Over the years, I have worked directly with one of the few pump foundries that specialize in producing large dredge pump impellers. These impellers are the heart of a dredger, and they have to survive extremely abrasive slurry, high impact loads, and severe cavitation. In my daily work, I have come to understand that although sand casting is one of the most flexible and economical ways to produce such large impellers, it is also one of the richest sources of quality problems. The phrase sand casting defects covers many different failure modes: shrinkage cavities, micro-porosity, gas holes, sand inclusions, slag inclusions, cold shuts, hot tears, dimensional deviations, and even hidden internal cracks that are only discovered during machining or dynamic balancing.
This article is a first-person account of the sand casting defects that I have observed in large dredge pump impellers, the metallurgical and moulding mechanisms behind them, and the practical countermeasures that we implemented. Just a few years ago, our first-pass qualification rate was far from acceptable. After analyzing every defect systematically, we changed riser design, moulding practice, gating practice, pouring practice, shakeout practice, and inspection practice. The results were dramatic. Today, more than ninety percent of our large dredge pump impeller castings pass the complete quality inspection system. This is not an abstract achievement. It means lower energy consumption, fewer rework cycles, shorter delivery times, and much greater confidence in the pump’s operational life.
1. The Nature of Large Dredge Pump Impeller Castings
Before describing individual sand casting defects, I want to explain why these defects are so frequent in large impellers. A dredge pump impeller is not a simple casting. It has a thick central hub, a long shaft head, a front cover plate, a back cover plate, and several curved blades that connect the two cover plates. The flow passages are narrow, curved, and often highly twisted. The outside diameter can be large, sometimes exceeding one meter. The wall thickness can change very abruptly from the thin blade tip to the very thick shaft head. These abrupt geometrical changes create large hot spots. The hot spots, in turn, favor shrinkage porosity and gas porosity.
Another important fact is that the casting must resist severe wear. Therefore, the iron composition is often alloyed for hardness, wear resistance, and corrosion resistance. Such alloyed iron has a relatively narrow solidification range and a strong tendency to form brittle phases if the cooling rate is not controlled. The mould material, core material, coating, pouring temperature, and shakeout time all influence the final microstructure. If any one of them is wrong, one or more sand casting defects appear.
The sand mould itself is also a major source of defects. Large cores are assembled to form the flow passages. Each core has a complex shape. In the worst cases, the core assembly leaves gaps, loose sand, sharp sand edges, or low compactness zones. When molten iron enters the mould, these weak zones can be eroded and washed into the casting. This produces sand inclusions and slag inclusions. In my experience, most sand casting defects are not caused by a single mistake. They are caused by the interaction between part geometry, mould design, sand properties, and pouring conditions.
| Defect category | Typical location in impeller | Main cause family |
|---|---|---|
| Shrinkage cavity and porosity | Shaft head riser, cover plate riser neck, thick-to-thin transitions | Insufficient feeding, wrong riser modulus, premature solidification of riser neck |
| Sand sticking, blowholes, gas porosity | Fillet roots between blades and cover plates | Sharp corners, high sand temperature, excessive resin, poor coating |
| Sand inclusions and slag inclusions | Machined cover plate face, shaft head end face | Loose sand in core assembly, erosion of mould surface, poor runner cleanliness |
| Cold shuts | Outlet edge, outside edge of front cover plate | Long pouring time, low molten metal temperature, interrupted pouring |
| Hot tears and cracks | Outer edge of cover plate, thin sections near hot spots | Impurities, early shakeout, excessive restraint during cooling |
| Dimensional and balance deviations | Flow passage width, blade thickness, cover plate thickness | Core box deformation, pattern wear, inconsistent core placement |
In the sections that follow, I will discuss each of these sand casting defects in the order in which we encountered them. For each defect, I will present the root-cause analysis, a practical countermeasure, and the technical reasoning behind the countermeasure.
