10th International Symposium on the Science and Processing of Cast Iron – SPCI10

Feeding Against Gravity with Spot Feeders in High Silicon Ductile Iron
N.K. Vedel-Smith1 and N.S. Tiedje1
1
Department of Mechanical Engineering, Technical University of Denmark, Kgs. Lyngby, Denmark

A test pattern, with three different moduli castings was developed to investigate methods to
optimise feeding of high silicon ductile cast irons. Different feeder types, modulus, and locations
were investigated using both an insulating and an exothermal sleeve material. Porosities were
analysed and classified using X-ray imaging and ultrasound analysis. The effect of the different
feeder configurations were classified in reference to defect location, sleeve material, and feeder
type, modulus, and location.
The analysis showed that exothermal feeder sleeves with the right configurations can feed up-
hill against gravity. This effect may contribute to the thermal expansion created by the exothermal
reaction. It was also found that the optimum feeder size does not scale linearly with the casting
modulus but that larger casting modulus requires relatively smaller modulus feeders. The thermal
gradient created by the feeders made with the insulating sleeve material was not sufficient to
significantly improve feeding.

Keywords: ductile iron, spot feeding, risering, solidification, high silicon, ram-up sleeves.

Introduction
Feeding complex castings with different moduli sections is a challenge for the foundries, as customers require improved
yield, less machining, and sound castings. Optimisation of cast components is an essential driver for many industries in
order to improve their products. Thus the foundries are met with an ever growing requirement to improve methods and
increase yield. The location and orientation of the casting is determined by casting geometry, location of cores and
feeders. However, new designs with great variation between thin and thick walled sections, and highly complex designs
limit the use of traditional feeders.
In vertically parted moulds geometrical feeders are normally located at the top of the casting on the parting plane.
All feeders require an unbroken feeding path from the feeder to the section that must be fed. This makes it difficult, if
not impossible, to feed heavy sections that are disconnected from the feeding path by a low modulus section.
Additionally, the feeder requires a driving force to move the melt from the feeder into the casting. Traditionally this
driving force is gravity, but other forces also act on the melt during solidification. E.g. solutions like the William wedge
and similar geometries are a part of almost all feeder designs to ensure that the feeder is kept open to the atmosphere
(punctured) and thus prevent the negative pressure gradient retaining the melt inside the feeder. Other natural forces
working on the melt can be the contraction and expansion of the melt itself as different sections of the casting goes
through the different stages of solidification at different intervals depending on the modulus, cooling rate, and alloy
composition. The movement, deformation, expansion, and the reduction in strength of the green sand mould also
influence these factors.
Descriptions and guidelines to the application and effect of feeders that make use of these naturally occurring
driving forces to move the melt from a feeder located at the bottom of the casting into a section at a higher elevation are
sparse at best. The study presented in this paper represents an experimental design comprising 9 different feeder
configuration tested on a scalable casting geometry in three different sizes of casting moduli—8 mm, 10 mm and 15
mm. The study quantified the effect of different modulus spot feeders for different modulus castings. The trial was
made with different insulating and exothermic ram-up sleeves together with high silicon ductile iron castings.

Experimental Procedure
Casting Geometry and Pre-feeder Design
The casting geometry was designed to be parametric in order to represent different moduli sections with the same
geometry (see Figure 1). The casting itself was a rectangular cuboid. A square footprint was chosen because the square
design allow for a high geometrical modulus and was better suited than a round design for ultrasound and x-ray
analysis. The height of the casting was chosen as 3 times the width and depth of the casting. The basic idea with the
design was to have one uniform section that would create a significant amount of shrinkage by itself. The height of the
casting should be great enough that a feeder at the top and a feeder at the bottom would influence the casting differently
due to the difference in ferrostatic pressure. For steel bars cast in a horizontal orientation the maximum feeding distance
between two feeders was reported as varying between 1-4 times the thickness of the bar1. Though the trial castings are


Corresponding author, email: nikvs@mek.dtu.dk
 10th International Symposium on the Science and Processing of Cast Iron – SPCI10
cast vertically and in SGI, it was chosen to uphold as great a distance between the two feeders on the casting as
possible. The height of the bar was governed by the pattern size and allowed for a bar length of 3 times the thickness
(a). In turn this allowed for a feeder distance of approximately 2.5 times the thickness (a).

a b

Fig.1: Schematic of the parametric casting design (a), and pattern layout (b). Measurements in mm.

