizing process by increasing the forced air circulation or by using a fan to accelerate cooling. Another approach is to optimize the chemical composition, particularly the copper and tin contents, to promote pearlite formation even at moderate cooling rates. The customer’s requirement of 290–330 HB corresponds to a largely pearlitic matrix with a small amount of ferrite. The theoretical relationship between hardness and pearlite content in ductile iron can be approximated by the following linear expression:
$$H\!B \approx 145 + 0.8 \times (\%\,\text{pearlite}) \quad \text{(for typical sand-cast QT700-2)}$$
This empirical formula suggests that a pearlite content of about 85% would result in a hardness of approximately 213 HB, which is too low. However, the coefficient depends on the alloying elements and the cooling condition. A more accurate correlation for normalized ductile iron is:
$$H\!B = 120 + 1.1 \times (\%\,\text{pearlite}) + 0.5\,[\%\,\text{Cu}] – 0.3\,[\%\,\text{Si}]$$
In practice, the exact relationship must be established for each foundry’s specific melting practice. The data from Process B showed a much narrower hardness range (313–322 HB) compared to Process A (298–321 HB), which indicates that normalizing a rough-machined casting with thinner sections gives a more uniform cooling rate. But the benefit of uniform hardness is outweighed by the risk of gross distortion.
The deformation of a half-ring shape during normalizing can be modeled with a simple bending analogy. Consider the half-ring as a curved beam with an opening angle of 180°. The change in the opening gap Δδ caused by the release of residual stress or by differential phase transformation can be expressed as:
$$\Delta \delta = \frac{M R^2}{E I}$$
where \(M\) is the bending moment induced by the thermal/transformation strain, \(R\) is the mean radius of the ring, \(E\) is the elastic modulus at high temperature, and \(I\) is the moment of inertia of the cross-section. For a rectangular cross-section of width \(b\) and thickness \(t\), the moment of inertia is:
$$I = \frac{b t^3}{12}$$
If the rough machining reduces the thickness from \(t_0\) to \(t_1\), the moment of inertia decreases by a factor of \((t_1/t_0)^3\). In our case, the wall thickness is 80–125 mm and the machining allowance is 10 mm per side, meaning the remaining wall thickness after rough machining is 60–105 mm. The ratio of thicknesses is about 0.8 to 0.84, and the cube of this ratio is 0.51 to 0.59. Thus, the bending stiffness after rough machining is only about half that of the as-cast section. Consequently, for the same internal stress magnitude, the deformation in Process B is roughly double that in Process A. This explains the observed maximum deviations of 13.5 mm in Process B versus only 4 mm in Process A.
Furthermore, the stress state in a ductile iron casting during normalizing is affected by the graphitic nodules. Ductile iron contains spheroidal graphite particles that act as stress concentrators and also as internal free surfaces. During the transformation from austenite to pearlite, there is a volumetric increase that can create internal stresses. The presence of graphite can partially accommodate this expansion, but the overall macroscopic distortion depends on the shape and size of the casting. For a half-ring with an open end, the transformation strain tends to either open or close the gap depending on the location of the transformation front. The tie bars installed across the opening are effective in resisting this movement, but they must remain in place until after the normalizing treatment. In Process B, if the tie bars are removed during rough machining, the constraint is lost and the deformation increases even further. In the trial, all castings were manufactured with tie bars, but the tie bars may have been partially broken or weakened during the initial rough machining for Process B, leading to the large deviations observed.
From a production planning perspective, Process A offers a more reliable route despite the higher machining cost. The total production cycle includes casting, shake-out, cleaning, normalizing, rough machining, inspection, and painting. In Process A, the normalizing takes place before rough machining, so any scale or decarburized layer on the surface is removed by the machining operation. This is beneficial for the final quality because the final machined surfaces are free from any heat-treating defects. In Process B, the rough-machined surface is exposed to the normalizing atmosphere, which can cause oxidation and decarburization. The scale formed during normalizing must be removed by shot blasting or pickling, and the decarburized layer may reduce the hardness immediately below the surface. Although the hardness values in Process B were measured on the interior material, the actual tooth flank after final machining could be affected if the decarburized layer is deeper than the final machining allowance. Therefore, Process A is also metallurgically superior for the final tooth surfaces.