2. Shrinkage Cavity and Porosity at the Shaft Head Riser and Cover Plate Riser Neck
The first major group of sand casting defects that troubled us was shrinkage cavities and shrinkage porosity at the shaft head riser and at the riser neck above the cover plates. In the early stage of production, ultrasonic testing and machining revealed cavities near the shaft head axis. The cavities were sometimes open to the surface, and sometimes internal. We also found fine shrinkage porosity at the riser neck region of the cover plate. The porosity was not always visible on the raw casting, but it appeared as soon as we started machining.
Let me explain the technical reason. The shaft head has a wall thickness of roughly twenty centimeters. This is a very large hot spot. In solidification theory, the solidification time of a casting section is controlled by its modulus, which is the ratio of volume to cooling surface area:
$$
M = \frac{V}{A_{\mathrm{cool}}}
$$
where \(V\) is the volume of the section and \(A_{\mathrm{cool}}\) is the surface area through which heat is lost. According to Chvorinov’s rule, the complete solidification time is proportional to the square of the modulus:
$$
t_{\mathrm{solidification}} = K \left( \frac{V}{A_{\mathrm{cool}}} \right)^2
$$
The thick shaft head has a large modulus, so it remains liquid for a long time. The thinner blade sections freeze earlier. If the riser does not have a modulus greater than the shaft head modulus, the riser itself will freeze before it can feed the hot spot. For a sound casting, the riser modulus must satisfy the following condition:
$$
M_{\mathrm{riser}} \ge 1.2\,M_{\mathrm{hot\,spot}}
$$
In our original production practice, the riser diameter and height were chosen without enough cooling-surface calculation. The riser was effectively a conventional sand riser, and its feeding capacity was far too low. The molten iron in the riser lost heat through the sand walls, and the feeding efficiency was poor. The result was shrinkage cavities in the shaft head.
Countermeasure 1: Exothermic Riser and Controlled Riser Height
The first solution that we adopted was to replace the ordinary sand riser with an exothermic riser sleeve. Exothermic materials generate heat when they react with the molten iron. This heat delays solidification of the riser metal. In effect, the exothermic riser increases the feeding time and feeding distance. The thermal insulation is especially valuable when the hot spot is thick and heavy, because the amount of liquid metal required to fill the shrinkage is large.
We also standardized the minimum riser height. For the 700WN class impeller shaft head riser, we require a minimum height of 50 centimeters. For the larger 800WN class impeller shaft head riser, we require a minimum height of 60 centimeters. After pouring, we wait five to ten minutes, then top up each riser with additional hot metal. We also apply an exothermic covering agent on the top of the riser immediately. This reduces the temperature gradient at the top surface of the riser and makes feeding more efficient.
For the riser neck, the problem was different. The riser neck is the passage through which liquid metal flows from the riser into the casting. If the riser neck freezes too early, it chokes off the feeding path. We observed that the moulding sand around the riser neck was strongly heated during pouring. The binder, especially resin, decomposed and generated gas. This gas created a phenomenon called “back pressure” or “choking fire” at the neck. As a result, the neck region contained both gas pores and shrinkage cavities.
To solve this, we eliminated the sand riser seat and used a preformed ceramic riser seat. Ceramic material has much higher refractoriness than silica sand. It does not soften, expand, or collapse at the high temperatures involved. This greatly reduces the formation of gas and sand erosion at the riser neck. The ceramic seat also provides a clean and stable passage for feeding. Combined with the exothermic riser, the ceramic riser seat makes the riser neck freeze later than the hot spot.