A pre-feeder was placed on top of the casting—designed to compensate for the liquid shrinkage that occur as the casting
cools from the pouring temperature to the solidus temperature, so that the variations in pouring temperature on total
shrinkage are eliminated. The design had to ensure an identical amount of shrinkage in all castings, regardless of
pouring temperature. If not done properly it would afterwards be impossible to prove that changes in porosities in the
castings were related to changes in feeder configurations and not attributable to a smaller or greater liquid shrinkage
caused by varying pouring temperatures.
The size of the pre-feeder neck was determined so that it closes and blocks feeding at the point in time where
solidification begins in the casting itself. Based on Chvorinov’s modulus law2 equation (1) was derived:

(1)

where Dneck was the diameter of the pre-feeder neck, Mneck was the modulus of the pre-feeder neck, and Mcasting was the
modulus of the casting. H was the enthalpy of the system, cp was the heat capacity, Tstart was the pouring temperature
and Teut was the eutectic temperature for the given alloy. Equation (1) gives the diameter of the pre-feeder neck.
However, the equation does not take into account the heat flux from the casting and pre-feeder but assume
unidirectional solidification. To determine the optimum pre-feeder neck height, which would reduce the amount of
liquid shrinkage as much as possible while still allowing for a timely solidification of the neck, numerical simulations of
the different pre-feeder neck geometries were used (see Figure 2).

a b

Fig.2: (a) Parametric study of varying pre-feeder neck lengths from 0.5a to 1.5a, and (b) a graph showing the
temperature at the centre of the casting (Closing temperature) as a function of the pouring temperature for three
different moduli castings and five different pre-feeder neck lengths. The dashed line indicates the eutectic temperature
of the alloy.
 10th International Symposium on the Science and Processing of Cast Iron – SPCI10
The solidification times obtained from the numerical simulation of the different parametrical geometries, at different
pouring temperatures, were plotted in Figure 2(b). The graph shows that the yellow line representing the shortest pre-
feeder neck was below the eutectic temperature line. This means that the solidification of the casting had begun before
the neck had closed off. All other pre-feeder neck lengths closed off before the casting centre reaches the eutectic
temperature. Hence the pre-feeder neck length of 0.75a was chosen because that length allowed for the best prevention
of liquid shrinkage and for the most uniform performance across all casting temperatures. Additionally, this analysis
was repeated for another alloy with a different eutectic temperature to make sure the design would function with
different alloy compositions.

Feeder Placement, Modulus, and Combinations

The trial setup consisted of 6 different feeder geometries, using either an insulating or an exothermic sleeve material,
and applying feeders to two different locations on the casting—at the top near the pre-feeder neck (upper) and at the
bottom near the ingate (lower). The possibility of placing feeders either at the top or bottom of the casting, or at both
locations simultaneously, enabled an analysis of the feeders’ performance in relation to the pressure height of the
location. The feeders themselves were mounted horizontally, which minimises the feeders own influence on the
ferrostatic pressure. The driving force for moving the melt from the spot feeder into the casting required other forces
than gravity to act on the liquid in order to feed the casting. The horizontal orientation also minimised the difference
between the upper and lower spot feeder location.
The different spot feeders were mounted onto the three different modulus castings as shown in Table 1. The first
group (0) was a control group without any spot feeders. Groups 1 and 2 had only feeders at the upper location, groups 3
and 4 only at the lower location. The remaining groups (5-9) all had feeders at both the upper and the lower location.
All combinations were cast in 2-3 duplicates to ensure repeatability.

Table 1: Trial combination overview. Numbers before the letter indicate melt volume [cm3], letters indicate I for
insulating and E for exothermic, and numbers after the letters indicate feeder modulus [mm].

# 0 1 2 3 4 5 6 7 8 9 Feeder Types

Dup. 3 3 3 2 3 2 2 2 2 2 Ins Exo

M08 U 28I08 28E10 28I08 07I05 28E10 08E08 07E06 07I05 07E06

L 08E08 28E10 28E10 08E08 28E10 07E06 28E10 08I06 08E08

M10 U 22I10 22E12 22I10 08I06 22E12 28E10 08E08 28I08 28E10

L 28E10 22E12 22E12 28E10 22E12 08E08 22E12 22I10 22E12

M15 U 121I16 121E19 121I16 22I10 121E19 82E15 22E12 82E15

L 82E15 121E19 121E19 82E15 121E19 22E12 121E19 121I16 121E19

The spot feeders used in the study were so called ram-up sleeves which are mounted on a pin on the pattern before the
moulding process begins. After the moulding process the ram-up sleeves are located inside the green sand mould as
described by Vedel-Smith et.al.3 Figure 3 show the spot feeders from group 6 mounted on the pattern, ready for
moulding.