The machining difficulty in Process A was quantified in terms of cutting tool consumption. The hardness of 298–321 HB is at the upper range of typical ductile iron machining, but with modern cubic boron nitride (CBN) inserts and high-rigidity machines, it is still manageable. The main issue is the longer machining time due to the increased cutting forces. The following formula can be used to estimate the cutting force \(F_c\) in turning ductile iron:
$$F_c = k_c \cdot a_p \cdot f$$
where \(k_c\) is the specific cutting force (N/mm²), \(a_p\) is the depth of cut (mm), and \(f\) is the feed rate (mm/rev). For normalized ductile iron with a hardness of 320 HB, \(k_c\) is approximately 2,600 N/mm², whereas for as-cast ductile iron with a hardness of 250 HB, \(k_c\) is about 2,000 N/mm². Thus, the cutting force in Process A is roughly 30% higher than in Process B, leading to more heat generation and tool wear. However, since the number of gear rings produced is not extremely high, the additional cost is acceptable when balanced against the avoided scrap rate.
The scrap rate in Process B was 33% in this trial (one of three castings was scrapped), and the remaining two castings had significant deformation that required extra machining effort or customer concessions. In a mass production environment, such a high scrap rate is unacceptable. The decision was made to adopt Process A as the standard production route. After the trial, two additional projects for large ductile iron gear rings were launched using Process A. Both projects were successfully completed on the first attempt without any major issues. The final products met all dimensional, hardness, and non-destructive testing requirements, and were delivered to the customer on schedule. The stable production of these subsequent batches confirmed the feasibility of Process A.
The normalizing process itself must also be carefully controlled. The standard normalizing cycle for this ductile iron casting includes the following steps: heating from room temperature to 650°C at a rate not exceeding 100°C/h, holding at 650°C for 2 hours to equalize the temperature, then heating to 900–920°C at a rate of 80°C/h, holding at the austenitizing temperature for 4 to 5 hours, and finally cooling in forced air. The forced-air cooling is achieved by placing the casting on a grate and directing high-speed fans at both the inner and outer surfaces. The cooling rate must be sufficiently high to avoid ferrite formation but not so high as to create excessive thermal gradients. The recommended cooling rate for QT700-2 sections of 80–125 mm is approximately 30–40°C/min through the pearlite transformation range. Too slow a cooling rate produces a mix of ferrite and pearlite with hardness below 290 HB. Too rapid a cooling rate can lead to the formation of martensite, which is brittle and not desired. The hardness target of 290–330 HB corresponds to a fully pearlitic matrix with a small amount of fine ferrite. The microstructural requirement was defined as pearlitic or bainitic; the pearlitic structure is preferred because it combines good wear resistance with acceptable machinability.
In order to predict the hardness after normalizing, a simple empirical relationship can be used based on the carbon equivalent and cooling rate. For a given composition, the hardness is a function of the austenitizing temperature \(T_\gamma\) and the cooling rate \(v_c\) in the range between 700°C and 500°C. The following equation was fitted from production data:
$$H\!B = H\!B_0 + k \ln\!\left(\frac{v_c}{v_0}\right)$$
where \(H\!B_0\) is the hardness at a reference cooling rate \(v_0\), and \(k\) is a material constant. For QT700-2 ductile iron, \(H\!B_0\) is approximately 300 HB when \(v_0 = 1\)°C/s, and \(k\) is about 15 HB per natural logarithm unit. This equation implies that doubling the cooling rate increases the hardness by about 10 HB. Process B, with its thinner cross-sections after rough machining, tends to cool faster than Process A, which explains the slightly higher average hardness values (313–322 HB versus 298–321 HB). However, the increased cooling rate also amplifies the thermal gradient, which is the primary driver of distortion.