| Parameter or practice | Before change | After change |
|---|---|---|
| Riser type at shaft head | Ordinary sand riser | Exothermic riser sleeve |
| Riser height, 700WN class | Not standardized, often too low | At least 50 cm |
| Riser height, 800WN class | Not standardized, often too low | At least 60 cm |
| Riser top management | No top-up, no covering agent | Top-up after 5–10 min, exothermic covering agent |
| Riser neck seat | Sand mould riser seat | Preformed ceramic riser seat |
The solidification of a riser can be roughly evaluated by the shrinkage volume that it must supply:
$$
V_{\mathrm{feed}} \geq \beta_{\mathrm{shrinkage}}\, V_{\mathrm{hot\,spot}}
$$
where \(V_{\mathrm{feed}}\) is the volume of liquid metal that the riser must provide, \(\beta_{\mathrm{shrinkage}}\) is the solidification shrinkage coefficient of the iron, and \(V_{\mathrm{hot\,spot}}\) is the volume of the hot spot being fed. For cast iron, \(\beta_{\mathrm{shrinkage}}\) is often between \(1\%\) and \(4\%\), depending on carbon equivalent and inoculation practice. If the riser sleeves are damaged during mould closing, or if the ceramic seat is cracked, the effective feeding volume drops sharply. Therefore, we inspect every riser sleeve and every ceramic seat before assembly.
After these changes, the internal quality of the shaft head improved dramatically. Cross-sections showed dense metal without central cavities. We still use ultrasonic testing to verify, but the number of defects detected has fallen by more than ninety percent compared with the previous practice.
3. Sand Sticking, Blowholes, Shrinkage Cavity and Porosity at the Blade Root Fillet
The second group of sand casting defects appears at the fillet roots where the blades meet the front and back cover plates. This is a geometrically complex zone. In the original mould design, these regions were often left with sharp corners or with very small fillet radii. The pattern did not faithfully reproduce the large rounded fillets that the drawing required. During moulding, the sand around these sharp corners was more difficult to compact. The result was a sand surface with low density and high permeability to gases.
When molten iron entered the mould, the sharp sand corners were heated very quickly. Because the surface-to-volume ratio of a sharp sand edge is high, the sand reached a very high temperature. The silica sand reacted with iron oxide to form a low-melting iron silicate. This is often called fayalite:
$$
2\text{FeO} + \text{SiO}_2 \rightarrow \text{Fe}_2\text{SiO}_4
$$
This reaction produced chemical sand sticking. Once the sand grains were fused together, they adhered strongly to the casting surface. On the machined and ground surfaces, these spots became visible as hard, dark inclusions. In addition, the high thermal load at the sharp sand corner generated a large amount of mould gas. The gas entered the molten metal and produced blowholes. Because the fillet root is also a thermal center, gas pores and shrinkage porosity often appeared together.
Our first corrective action was to increase the fillet radii at the intersections between the blades and the cover plates. We did this on the pattern according to the engineering design requirements. The object was to eliminate sharp corners and to create larger radii that would reduce sand overheating. I cannot overemphasize the importance of this step. The original sharp corners were a root-cause condition that made all other controls less effective.
Countermeasure 2: Chromite Sand, Resin Control, and Hot Air Drying
We then changed the moulding material in the fillet-root zone. Instead of ordinary silica sand, we now use chromite sand in the critical fillet locations. Chromite sand has higher refractoriness than silica sand. It also has better thermal conductivity, which means it acts as a chill. By extracting heat more quickly from the molten iron, it reduces the local temperature and accelerates solidification. This shortens the time available for gas nucleation and shrinkage formation.
We also placed strict limits on resin addition. More resin means more gas evolution during pouring. The relationship is straightforward: the gas generated in a sand mould is proportional to the gas generation capability of the binder:
$$
V_{\mathrm{gas}} = V_{\mathrm{sand}}\, G_{\mathrm{binder}}
$$
where \(V_{\mathrm{gas}}\) is the gas volume produced, \(V_{\mathrm{sand}}\) is the volume of sand at the mould wall, and \(G_{\mathrm{binder}}\) is the gas generation per unit volume of sand. If \(G_{\mathrm{binder}}\) is too high, the gas pressure in the mould cavity increases. This causes blowholes. Therefore, we reduce the resin content as much as possible while still maintaining the required sand strength. For impellers smaller than the 700WN class, I now require resin addition below two percent by weight of sand.