Fig.3: Spot feeders from group 6 mounted on the pattern, ready for moulding. Insulating spot feeders on the top.
Exothermic spot feeders at the bottom.
 10th International Symposium on the Science and Processing of Cast Iron – SPCI10

Production, Alloy and Thermal Measurements

The castings were produced on a vertically parted moulding machine—Disamatic 240B—as part of a production run in
an operating foundry. The mould size was 775 x 600 x 300 mm. The poured weight of the casting without any spot
feeders was approx. 16 kg—whereof the M08 weights ~1.1 kg, M10 ~3.7 kg, and M15 ~7.3 kg. The pouring time was
around 6 s. The castings were made with an EN-GJS-500-144 alloy and four alloy composition analyses were made
during the duration of the trails. See Table 2.

Table 2: Alloy composition [wt%] and casting temperature [°C] variation during the trials.

CE C Si Mn P S Mg Cr Ni Mo Cu Sn Temp [°C]
I 4.60 3.31 3.91 0.31 0.015 0.003 0.051 0.047 0.021 0.001 0.09 0.004 1,398 (±5)
II 4.57 3.31 3.81 0.31 0.015 0.003 0.045 0.046 0.023 0.001 0.09 0.005 1,387 (±5)
III 4.54 3.35 3.61 0.25 0.015 0.004 0.042 0.051 0.026 0.001 0.06 0.005 1,380 (±5)
IV 4.54 3.34 3.64 0.25 0.015 0.004 0.039 0.050 0.025 0.001 0.06 0.005 1,361 (±5)
Avg 4.56 3.33 3.74 0.28 0.015 0.004 0.044 0.049 0.024 0.001 0.08 0.005 1,382 (±5)

Four temperature measurements were made in one of the castings—4A—which was cast immediately before the series
II castings listed in Table 2 were made. Three thermo couples were placed at the centre of each parametric casting, and
a single thermo couple was placed at the lower spot feeder of the M15 parametric casting (see Figure 4). All thermo
couples were K-type and the data were recorded at a sampling rate of 1 Hz with a stand-alone 4-channel thermo couple
data logger.
The castings solidified and cooled in the mould for approx. 1½ hours, and were thereafter removed manually from
the moulds at the shake-out station. This ensured that all spot feeders remained attached to the castings. When the
castings had air cooled to room temperature they were cleaned by shot blasting.

Ultrasound Analysis and X-ray Analysis

All castings were analysed with ultrasound by the same, experienced operator using a Karl Deutsch Digital-Echograph.
The location and size of the porosities were painted on the castings. This gave a detailed picture of the size, location,
and direction of the porosities. All castings were photo documented for later analysis.
Following the ultrasound analysis selected groups of castings were analysed using x-ray imaging. The x-ray imaging
focussed on confirming the results obtained by the ultrasound analysis, but also on documenting the porosities located
in areas not suitable for ultrasound analysis—the pre-feeder neck and the spot feeders. Several images were taken of
each of the castings and then assembled into an overview, allowing for a more holistic analysis of the x-ray imaging
results.

Results
Ultrasound Analysis
The findings from the ultrasound analysis were classified with respect to porosity size (0-4 where 0 is no porosities and
4 is large porosities), porosity location (top, middle, bottom), and if porosities at different locations were connected or
disconnected. Additionally it was also registered when the porosities had an opening out unto the surface of the casting
(bold). See Table 3.

Table 3: Results of the ultrasound analysis

# 0 1 2 3 4 5 6 7 8 9 Nomenclature
A B C A B C A B C A B A B C A B A B A B A B A B 0 No porosity
T 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 0 2 4 4 4 3 1 Micro Porosities
M8 M 2 2 2 2 2 0 0 0 0 2 2 2 2 2 2 2 0 0 2 2 4 4 4 4 2 Small Porosities
B 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 3 Medium Porosit.
T 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 Large Porosities
M10 M 2 2 2 1 0 1 2 2 0 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
B 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
T 4 4 4 4 4 4 4 4 4 0 0 0 4 4 4 4 0 0 2 2 1 0 0 0 Connected
M15 M 0 4 2 0 2 1 0 0 0 1 3 2 4 4 0 2 2 1 0 0 0 0 1 0 Disconnected
B 0 0 0 0 0 0 0 0 0 0 0 2 4 4 0 0 0 0 0 0 0 0 0 0 Puncture at neck
 10th International Symposium on the Science and Processing of Cast Iron – SPCI10
The quantified results of the ultrasound analysis indicated that many of the different castings, especially the two
smallest modulus castings, displayed the same amount of porosity as the reference casting groups (0) without any spot
feeders. This was partly true, however it should be noted that the large porosity (4) indication has no upper limit,
meaning that the same indication can cover great differences (see Figure 4).