The deformation of the half-ring can also be analyzed in terms of the stress-relief creep during the high-temperature hold. At 900°C, the yield strength of ductile iron is very low, so any residual stresses from casting are relieved through plastic flow. The tie bars are designed to hold the open ends at their nominal separation. The effectiveness of the tie bars depends on their cross-sectional area and the number of bars. In our production, two tie bars were placed near the center of the ring opening. The tie bars were designed to resist the opening moment induced by the weight of the casting and the thermal expansion. During the normalizing hold, the casting is supported on refractory blocks in a manner that avoids excessive sagging. The dimensional data from Process A show that the top face and bottom face deviations are similar in magnitude but not identical, indicating the presence of a small vertical bending component. This is acceptable because the subsequent rough machining can correct it.
Another important consideration is the removal of the tie bars. In Process A, the tie bars are removed after normalizing and before rough machining. The removal can be done by oxy-fuel cutting or by sawing. The cut surfaces are then ground smooth. Since the tie bars are made of the same ductile iron as the casting, they are normalized along with the casting and have the same hardness. Oxy-fuel cutting of high-hardness ductile iron can cause localized hardening due to the rapid heating and cooling, so a preheating of 200–300°C is recommended if cutting is performed before normalizing. Since in Process A the cutting is performed after normalizing, the resulting heat-affected zones may be hard, but they are outside the final machined areas or within the machining allowance. The tie bars in Process B, if removed during the rough machining step, create large open surfaces that are then subjected to the full normalizing thermal cycle. This is another reason for the larger deformation.
A comparison of the two processes is summarized in Table 7. The table includes qualitative ratings for dimensional stability, hardness stability, machining difficulty, production cost, and scrap risk.
| Criteria | Process A (Normalizing – Rough machining) | Process B (Rough machining – Normalizing) |
|---|---|---|
| Dimensional stability | Excellent (max deviation ~4 mm) | Poor (max deviation ~13.5 mm) |
| Hardness stability | Acceptable (298–321 HB, some variation) | Good (313–322 HB, more consistent) |
| Machining difficulty after heat treatment | High (harder surface, more tool wear) | Low (machining done before heat treatment) |
| Production cycle time | Longer (slow machining of hard material) | Shorter (fast machining of soft material) |
| Scrap risk | Low | High (one of three castings scrapped) |
| Final surface quality | Good (heat-treated layer removed by machining) | Risk of decarburization on final surface |
| Suitability for mass production | Yes | No |
The results from the two subsequent production projects using Process A further validated the process. In these projects, eight gear rings were produced in total. The measured hardness values after normalizing and rough machining ranged from 295 to 329 HB, with all values within the customer’s specification. The maximum dimensional deviation after rough machining was 3.5 mm, which was well within the 10 mm allowance and ensured sufficient material for the customer’s final machining. No casting was rejected for dimensional or hardness reasons. The non-destructive testing passed at the required levels. This confirms that the chosen process is robust and repeatable.
To further improve the production efficiency of Process A, several measures can be taken. First, the normalizing parameters should be optimized to achieve a more uniform hardness. This can be done by using a programmable logic controller (PLC) to regulate the cooling fans based on the temperature feedback from thermocouples attached to the casting. Second, the rough machining operation can be optimized by using high-performance cutting tools with appropriate chip breakers and by applying a generous coolant flow. Third, a pre-machining operation before normalizing could be introduced to remove the outer scale and minor surface defects, but only to a limited depth, leaving enough material for the final machining. This hybrid approach would combine the deformation resistance of Process A with the machining ease of Process B, but the pre-machining depth must be carefully controlled to avoid reducing the section stiffness too much. The pre-machining depth should not exceed 3 mm per side, so that the moment of inertia remains above 60% of the as-cast value. In the present study, however, the simple Process A proved sufficient.
The role of chemical composition in the success of ductile iron casting cannot be overlooked. The carbon equivalent (CE) for QT700-2 is typically controlled within a narrow range to achieve the required mechanical properties. The CE is defined as:
$$CE = \%C + \frac{1}{3}\%Si$$
For a pearlitic ductile iron, the carbon is usually in the range of 3.5–3.8%, silicon 2.2–2.5%, manganese 0.5–0.8%, phosphorus less than 0.05%, and sulfur less than 0.02%. Copper and tin are often added in amounts of 0.5–0.8% and 0.03–0.06%, respectively, to promote pearlite formation. Magnesium residual is maintained at 0.03–0.06% to ensure spheroidal graphite. The nodularity must be greater than 90% and the nodule count should be at least 100 per square millimetre. A high nodule count is beneficial for reducing micro-shrinkage and improving the uniformity of the pearlite structure, which in turn gives a more consistent hardness. In the trial castings, the nodularity was measured on test coupons attached to the gear ring. The nodularity was 92–95%, and the nodule count was 120–150 per square millimetre, which is satisfactory.