Another important countermeasure is hot air drying after mould closing. In many foundries, the mould is closed and poured as quickly as possible. But in our large impeller moulds, the core assembly process is long. During this time, the sand absorbs moisture from the atmosphere and from the core paste. At the fillet roots, the local moisture can be high. To remove this moisture, we insert a hot air blower pipe into the mould cavity through the riser root. We then dry the cavity with hot air for four to six hours. The outlet temperature is controlled between \(120\,^{\circ}\mathrm{C}\) and \(150\,^{\circ}\mathrm{C}\). About half an hour before pouring, we remove the drying equipment and seal the riser to keep the cavity clean.
This practice is very effective for reducing gas-related sand casting defects. The hot air also dries the coating and hardens the sand surface. During drying, we take care not to let the air blast directly impinge on a single spot, because that could erode the sand and create a new defect. Instead, we guide the pipe so that the airflow circulates gently inside the cavity.
Finally, we strengthened the coating procedure at the fillet roots. The coating must have high refractoriness and good resistance to iron oxide attack. We apply the coating in multiple layers, and after each layer we carefully check for missing spots. At the exact fillet root, we also hand-grind the sand fillet until it is smooth. To verify the fillet radius, we make arc gauges with the correct radius value. Every fillet is checked with the gauge before mould closing.
| Factor | Original condition | Improved condition |
|---|---|---|
| Fillet radius at blade root | Sharp corners, small radius | Increased fillet radius, checked by arc gauge |
| Mould sand at fillet root | Silica sand | Chromite sand with higher refractority |
| Resin content | Uncontrolled | Less than 2% for 700WN class and smaller |
| Mould drying after closing | None | Hot air dry 4–6 h at 120–150 °C outlet |
| Coating quality | Light brushing, uneven | Multi-layer brushing, hand-finished fillets, arc gauge inspection |
These changes did not completely eliminate all sand casting defects, but they reduced the defect frequency in the fillet-root region by a large margin. In particular, the combination of chromite sand and controlled resin addition reduced both blowholes and chemical sand sticking. The hot air drying step added a small cost in time, but it paid for itself in fewer scrapped castings and much less grinding time.
4. Sand Inclusions and Slag Inclusions on Machined Faces
The third important group of sand casting defects appears on machined surfaces. After machining the cover plate plane and the shaft head end face, our operators frequently found dark sand spots and glassy slag spots. These spots were not visible on the raw casting surface, because they were hidden below the machining allowance. Only after the first cut did they appear. This was a serious issue because it often caused the entire casting to be rejected after valuable machining time had already been spent.
We investigated the root cause by sectioning several rejected castings. We found that the inclusions were a mixture of loose silica sand, decomposed binder, and oxidized metal. The main source was the core assembly process. A large impeller mould is made by assembling several individual cores. During assembly, small particles of sand fall from the core surfaces, the core prints, and the locators. Some of this loose sand remained in the lower cavity after closing. When the molten iron entered the mould, it floated these sand particles upward and carried them into the solidifying surface. Because the cover plate and shaft head are horizontal, these sand particles accumulated at the upper surface of the casting, exactly where the machining operation later exposed them.
We also discovered that the mould surface in some areas was not compacted to a sufficient density. When the iron velocity was high, the iron eroded loose sand from the mould wall. This is especially common at the ingate and near the runner. To reduce this, we studied the metal flow velocity in the gating system. The velocity of metal leaving the gating system can be approximated by:
$$
v = \mu \sqrt{2 g h_{\mathrm{metal}}}
$$
where \(v\) is the velocity, \(\mu\) is the friction coefficient, \(g\) is gravity, and \(h_{\mathrm{metal}}\) is the effective metal head. A high metal head produces a high velocity. If the velocity is too high, the metal jet becomes turbulent and erodes the sand. Therefore, we reduced the pouring head height where possible and designed the gating to distribute the metal more gently into multiple gates.
Countermeasure 3: Sand Collection, Ceramic Gating, and Slag Traps
We adopted several practical countermeasures. First, we improved the ramming procedure for all sand surfaces. In particular, the bottom face of the lower mould cavity must be compacted until the surface has a hard, uniform feel. We check the compactness with a hardness tester. If the surface is too soft, we add more sand and re-ram before closing.