Fig.4: Porosity markings from the ultrasound analysis of casting groups 5A and 5B.

Additionally, some of the reference castings without spot feeder displayed clear signs of surface shrinkage, indicating
that some of the shrinkage for these castings have occurred in location that have not been covered by this analysis.
Surface shrinkage was not observed in any of the castings with spot feeders.

X-Ray Imaging and Analysis

The x-ray images are greyscale images produced by irradiating the casting from one side, and recording the radiation
that reaches the sensor at the opposite side. This gives an image that in greyscale contrast show areas with little material
in between (light), and areas with a lot of material in between (dark). Porosities show up because they are holes in the
bulk material, and thus areas with porosities absorb less radiation, resulting in a brighter area of the image. However,
the radiation does scatter and diffuse, rendering the images a little fuzzy at the edges. Because of this effect there was a
limit to the difference between the size of the casting and the size of the porosity that could be imaged. This made it
difficult to obtain good images of the defects in the M15 castings (see Figure 5).

Fig.5: X-ray images of casting 6A, 7A, 8B, and 9B—all that largest casting with a modulus of 15 mm.
 10th International Symposium on the Science and Processing of Cast Iron – SPCI10

Thermal Measurements and Cooling Curves

The cooling curves showed that the three parametric castings solidified and cooled in a comparable manner, confirming
that the parametric design provide the intended comparison between the different moduli castings (see Figure 6). The
maximum temperature measured was 1,325 ±5 °C indicating that the gating filling of the mould has lowered the
temperature of the melt by 62 ±10 °C from the pouring temperature of 1,387 ±5 °C. However, the specified temperature
limit of 1.300 °C for the K-type thermo couples for short duration readings should be taken into consideration5.
The cooling curves showed that in the bottom part of the M15 casting ended solidification approx. 320 s before the
centre of the casting, albeit 400 s later the curves met again. This was caused by the thermal influence of the spot
feeder, and the area between the two curves indicates the energy that the spot feeder provides transferred to the casting.

Fig.6: Thermal measurement of the casting, cooling, and solidification of casting group 4A.