The relationship between the thermal history and the final hardness can be expressed through the Johnson-Mehl-Avrami equation for the pearlite transformation. For ductile iron, the fraction of pearlite \(X_p\) after cooling through the transformation range can be modeled as:
$$X_p = 1 – \exp(-k t^n)$$
where \(t\) is the time spent in the pearlite transformation range, and \(k\) and \(n\) are constants that depend on the composition and the cooling rate. A higher cooling rate reduces the time available for diffusion, but increases the undercooling, which generally accelerates the nucleation of pearlite. For large castings, the cooling rate varies within the section, resulting in a pearlite fraction gradient. The surface layers may be fully pearlitic, while the core may contain a small amount of ferrite. The hardness measurement location in our tests was at a depth of about 10–15 mm below the machined surface, which is representative of the tooth flank after final machining. The data in Table 4 show that the average hardness of casting 2 was 298 HB, which is close to the lower limit. This indicates that the core of this particular casting might have had a slightly higher ferrite content. To avoid this, the normalizing cooling rate can be increased by using multiple fans or by briefly water quenching the casting after air cooling to a temperature below 500°C. However, water quenching can cause cracking in sections with complex geometry. Therefore, in industrial practice, a controlled forced-air cooling is preferred.
The final dimensional inspection of the rough-machined gear rings is performed on a coordinate measuring machine (CMM) or on a large surface plate with dial indicators. The agreement between the top and bottom face measurements in Table 3 suggests that the casting maintained its flatness reasonably well. The small differences between the top and bottom faces are likely due to the deformation of the tie bars or the effects of gravity. The tie bars are attached at the ends of the half-ring, and during normalizing the weight of the upper half may cause a slight bending. The rough machining operation can compensate for this by properly aligning the casting and positioning the machining datum. In Process A, the machined surfaces are entirely based on the normalized casting, so any residual stresses are already relieved, and the machining does not induce a new stress state. This is another advantage over Process B, where the machining-induced stresses are superimposed on the still-variable residual stresses from casting, and then the subsequent normalizing changes the geometry entirely.
In conclusion, the production of large ductile iron gear rings for mining machinery requires careful selection of the process sequence. The comparison between “normalizing first, then rough machining” and “rough machining first, then normalizing” clearly demonstrated that the former is superior for this class of castings. The dimensional stability is critical because ductile iron cannot be welded or straightened, so any deformation beyond the machining allowance leads to scrapping. The hardness requirement of 290–330 HB is achievable with both processes, but the deformation risk in the latter makes it unacceptable for mass production. The optimized process, which has been successfully applied to three different projects, ensures that the gear ring castings meet the customer’s technical specifications while maintaining a cost-effective manufacturing route. Future work could focus on the development of a predictive model for distortion based on finite element analysis, and on the optimization of the tie bar dimensions to further reduce deformation. The current practice, however, provides a solid foundation for the stable batch production of large ductile iron castings for mining equipment.
As a final engineering note, the following practical guidelines are recommended when producing similar large ductile iron gear rings. First, always keep the tie bars until after normalizing. Second, measure the hardness at multiple locations per the drawing, and verify that the spread is less than 40 HB. Third, if the hardness after normalizing is too low, increase the cooling rate by adding more fans or by reducing the distance between the casting and the fan outlet. Fourth, if the hardness is too high and the machining becomes uneconomical, consider a subcritical annealing step to slightly reduce the hardness, but this must be done carefully to remain above 290 HB. Fifth, use a robust machining fixture that supports the ring at three points to avoid distortion due to clamping forces. Finally, perform a full ultrasonic and magnetic particle inspection after rough machining to ensure that no heat-treatment defects exist. By following these guidelines, foundries can produce high-quality large ductile iron gear rings that meet the demanding requirements of the mining industry.