Second, we added three sand-collecting holes at the lowest points of the lower mould cavity. These are small pockets that trap loose sand coming from core assembly. During pouring, the first metal enters these trap pockets, where the loose sand is captured. The trap pockets are positioned at the low points, because gravity causes loose sand to settle there. Before closing, the operator blows out the cavity, but not every small particle can be removed. The sand traps act as a safety net.
Third, we replaced the conventional sand gating system with a ceramic gating system. The ceramic runner and ceramic sprue bush sharply reduce the contact between molten iron and sand. Because the ceramic is smooth and non-erodible, the gating system does not generate sand inclusions. The ceramic tubes also improve flow control and prevent aspiration of sand from the runner walls.
Fourth, we added a slag collector at the end of the runner. The slag collector is a small cavity beyond the last ingate. It is designed to hold the first, dirtiest metal that enters the gating system. This prevents slag and oxidised metal from reaching the casting cavity. The design follows the principle of steady flow:
$$
A_{\mathrm{slag\,trap}} > A_{\mathrm{ingate}}
$$
The larger area reduces the flow velocity in the trap, allowing inclusions to float upward instead of being carried into the casting.
| Countermeasure | Purpose |
|---|---|
| Intensive ramming of mould surface | Avoid erosion by molten iron and keep loose sand out |
| Three sand-collecting pockets at lowest cavity points | Capture loose sand from core assembly |
| Ceramic gating system | Prevent sand erosion in runner and sprue |
| Slag collector at runner end | Capture slag and oxidized metal before it reaches the casting |
After these changes, the number of sand inclusions found on machined surfaces dropped significantly. We also noticed a secondary benefit: the casting surface near the ingates became much cleaner, which reduced the amount of grinding required before shipment.
5. Cold Shuts at the Outlet Edge and Front Cover Plate Periphery
Cold shuts are one of the most visible sand casting defects. They appear as irregular cracks or seams on the surface, often near the thin outlet edge of the impeller blades and along the outside edge of the front cover plate. In a cold shut, two streams of molten iron meet but do not merge completely because one or both streams have become too cold. The interface remains as a weak, oxidized film.
In our production data, we identified three causes of cold shuts. First, the pouring time was too long. For a large impeller, the pouring time must be short enough that the whole mould cavity is filled while the metal is still fluid. If the pouring time is too long, the metal loses heat to the sand and its fluidity becomes poor. Second, the metal head pressure was not sufficient. A low pouring head means slower filling of the lower parts of the mould. Third, there were occasional interruptions of the pouring stream. When the pouring ladle was emptied too quickly, or when the stopper was closed and reopened, the flow stopped. This changed the filling pattern and caused cold shuts at the final meeting fronts.
The classic filling time equation can be used to express this. The total pouring time is:
$$
t_{\mathrm{pour}} = \frac{W}{\rho_{\mathrm{iron}}\, A_{\mathrm{choke}}\, v_{\mathrm{metal}}}
$$
where \(W\) is the total casting weight, \(A_{\mathrm{choke}}\) is the choke area of the gating system, \(v_{\mathrm{metal}}\) is the metal velocity at the choke, and \(\rho_{\mathrm{iron}}\) is the density of the molten iron. The metal velocity itself depends on the effective head height:
$$
v_{\mathrm{metal}} = \sqrt{2 g h_{\mathrm{head}}}
$$
From these equations, it is clear that a longer pouring time is associated with a smaller choke area or a lower head. In both cases, the last metal to enter the mould is slower and colder.
Countermeasure 4: Pouring Discipline and Controlled Head Height
We now specify the allowable pouring time on the process card. This may sound simple, but it was not always done in our foundry. Every pouring crew must read the process card before starting. A supervisor records the actual pouring time and compares it with the specified range. If the pouring time is outside the range, the casting is automatically flagged for inspection.