Discussion
Examining Table 3 it was seen that most of the castings, regardless of modulus size and spot feeder combinations, had a
large (size 4) porosity at the top and a small (size 2) porosity at the middle. As mentioned in the results section the large
porosity characteristic was open ended, and covered many gradually increasing porosities. However, the consistent
results show the stability of the model and the production conditions. Thus, the most interesting castings were the ones
that differentiate from the stable and repeatable porosities formed in the other castings.
Castings 8B-M15 and 9B-M15 were classified as porosity free and the castings with the same spot feeder
configuration—8A-M15 and 9A-M15—only displayed micro (size 1) porosities. 7A-M15 and 7B-M15 were classified
with small porosities at the top, but no porosities at the middle. Finally, 3A-M15 and 3B-M15 only displayed porosities
at the middle of the casting—ranging from micro (size 1) to medium (size 3) porosities.
The smaller castings—M08 and M15—did not show the same effect for these feeder configurations even though the
modulus of the spot feeders were scaled according to the changes in casting modulus. This indicated that the
solidification of the three different modulus castings was different as well. These changes in solidification can be
caused by the slower solidification of the large modulus castings, which provide the longer time for the graphite nodules
to grow and inhibit pearlite formation which would have reduced the effect of the graphite expansion. However, the
high Si content of the alloy greatly limits pearlite formation already, and none of the three castings are small enough to
be considered thin walled sections. Hence other factors were needed to fully explain the solidification differences
between the different modulus castings.
Additionally several of the castings had ‘punctures’ at the bottom of the pre-feeder neck, opening into a large
porosity in the casting. This effect seemed to have been dominant for the smaller modulus castings, but it also seemed
to be unrelated to the amount of porosities recorded and the effectiveness of the spot feeders.
The most likely explanation was that the large modulus castings had a greater tendency to form a solid shell early
during solidification, so that the low pressure that occurred inside the casting towards the end of solidification had
enough force to move the melt from the spot feeders and into the casting itself.
Some melt may have been provided by the pre-feeder regardless of the intention that this should not happen. The x-
ray images showed that the pre-feeders contained porosities. It was not possible to determine how much of the
porosities in the pre-feeders that was a result of feeding and how much were related to the liquid shrinkage that the pre-
feeders were designed to handle. However, it was noted that most pre-feeders displayed the some amount of porosities.
 10th International Symposium on the Science and Processing of Cast Iron – SPCI10
No correlation was found between the amount of porosities in the pre-feeder and the amount of porosities in the casting
itself.
However, the negative pressure gradient from the casting itself could completely explain the results shown in Table
3. If group 3 and 4 were compared for the M15 castings it was noted that the first displayed a few and small porosities
whereas the latter displayed more and larger porosities. However, it was the latter—group 4—that had the largest spot
feeder. If the main driving force for feeding from the spot feeder into the casting was the negative pressure in the
casting, then the larger spot feeder should have produced a porosity free casting. Instead it was seen that the smaller
spot feeder reduced the porosities in the casting significantly compared to the large one.
To explain this phenomenon other forces than a negative pressure caused by shrinkage of the material in the casting
must be taken into account. The graphite expansion was assumed to be the same for both configurations as they were
cast with the same alloy, and only minor differences would occur as an effect of the small changes in solidification
between the two spot feeder configurations. However, the timing of the graphite expansion, and particularly in relation
to the timing of the negative pressure inside the casting, seemed to reach an optimum for this configuration. Thus, the
two different forces come to act together, rather than against each other. If so, this could be seen as a special case of
John Campbell’s feeding rule no. 6 regarding the pressure gradient requirement6.
External forces can also have occurred and gas development from exothermic feeder sleeves was a known concern.
However, if gas development from the exothermic material was a significant driving force for the movement of the
melt, then group 4 and 5 should have displayed fewer porosities than what was recorded.
Finally, examining the results of the M15 casting groups 3, 6, 8, and 9, in comparison with the other five groups, it
was shown that a feeder located at the lower part of a casting section can feed the section with equal efficiency
compared the same feeder located at the upper part of the casting section. This showed that the horizontally oriented
spot feeders with exothermic sleeve material depend little upon the gravity as a driving force for feeding.

Conclusions
1. The optimum feeder size did not scale linearly with casting modulus. Larger casting modules required relatively
smaller modulus feeders.
2. The timing of the negative pressure from solidification shrinkage combined with the timing of the graphite
expansion seemed to be important in order to achieve the best possible feeding conditions. Similarly a larger feeder
may shift the time enough for the effects to counteract each other and thus cancel most of the feeding effect if not
directly developing more shrinkage.
3. The location for horizontally oriented spot feeders was relatively unaffected by the difference between the high and
low location. The spot feeders that functioned at one location also functioned at the other location. The spot feeders
that did not function at one location did not function at the other location either. In special cases it was possible to
feed against gravity.

References
1. ASM International: Casting Design and Performance–Part II: Process Design, Riser Design, 2009,
61-72.
2. N. Chvorinov: Theory of the Solidification of Castings, Giesserei, Vol. 27, 1940, 177-186.
3. N.K. Vedel-Smith, N.S. Tiedje, K.T. Maza, and J. Sällström: Quantification of Feeding Effects of
Spot Feeding Ductile Iron Castings made in Vertically Parted Molds, AFS Proceedings, 13-1310,
2013.
4. EN 1563:2012-3: Founding: Spheroidal Graphite Cast Irons, 2012.
5. L. Michalski, K. Eckersdorf, J. Kucharski, and J. McGhee: Temperature Measurements, 2nd Edn,
2001 John Wiley & Sons Ltd.
6. J. Campbell: ‘Casting Practice – The 10 Rules of Castings’, 120-156, 2004, Burlington, Elsevier
Butterworth-Heinemann.

Acknowledgement
This work was funded by the Public Service Obligation (PSO) funds made available by the Danish Government. The
project was made in collaboration with FOSECO Ltd., MAGMA GmbH, DISA Industries A/S, and Vald. Birn A/S.
Rune Engelberth Hansen carried out the development of the parametric casting geometry.