We also set a minimum pouring head height. In our practice, the effective head height above the ingate must be at least 200 millimeters during pouring. This ensures that the metal velocity remains high enough to fill the thin blade outlets before solidification starts. We made a special pool-shaped pouring basin with a large metal reservoir. The puddle basin reduces vortices and prevents the pouring stream from being interrupted as the ladle is repositioned. The basin also helps to separate slag, because the slag floats on top of the metal while clean metal enters the sprue from below.
I also instructed the melt department to raise the pouring temperature slightly for impellers with very thin cover plate edges. The exact temperature depends on the composition and section thickness, but the key principle is that the final metal that reaches the thin edges must still have enough superheat to merge with the earlier metal. The proper pouring temperature is a balance: too high creates gas and shrinkage problems, too low creates cold shuts. We therefore use a narrow pouring window and measure each ladle temperature before pouring.
| Cause of cold shut | Corrective action |
|---|---|
| Pouring time too long | Specify pouring time on process card, supervise every pour |
| Low metal head pressure | Keep effective head height at least 200 mm |
| Interrupted pouring stream | Use pool-shaped pouring basin, avoid repositioning ladle during pour |
After implementing these measures, the frequency of cold shuts fell noticeably. In fact, we now rarely see cold shuts on the outlet edge. When one does occur, we can usually trace it to an abnormally long pouring time or to a failure to maintain the head height.
6. Other Sand Casting Defects and Their Countermeasures
In addition to the major defect groups described above, I have encountered several other problems that are less frequent but still very important. Each one is a typical example of how sand casting defects can appear in different forms. I will describe them in detail because they often go unnoticed until final inspection or assembly.
6.1 Uneven Flow Channel Width
The first additional problem is uneven flow channel width. The impeller has several curved flow passages. The core boxes are made of glass-fiber reinforced plastic. In the original practice, the core maker did not always verify the pattern dimensions before making each core. After many cycles, the plastic core box began to wear or deform. As a result, the patterns produced cores of different sizes. When these cores were assembled in the mould, the flow passage width varied from one passage to another. In some places it was larger than the drawing tolerance; in other places it was smaller.
This defect directly affects pump performance and impeller balance. The flow passage width must be uniform to maintain uniform hydraulic loading and to avoid cavitation. I introduced a simple dimensional check. Before making each core, a technician measures the critical dimensions of the core box, including the blade positions, core thickness, and draft angles. If any key dimension is outside the allowable range, the core box is corrected or replaced before production. In addition, during the mould closing step, a process inspector records the core size and position of every core. The inspector verifies that the width between adjacent core surfaces matches the drawing.
The dimensional deviation can be expressed as:
$$
e_{\mathrm{width}} = \frac{b_{\mathrm{actual}} – b_{\mathrm{nominal}}}{b_{\mathrm{nominal}}} \times 100\%
$$
where \(e_{\mathrm{width}}\) is the percentage width error, \(b_{\mathrm{actual}}\) is the measured flow passage width, and \(b_{\mathrm{nominal}}\) is the design width. We now require every assembled core group to have a width error within the drawing tolerance. If an error is found, the mould is opened and corrected before pouring.
6.2 Cracks on the Outer Edge of the Cover Plate
The second additional problem is cracking on the outer edge of the cover plate. This is one of the most feared sand casting defects because it can lead to complete scrapping. In our analysis, cracks appeared mainly for two reasons. The first was a high concentration of harmful elements, especially sulfur and phosphorus. Sulfur and phosphorus reduce the hot ductility of the iron and make it brittle in the solidification and cooling stage. If the iron chemistry is not controlled, the thermal stress generated during cooling will exceed the hot strength of the casting and cracks will form.
The thermal stress can be expressed simply as:
$$
\sigma_{\mathrm{therm}} = E \alpha \left( T_{\mathrm{surface}} – T_{\mathrm{core}} \right)
$$
where \(E\) is the elastic modulus, \(\alpha\) is the coefficient of thermal expansion, and \(T_{\mathrm{surface}} – T_{\mathrm{core}}\) is the temperature difference between the surface and core. If \(\sigma_{\mathrm{therm}}\) is greater than the hot tensile strength \(\sigma_{\mathrm{hot}}\), cracking occurs:
$$
\sigma_{\mathrm{therm}} > \sigma_{\mathrm{hot}}
$$
Therefore, we now check the sulfur and phosphorus content before every tap. If the melt does not meet the internal specification, it is not allowed to be poured. This is a hard rule.
The second reason for cracks was premature shakeout. In the original practice, the casting was sometimes knocked out of the mould too early. The outer edge of the cover plate cooled much faster than the thick hub. This created a steep temperature difference and severe thermal stress. To solve this, we now follow a strict pressing and slow-cooling procedure. For impellers larger than the 700WN class, the casting must remain in the mould under pressure for at least eight days. After pouring, we wait five hours, then place the weights on the top box. After twenty-four hours, we loosen the sand around the mould. After forty-eight hours, we loosen the upper box. The actual shakeout is not allowed until the casting temperature reaches a safe value. We require the casting to be below \(300\,^{\circ}\mathrm{C}\) before removing it from the mould. During shakeout, we prohibit any collision with the casting. The shakeout hammer and lifting chains must be used carefully.
We also control the flash thickness at the mould joint. If the flash is too thick, it acts as an unwanted structural restraint during cooling. The flash thickness must not exceed 5 millimeters. We also avoid using excessive asbestos rope in the mould joint. Too much rope creates a hard, incompressible joint that restricts contraction and causes cracks.
| Parameter | Requirement |
|---|---|
| Sulfur and phosphorus content | Must meet internal specification, otherwise no pouring |
| Flash thickness at mold joint | Not more than 5 mm |
| Asbestos rope at mold joint | Only minimal amount, avoid excessive packing |
| Pressing time for impellers above 700WN class | At least 8 days |
| Time to apply weights after pouring | 5 hours |
| First loosening of surrounding sand | 24 hours after pouring |
| Second loosening of upper box | 48 hours after pouring |
| Maximum shakeout temperature | 300 °C |
6.3 Large Unbalance in Static Balancing
The third additional problem is excessive unbalance during static balancing. Dredge pump impellers are usually subject to a balancing check before machining or after machining. A large unbalance means that the center of mass of the impeller does not coincide with its axis of rotation. This causes vibration, bearing wear, and reduced pump life. The unbalance value is related to the mass \(m\) and the eccentricity \(r\):
$$
U = m \cdot r
$$
If the unbalance is too large, it is often necessary to remove metal from the heavy side. In severe cases, the impeller is rejected. Our analysis showed that the main causes were uneven thickness in the cover plates and blades, and inconsistent placement of the machining reference. The pattern dimensions were not controlled sufficiently, so some castings had slightly thicker sections on one side. This did not show up as a visual defect, but it became obvious during balancing.
We now take several countermeasures. The first is to control the pattern dimension accuracy. After the pattern is repaired or changed, we measure the critical thickness dimensions and record them. The second is to check the core placement during mould closing. Each blade core must be positioned so that the blade thickness is uniform from one side to the other. The third is to communicate with the machining shop before production. We agree on the locating datum and the permitted metal-removal zones. The machining shop knows whether the balance correction can be made by grinding a certain boss or by machining the cover plate. This avoids surprises after the casting is fully machined.
The residual unbalance limit is sometimes expressed by the well-known formula:
$$
U_{\mathrm{permissible}} \le \frac{6350 \cdot W}{n}
$$
where \(W\) is the impeller weight and \(n\) is the rotational speed in revolutions per minute. Although this is a simplified approximation, it helps our production team understand that heavier and faster impellers require tighter balance control.
7. Systematic Quality Control and the Role of Standardization
All of the above countermeasures are effective only if they are embedded in the daily quality system. In my experience, sand casting defects have a habit of returning if the process is not controlled. For this reason, I introduced several system-level procedures in our foundry.
The first is the process card. Every impeller has a process card that lists the pouring temperature range, pouring time, riser specifications, mould drying time, resin content, shakeout temperature, and all inspection steps. The process card is signed by the relevant operators. If any step is missed, the casting cannot proceed to the next operation.
The second is the first-article inspection. Before each production batch, we make one trial core and one trial mould under the same conditions as the main batch. We measure the key dimensions and check the fillet radii. This has prevented many sand casting defects before they reach the pouring stage.
The third is the defect traceability system. Every casting is assigned a unique identification number. The operators record the melt number, sand batch, core batch, mould assembly time, pouring time, and shakeout time. If a defect appears later, we can trace the casting back to the exact process condition. This has been extremely helpful in identifying small changes that cause sand casting defects.
The fourth is education and training. I have found that many operators do not fully understand why a riser height matters or why a core box dimension must be checked. We now hold short toolbox talks before every difficult impeller order. I explain the solidification mechanism and the consequence of not following the process. The operators have become much more aware of sand casting defects and are more willing to stop the line when they see something unusual.
| Control element | What we do | Defect prevented |
|---|---|---|
| Process card | Written parameters for pouring, drying, shakeout | Cold shuts, gas porosity, cracks |
| First-article inspection | Trial core and mould measured before batch release | Dimensional deviations, sand inclusion |
| Traceability system | Unique ID, melt and sand records | All sand casting defects |
| Operator training | Explain metallurgy and moulding principles | All sand casting defects |
8. Achieved Results
I can say with confidence that the combination of engineering corrections and system-level control has transformed our foundry. In the period before the changes, large dredge pump impellers often required expensive repairs, and many were completely scrapped. The main causes, as I have discussed, were insufficient riser feeding, sharp fillet roots, excessive resin, loose sand contamination, cold shuts, and premature shakeout. These were all classic sand casting defects, but each one had a unique solution.
After implementing the countermeasures described in this article, our large dredge pump impeller casting qualification rate now exceeds ninety percent. The improvement is visible not only in the inspection reports but also in the machining shop. Machinists report far fewer blowholes and inclusions on finished surfaces. The internal quality, confirmed by ultrasonic and radiographic testing, is much more uniform. The balancing department now spends much less time correcting unbalance. Customers report fewer vibration problems and longer pump life.
From a financial point of view, the reduction in sand casting defects has reduced scrap and rework costs by a large amount. The production capacity is higher because we do not waste time repairing bad castings. The company has also gained a stronger position in the market. We can now accept larger and more complex dredge pump impeller orders with confidence.
I must be honest and say that no foundry can completely eliminate every defect. Sand casting is a complex process with many variables. The temperature of the molten iron, the humidity of the sand, the speed of the crane, and the skill of the operator all influence the result. However, the systematic approach that I have described can reduce the frequency and severity of sand casting defects to an acceptable level.
9. Conclusion
In conclusion, large dredge pump impeller castings are challenging because of their complex flow passages, thick-thin transitions, and severe service conditions. The main sand casting defects that I have encountered are shrinkage cavities at the shaft head and riser neck, sand sticking and blowholes at the fillet roots, sand and slag inclusions on machined surfaces, cold shuts at the thin outer edges, cracks on the cover plate periphery, and dimensional or balance deviations.
Each of these sand casting defects has a clear root cause and a practical solution. I have learned that the first step is to analyze the solidification and filling behavior, then to correct the mould design, gating design, and pouring practice. The second step is to standardize the process parameters and to train every operator to follow them. The third step is to close the loop by inspecting every part and using the defect data to improve the next batch. When these steps are done together, the quality of large dredge pump impellers improves dramatically.
The benefits of reducing sand casting defects are not limited to the foundry. They extend to the dredging industry as a whole, because a reliable impeller means less downtime, lower maintenance costs, and more efficient dredging operations. In my own experience, the journey toward zero defects is never complete. But the progress that we have made proves that a careful, scientific approach to sand casting quality can yield outstanding results.